Individual and combined ecotoxic effects of water-soluble polymers

Test chemicals

The four WSPs used as test compounds for the single substance tests and as mixture components in our study were purchased from Sigma-Aldrich (Schnelldorf, Germany): polyacrylamide (PAM; CAS No: 9003-05-8; average MW: 10,000 Da), polyethylene glycol (PEG-20; CAS No: 25322-68-3; average MW: 200 Da); anionic homopolymer of acrylic acid (P-AA; CAS No: 9003-04-7; average MW: 2,100 Da); cationic polyquaternium-6 (PQ-6; CAS No: 26062-79-3, average MW: 100,000 Da). The chemical properties and structures of the WSPs are given in Table 1. Generally, water solubility, charge and MW are the polymer properties considered most relevant with regard to the environmental fate and effects of polymers (Duis, Junker & Coors, 2021). In the selection of the WSPs, attention was paid to obtain polymers with the lowest MW belonging to the non-ionic, anionic and cationic polymer classes. The test compounds were dissolved in water and stored at 4 °C. The same stock solutions were used throughout the study. A spacing factor of 2 was used to delineate the concentration of the WSPs in the media used for the test (e.g., 1.5, 3, 6, 12, 25, 50, 100, 200 mg/L). All tests were performed using at least six different concentrations. The concentration range tested was generally 0.07 to 100 mg/L, but a range of 0.01 to 0.24 was used when preliminary data for the particular WSP indicated a higher toxicity, while a range of 100 to 1,000 mg/L was used when a lower toxicity was observed.

Table 1:

Chemical properties and structures of the water-soluble polymers used as test compounds. n.a. = not available.

Polymer class	Technical name	Systematic name	Average molecular weight and density	Chemical structure
Polyacrylamides	Polyacrylamide (PAM)	poly(2-propenamide)	10,000 Da n.a.
Polyethylene glycols	Polyethylene glycol(PEG-200)	Poly(oxy-1,2-ethanediyl), α-hydro- ω-hydroxy-	200 Da 1.16 g/mL
Polycarboxylates: polyacrylates	Anionic homopolymer of acrylic acid(P-AA)	2-Propenoic acid, homopolymer, sodium salt	2,100 Da 0.55 g/mL
Polyquaterniums	Cationic polyquaternium-6(PQ-6)	2-Propen-1-aminium, N,N-dimethyl-N-2-propen-1-yl-, chloride (1:1), homopolymer	100,000 Da 1.09 g/mL

DOI: 10.7717/peerj.16475/table-1

Microtox assay

The Microtox assay with the bioluminescent bacterium Aliivibrio fischeri assessed the baseline toxicity (ISO-Guideline 11348-3, 2017). The assay was performed in a miniaturized 96-well plate format according to Völker et al. (2017). Luminescence was measured prior to and 30 min after sample addition using a Spark 10 M microplate reader (Tecan, Crailsheim, Germany). In total, three independent experiments were conducted for every single compound and their binary mixtures in the concentration range between 7.81 and 1,000 mg/L with 8 concentrations and a spacing factor of 2.

Ames fluctuation assay

The Ames fluctuation assay was applied to evaluate the mutagenic potential of the WSPs used in the study. The assay was performed according to Magdeburg et al. (2014) in accordance with ISO guideline 11350 (ISO-Guideline 11350, 2012) with Salmonella typhimurium strains YG 1041 and YG 1042 (Hagiwara et al., 1993) with and without metabolic activation by rodent liver enzymes (S9-mix; Envigo CRS, Roßdorf, Germany). Mutagenicity was photometrically determined by a colour change of the indicator bromocresol purple (CAS: 115-40-2; Merck, Darmstadt, Germany) at 420 nm (Spark M; Tecan, Crailsheim, Germany). Wells with an optical density of >0.45 (and a clear colour change) were considered mutagenic. For every single compound and their binary mixtures, triplicate experiments were conducted as a limit test at a concentration of 1,000 mg/L.

Yeast-based reporter gene assays for endocrine and dioxin-like activities

Yeast-based reporter gene (lacZ encoding for β-galactosidase) assays were conducted to assess the endocrine and dioxin-like activity. We examined the agonistic activities at the human estrogen receptor α (hERα) (Yeast Estrogen Screen = YES; Routledge & Sumpter, 1996) and human androgen receptor (hAR) (Yeast Androgen Screen = YES; Sohoni & Sumpter, 1998) as well as the activation of the human aryl hydrocarbon receptor (Yeast Dioxin Screen = YDS; Miller, 1997) according to the protocol of Abbas et al. (2019). Activities were photometrically determined at 540 nm (Multiskan Ascent, Thermo Labsystems) through cleavage of chlorophenol red- β-D-galactopyranoside (CAS: 99792-79-7; Sigma-Aldrich, Steinheim, Germany) by β-galactosidase over a period up to 60 min at 5-10 min intervals. In total, three independent experiments were conducted for every single compound and their binary mixtures as a limit test at a concentration of 1,000 mg/L.

72-h growth inhibition assay with Raphidocelis subcapitata

The 72 h growth inhibition experiment was performed using R. subcapitata algae in the exponential growth phase. This experiment was carried out according to OECD guideline 201 adapted to 24-well microplates (Moreira-Santos, Soares & Ribeiro, 2004). The algae were exposed to six concentrations of each polymer (for PAM, PEG-200, and P-AA ranging from 3.2 to 100 mg/L; for PQ-6 ranging from 0.015 to 0.48 mg/L), a positive control (potassium chromate at a concentration of 8.262 ×10⁻⁹ mol/microplate-well) and a negative control for 96 h. Three replicates were performed for each concentration and control with two mL of test solution and a cell density of 20,000 cells/mL. The experiments were carried out under continuous illumination (4 klx/m²/s), at 26 ± 1 °C, 60% humidity. To prevent cell clumping, an automatic stirrer at 220 rpm was used. Mixture tests were carried out by mixing polymers which showed inhibitory effects on growth in R. subcapitata. Nine concentrations were prepared for each polymer mixture with 5 replicates by following the procedure described above. Absorbance (ABS) values were measured at 595 nm after 24, 48, 72 and 96 h of exposure using a Tecan microplate reader (Tecan, Crailsheim, Germany). After each measurement, plates were sealed with Breathe-Easy-membrane.

Statistical analysis and mathematical modelling

GraphPad Prism (versions 5 and 8; GraphPad Software, San Diego, CA, USA) were used for statistical and nonlinear regression analyses of the data obtained from the experiments. One-way analysis of variance (ANOVA) followed by Dunnett’s test was applied to analyse significant differences among treatments at the p < 0.05 significance level.

Concentration-response curve fits and EC₅₀ values with 95% confidence intervals were calculated only for chemicals that showed significant effects in the R. subcapitata growth inhibition test. The top value was set to the data point with the greatest inhibitory effect in the test. Top values and EC₅₀ values were used to predict mixture toxicity with the models in an Excel spreadsheet.

Equation (1) was used for CA modelling. (1)${\displaystyle \mathrm{EC}{x}_{\mathit{mix}}={\left(\sum _{i=1}^n\frac{p_i}{EC{x}_i}\right)}^{-1}}$ where ECx_mix is the total concentration of the mixture at which x% effect occurs and p_i is the fraction of component i in the mixture. EC_xi is the concentration of i mixture component that provokes x% effect when applied singly.

Equation (2) was used to calculate the mixture effects according to IA. The concentration–response relationships F_i of the individual components were used to calculate their effects E(c_i) as (2)${\displaystyle E\left(c_{mix}\right)=1-\prod _{i=1}^n\left[1-F\left(c_i\right)\right]}$

If the concentrations of the individual components are expressed as fractions of the total concentration, the overall effect of any given total mixture concentration can be calculated as according to Eq. (3) (3)${\displaystyle E\left(c_{mix}\right)=1-\prod _{i=1}^n\left[1-F\left(p_i\ast c_{mix}\right)\right]}$

where p_i is the fraction of component i in the total mixture concentration and ∏ stands for the product (multiplication).

Equation (4) was used to calculate the mixture effects according to GCA. (4)${\displaystyle E=\frac{\left(\frac{{\text{max effect level}}_A\left[A\right]}{E{C}_{50A}}+\frac{{\text{max effect level}}_B\left[B\right]}{E{C}_{50B}}+\dots \right)}{1+\frac{\left[A\right]}{E{C}_{50A}}+\frac{\left[B\right]}{E{C}_{50B}}+\dots }}$

In Eq. (4), ‘E’ represents the effect of the mixture at the given concentration of the chemicals A ([A]) and B ([B]). The term ‘max effect level’ stands for the maximum effect levels of the chemicals A and B as individual compounds, respectively, while ‘EC_50A’ and ‘EC_50B’ represent their median effect concentrations as individual compounds. Using Eq. (4), concentration-response curves were calculated for various concentrations of the two compounds in binary mixtures.

In vitro assays

None of the four individual substances and none of their binary mixtures showed statistically significant effects in the investigated concentration range in the in vitro assays. In the Microtox assay to assess baseline toxicity, the maximum inhibition of luminescence was only 1.4% and 5.5% for PAM and PEG-200, respectively, at the highest test concentration of 1,000 mg/L. These inhibitions were not statistically significantly different from the control (one-way ANOVA, p > 0.05). Likewise, none of the other WSPs or their mixture showed an inhibition of luminescence in the Microtox assay.

The results of the Ames assay with strains YG 1041 and YG1042, both with and without S9 mix, also provided no evidence of mutagenic activity of the tested individual WSPs or their binary mixtures at a concentration of 1,000 mg/L, since the revertant proportions remained well below the required threshold of 20.8% for a positive test outcome according to ISO-Guideline 11350 (2012). Revertant proportions above 10% were only observed in two tests: 14.0% for PEG-200 with strain YG 1041 without S9 and 15.8% for the PEG-200/P-AA mixture with strain YG 1042 without S9, while the revertant proportions for all other individual WSPs and their binary mixtures were below 10%. In assays with S9 mixture, no revertants appeared for any of the substances or mixtures.

The agonistic activities determined at the estrogen receptor α, the testosterone and aryl hydrocarbon receptors were well below the limit of detection (LOD) of the respective yeast-based reporter gene assays (LOD: YES = 0.002 µg 17 β-estradiol equivalents/L; YAS = 0.176 µg testosterone equivalents/L; YDS = 11.2 µg β-naphthoflavone equivalents/L). Accordingly, even at the tested concentration of 1,000 mg/L, the results do not indicate any receptor mediated estrogenic, androgenic or dioxin-like activity for the individual WSPs and their binary mixtures.

72-h growth inhibition assay with Raphidocelis subcapitata

In contrast, P-AA at concentrations of 25, 50 and 100 mg/L and PQ-6 at concentrations of 0.05 and 0.1 mg/L as single substances caused a statistically significant (p < 0.05 significance level) reduction of specific growth rate in the microalga R. subcapitata, but no toxicity effects were detected for the non-ionic WSPs PEG-200 and PAM (Fig. 1 and Fig. 2A). The toxic effect of P-AA in binary mixtures with PAM and PEG-200 on algal growth was the same as that of P-AA alone, but the effect of PQ-6 in binary mixtures with the non-ionic WSPs on algal growth was only observed when PQ-6 was at the highest concentration (0.1 mg/L). Among the binary mixtures, the mixture of P-AA and PQ-6 with a concentration ratio of 1/1,000 had the highest toxic effect (Fig. 2B).

P-AA, PQ-6 and their mixture exerted different toxicities to R. subcapitata with EC₅₀ values after 72 h ranging from 0.04 to 35.5 mg/L (95% confidence limits). PQ-6 showed to be more toxic with a 72 h EC₅₀ (95% CL) of 0.06 (0.04–0.08) mg/L than P-AA with 24.2 (16.5–35.5) mg/L. The 72 h EC₅₀ value determined for their mixture (9.74 mg/L; 95% CL 8.06–11.8 mg/L) is lower than for P-AA and higher than for PQ-6 (Table 2 and Fig. 3).

The results of the experimental determination of the mixture toxicity of P-AA and PQ-6 as well as the predictions made by CA, IA, and GCA concepts are shown in Fig. 4. The CA model predicts a higher mixture toxicity than the IA and GCA models. The observed EC₅₀ value of the mixture was 9.74 mg/L, while the values estimated by CA, GCA and IA were 4.68 (95% CL 4.60–4.77), 7.10 (6.49–7.76) and 9.28 (8.65–9.96) mg/L, respectively. Accordingly, only the IA model exhibited overlapping 95% confidence limits of the calculated EC₅₀ value with the experimentally determined EC₅₀.