Internal exposure dynamics drive the Adverse Outcome Pathways of synthetic glucocorticoids in fish

The Adverse Outcome Pathway (AOP) framework represents a valuable conceptual tool to systematically integrate existing toxicological knowledge from a mechanistic perspective to facilitate predictions of chemical-induced effects across species. However, its application for decision-making requires the transition from qualitative to quantitative AOP (qAOP). Here we used a fish model and the synthetic glucocorticoid beclomethasone dipropionate (BDP) to investigate the role of chemical-specific properties, pharmacokinetics, and internal exposure dynamics in the development of qAOPs. We generated a qAOP network based on drug plasma concentrations and focused on immunodepression, skin androgenisation, disruption of gluconeogenesis and reproductive performance. We showed that internal exposure dynamics and chemical-specific properties influence the development of qAOPs and their predictive power. Comparing the effects of two different glucocorticoids, we highlight how relatively similar in vitro hazard-based indicators can lead to different in vivo risk. This discrepancy can be predicted by their different uptake potential, pharmacokinetic (PK) and pharmacodynamic (PD) profiles. We recommend that the development phase of qAOPs should include the application of species-species uptake and physiologically-based PK/PD models. This integration will significantly enhance the predictive power, enabling a more accurate assessment of the risk and the reliable transferability of qAOPs across chemicals.

. Chemical structures of BDP and three of its transformation products.

Study rationale
Exposure of fish to beclomethasone dipropionate (BDP) for 21 days in Experiment 1 caused modeof-action driven effects at nominal concentrations ≥100 ng/L (i.e. hyperglycaemia, skin androgenisation, decreased lymphocyte population); however, unexpected time-dependent drug stability problems identified subsequently prevented an exact determination of NOEC and LOEC values. Due to the observed in vivo potency of BDP, we then performed a number of studies to characterize the chemical behaviour of BDP in the flow-through exposure system used in Experiment 1. The results of these studies showed an highly reproducible degradation dynamics of BDP in water that allowed a retrospective reliable determination of exposure dynamics in Experiment 1. Finally, the same results were used to drive the design of Experiment 2 and aid the interpretation of the results of our previous exposure experiments (12).
Dechlorinated tap water (5 and 10 μm carbon filtered) was used as dilution water, and general parameters (pH, temperature and dissolved oxygen) were monitored daily throughout the study.
Water pH ranged from 7.3 to 7.7, temperature from 24.5 to 25.7 °C, and dissolved oxygen from 7.2 to 7.9 mg/L.

Flow-through exposure system
Two 21-day in vivo exposure studies were carried out at Brunel University London using a flowthrough system. A generalised scheme of the system is represented in Figure S2. Thermostatically heated (25±1°C) dechlorinated tap water flowed into individual glass mixing chambers (12 in Experiment 1; 9 in Experiment 2) at a rate of 200 mL/min. The same chambers also received the stock solution of the test chemical via peristaltic pump at a rate of 0.2 mL/min (Experiment 1) and 0.02 mL/min (Experiment 2), in order to achieve the desired nominal concentrations. From each mixing chamber water flowed into one glass tank (20 L) hosting 12 fish. Each fish tank underwent approximately volume changes per day. In Experiment 1, BDP concentrated stock solutions were prepared in double-distilled water every four days. In Experiment 2, BDP concentrated stock solutions were prepared in N,N-dimethylformamide (DMF; CAS number 68-12-2; ≥99%) (Sigma, Poole, UK) and replaced every seven days. Experiment 1 included three treatment groups exposed to nominal concentrations of 10, 100 and 1000 ng BDP/L and one control group receiving only clean water. Experiment 2 included two treatment groups exposed to 10 and 1000 ng BDP/L and one solvent control group. Water in all test vessels contained DMF at 0.0095%, within the limit recommended by OECD guidelines. Each treatment group included three replicate tanks. The stability of BDP and its two major metabolites, 17-BMP and BOH ( Figure S1), was investigated both in concentrated solutions and in fish tanks, two critical sites of chemical transformation. To ensure that the observed chemical behaviour of BDP was not due to laboratory-specific conditions, the studies on BDP stability were replicated independently at Brunel University London

Stability studies.
The possible transformation of BDP in concentrated stock solutions was investigated by incubating concentrated BDP solutions (10,000 µg/L) prepared in different solvents (water and DMF) and at different pH (6.0, 7.4, 8.4) at 25°C for 96h in amber glass bottles. The concentrations of BDP and its potential transformation products (17-BMP, BOH) were quantified every 24h. Additional experiments were conducted to confirm the initial results and to investigate not only the stability of BDP, but also of 17-BMP and BOH. Concentrated stock solutions were prepared in different solvents (water, methanol, water/methanol (90:10, v/v) and incubated for 72h at 25°C. Samples were collected every 8 h and analysed immediately by LC-MS/MS. Additionally, the stability of BDP, 17-BMP and BOH was assessed in samples preserved under different conditions and analysed at different time points after collection, this being the most realistic scenario in ecotoxicology and environmental chemistry studies (43). Two sample storage strategies were employed during a period of 21 days using water samples collected from control fish tanks during an in vivo study. This water likely contained the typical load of biological material, such as bacteria and microalgae, associated with fish tanks, but did not contain any xenobiotic chemical other than the one spiked during the procedure. In the first strategy, water samples were spiked, separately, with 1 µg/L of BDP, 17-BMP and BOH and stored at −20°C in PET containers for up to 21 days. In the second preservation strategy, water samples were spiked with 1 µg/L of BDP, 17-BMP and BOH, loaded immediately onto solid phase extraction (SPE) cartridges and stored at −20°C. All samples were analysed on Day 0, 4, 14, and 21.

Pharmacological activity of transformation products.
To investigate if the transformation products generated in water retained pharmacological activity, the GR activity of water solutions of BDP, 17-BMP and BOH was quantified using a Human GR Reporter Assay System (Indigo Biosciences) following the manufacturer's instructions. Equipotent solutions of dexamethasone (DEX, used as reference compound), BDP, 17-BMP and BOH were prepared on the basis of their relative potencies (44, 45). The relative potency factors used were DEX : BDP : 17-BMP : BOH = 1 : 0.4 : 13.5 : 0.8. BDP, 17-BMP, and BOH water solutions were assayed immediately after preparation (Time 0) and after 4 days of incubation at 25°C. At each time point, the concentrations of the test compounds were quantified by LC-MS/MS.

Preparation of water and plasma samples.
Samples collected from concentrated stock solutions and fish tanks were diluted with the appropriate volume of methanol (MeOH) or water to reach a MeOH:water ratio of 50:50 (v/v) and spiked with internal standard (BDP-d10, 10 µg/mL) before analysis by LC-MS/MS. SPE was applied to all the other samples (46). Extracts were reconstituted in 1 mL of methanol/water (50:50, v/v) and 10 µL of BDP-d10 solution (1 mg/mL). Concentrations of BDP, 17-BMP and BOH were quantified by LC-

MS/MS.
Plasma samples (10-50 µl) were added to 400 µl of acetonitrile to achieve protein precipitation and vortexed for 15 s. Samples were transferred to an Ostro 96-well plate (Waters, USA) connected to a vacuum system for the removal of phospholipids. An aliquot of 300 µl was collected from each extract and was placed under a N2 current to dry completely. Finally, extracts were dissolved in 100 µl of methanol/water (1:1). Concentrations of BDP, 17-BMP and BOH were quantified by LC-
The UHPLC instrument was coupled with a 5500 QTRAP hybrid triple quadrupole-linear ion trap mass spectrometer (Applied Biosystems) with an electrospray interface. Compound dependent MS parameters (declustering potential (DP), collision energy (CE) and collision cell exit potential (CXP)) as well as compound selected reaction monitoring (SRM) transitions were optimized by direct infusion of individual standard solution of each analyte at 10 µg/L (Table S1). All transitions were recorded in Scheduled MRM algorithm with 30 s detection window. Source dependent parameters were: curtain gas (CUR): 30 V; nitrogen collision gas (CAD): medium; source temperature: 300 °C; ion spray voltage: 5500 V; ion spray gases GS1: 60 V and GS2: 70 V. Instrument control data acquisition and data analysis were carried out using Analyst software (Applied Biosystem). Two SRM transitions between the precursor ion and the two most abundant fragment ions were monitored for each compound. The first transition was used for quantification purposes, whereas the second one was used to confirm the identity of the target compounds.

Stability of concentrated solutions of BDP, 17-BMP and BOH in water and solvent.
The first set of experiments was conducted to assess the chemical stability of BDP in the concentrated stock solutions (10,000 µg/L). The results ( Figure S3A) showed that BDP in water decreased to 22 ± 3% (SD, n=3) of the initial concentration after 24h in a room at 25°C. Thereafter, the rate of decrease slowed markedly to reach 11 ± 12% of the initial BDP concentration at 96h. The pH of the water solution did not significantly affect the degradation rate, which was similar for all the pH values used in the experiment (6, 7.4, and 8.4). BDP solutions prepared in 100% DMF showed minor degradation after 48h, but BDP concentrations remained within ± 15% of the starting value for the duration of the experiment ( Figure S3A). Subsequent experiments were performed by increasing the sampling frequency (from 24h to 2-8h) to better characterize the chemical behaviour of BDP in the first 24h ( Figure S3B). BDP stability was compared in solutions prepared in 100% water, MeOH:water (10:90, v:v) and 100% MeOH at pH 7.4. BDP solutions prepared in 100% MeOH showed minor degradation after 48h, but, as with DMF, BDP concentrations remained within ± 15% of the starting value for the duration of the experiment. However, 10% MeOH did not produce any stabilising effect on BDP concentrations, which had decreased to approximately 50% of the initial BDP concentration after 4h of incubation (t 1/2 =4h), and to 25% after 8h. BDP prepared in water was not detected anymore after 54h of incubation (t 1/2 =4h).
The BDP solutions were also tested for the formation of the metabolites 17-BMP and BOH. 17-BMP was detected at a concentration of <0.2 µg/L after 32h of incubation and at all following sampling times. Thus 17-BMP represented 0.002% of the parent compound BDP in the sample. BOH was never detected at any time point. This result indicates that cleavage of one or both of the propionate groups does not explain the rapid BDP degradation in water and confirms that metabolic activation in the organism is required to efficiently convert BDP into 17-BMP and 17-BMP into BOH. This reaction does not seem to occur abiotically in sterile water (e.g. via hydrolysis), or at least not above negligible rates.
To test the hypothesis that the propionate groups of BDP and 17-BMP increase their susceptibility to degradation, stability experiments were also performed for 17-BMP ( Figure S3C) and BOH ( Figure   S3D). As expected, 17-BMP degraded in 100% water (t 1/2 =40h), but at a slower rate than BDP. After 8h, 17-BMP was present at 87 ± 2% of the starting concentration versus 21 ± 4% of BDP. At subsequent sampling times the degradation continued in a linear manner, the concentration reaching 31 ± 1% after 72h. The complete absence of degradation of BOH in both 100% water and MeOH solutions (10% and 100%) ( Figure S3D) suggests that the propionate groups increase the susceptibility to degradation.

Pharmacological activity of BDP, 17-BMP and BOH and their transformation products.
To investigate if unidentified transformation products of BDP, 17-BMP and BOH retained pharmacological activity, a GR assay was conducted on concentrated solutions prepared in 100% water immediately after preparation and after 96h. The GR activities of the BDP, 17-BMP and BOH solutions after 96h were, respectively, 30.3 ± 9.7%, 53.1 ± 8.6 %, and 91.0 ± 14.9% (mean ± SD; n=3) of the activity quantified on Day 0 ( Figure S3E). The concentrations of BDP, BMP and BOH in the same samples after 96h of incubation were 47% ± 4.7%, 60% ± 7.1% and 93% ± 2.8% of the initial concentrations. This result indicates that the pharmacological activities of BDP and 17-BMP decrease simultaneously with the decrease in concentrations of the parent compounds, and that the transformation products of both BDP and 17-BMP do not contribute to the pharmacological activity of the sample to any significant degree. degradation of all three compounds was observed when samples were preserved frozen at -20°C (Table S2). Relatively little degradation was observed for samples stored on SPE cartridges after 14 days, with concentrations remaining between 80% and 100% of the initial concentrations. However, significant loss of BDP and 17-BMP had occurred by Day 21 (Table S2). BDP and 17-BMP stored in water showed significant degradation after 4 days and by Day 21 the concentrations had decreased to approximately 50% of the initial value (Table S2). BOH confirmed its higher stability compared to BDP and 17-BMP and degradation was observed only after 21 days in both storage conditions.

Quantification of BDP, 17-BMP and BOH in water samples collected from fish tanks.
BDP concentrations in water sample collected during Experiment 1 were approximately 10% of the nominal values for all the three treatment groups (Table S3). In the same water samples, 17-BMP and BOH were also detected at concentrations between 21 and 30 ng/L. At each sampling time, water was collected before the replacement of the stock solutions. This result, combined with those obtained in the stability studies, indicated that, in Experiment 1, fish were exposed to oscillatory concentrations of BDP in water that, in turn, likely produced oscillatory concentrations of BDP, 17-BMP and BOH in fish plasma. In this scenario, fish were exposed to the peak concentrations of BDP (equal to the nominal concentrations) only for 10 hours out of the total 504 hours of exposure. In Experiment 2, measured water concentrations of BDP were consistently approximately 100% of the nominal concentrations, indicating that fish were exposed to constant concentration of the drug over the 21 days of the experiment. The results obtained in all the described studies were used to estimate the exposure dynamics in both Experiment 1 and Experiment 2 ( Figure S3).  Figure S4. Water concentrations of BDP in Experiment 1 and Experiment 2.