Defining the chronic impacts of atenolol on embryo-larval development and reproduction in the fathead minnow (Pimephales promelas)

Defining the chronic impacts of atenolol on embryo-larval development and reproduction in the fathead minnow (Pimephales promelas)

Aquatic Toxicology 86 (2008) 361–369 Defining the chronic impacts of atenolol on embryo-larval development and reproduction in the fathead minnow (Pi...

154KB Sizes 0 Downloads 25 Views

Aquatic Toxicology 86 (2008) 361–369

Defining the chronic impacts of atenolol on embryo-larval development and reproduction in the fathead minnow (Pimephales promelas) Matthew J. Winter a,∗ , Adam D. Lillicrap a , John E. Caunter a , Christian Schaffner b , Alfredo C. Alder b , Maria Ramil c , Thomas A. Ternes c , Emma Giltrow d , John P. Sumpter d , Thomas H. Hutchinson a a

AstraZeneca Global Safety, Health and Environment, Brixham Environmental Laboratory, Freshwater Quarry, Brixham, Devon TQ5 8BA, United Kingdom b Swiss Federal Institute of Aquatic Science and Technology (Eawag), Ueberlandstrasse 133, CH-8600 D¨ ubendorf, Switzerland c Federal Institute of Hydrology (BfG), Am Mainzer Tor 1, 56068 Koblenz, Germany d Institute for the Environment, Brunel University, Kingston Lane, Uxbridge, Middlesex UB8 3PH, United Kingdom Received 12 September 2007; received in revised form 24 November 2007; accepted 26 November 2007

Abstract Atenolol is a ␤-adrenergic receptor antagonist (‘␤-blocker’) widely used for the treatment of angina, glaucoma, high blood pressure and other related conditions. Since atenolol is not appreciably metabolized in humans, the parent compound is the predominant excretory product, and has been detected in sewage effluent discharges and surface waters. Consequently, atenolol has been chosen as a reference pharmaceutical for a European Union-funded research consortium, known as ERAPharm (http://www.erapharm.org), which focused on the fate and effects of pharmaceuticals in the environment. Here, we present data generated within this project from studies assessing population-relevant effects in a freshwater fish species. Using fathead minnows (Pimephales promelas) as a standard OECD test species, embryo-larval development (early life stage or ELS) and short-term (21 d) adult reproduction studies were undertaken. In the ELS study, the 4 d embryo NOEChatching and LOEChatching values were 10 and >10 mg/L, respectively, and after 28 d, NOECgrowth and LOECgrowth values were 3.2 and 10 mg/L, respectively (arithmetic mean measured atenolol concentrations were >90% of these nominal values). In the short-term reproduction study, NOECreproduction and LOECreproduction values were 10 and >10 mg/L, respectively (mean measured concentrations were 77–96% of nominal values), while the most sensitive endpoint was an increase in male fish condition index, giving NOECcondition index and LOECcondition index values of 1.0 and 3.2 mg/L, respectively. The corresponding measured plasma concentration of atenolol in these fish was 0.0518 mg/L. These data collectively suggest that atenolol has low chronic toxicity to fish under the conditions described, particularly considering the low environmental concentrations reported. These data also allowed the assessment of two theoretical approaches proposed as predictors of the environmental impact of human pharmaceuticals: the Huggett ‘mammalian-fish leverage model’; and the acute:chronic ratio (ACR). The Huggett model gave a measured human: fish effect ratio (ER) of 19.3 for atenolol, which compared well with the predicted ER of 40.98. Moreover, for an ER of 19.3, the model suggests that chronic testing may be warranted, and from our resultant effects data, atenolol does not cause significant chronic effects in fathead minnow at environmentally realistic concentrations. The calculated ACR for atenolol is >31.25, which is far lower than that of 17␣-ethinylestradiol and other potent steroidal oestrogens, thus further supporting the observed low toxicity. The data produced for atenolol here fit well with both approaches, but also highlight the importance of generating ‘real’ experimental data with which to calibrate and validate such models. © 2007 Elsevier B.V. All rights reserved. Keywords: Aquatic; ␤-blocker; Environment; Fish; Pharmaceutical

1. Introduction The ␤-adrenergic receptor antagonists, or ␤-blockers, are an important class of human pharmaceuticals used in the manage-



Corresponding author. Tel.: +44 1803 882882; fax: +44 1803 882974. E-mail address: [email protected] (M.J. Winter).

0166-445X/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2007.11.017

ment of various cardiovascular disorders, such as heart failure, cardiac arrhythmias, hypertension and Angina pectoris. Collectively, the ␤-blockers are widely prescribed drugs (e.g. Fraysse and Garric, 2005; Fent et al., 2006; Maurer et al., 2007), and as such, have been detected in sewage treatment works effluents and surface waters in a number of countries. Typically, surface water concentrations have been reported as in the ng/L to low ␮g/L levels (Ternes, 1998; Fent et al., 2006). For example, Zuccato et

362

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

al. (2005) measured atenolol concentrations of 241 ng/L in the river Lambro and 17.2 ng/L in the river Po, near Milan, Italy. Despite all sharing a basic pharmacodynamic mode of action, namely the competitive antagonism of ␤-adrenergic receptors, this group of compounds possess a diverse range of specific receptor targets, potencies and specificities, as well as pharmacokinetic and physicochemical properties (see Maurer et al., 2007 and Owen et al., 2007). Consequently, the potential environmental fate and effects of this group of compounds also has the potential for large diversity. For example, two of the most extensively prescribed ␤-blockers, propranolol and atenolol, exhibit widely different octanol–water partition coefficients (>3 and <1, respectively; Maurer et al., 2007), levels of human metabolism (approximately 90 and 10%, respectively; Maurer et al., 2007), and receptor specificities (␤1 and ␤2, and ␤1 only, respectively; Rang et al., 2003). There are a limited number of published reports on the effects of ␤-blockers in environmentally relevant species groups (see Owen et al., 2007, for a review). Of these, even fewer are from in vivo studies, none of which, to our knowledge, focuses on the environmental toxicology of atenolol. Of the few on ␤blockers in general, there is one published study reporting the chronic effects of propranolol on population-relevant endpoints in fish. Huggett et al. (2002) reported that propranolol significantly inhibited the egg production of Japanese medaka at 1.0 and 0.5 ␮g/L when compared with controls, although this effect was not seen at higher concentrations. Consequently, the current study served two main purposes. Firstly, it aimed to fill the knowledge gap that exists regarding the effects of chronic atenolol exposure in an OECD regulatory fish species as part of the European Union-funded research project, ERApharm (Knacker et al., 2005). Secondly, it provided data with which to assess two theoretical approaches proposed as predictors of the environmental impact of human pharmaceuticals: the Huggett ‘mammalian-fish leverage model’ (Huggett et al., 2004); and the acute:chronic ratio (ACR; Ankley et al., 2005). Both aims were addressed by undertaking two proposed regulatory chronic exposure studies using the fathead minnow (Pimephales promelas): the first, a fish early life stage study (ELS) assesses the effects of atenolol on embryo-larval development; and the second, a 21 d adult fish reproduction study assesses the effects of atenolol exposure on the reproduction of this species. 2. Materials and methods 2.1. Test substance and dilution water The test substance, atenolol (free base; CAS number 2912268-7), was obtained from AstraZeneca as milled, 99.9% pure material (optical rotation 0.00, material identification number E000881). Concentrated stock test solutions were prepared directly in water, and then dosed into a mixing chamber also receiving dilution water. In the mixing chamber, further dilution occurred to achieve the desired test concentrations: 10, 3.2, 1, 0.32 and 0.1 mg/L. These nominal exposure concentrations were

used in both studies and were based on the results of a range finding experiment to assess compound toxicity to a limited sample size (data not shown). The dilution medium was dechlorinated tap water, which had been passed through activated carbon, coarsely filtered to remove particulate material and dechlorinated with sodium thiosulphate. Salts were added, as required, to maintain minimum hardness levels, and the treated water was then passed through an ultraviolet steriliser to a second set of 20 and 10 ␮m filters. The supply was then delivered to a temperature controlled header tank in the test laboratory and finally re-filtered at 5 ␮m before use. This was used as the dilution water for all test concentrations following addition of the appropriate amount of atenolol, and in isolation for the dilution water control (DWC) tanks. The dilution water was monitored for general parameters (e.g. pH, hardness, conductivity) and also for specific trace metal and organic contaminants (including polychlorinated biphenyls (PCBs) and organochlorine pesticides) routinely during the test period. Additionally, alkalinity, hardness and conductivity were measured in one replicate of the control and one replicate of the test substance per week. Across both studies, the mean tank water pH ranged from 7.64 to 7.76, temperature from 24.93 to 25.06 ◦ C, dissolved oxygen from 7.62 to 8.06 mg/L, conductivity from 240.50 to 259.00 ␮S/cm, hardness from 47.00 to 52.71 mg/L CaCO3 , 27.11 to 27.95 mg/L CaCO3 , and chlorine was <2 in all cases. During the experiments themselves, all exposures were based on nominal concentrations, as preliminary chemical analyses suggested that measured concentrations would be close to 100% of nominal concentrations. After completion of the studies, frozen water samples were sent to Eawag, D¨ubendorf, Switzerland, to allow confirmation of atenolol concentrations (see below for details). 2.2. Experimental animals In both studies, the experimental animals were fathead minnows, P. promelas, a temperate cyprinid fish used as an OECD model species in developmental and reproductive test protocols (Ankley and Villeneuve, 2006). For the ELS study, fish were supplied as newly fertilised embryos (collected November 2005, at <24 h post fertilization) that were derived from a stock held under optimal breeding conditions at AstraZeneca, Brixham. For the 21 d adult reproduction test, the experimental subjects were adult virginal stock of approximately 180 d (post-hatch), which were raised from eggs produced by an established culture at AstraZeneca, Brixham. These fathead minnows were supplied for this study in February 2006. In the ELS study, fry were first fed rotifers (Brachionus plicatilis) then Artemia salina nauplii, and as the fry developed, A. salina were supplemented with a proprietary brand of pelleted fish food. Each batch of pelleted food was routinely analysed for pesticides and heavy metals. In the 21 d adult fish reproduction test, adult fish were fed ad libitum with frozen adult A. salina, twice daily during the working week, and once daily at the weekends. This was supplemented with a proprietary brand of pelleted fish food once daily, which

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

was also routinely analysed for a standard set of heavy metals and organochlorines. 2.3. Experimental procedures 2.3.1. Fish early life stage study (ELS) The fish ELS study was conducted using a protocol based on OECD Test Guideline 210 (OECD, 1992). The test apparatus was comprised of glass test vessels (9.5 L working volume, measuring 305 mm × 205 mm × 210 mm), and egg incubation cups (8 cm lengths of 5 cm external diameter glass tubing with 0.47 mm gauge nylon mesh fastened to the bottom of each cup using silicone sealant), which were suspended in the test chambers and oscillated vertically over a distance of 2–5 cm at a rate of approximately 2 oscillations/min. The dilution water was fed from a temperature controlled constant head tank via a flow control device to mixing/flow splitting chambers. These chambers also received, via a peristaltic pump, the calculated volume of concentrated atenolol stock solution required to attain the desired nominal exposure concentration in the tanks. Calibration of the test substance and dilution water delivery systems was by direct measurement of flow rates into and out of the mixing/flow splitting chamber. Direct measurements of the above flow rates were made at least twice per week throughout the test period, with daily checks to ensure the correct operation of the dosing system. Separate lines from each mixing/flow splitting chamber supplied at least 5-tank volumes/d to each of four replicate test vessels, and these flow rates were maintained within ±10%. Test vessels, and thus exposure concentrations, were randomly distributed on the test rig to avoid positional bias. The photoperiod used was 16 h of light followed by 8 h of dark, with 20 min dawn and dusk transition periods. The mean (±standard deviation) light intensity at the surface of the solutions was 440 ± 60 lux (n = 3). The sides and bottoms of all tanks were cleaned, and the tanks siphoned to remove debris, at least twice per week. To initiate the test, embryos 2–24 h old, from at least three females, were combined in a dish filled with dilution water. Incubation cups were randomly placed in a separate tray containing dilution water, and embryos consecutively distributed, in batches of five, to each of the cups. This process was repeated until each cup contained 20 embryos. Exposure was initiated by placing one incubation cup into each of 4 replicate tanks, giving a total of (approximately) 80 embryos per concentration. During the experimental period, the number of live embryos was recorded daily, and dead embryos were discarded. The % hatched embryos in each concentration was assessed daily and when >90% complete, or 24 h after the first hatch, embryo-larvae were released from the incubation cup into the test vessel for further development. Daily observations of fry mortality, behaviour and appearance were made and any abnormal effects recorded. At the end of the exposure period (28 d post-hatch), all surviving larvae were humanely terminated (0.5 mg/L MS222, adjusted to pH 7.5 using 1 M NaOH), and the mean wet weight and standard length of the population in each test vessel was determined, and any physical abnormalities recorded. Fish were not fed for 24 h prior to termination.

363

2.3.2. Short-term (21 d) fish reproduction study The 21 d adult fish reproduction study protocol was based on that of Harries et al. (2000) and Ankley et al. (2001). The test apparatus comprised of glass ‘adult’ tanks (13 L: measuring 300 mm × 150 mm × 350 mm), each of which contained a spawning tile. The spawning tiles consisted of an 80 mm length of 110 mm diameter PVC half guttering above a screened collection tray. The rectangular stainless steel collection tray measured 120 mm × 110 mm × 20 mm, and the ‘screen’, which was inserted in the tray below the tile, was a 0.5 cm2 stainless steel mesh. This spawning tile design was adapted from that of Thorpe et al. (2007), and allowed eggs to fall through the mesh and prevent adult fish predation, thus providing a more accurate indication of spawning performance. The egg incubation apparatus (progeny tanks), in which fertilised eggs were incubated to assess hatchability, were identical to those used in the ELS study (see previous section). The apparatus used to supply the dilution water and atenolol stock concentrations was also identical to that used in the ELS study, except that each test vessel (adult and progeny tanks) was supplied with approximately 9 tank volumes/d, to cope with the additional water quality issues exerted by using adult fish. Flow rate measurements and calibrations were undertaken in a manner that was identical to that used in the ELS study. Test vessels, and thus exposure concentrations, were randomly distributed on the test rig to avoid positional bias. The photoperiod and cleaning regime used were the same as for the ELS study, except that spawning tiles were also removed from the tanks daily and, after spawning assessment, were cleaned of eggs and debris. Where possible, all tanks were cleaned at the same times so that they received approximately equal levels of ‘disturbance’. The mean (±S.D.) light intensity at the surface of the test vessels was 740 ± 120 lux (n = 6). To initiate the test (initiation period), sexually mature male and female fish were selected at random, and one of each sex was placed into each adult test vessel (n = 8 pairs per treatment), receiving dilution water only. The fish were then left to acclimate with minimal disturbance for 3 d, during which the fish were just fed (no cleaning or measurement of any parameters were undertaken). After 3 d, the fish were then left for a further 7 d to assess breeding compatibility via observations of normal breeding behaviour, egg number, egg quality (e.g. viability and the presence of abnormalities) and spawning success (e.g. whether the female laid eggs on the spawning tile). During this pre-acclimation period, fish from the same batch were held under identical conditions as ‘spare’ fish, until successful breeding was confirmed in all test vessels. Any fish that did not meet these requirements was replaced by one of these spare fish, with all movements being recorded. The spawning data from this period, although recorded, were not used in the final analysis to assess the effect of the test chemical on reproductive performance. After initiation and acclimation, the 21 d baseline reproduction period commenced, during which all tanks received dilution water only, and baseline reproductive performance was assessed (thus allowing each pair to act as its own control). Following the baseline period, a transitional period was undertaken when the test substance was introduced into the test vessels. The transitional period commenced, after all baseline hatcha-

364

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

bility trials had been completed. Once the test substance was introduced, the system was left for a further 2 d before the exposure period was started, to ensure that the test vessels reached the designated exposure concentrations. Following the transitional period, the exposure period then continued for a further 21 d. This allowed the assessment of reproductive performance under conditions of test substance exposure, in the same breeding pairs in which baseline reproduction data had been collected. During the baseline and exposure periods, reproductive performance was assessed daily via numbers of spawnings, site of spawning (tile, mesh or base), numbers of eggs produced (recording any dead eggs and from which site), number of eggs per brood (egg batch size), and stage of embryo development. In addition, daily observations of the adults for mortality, behaviour and appearance (including observation of secondary sexual characteristics) were also made, and any abnormal effects recorded. In addition to adult fish reproductive output, specific batches of resultant eggs were subjected to hatchability trials to assess early embryo-larval development, during both the baseline and exposure periods. For each hatchability trial, 50 viable eggs were selected at random from either the tile, mesh or base (in order of preference) and these were split between 2 and 3 incubation cups each containing a maximum of 25 eggs, up to a maximum total of 50 for each trial. Wherever possible, the eggs were taken from the tile. However, if this was not possible due to insufficient egg numbers, the eggs were taken from alternative sites, in the order detailed above. On such occasions, the eggs were taken from more than one site, and the study records were used to record the site from which the eggs were obtained. In these instances, separate incubation cups were used for each site (e.g. cup 1: 20 from tile, cup 2: 25 from mesh, cup 3: 5 from base). Ideally for the hatchability trials, each pair was left for at least one spawning before any hatchability trials were undertaken. Following this first spawning, a hatchability trial was undertaken on the next batch of ≥50 eggs, and this was ideally repeated twice (with a gap of one spawning between) per breeding pair during both experimental periods. Additionally, if any two hatchability trials from one breeding pair differed by more than 30% (as % hatched), then an additional trial was undertaken, subject to egg availability. During each hatchability trial, daily observations were made to assess mortalities, abnormalities and stage of development. Eggs were incubated until all eggs hatched, or at least 2 d elapsed since the final hatching. Following the assessment of hatchability, all larvae were sacrificed using an overdose of anaesthetic (0.5 g/L MS222 in dilution water, adjusted to pH 7.5 with 1 M NaOH). At the end of the exposure period, adult F0 fish were terminated via an overdose of anaesthetic (0.5 g/L MS222 in dilution water, adjusted to pH 7.5 with 1 M NaOH), followed by destruction of the brain, and wet weight and standard length obtained. Blood samples were taken (chilled monoject syringes heparinized with 2500 units of ammonium heparin per mL of 0.9% saline) and plasma separated via centrifugation (7000 × g for 5 min), frozen in liquid nitrogen, and stored at −80 ◦ C until plasma atenolol analysis. The gonads of all fish were removed and weighed to determine the gonadosomatic index (GSI. Gonad weight/body weight × 100). Secondary sexual characteristics

of males and females were also assessed using male fish fatpad weight, fatpad index (fatpad weight/body weight × 100) and tubercle number and prominence. Tubercle prominence was assessed using a five-point scale adapted from the work of Smith (1977). 2.4. Chemical analyses of atenolol exposure concentrations For the ELS study, triplicate 20 mL water samples were taken from each of the test concentrations on study days 0, 4, 7, 14, 21 and 32 (equivalent to 28 d post-hatch), for analysis of atenolol concentration. For the 21 d adult fish reproduction study, triplicate 20 mL water samples were taken from each of the test concentrations on exposure days 0, 4, 7, 14, and 21. All 20 mL water samples were collected in 40 mL amber glass vials, acidified with 4 drops of 1 M hydrochloric acid, and stored at approximately −20 ◦ C until dispatch and analysis at Eawag, D¨ubendorf, Switzerland. Water atenolol concentrations were measured using the following method, adapted from that of Maurer et al. (2007). Water samples were filtered through 0.45 ␮m glass-fibre filters (GF 8, Schleicher and Schuell, Dassel, Germany), and spiked with 150 ng of deuterated surrogate standard (atenolol-d7, courtesy of Dr. Ehrenstorfer, Augsburg Germany). The volume of sample to be analysed was selected, such that the amount of surrogate standard present reflected the nominal concentration present in the vial. After evaporation to dryness, the extracts were reconstituted in 400 ␮L of 10% acetonitrile in water (v/v). Separation was achieved by reverse-phase liquid chromatography (250 mm × 2 mm Nucleodur C18 Gravity ␮m, Macherey-Nagal, D¨urnten, Switzerland), prior to detection by tandem electrospray mass spectrometry (API 4000, Applied Biosystems, Rotkreuz, Switzerland) in the positive ionization mode. This work was undertaken following confirmation in a preliminary study that atenolol in dilution water did not significantly degrade upon short-term freezer storage. 2.5. Chemical analyses of blood plasma atenolol concentrations Fish plasma atenolol concentrations were determined at BfG, Koblenz, Germany, using the following method. To provide enough plasma for analysis, samples from the same exposure concentrations were pooled as male or female groups. Initially, samples were vortexed with 1 mL of MeOH to precipitate plasma proteins, following which the resultant supernatant extract was diluted in 100 mL of clean groundwater (free of anthropogenic contaminants). The diluted extracts (approximately pH 7) were spiked with the surrogate standard (atenolol-d7) and enriched using endcapped C18 (500 mg) solid phase extraction (SPE) cartridges at a flow rate of 10–20 mL/min (200 mbar). Cartridges were previously conditioned with 1 mL heptane, 1 mL acetone, 3 mL × 1 mL methanol and 5 mL × 1 mL of groundwater. After the extraction of the samples, cartridges were dried with nitrogen for 1 h and the analytes eluted using 4 mL × 2 mL of methanol. After evaporation to dryness, extracts were reconstituted in 1 mL of an aqueous buffer and measured by liquid

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

365

Table 1 Summary of the biological endpoint data obtained from the ELS study Treatment (mg/L)

% Survival (at 28 d)

% Hatchability

Wet weight (g) at 28 d

DWC 0.1 0.32 1.0 3.2 10

94 95 91 100 100 94

100 (4) 100 (4) 100 (4) 98 ± 2.6 (4) 99 ± 2.5 (4) 98 ± 2.9 (4)

0.20 0.20 0.20 0.20 0.20 0.18

± ± ± ± ± ±

Standard length (mm) at 28 d

0.03 (77) 0.03 (74) 0.03 (77) 0.03 (82) 0.03 (79) 0.03 (71)*

22.57 22.74 22.60 22.17 22.21 21.75

± ± ± ± ± ±

1.32 (77) 1.11 (74) 1.15 (77) 1.46 (82) 1.11 (79) 1.21 (71)*

DWC, dilution water control (equivalent to 0 mg/L). % Survival data are mean of raw data as percentages, based on approximately 80 embryos per concentration at start, minus any mortalities. % Hatchability presented as mean ± S.D. (n) of the % eggs hatched per replicate tank (4 tanks per treatment). Wet weight and standard length data are shown as the mean ± S.D. (n) of all of the larvae measured at each treatment level, across 4 replicates. * Denotes a statistically significant difference compared with the DWC (p < 0.05).

chromatography-tandem mass spectrometry (LC-MSMS) in the positive ion mode, as described in Ramil, M., El Aref, T., Fink, G., Ternes, T.A. (unpublished). The limit of quantification (LOQ) for atenolol was 0.5 ng per pooled plasma sample. 2.6. Statistical analyses All dichotomous (% hatch, % survival) data were analysed using an exact 2 × 2 contingency table test to identify significant differences between the controls and each test substance treatment, or in the case of adult fish survival, by Fisher’s exact test. Adult fish and embryo-larval length and weight data were tested for normality, followed by Bartlett’s or Levene’s tests for homogeneity of variance. If the data were normally distributed and the variances homogenous, then the individual replicate data were analysed using ANOVA, with subsequent comparison of the DWC group to each treatment group using Dunnett’s test. If the data failed to meet the assumptions required for parametric testing, then the non-parametric Kruskal–Wallis test, followed by Dunn’s comparison of treated and DWC groups, was applied. Adult fish egg production data (total eggs per pair, total eggs per brood, and number of broods) were initially plotted (regression analysis and scatterplot) to assess if any relationship between pre and post exposure data (within the same pairs of fish) was apparent. As no relationship was observed, it was considered inappropriate to perform an analysis of co-variance (ANCOVA), and then treatments were compared with DWC groups within pre and post exposure datasets using ANOVA, or Kruskal–Wallis tests, as appropriate. Egg production data for

pairs in which a mortality occurred during any experimental period were excluded from subsequent statistical analysis. These data were, however, included in the hatchability analyses, as it was considered that parental health was an integral and acceptable factor in determining progeny hatching success. For some datasets, the coefficient of variation (CV) was calculated, and this is defined as the (mean/standard deviation) × 100. 3. Results 3.1. Fish early life stage study (ELS) The results of the biological endpoints measured in the ELS study are summarised in Table 1. Survival at the end of the study ranged between 91 and 100%, and analysis of these data concluded there were no treatment-related mortalities. Mean hatchability was greater than 98% at all concentrations, with no discernable association with treatment. Across all treatment groups, mean wet weight ranged from 0.18 to 0.20 g, and mean standard length between 21.75 and 22.74 mm. Statistical analyses revealed that there was significant reduction in both wet weight (p < 0.05) and standard length (p < 0.05) in the 10 mg/L group, when compared with the DWC larvae. 3.2. Short-term fish reproduction study 3.2.1. Adult fish (F0) mortality and morphology The results of the morphological endpoint analyses, along with mortality data measured in the 21 d adult fish reproduction

Table 2 Summary of the mortality and morphological endpoint data obtained from F0 fish in the 21 d adult fish reproduction study Treatment (mg/L)

DWC 0.1 0.32 1.0 3.2 10

Mortalities

Terminal wet weight (g)

Terminal standard length (mm)

Condition index

M

F

M

M

M

1 1 0 0 0 1

1 2 1 5 2 2

3.67 3.74 3.28 4.17 4.18 4.34

F ± ± ± ± ± ±

0.95 (7) 0.88 (7) 0.55 (8) 0.81 (7) 0.73 (8) 0.73 (7)

1.88 1.97 2.13 1.66 1.61 1.77

± ± ± ± ± ±

0.42 (7) 0.38 (6) 0.41 (7) 0.30 (3) 0.23 (6) 0.30 (6)

54 54 52 55 54 57

F ± ± ± ± ± ±

4.97 (7) 4.19 (7) 2.90 (8) 4.43 (7) 2.97 (8) 5.26 (7)

45 47 47 45 44 45

± ± ± ± ± ±

3.3 (7) 2.99 (6) 2.43 (7) 2.08 (3) 2.10 (6) 2.17 (6)

2.28 2.30 2.33 2.43 2.58 2.36

F ± ± ± ± ± ±

0.13 (7) 0.16 (7) 0.15 (8) 0.19 (7) 0.11 (8)* 0.30 (7)

2.00 1.90 2.00 1.85 1.88 2.0

± ± ± ± ± ±

0.21 (7) 0.09 (6) 0.11 (7) 0.10 (3) 0.13 (6) 0.27 (6)

DWC, dilution water control (equivalent to 0 mg/L). M, male; F, female. All treatments started with 8 pairs (8 M and 8 F, total 16 fish). Condition index calculated as CI (wet weight (mg) × 100/standard length (mm3 )). Data, where applicable, are presented as mean, ±S.D. (n). * Denotes a significant difference compared with the DWC at the p < 0.05 level.

366

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

study are summarised in Table 2. From these, mean adult fish survival ranged between 94% (0.32 mg/L) and 69% (1 mg/L), and despite the relatively high numbers of mortalities in certain treatment groups, there was no discernable relationship with concentration, or any significant differences between any treatment and the DWC for males, females or all fish combined. This conclusion was supported by the fact that in the majority of cases, mortalities were explainable (e.g. egg binding of females). Across all treatments, mean F0 male wet weights ranged from 3.28 to 4.34 g, and female wet weights from 1.66 to 2.13 g; mean F0 male standard lengths ranged from 52 to 57 mm, and females from 44 to 47 mm. There were no statistically significant differences between any of the treatments and the DWC group in either male or female fish wet weights or standard lengths, and no obvious relationship associated with treatment in either sex for either parameter. The condition index of the animal is widely used as a non-specific measure of physiological fitness in both field and laboratory studies of fish health (Weatherley and Gill, 1987). Mean F0 male condition index ranged from 2.28 to 2.58, and female fish from 1.85 to 2.00 across all treatments. In common with the weight and length data, there was no discernable pattern associated with treatment in either the males or females. There was, however, a statistically significant increase in male condition index between the DWC and the 3.2 mg/L treated group (p < 0.05), although there was an absence of any concentration-related pattern with respect to this endpoint. 3.2.2. Adult fish (F0) reproductive status Across all treatments, mean F0 male gonad weights ranged from 0.03 to 0.06 g, and the ovaries of females from 0.19 to 0.27 g. There was no discernable relationship with treatment in either sex, and this was supported by the absence of any statistically significant differences between treatments and the DWC in either sex. Gonadosomatic index can give an indication of the reproductive status of the animal by expressing gonad weight as a proportion of body weight. Mean F0 Male GSI ranged from 0.96 to 1.39%, and individual female GSI from 10.61 to 13.66% across all treatments. In common with gonad weights, there was no obvious treatment-related pattern in GSI in either males or females. This conclusion was supported by the absence of any statistically significant differences between treated groups and the DWC for this parameter. Across all treatments, mean F0 Male fatpad weight ranged from 0.16 to 0.22 g, and mean fatpad index from 3.80 to 5.09%. The mean number of tubercles counted on male fish ranged from 14.63 to 15.88, and the tubercle prominence score from 2.88 to 3.75. There was no consistent pattern associated with treatment with respect to any of the secondary sexual characteristics observed, and this was supported by the absence of statistically significant differences between any treatments and the DWC. In addition, no male secondary sexual characteristics were observed in any of the female fish sampled. 3.2.3. Adult fish (F0) egg production All datasets in which mortality occurred were excluded from the following figures or statistical analysis. Baseline period mean total egg production ranged from 3610 to 7382 across all treat-

ments, and the mean total eggs per female data from 516 to 1055, with no discernable relationship with treatment in either case. The mean total eggs per female per reproductive d data ranged from 24.6 to 50.2 in the baseline period, and between 27.6 and 34.2 in the exposure period. Similarly, the mean total eggs per female per brood ranged from 97.1 to 159.7 in the baseline period to between 106.3 and 150.6 during the exposure period. The change in egg production from baseline to exposure period ranged from −32% to +39% for eggs per female, from −35% to +39% for eggs per female per reproductive d, and from −15% to +33% for eggs per female per brood. None of the egg production parameters showed a discernable relationship with atenolol concentration, and this was supported by the absence of any statistically significant differences between treatments and the DWC. Although egg production appeared low in all experimental groups in both the baseline and exposure periods, the mean number of eggs per female per reproductive d compared favourably with DWC data from previous fathead minnow chronic studies conducted at AstraZeneca, Brixham (unpublished data). 3.2.4. F1 egg hatchability Across all treatment concentrations, mean F1 egg hatchability ranged between 81 and 89% during the baseline period, and 48 and 83% during the exposure period. Although the mean% hatchability was generally lower during the exposure period when compared with the baseline period, this was the case in all treatments including the DWC, and there was no discernable association with atenolol concentration. Statistical analysis failed to reveal any significant differences in hatchability between treated and DWC groups, within either the baseline or exposure experimental periods. 3.3. Summary of NOECs and LOECs obtained from both studies A summary of all no observed effect concentrations (NOECs) and lowest observed effect concentrations (LOECs) obtained for the endpoints measured in the current studies are presented in Table 3. From these data, the lowest NOEC and LOEC obtained, and as such the data that would be used in an environmental risk assessment for atenolol, can be seen. These were for the 28 d wet weight and standard lengths of larval fish from the ELS study (NOEC = 3.2 mg/L and LOEC = 10 mg/L), and for the F0 adult male fish condition index from the 21 d adult fish reproduction study (NOEC = 1.0 mg/L and LOEC = 3.2 mg/L), although the biological relevance of the latter is questionable, as mentioned earlier and discussed later. 3.4. Chemical analyses of water and fish plasma atenolol concentrations 3.4.1. Water atenolol concentrations In the ELS study, from the time-weighted mean data, measured atenolol concentrations ranged from 90.05 to 99.83% of nominal (89.92–100% for arithmetic means). No atenolol was detected in any of the dilution water control samples analysed. From the time-weighted mean data in the 21 d adult

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

367

Table 3 Summary of all of the no observed effect concentrations (NOECs) and lowest observed effect concentrations (LOECs) measured in both the ELS and 21 d adult fish reproduction studies Study

Endpoint

NOEC (mg/L)

ELS study

Hatchability % Survival Wet weight Standard length

10 10 3.2 3.2

21 d adult fish reproductive study

LOEC (mg/L) >10 >10 10 10

Sex

Male

Female

Male

Female

Survival Standard length Wet weight Condition index Gonad weights GSI Fatpad weight Fatpad index Tubercle number Tubercle grade Egg production F1 hatchability

10 10 10 1.0 10 10 10 10 10 10 10 10

10 10 10 10 10 10 10 10 10 10 10 10

>10 >10 >10 3.2 >10 >10 >10 >10 >10 >10 >10 >10

>10 >10 >10 >10 >10 >10 >10 >10 >10 >10 >10 >10

NOEC, no observed effect concentration and LOEC, lowest observed effect concentration.

fish reproduction study, the measured atenolol concentrations ranged from 77.10 to 95.67% (76.86–96.03% for arithmetic means). Again, no atenolol was detected in any of the dilution water control samples analysed. 3.4.2. Fish plasma atenolol levels The results of the fathead minnow plasma atenolol concentration analyses are summarised in Table 4. From these data, plasma atenolol concentrations were between 1.8 and 6.2% (males) and between 0 and 12.2% (females) of the corresponding time-weighted mean measured water concentrations. 4. Discussion Two experiments have been successfully completed to address, firstly, the effect of atenolol on fish embryo-larval development, and secondly on adult fish reproduction. Using fathead minnows as a standard OECD test species, the ELS study gave 28 d NOECgrowth and LOECgrowth values of 3.2 and

10 mg/L, respectively (time-weighted mean measured values were between 90 and 100% of these nominal exposure values). Survival and hatching were both approximately 3-times less sensitive than growth in the fish ELS study, agreeing with data from other studies (McKim, 1977). It was, however, not possible to associate these effect levels with plasma concentrations of atenolol, as the animals were too small to obtain blood samples from. For the 21 d adult fish reproduction study, the most sensitive endpoint was male fish condition index, giving a NOECcondition index and LOECcondition index of 1.0 and 3.2 mg/L, respectively. In the same study, the NOECreproduction and LOECreproduction were 10 and >10 mg/L, respectively (timeweighted mean measured values were between 77 and 96% of these nominal exposure values). Collectively, these data suggest that atenolol is far less toxic in fish than propranolol, another commonly prescribed ␤-blocker (Owen, S.F., Huggett, D.B., Hutchinson, T.H., Hetheridge, M.J., Kinter, L.B. Ericson, J.F., Sumpter, J.P., unpublished

Table 4 Atenolol concentrations measured in fathead minnow plasma from the 21 d adult fish reproduction study Nominal concentration (mg/L)

DWC 0.1 0.32 1.0 3.2 10.0

Male

Female

Plasma volume (␮L)

Atenolol concentration (␮g/L)

% Of measured water concentration

Plasma volume (␮L)

Atenolol concentration (␮g/L)

90 110 225 175 2 × 70 50


– 4.4 4.4 6.2 1.9 and 1.8 1.8

50 75 85 45 70 50

1.7
% Of measured water concentration – – 12.2 – 3.9 3.2

DWC, dilution water control (equivalent to 0 mg/L). All values are derived from just one pooled sample (due to low sample volumes), except (* ). All concentrations are corrected for sex-specific blank (DWC) values. % Of measured concentrations related to time-weighted mean water concentrations measured. LOQ, limit of quantification (0.5 ng per pooled plasma sample).

368

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

data). In addition, in the current study, fish reproduction was approximately 3-times less sensitive than the fish embryo-larval NOECgrowth . This is in contrast with the findings of Huggett et al. (2002) for propranolol, who reported that fish reproduction endpoints were of the highest sensitivity. Although the specific reason for the differences in the toxicological profiles of these two closely related compounds is not clear, in the first instance species sensitivity could certainly have played a part (for example a cyprinid was employed here, versus a salmonid in Owen, S.F., Huggett, D.B., Hutchinson, T.H., Hetheridge, M.J., Kinter, L.B. Ericson, J.F., Sumpter, J.P., unpublished data). In addition, as stated earlier, although sharing the same basic mode of pharmacological action, these two compounds exhibit distinctly different specific receptor targets, potencies and specificities, as well as pharmacokinetic and physicochemical properties (see Maurer et al., 2007; Owen et al., 2007). Consequently, all of these factors are likely to impact on the overall toxicological profile of these two ␤-blockers, and could have contributed to the different reported effects in the various fish species investigated, especially given the apparent differences in plasma exposure between the two compounds at a given water concentration (see below). The corresponding male fish plasma concentration of atenolol measured at the male fish LOECcondition index of 3.2 mg/L (nominal), was 0.0518 mg/L, and these data can be used to ‘test’ the conceptual ‘mammalian-fish leverage model’ described by Huggett et al. (2004). This model proposes the calculation of a theoretical fish steady-state plasma concentration based on its log Kow and predicted or measured environmental concentration (EC), which can then be compared with the human therapeutic plasma concentration to provide an effect ratio (ER). As we have both measured ‘environmental’ (tank) and fish plasma concentrations, we can use these data to evaluate the validity of this approach with respect to atenolol. The first part of the model uses a water:blood partition coefficient (log PBlood:Water ) calculation based on the log Kow of the compound (Fitzsimmons et al., 2001). In this respect, the most appropriate value to use is the log D7.4 of −1.61 (Barbato et al., 2005), as the partition coefficient at a physiologically relevant pH (equal to log Kow at pH 7.4). Therefore, the blood:water partition coefficient: log PBlood:Water = 0.73 × log Kow − 0.88 where log D7.4 for atenolol = −1.61. From this, the fish steady-state plasma concentration (Fss PC) is calculated by: (Fss PC) = EC × (PBlood:Water ) where the environmental concentration is 2.764 mg/L (timeweighted mean measured concentration at nominal 3.2 mg/L), and PBlood:Water = 8.81−3 . From this, the estimated Fss PC for atenolol would be 0.0244 mg/L compared with the measured plasma concentration of 0.0518 mg/L at the same ‘environmental’ concentration. The measured fish plasma concentration, therefore, compares favourably with the model’s predicted fish plasma concentration, particularly considering the model is calibrated for trout plasma,

and takes no account of absorption, distribution, metabolism or excretion (ADME). The plasma concentrations measured here, in relation to the corresponding measured water levels (ranging from 1.8 to 12.2%), are low compared with the few other published reports of fish plasma concentrations of human pharmaceuticals. For example, Owen et al. (2007) and Owen, S.F., Huggett, D.B., Hutchinson, T.H., Hetheridge, M.J., Kinter, L.B. Ericson, J.F., Sumpter, J.P. (unpublished data) measured plasma propranolol concentrations typically in the region of 40–80% of the measured water levels in rainbow trout (Oncorhynchus mykiss) after 40 d (although levels were variable); and Mimeault et al. (2005) measured plasma gemfibrozil concentrations of between 92 and 500 times the corresponding measured water levels in goldfish (Carassius auratus) after 14 d. Although the number of published examples is still small, this serves to highlight the great variability in fish plasma levels following waterborne exposure to human pharmaceuticals, and this is probably due, at least in part, to the wide differences in physicochemical properties and ADME shown by such compounds. Next, we can apply the predicted and measured fish plasma concentration at the LOEC, along with the human therapeutic concentration (from Schulz and Schmoldt, 2003), to obtain the ER for atenolol according to the next part of the model: Effect ratio (ER) =

HT PC Fss PC(measured or

predicted)

where HT PC = maximum human therapeutic blood plasma concentration (for atenolol = 1 mg/L), Fss PC(measured) = 0.0518 mg/L for atenolol, Fss PC(predicted) = 0.0244 mg/L for atenolol. Hence, for atenolol the measured ER = 19.3, and the predicted ER = 40.98. Interestingly, Huggett et al. (2004) proposed that if the ER is >1000 then theoretically there is a low likelihood of a drug causing significant chronic effects in fish. If the ER is between 1 and 1000 then it was suggested that fish chronic toxicity testing might be warranted, while if the ER is <1 then it was proposed that fish chronic toxicity testing should be conducted in case of potent protein receptor-mediated effects. From our data, a measured ER of 19.3 would warrant fish chronic testing, as has been undertaken, and from the resultant data, atenolol does not cause significant chronic effects in fathead minnow at environmentally realistic concentrations. The data produced for atenolol here, therefore, fit well with this approach but also serve to highlight the importance of generating ‘real’ experimental data with which to calibrate and validate such models. Moreover, the absence of significant chronic effects in the in vivo studies also reinforces the importance of such data as central to the environmental risk assessment of pharmaceuticals, rather than just a reliance on what should be considered adjunct, in silico, approaches. Finally, these fish chronic effects data can also be used to address the issue of the acute-chronic ratio of atenolol; a concept that has been identified as particularly important in respect of the potential environmental impact of human pharmaceuticals (Ankley et al., 2005). For atenolol, the 96-h LC50 for rainbow trout is >100 mg/L (AstraZeneca, unpublished data) whereas the most sensitive chronic endpoint (from the current studies)

M.J. Winter et al. / Aquatic Toxicology 86 (2008) 361–369

gave a LOECcondition index of 3.2 mg/L (nominal). Therefore, the acute LC50 :chronic LOEC value gives an ACR of >31.25 for atenolol. This value is somewhat higher than the ACRs of <10 typical for industrial chemicals, but far lower than that of 17␣ethinylestradiol and other potent steroidal oestrogens (Ankley et al., 2005). Overall this reflects the relatively low chronic fish toxicity of atenolol confirmed under the conditions of the current studies. In conclusion, these data suggest that atenolol has low chronic toxicity to fish under the conditions described, particularly when consideration is given to the low environmental concentrations of this compound that have been detected in surface waters. In addition, these data have proven valuable for testing the conceptual ‘mammalian-fish leverage model’ of Huggett et al. (2004), and also for calculating an ACR, thus evaluating these approaches as useful indicators (along with in vivo data) of the possible long-term effects of ␤blockers, and other human pharmaceuticals, in wild fish populations. Acknowledgements The authors are grateful to Mr. Gareth Readman for supplying fathead minnows and Mr. Alan Sharpe for the statistical analyses. We also thank Drs. Grace Panter and Tim Williams for their scientific advice, and technical input on this work. This work was funded by the AstraZeneca Global Safety, Health and Environment Research Program, and the European Union (ERApharm project, contract no. 511135). References Ankley, G.T., Jensen, K.M., Kahl, M.D., Korte, J.J., Makynen, E.A., 2001. Description and evaluation of a short-term reproduction test with the fathead minnow (Pimephales promelas). Environ. Toxicol. Chem. 20, 1276–1290. Ankley, G.T., Black, M.C., Garric, J., Hutchinson, T.H., Iguchi, T., 2005. A framework for assessing the hazard of pharmaceutical materials to aquatic species. In: Williams, R.T. (Ed.), Human Pharmaceuticals—Assessing the Impacts on Aquatic Ecosystems. SETAC Press, Pensacola, pp. 183–237. Ankley, G.T., Villeneuve, D.L., 2006. The fathead minnow in aquatic toxicology: past, present and future. Aquat. Toxicol. 78, 91–102. Barbato, F., di Martino, G., Grumetto, L., La Rotonda, M.I., 2005. Can protonated ␤-blockers interact with biomembranes stronger than neutral isolipophilic compounds? A chromatographic study on three different phospholipid stationary phases (IAM-HPLC). Eur. J. Pharma. Sci. 25, 379–386. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Toxicol. 76, 122–159 (with erratum in Aquat. Toxicol. 78, 207).

369

Fitzsimmons, P.N., Fernandez, J.D., Hoffman, A.D., Butterworth, B.C., Nichols, J.W., 2001. Branchial elimination of superhydrophobic organic compounds by rainbow trout (Oncorhynchus mykiss). Aquat. Toxicol. 55, 23–34. Fraysse, B., Garric, J., 2005. Prediction and experimental validation of acute toxicity of beta-blockers in Ceriodaphnia dubia. Environ. Toxicol. Chem. 24, 2470–2476. Harries, J.E., Runnalls, T., Hill, E., Harris, C.A., Maddix, S., Sumpter, J.P., Tyler, C.R., 2000. Development of a reproductive performance test for endocrine disrupting chemicals using pair-breeding fathead minnows (Pimephales promelas). Environ. Sci. Technol. 34, 3003–3011. Huggett, D.B., Brooks, B.W., Peterson, B., Foran, C.M., Schlenk, D., 2002. Toxicity of select beta adrenergic receptor-blocking pharmaceuticals (betablockers) on aquatic organisms. Arch. Environ. Contam. Toxicol. 43, 229–235. Huggett, D.B., Ericson, J.F., Cook, J.C., Williams, R.T., 2004. Plasma concentrations of human pharmaceuticals as predictors of pharmacological responses in fish. In: K¨ummerer, K. (Ed.), Pharmaceuticals in the Environment: Sources, Fate, Effects and Risks. Springer, Berlin, pp. 373– 386. Knacker, T., Duis, K., Ternes, T., Fenner, K., Escher, B., Schmitt, H., Rombke, J., Garric, J., Hutchinson, T.H., Boxall, A.B., 2005. The EU-project ERAPharm, Incentives for the further development of guidance documents? Environ. Sci. Pollut. Res. Int. 12, 62–65. McKim, J.M., 1977. Evaluation of tests with early life-stages of fish for predicting long term toxicity. J. Fish. Res. Board Can. 34, 1148–1154. Maurer, M., Escher, B.I., Richle, P., Schaffner, C., Alder, A.C., 2007. Elimination of beta-blockers in sewage treatment plants. Water Res. 41, 1614–1622. Mimeault, C., Woodhouse, A.J., Miao, X.S., Metcalfe, C.D., Moon, T.W., Trudeau, V.L., 2005. The human lipid regulator, gemfibrozil bioconcentrates and reduces testosterone in the goldfish, Carassius auratus. Aquat. Toxicol. 73, 44–54. Organisation for Economic Co-operation and Development (OECD), 1992. Fish, early-life stage toxicity test–test guideline 210. Adopted by the council on 17th July 1992. Owen, S.F., Giltrow, E., Huggett, D.B., Hutchinson, T.H., Saye, J., Winter, M.J., Sumpter, J.P., 2007. Comparative physiology, pharmacology and toxicology of ␤-blockers in fish: a critical review. Aquat. Toxicol. 82, 145–162. Rang, H.P., Dale, M.M., Ritter, J.M., Moore, P.K., 2003. Pharmacology, fifth ed. Elsevier, Churchill Livingstone, p. 797. Schulz, M., Schmoldt, A., 2003. Therapeutic and toxic blood concentrations of more than 800 drugs and other xenobiotics. Pharmazie 58, 447–474. Smith, R.J.F., 1977. Seasonal changes in the histology of the gonads and dorsal skin of the fathead minnow (Pimephales promelas). Can. J. Zool. 56, 2103–2109. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32, 3245–3600. Thorpe, K.L., Benstead, R., Hutchinson, T.H., Tyler, C.R., 2007. An optimised experimental test procedure for measuring chemical effects on reproduction in the fathead minnow, Pimephales promelas. Aquat. Toxicol. 81, 90–98. Weatherley, A.H., Gill, H.S., 1987. The Biology of Fish Growth. Academic Press, London, p. 443. Zuccato, E., Castiglioni, S., Fanelli, R., 2005. Identification of pharmaceuticals for human use contaminating the Italian aquatic environment. J. Hazard. Mater. 122, 205–209.