Influence of pH on the toxicity of ionisable pharmaceuticals and personal care products to freshwater invertebrates.
TL;DR: The results of this study show that pH fluctuations can have a considerable influence on toxicity thresholds, and should be taken into account for the risk assessment of ionisable pharmaceuticals and personal health-care products.
Abstract: The majority of pharmaceuticals and personal health-care products are ionisable molecules at environmentally relevant pHs. The ionization state of these molecules in freshwater ecosystems may influence their toxicity potential to aquatic organisms. In this study we evaluated to what extent varying pH conditions may influence the toxicity of the antibiotic enrofloxacin (ENR) and the personal care product ingredient triclosan (TCS) to three freshwater invertebrates: the ephemeropteran Cloeon dipterum, the amphipod Gammarus pulex and the snail Physella acuta. Acute toxicity tests were performed by adjusting the water pH to four nominal levels: 6.5, 7.0, 7.5 and 8.0. Furthermore, we tested the efficiency of three toxicity models with different assumptions regarding the uptake and toxicity potential of ionisable chemicals with the experimental data produced in this study. The results of the toxicity tests indicate that pH fluctuations of only 1.5 units can influence EC50-48 h and EC50-96 h values by a factor of 1.4–2.7. Overall, the model that only focuses on the fraction of neutral chemical and the model that takes into account ion-trapping of the test molecules showed the best performance, although present limitations to perform risk assessments across a wide pH range (i.e., well above or below the substance pKa). Under such conditions, the model that takes into account the toxicity of the neutral and the ionized chemical form is preferred. The results of this study show that pH fluctuations can have a considerable influence on toxicity thresholds, and should therefore be taken into account for the risk assessment of ionisable pharmaceuticals and personal health-care products. Based on our results, an assessment factor of at least three should be used to account for toxicity differences between standard laboratory and field pH conditions. The models evaluated here can be used to perform refined risk assessments by taking into account the influence of temporal and spatial pH fluctuations on aquatic toxicity.
Summary (4 min read)
- Residues of pharmaceuticals and chemicals contained in personal health care products , have been monitored in a wide range of aquatic ecosystems across the world (Boxall et al.
- Bioaccumulation and toxicity predictive models used for the ecological risk assessment of pharmaceuticals and PHCPs are generally based on the hydrophobic nature of chemicals and may therefore provide less accurate predictions when applied for ionisable substances.
- The second model is based on the ion trap effect and assumes a preferential uptake of the neutral form of the chemical followed by a fast intracellular dissociation.
- The main objectives of the present study were to assess the toxicity of a pharmaceutical and a PHCP ingredient to three aquatic invertebrates under a gradient of environmentally relevant pH conditions, and to evaluate the suitability of the aforementioned pH-dependent toxicity models for them.
- The selected compounds were enrofloxacin (ENR) and triclosan (TCS).
2.1. Study chemicals
- ENR (active ingredient ≥ 98%) and TCS (active ingredient ≥ 97%) were purchased from Sigma Aldrich (St Louis USA).
- Separate stock solutions of ENR (50 g/L) and TCS (2 g/L) were prepared by diluting the pure substances in Milli-Q water with the help of NaOH, and were stored at −20 °C until their use in the experiments.
2.2. Test organisms
- The toxicity of ENR and TCS was evaluated on three invertebrate species: the amphipod crustacean Gammarus pulex, the insect nymphs of Cloeon dipterum and the freshwater snail Physella acuta.
- G. pulex were collected from an uncontaminated stream in Heelsum, the Netherlands.
- Prior to the experiments the water content, the lipid content and the internal pH of the test organisms was evaluated (Table 1).
- After evaporation, the vials were weighed again and the total lipid content of the sample was determined to calculate the lipid content of the aquatic organisms.
- Then, both micro sensors were inserted into the solution formed and the pH was read from this sample.
2.3. Toxicity experiments
- Toxicity experiments were performed following a 4 × 6 factorial design, with 4 different pHs (6.5, 7, 7.5 and 8), one control and 5 chemical concentrations.
- The experiments were performed following some general recommendations provided in the Organisation for Economic Co-operation and Development (OECD): test guideline No. 202 (OECDOrganization for Economic Cooperation and Development, 2004).
- The chosen temperature and light:dark regime was 20 °C and 12:12 h, respectively.
- Temperature, conductivity and dissolved oxygen concentration in the exposure media were measured at the beginning and at the end of the toxicity experiment (Table S3).
- G. pulex and C. dipterum individuals were counted as immobile when they showed inability to move after a tactile stimulus provided with a glass Pasteur pipette.
2.4. Chemical analyses
- ENR and TCS concentrations were measured in the test medium at 2 h and 96 h after the application of the test compounds to verify the nominal concentrations and to assess the dissipation of the test compounds (Table S4).
- Water samples were filtered through a 0.22-μm cellulose acetate membrane.
- Chemical quantification was performed by injecting the amber glass vials into a triple quadrupole LC/MS system equipped with an ESI+.
- A full description of the equipment and conditions used for the analysis of ENR and TCS are provided in the Supporting Information (see also Tables S5 and S6).
- Additional tests were performed to evaluate the recovery of ENR and TCS from the test medium, using a concentration of 1 mg/L of ENR and 634 μg/L of TCS, which are in the low-to-middle range of the concentrations used in the toxicity tests.
2.5.1. Model 1: Only the neutral chemical form is active
- The model considers the speciation of compounds in the exposure medium, and assumes that the neutral chemical form is taken up faster than the charged, so that the charged form does not contribute at all to the observed effect and can be neglected (Boström and Berglund, 2015).
- Hence, the slope coefficient ( )1 N is calculated and used as independent variable in a linear regression, and the EC50 is determined from the regression slope coefficient.
2.5.2. Model 2: Both chemical forms are active and act additively
- The model assumes that both the and the forms are biologically active but with different effect concentrations, EC50 and EC50 , and that the and the concentration act additively in the mixture, i.e., using the concentration addition model (Neuwoehner and Escher, 2011).
- For simplicity, the authors assume that the cationic chemical form (in the case of ENR) does not contribute to the overall effect and consider only the anionic form.
2.5.3. Model 3: Only the neutral chemical fraction is active and results in an ion-trap effect
- Similarly to model 1, this model assumes that the uptake of neutral chemical form by the aquatic organisms is much faster than that of the charged one, and therefore assumes permeability of the neutral chemical form only.
- Moreover it considers dissociation of the chemical inside the organisms due to a difference between the pH of the exposure medium and the internal pH of the organisms, leading to an ion trap effect.
2.5. Data analyses
- The data obtained from the toxicity experiments were used to calculate EC50 values, and their 95% confidence intervals, after an exposure period of 48 h and 96 h.
- The calculations were performed using a log-logistic regression model as described by Rubach et al. (2011), and using the GenStat 11th edition software (VSN International Ltd., Oxford, UK).
- All calculations were done on the basis of the average measured exposure concentrations during the experimental period.
- Models 1–3 were implemented in Mathematica 12.0 (Wolfram Research) and fitted to experimental data.
- Linear regression coefficients (R2) and Pearson p-values were calculated using the method “LinearModelFit”, and were used as indicators of correspondence between the calculated experimental data and the fitted models.
3.1. Invertebrate's sensitivity at different pH levels
- Toxicity tests were performed to evaluate the sensitivity of the three invertebrate species to ENR and TCS at four different nominal pH levels.
- Differences between the measured pH values and the nominal pH in the test medium of the toxicity experiments were generally within 0.2 units, with few exceptions going up to 0.3 units (Table 2).
- According to Aranami and Readman (2007), the fast water dissipation of this compound is explained by its photolytic nature, its high sorption capacity to organic matter, and to a lower extent by hydrolisis.
- The dissociation of TCS in the tested pH range was a bit lower than for ENR, and ranged from 3% to 35%, approximately (Table 2).
- For G. pulex, TCS EC50-96 h values were low and showed less marked differences; however EC50-48 h values showed the same trend as for the other invertebrates, with a toxicity value that was 1.5 times higher in the pH 8 treatment as compared to the 6.5 treatment (Table 2).
3.2. pH-dependent toxicity models
- Model 1 showed a good representation of the variability in the pHvariable toxicity values for both tested compounds (Figs. 1 and 2, Table 3), with R2 values above 94% and 85% for ENR and TCS, respectively, and significant Pearson correlations (p-values < 0.05).
- From a theoretical point of view, Model 2 would be the preferred option as compared to Model 1 since it assumes that both the charged and the neutral chemical forms are active, and altough have different toxic potency, they act additively.
- Model 2 showed the poorest fit for ENR and TCS, with Pearson correlation p-values above 0.05 (Table 3).
- The latter confirms that for ENR the EC50s is more toxic than the EC50s .
- These results must be interpreted taking into account that only a narrow pH range could be tested, the internal pH values of the tested organisms were close to neutrality, and the variability in the EC50 values was comparatively large.
- This study supports the need to take into account the variability in pH conditions of aquatic ecosystems for the risk assessment of ionisable pharmaceuticals and PHCPs.
- Moreover, this study shows the efficiency of three models that can be used to extrapolate toxicity values under different pH conditions.
- Conceptualization, Investigation, also known as Frits Gillissen.
- Paul J. Van den Brink: Conceptualization, Writing - original draft.
- The authors declare no conflicts of interest.
Did you find this useful? Give us your feedback
Related Papers (5)
Frequently Asked Questions (1)
Q1. What are the contributions mentioned in the paper "Influence of ph on the toxicity of ionisable pharmaceuticals and personal care products to freshwater invertebrates" ?
In this study the authors evaluated to what extent varying pH conditions may influence the toxicity of the antibiotic enrofloxacin ( ENR ) and the personal care product ingredient triclosan ( TCS ) to three freshwater invertebrates: the ephemeropteran Cloeon dipterum, the amphipod Gammarus pulex and the snail Physella acuta. Acute toxicity tests were performed by adjusting the water pH to four nominal levels: 6. 5, 7. 0, 7. 5 and 8. 0. Furthermore, the authors tested the efficiency of three toxicity models with different assumptions regarding the uptake and toxicity potential of ionisable chemicals with the experimental data produced in this study. The results of this study show that pH fluctuations can have a considerable influence on toxicity thresholds, and should therefore be taken into account for the risk assessment of ionisable pharmaceuticals and personal health-care products.