Different factors determined the toxicokinetics of organic chemicals and nanomaterials exposure to zebrafish (Danio Rerio)

Different factors determined the toxicokinetics of organic chemicals and nanomaterials exposure to zebrafish (Danio Rerio)

Ecotoxicology and Environmental Safety 186 (2019) 109810 Contents lists available at ScienceDirect Ecotoxicology and Environmental Safety journal ho...

1MB Sizes 0 Downloads 39 Views

Ecotoxicology and Environmental Safety 186 (2019) 109810

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety journal homepage: www.elsevier.com/locate/ecoenv

Different factors determined the toxicokinetics of organic chemicals and nanomaterials exposure to zebrafish (Danio Rerio)

T

Yongfei Gaoa, Zhicheng Xieb, Jianfeng Fenga,∗, Weiqi Maa, Lin Zhua a

Key Laboratory of Pollution Process and Environmental Criteria of Ministry of Education and Tianjin Key Laboratory of Environmental Remediation and Pollution Control, College of Environmental Science and Engineering, Nankai University, Tianjin, 300071, China b Tianjin Academy of Environmental Sciences, Tianjin, 300191, China

A R T I C LE I N FO

A B S T R A C T

Keywords: Organic chemicals Nanomaterials Toxicokinetics BCF Zebrafish (Danio Rerio)

Little is known about how the chemical properties (molecular structure, such as the hydrophobic and hydrophilic end group for organic chemical, and particle size for nanomaterials (NMs)) quantitatively affect the toxicokinetics (TK) in organisms especially in short-term, single-species studies. A novel method based on a firstorder one compartment TK model which described the monophasic uptake pattern and two-compartment TK model which adequately described the biphasic metabolism pattern was used to determine the bioconcentration and TK rate constants of organic compounds (n = 17) and nanomaterials (NMs, n = 7) in zebrafish. For both one and two compartment model, the uptake (kin) and elimination (kout) rate constants were fitted using a one- and two-compartment first-order kinetic model, and bioconcentration factors (BCF) and 95% depuration times (t95) for all tested chemicals were calculated, respectively. The results showed that there was significant difference in TK parameters kin, kout, and BCF between organic chemicals and nano metal oxides. For organic compounds, significant correlations were found between the kin and BCF and the octanol-water partition coefficient (Kow) and molecular mass. For nano metal oxides, there was a significant negative correlation between the kin or BCF and particle size, but a positive correlation between kin and Zeta potential of nanoparticles and also a significant positive correlation between kout and particle size or specific surface area. Those findings indicated that NMs particle size does matter in biological influx and efflux processes. Our results suggest that the TK process for organic compound and NMs are correlated by different chemical properties and highlight that the Kow, the absorption kin, metabolism k12 and k21, elimination rate kout, and all the parameters that enable the prediction and partitioning of chemicals need to be precisely determined in order to allow an effective TK modeling. It would therefore appear that the TK process of untested chemicals by a fish may be extrapolated from known chemical properties.

1. Introduction Different chemicals with different properties have revealed considerable variation in toxicity, which was shown in short-term, singlespecies studies (Baird and Van den Brink, 2007; Maltby et al., 2005). For certain lipophilic compounds such as polychlorobiphenyls (PCB), only small amount were metabolized and excreted by organisms, leading to the concentrated compounds in the tissues or various organs of the organisms compared with the surrounding water (Venquiaruti Escarrone et al., 2016). This phenomenon is termed biological concentration (bioconcentration), which was considered as the most important problem among the chronic toxicity of chemicals to non-target organs (Pittinger and Pecquet, 2018). Therefore, it is challenging for predicting the bioconcentration of a chemical before discharging it as a ∗

toxicant into the environment. The chronic toxicity of NMs to fish in long-term exposure, such as biological accumulation, attracted attention (Hou et al., 2013), and the chronic exposure of aquatic organisms to metal oxide NMs could affect the uptake, biotransformation, and elimination of NMs by aquatic organisms and then the overall toxicity of NMs (Du et al., 2019; Sousa et al., 2019; Tiwari et al., 2019). The accumulation and toxicity of NMs was found to correlate with several physicochemical properties such as the shape, specific surface area, the primary particles size, surface potential (Zeta-potential), and surface chemistry of the nanoparticles (Nel et al., 2006). Thus, the exceptional physicochemical characteristics and the toxicokinetic (TK) process of NMs in biological liquid should be taken into account in toxicity studies (Grech et al., 2017; Lamon et al., 2019; Yang et al., 2017). In environmental risk assessment (ERA), it is challenging how to account for

Corresponding author. E-mail address: [email protected] (J. Feng).

https://doi.org/10.1016/j.ecoenv.2019.109810 Received 26 June 2019; Received in revised form 28 September 2019; Accepted 11 October 2019 0147-6513/ © 2019 Elsevier Inc. All rights reserved.

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

used in a lot of biological toxicology (Scholz, 2012). In the last decade, zebrafish toxicity tests on EDCs, PAHs, and NMs have great advances to assess modes of toxic action of EDCs on steroid receptor-regulated pathways both in vitro and in vivo (Cosnefroy, 2012; Tokarz, 2013; Vosges, 2010). In this study, by extracting uptake and elimination data in freshwater zebrafish of EDCs including insecticides, fungicides, herbicides, chlorinated or brominated flame retardants, and PAHs and nano-metal oxides, the variations in one- and two-compartment TK for different chemicals were quantified by estimating uptake and elimination rate constants (kin and kout, respectively) from the extracted experimental data using a first-order one-compartment model. We wanted to address the question “Does the constant TK model apply to PAH, EDCs, and nano metal particles toxicity in the freshwater zebrafish?”, and the utility of the first- order TK model was evaluated to predict the chemical bioconcentration factors to estimate the internal effective concentration of the tested chemicals and the impact factors of chemical metabolism in zebrafish. Furthermore, we explored the physicochemical factors influencing kin, kout, 95% depuration times (t95), and bioconcentration factors (BCF) with prediction intervals, and discussed the observed variation across freshwater organisms for ERA. Our results provide foundation for extrapolation TK between chemical to chemical, even for extrapolation TD.

these differences in the TK process in a species exposed to the different chemicals. We need a tool to relate local TK (absorption, metabolism at tissue or organ, elimination) and exposure at whole body level in order to further exploit the TK process in zebrafish exposed to different chemicals. Conventionally, Quantitative Structure–Activity Relationships (QSAR) model was used to extrapolate toxicity of different chemicals, including organic chemicals and metal oxides NMs (Babic et al., 2018; Choi et al., 2018; Wang et al., 2019). However, QSAR did not provide biological information (e.g. absorption, metabolism, and elimination, AME) linked with chemical properties (e.g. hydrophobicity and hydrophilicity). For a lot of chemicals, QSAR was found suitable for the protection of aquatic life (Maltby et al., 2005), but a comprehensive understanding of effect of the chemical properties on accumulation variation in a species across organic compounds and NMs is still lacking (Chen et al., 2017; Liu et al., 2019). Moreover, a large part of difference in species sensitivity towards toxicants was explained mostly by differences between species themselves (Vaal et al., 2000), the remaining difference toxicity must lie physicochemical properties and their mode of metabolism. Toxicokinetic–toxicodynamic (TK-TD) modeling has been proposed as a risk assessment tool and aims to evaluate the toxic effects in organisms exposed to Plant Protection Products at the Society of Environmental Toxicology and Chemistry (SETAC) European Workshop (Brock et al., 2010). Conceptually, TK and TD are two processes of a chemical entering the organisms, while TK include absorption, distribution, metabolism and elimination (ADME) of the chemical and TD include the damage, damage recovery and repair processes and subsequent toxicity induced by the chemical in organism themselves, which was integrated by Ashauer et al. (2006). Toxicity variation in a species across chemicals is systematically driven by TK processes. One compartment TK model including absorption and elimination has no capability to quantify the processes of metal metabolism, especially in adult animals, and subsequently cannot accurately define the biological effective dose and interpret the mode of toxic action due to the detoxification mechanisms (Stadnicka-Michalak et al., 2014). Therefore, it is necessary to consider the detoxification process of chemicals in organisms and use a two-compartment TK model for chemical which separated metabolically available and detoxified metals, and further toxicity can thus be successfully interpreted through relating to the concentration of the metabolically available metals rather than detoxified metals (Tan and Wang, 2012). Species sensitivity towards toxicants can be explained by TK parameters. There exists a solid body of literature on how TK and TD process of hydrocarbons relate to the octanol-water partition coefficient (Kow) (Baas et al., 2015; Klok et al., 2014). However, these relationships have so far only been derived from empirical data on Daphnia magna and Pimephales promelas, and there is a lack of quantifying TK and TD process of hydrocarbons relate to the octanol-water partition coefficient (Kow). In previous studies, researchers revealed the TK process of several trace metals and found the suitable freshwater fish to monitor trace metal bioavailability (Gao et al., 2015, 2016; Tan and Wang, 2012; Wang et al., 2013). Further studies investigated TK processes of Polycyclic Aromatic Hydrocarbons (PAHs), endocrine disrupting chemicals (EDCs), and nano metal particles in freshwater fish (Lee et al., 2002; Segner, 2009; Zhu et al., 2009), and indicated that biotransformation of above compounds was not negligible. Little is known on the ability of freshwater fish to concentrate aqueous PAHs, EDCs, and nano metal particles under laboratory conditions, to derive TK parameters (BCF, uptake and elimination rate constants, biological half-lives), and to investigate the potential suitability of first-order compartment TK models to describe uptake and elimination processes. Moreover, little is known about how the physicochemical properties (e.g., partitioning coefficients and water solubility) affect protein binding and biotransformation. Studies which focus on biotransformation in relation to bioconcentration in fish species are scarce. Zebrafish (Danio rerio) as toxicology model organism is worldwide

2. Materials and methods 2.1. Data sets The data sets collected from literatures by extracting time-course accumulation and elimination data in zebrafish during freshwater experiment for EDCs, PAHs and nano-metal and metal oxides. Data should meet requirements: (1) the chemicals for ZFE exposure were dissolved in ISO standard dilution water. All substances were well below their maximum water solubility; (2) quality assurance (QA) and quality control (QC) requires parallel experiments and continuous experimental data. For more details are shown in the Supporting Information Tables S1–S5. The data concerns the organic compounds (Brox et al., 2014; Kuehnert et al., 2013; Pery et al., 2014; Sanz-Landaluze et al., 2015) including benz(a)anthracene, fluoranthene, fluorene, naphthalene, PBDE154 (2,2′,4,4′,5,6′-hexabromo diphenyl ether), PCB104 (2,2′,4,6,6′-pentachlorobiphenyl), PCB136 (2,2′,3,3′,6,6′-hexachlorobiphenyl) (embryo), and atrazine, chlorpyrifos, dicofol (larvae), Benzocaine, Clofibric acid, Metribuzin, BPA (Bisphenol A), diazinon, EE2 (Ethinyl-estradio), endosulfan alpha, endosulfan beta, tebuconazole (adult), nano-metal oxides (Zhang et al., 2015) (Fe2O3, CuO, Fe3O4, ZnO, different crystal structure A-TiO2 and R–TiO2) and nano Ag (adult) (Griffitt et al., 2009; Ma et al., 2018). Above data sets corresponded to a constant exposure scenario under different experimental designs, ranging from 1 or 2 tested concentrations, 3 to 10 time points and 10 to 30 organisms in each treatment to measure survival. 2.2. Model First-order compartment models were used to describe the timedependent tissue residues in zebrafish. Examination of the elimination patterns revealed monophasic elimination for individual chemicals. The basic concept of a one or two-compartment model for relating metal accumulation to toxicity has been described by Swietaszczyk and Jodal (Swietaszczyk and Jodal, 2019). In modeling, a compartment is used to specify where (e.g. in the plasma) or in what state (e.g. free or bound) the toxicant is distributed. Compartment modeling assumes uniform distribution within each compartment, i.e. that each compartment can be assigned a concentration. As shown in Fig. 1, One- and two-compartment model were assumed to better described uptake and elimination patterns of toxicant (Tan and Wang, 2011). One-compartment assumes only a metabolized compartment. The two-compartment model assumes a metabolized compartment and an exchanging 2

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

dC2 (t ) = k12 × C1 (t ) − k21 × C2 (t ) dt

(4)

where C1(t) and C2(t) are the metabolically available and detoxified metal concentration (amount mass−1), respectively. kout is the efflux rate constant of metabolically available metals (time−1), and k12 and k21 are the metal transfer rate constants from metabolically available to detoxified metal and detoxified to metabolically available metal (time−1), respectively. 2.5. Parameterization In this study, using the Nelder-Mead method with the “fitode” package in R 3.5.0. (http://www.R-project.org/), the one compartment TK parameters were estimated using chemical accumulation datasets with the Nelder-Mead method Levenberg–Marquardt (L-M) algorithm. The accumulation data were used to estimate kin, and the elimination data were used to estimate kout. From the kout estimated by L-M method, 95% depuration times (t95) was derived using the equation: t95 = -ln (0.05)/kout. For theoretical steady-state conditions at constant exposure concentrations, the net transport between adjacent compartments is assumed to be zero. For one compartment, the bioconcentration factor (BCF, L g−1) can be calculated as BCF = kin/(kout + kg), kg is the fish's growth rate (h−1) and is set at 0 in this study (Rubach et al., 2010). For two compartment, BCF = fw1(k1/k2) + (1-fw1)(k3/k4)(k1/k2). Further, we assumed the weight fraction of the metabolized compartment as fw1 (dimensionless ratio): fw1 = w1/(w1+ w2). In this study, the fw1 was arbitrarily chosen as 0.5 without definite information on internal distribution species of all chemicals (Van Hattum and Montanes, 1999).

Fig. 1. First-order rate constant one and two-compartment model applied in this study. Symbols and units are explained in the text.

detoxified compartment. Detoxified processes meant that metal was bound to protein (like metallothionein MT) which was not toxic to organisms (Wallace et al., 2003; Wallace and Luoma, 2003). Metabolic processes meant that metal was brought into organelle (like mitochondria and lysosome) (Wallace et al., 2003; Wallace and Luoma, 2003). In the models considered here, the rate of the transport of toxicant (quantity per time) is proportional to its concentration in individual compartment. The transport rate constants (k) are describing the direction of the transport, i.e., kij denotes transfer rate constant from compartment i to j, kji denotes transfer rate constant from j to i (Swietaszczyk and Jodal, 2019). The weights of the metabolized compartment and exchanging metabolized compartment were defined as w1, w2, and w1+w2 (in g) of the whole organism.

2.6. Statistical and analysis We evaluated normality of distribution for all experimental data using the Kolmogorov–Smirnov test. All linear regression between kin, kout, t95, and log BCF and log Kow or molecular mass was done using Origin 9.1 and the coefficient of determination (R2) and the cut-off level for statistical significance was taken as p < 0.05. Then we compared the estimated kin, kout, t95, and log BCF of all tested chemicals, and try to explore the physiochemical factors of chemicals resulting in difference among these TK parameters. The log BCFs of all tested chemicals was often compared to their octanol-water partition coefficient (log Kow) in publications (Brox et al., 2014; Kuehnert et al., 2013; Pery et al., 2011, 2014; Rapaport and Eisenreich, 1984; Sanz-Landaluze et al., 2015; Zhou et al., 2005). Log Kow represents the lipophilicity and the hydrophobicity of a chemical and how it thermodynamically distributes, i.e., partitions, between aqueous and organic phases. Log Kow is generally considered to be a reasonable surrogate phase for lipids in biological organisms (Mackay, 1982). For all chemicals, t95 and log BCF individually were related to log Kow or molecular mass. For nano metal oxides, kin, kout, t95, and log BCF individually were related to particle sizes. Here, NP particle size is referring to the mean hydration size in the test media.

2.3. One compartment TK model A simple one-compartment TK model with a first-order kinetic equation was applied to simulate the toxicant concentration in organisms which related the exposure concentrations. The time-course of the metal concentration in the whole-body Cint(t) [amount mass−1] is described as the following equation:

dCint (t ) = kin × C (t ) − kout × Cint (t ) dt

(1)

where C(t) [amount volume−1] is the chemical concentration in exposure over time. kin is the uptake rate constant (volume mass−1 time−1), kout is the elimination rate constant (time −1). t is time (time).

3. Results and discussions 2.4. Two compartment TK model 3.1. Measured and modeled toxicokinetics Metals in fish were divided into two parts: metabolically available (compartment one) and detoxified (compartment two). Briefly, timecourse metal concentration in fish Cint(t), (amount mass−1) is written as:

Cint (t ) = C1 (t ) + C2 (t )

(2)

dC2 (t ) = kin × CW (t ) − (ke1 + k12) × C1 (t ) + k21 × C2 (t ) dt

(3)

Internal concentrations increased with prolonged time, indicating uptake of the organic compounds and nano-metal oxides into the body, and then transfer of the exposed organisms into ultrapure water, decreasing internal concentrations were observed, showing that the toxicant was being eliminated (See in the Supplemental Information). Oneand a two-compartment TK model were used to fit the measured concentrations in zebrafish in order to calculate uptake and elimination rates, i.e., kin and kout (Supporting Information Figs. S1–50 and Table 1). 3

zebrafish

adult adult adult adult adult adult adult larvae larvae larvae adult larvae larvae adult larvae larvae larvae adult adult larvae larvae larvae larvae larvae larvae adult adult adult adult adult adult adult

chemical

ATiO2 CuO Fe3O4 FeO RTiO2 ZnO Ag atrazine atrazine benz(a)anthracene Benzocaine chlorpyrifos chlorpyrifos Clofibric acid dicofol Fuoranthene fluorene Metribuzin naphthalene PBDE 154 PBDE 154 PCB 104 PCB 104 PCB 136 PCB 136 BPA diazinon diazinon EE2 endosulfan apha endosulfan beta tebuconazole

4000 4000 4000 4000 4000 4000 10 100 500 3.386 10,000 1 10 50,000 1 92.02 67.48 50,000 578.86 1 5 1 10 4 12 97.5 100 400 1 0.2 0.1 200

0.0483 0.1450 0.0287 0.0064 0.0077 0.0146 0.7181 0.0164 0.0031 0.5285 0.0001 0.0781 0.1903 0.0000 0.1276 0.7346 0.3441 0.0000 0.0806 0.3359 0.0480 0.0748 0.0327 1.6464 0.1725 0.0006 0.0181 0.0264 0.0179 0.8877 0.1598 0.0002

0.6693 0.977 0.077 0.117 0.529 0.077 0.024 1.85 0.852 0.849 0.052 0.077 0.163 0.047 0.027 0.836 0.842 0.0862 0.856 0.075 0.047 0.038 0.837 0.0811 0.0852 0.008 0.117 0.089 0.015 0.387 0.36 0.002

4.4759 3.0663 38.9056 25.6045 5.6630 38.9056 124.8222 1.6193 3.5161 3.5285 57.6102 38.9056 18.3787 63.7390 110.9530 3.5834 3.5579 34.7533 3.4997 39.9431 63.7390 78.8351 3.5791 36.9387 35.1612 374.4665 25.6045 33.6599 199.7155 7.7409 8.3215 1497.8661

t95(h) 0.0721 0.1484 0.3728 0.0544 0.0146 0.1897 29.9193 0.0089 0.0036 0.6225 0.0013 1.0149 1.1673 0.0004 4.7253 0.8787 0.4086 0.0005 0.0942 4.4781 1.0204 1.9677 0.0390 20.3015 2.0243 0.0729 0.1543 0.2966 1.1932 2.2937 0.4438 0.1060

BCF (L·g−1) 1.8579 2.1713 2.5714 1.7355 1.1639 2.2780 4.4760 0.9486 0.5616 2.7941 0.1263 3.0064 3.0672 −0.4088 3.6744 2.9439 2.6113 −0.3170 1.9738 3.6511 3.0088 3.2940 1.5914 4.3075 3.3063 1.8628 2.1884 2.4721 3.0767 3.3605 2.6472 2.0253

logBCF (L·g−1) 89 67 70 89 91 92 74 94 95 20 57 87 65 39 95 69 72 64 83 76 85 71 73 81 89 44 94 75 98 69 85 72

R2 (%) 0.0116 0.0333 0.0289 0.0063 0.0039 0.0146 0.0378 0.0234 0.0040 0.3075 0.0005 0.0110 0.3763 0.0000 0.1289 0.4329 0.2015 0.0000 0.0146 0.9332 0.2538 0.3645 0.1152 0.7991 0.6190 0.1500 0.0231 0.0289 0.0257 2.2798 1.5979 0.0020

kin (L·g−·1h−1)

kM,out (h−1)

Cw (μg·L−1)

kin (L·g−1·h−1)

Two compartment

One compartment

0.047 0.032 12,200 2890 2,580,000 2323 1118.59 0.349 3.873 142,706 0.087 3.873 0.255 0.126 331,657.73 0.08 0.241 66,034.16 346,827.1 0.076 0.038 0.0012 0.056 0.076 0.203 0.024 3.334 1.197 1.448 0.474 0.07 0.105

k12 0.016 0.021 8159 4060 4,090,000 1537 797.54 2.678 2.263 135,702.8 0.048 2.263 0.118 0.016 9447.21 0.04 0.198 35,559.32 381,312.4 0.003 0.012 0.138 0.007 0.017 0.089 0.003 0.523 0.145 0.033 0.237 0.169 0.0015

k21 0.163 0.188 0.194 0.191 0.184 0.195 0.160 2.99 2.994 0.993 0.994 2.896 0.994 0.42 0.996 1.005 0.993 0.194 0.294 1.007 1.022 1.04 1.02 1.014 1.006 0.089 0.991 0.981 0.794 2.824 1.671 0.983

kM,out (h−1)

18.3787 15.9347 15.4419 15.6845 16.2812 15.3627 16.6365 1.0019 1.0006 3.0169 3.0138 1.0344 3.0138 7.1327 3.0078 2.9808 3.0169 15.4419 10.1896 2.9749 2.9312 2.8805 2.9370 2.9544 2.9779 33.6599 3.0229 3.0538 3.7730 1.0608 1.7928 3.0475

t95(h)

0.1233 0.3542 0.1898 0.0278 0.0162 0.0966 0.9066 0.0041 0.0019 0.3184 0.0007 0.0053 0.6202 0.0002 2.5575 0.6677 0.2272 0.0003 0.0473 13.3289 0.5442 0.1594 0.5479 2.2922 1.0487 8.1766 0.0923 0.1471 0.7966 1.2513 0.6482 0.0786

BCF (L·g−1)

2.0908 2.5492 2.2782 1.4439 1.2093 1.984 4.1731 0.6106 0.2744 2.5030 −0.1634 0.7246 2.7925 −0.7085 3.4078 2.8245 2.3564 −0.4674 1.6745 4.1247 2.7357 2.2024 2.7387 3.3602 3.0206 3.9125 1.9652 2.1675 2.9012 3.0973 2.8116 1.8953

logBCF (L·g−1)

41 35 70 90 78 92 64 97 95 55 50 75 73 82 96 87 85 53 67 94 92 94 90 86 84 76 93 91 98 83 66 70

R2 (%)

Table 1 Toxicokinetic parameter estimates obtained with an iterative numerical integration technique: uptake, elimination rate constant, Rate Constant-Based Wet Weight BCFs According to Eq 5 and 7 for one and two compartments, respectively, and 95% depuration times calculated using the formula: t95 = − ln(0.05)/kout, and R2 is determined coefficient (corrected for degrees of freedom) between predicted and observed chemical concentrations in body.

Y. Gao, et al.

Ecotoxicology and Environmental Safety 186 (2019) 109810

4

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

Fig. 2. Comparison between the published and predicted BCFs from the one compartment (A) and two compartment (B) TK model parameters kin and kout. Plain squares in red and green represent organic compounds and nano-metal oxides, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The model fitting presented here clearly show that the applied one compartment TK model is applicable to a broad range of chemicals (Figs. S1–S50). Judging from the explained variance (R2), for Benzocaine, chlorpyrifos, Metribuzin, and naphthalene, the one-compartment TK model adequately described the internal concentration-time data, and predicted a constant increase in internal concentration and then reached a steady-state (Table 1, and Fig. S19, S20-S23, S29, S30). For atrazine, benz(a)anthracene, Clofibric acid, Fuoranthene, PBDE 154, PCB 104, PCB 136, BPA, diazinon, endosulfan alpha, and endosulfan beta, twocompartment TK model with inclusion of metabolism addressed the curvature and gave a better fit to the internal concentration-time data (Table 1, and Fig. S14–17, S18, S24, S27, S31-32, S33-34, S35-36, S38, S39, S45-46, S41, S48-49). For dicofol, EE2, and tebuconazole, both models gave a better fit to the internal concentration-time data (Table 1, and Fig. S25 and 26, S40, S50). For nanoparticles (Fe2O3, CuO, Fe3O4, ZnO, A-TiO2 and R–TiO2 and Ag), one compartment model explained better the internal concentration-time data than two compartment model (Table 1, and Figs. S1–S13). We have no explanation for the monophasic metabolism of Benzocaine, chlorpyrifos, Metribuzin, and naphthalene. Probably incomplete elimination from the metabolized compartment could have covered up the existence of the detoxified compartment during the limited exposure duration. Large differences between uptake and elimination data for nano metal oxides and TK model fits data were obvious in Figs. S1–S50, indicating differences in the kinetics among chemicals in zebrafish, i.e., kin and kout of the chemical, which can be compared quantitatively as shown in Table 1. The theoretical equilibrium BCFs derived from kin and kout were also indicated in Table 1. For the different chemicals fits were obtained and explained variance (R2) values ranged from 20% for benzo[a]anthracene to 98% for EE2 (Table 1). Individual chemical was first estimated for their kout, followed by estimate of kin. The kin and kout values were theoretically independent of exposure concentrations (Rubach et al., 2010) and there were large differences in both kin (by a factor of 25) and kout (by a factor of 10) among organic compounds and nano-metal oxides (Figs. S51 and S52). As a whole, either in one or in two compartment model, kin and kout for organic compounds were significantly higher than kin and kout for nanometal oxides (Figs. S51 and S52). The derived BCFs (kin/kout) considering both uptake and elimination processes showed how much chemicals accumulates at steady state in zebrafish, with obviously order of organic compounds > nano-metal oxides (Figs. S51 and S52). The range of t95 for tested chemicals considering elimination processes were

the highest in organic compounds, and then in nano-metal oxides (Figs. S51 and S52). Our data kin and kout of examined chemicals have been only collected for fish in Table 1. For atrazine, chlorpyrifos, PBDE 154, PCB 104, PCB 136, and diazinon, kin was directly proportional to chemical concentration in water (Cw). Previous studies showed that the kin was directly proportional to chemical concentration in water (Cw), as biomonitors of ambient chemical bioavailability (Lee et al., 1998; Wang et al., 1996; Wang and Dei, 1999). In most previous studies, kin was calculated at different Cw values (Reeves et al., 2007). When modeling chemical bioaccumulation, it can be used at one chemical ambient concentration (Eqn. 1). In the latter approach, kin was relatively independent of Cw and, thus, was applicable to a wide range of ambient chemical concentrations (Wang et al., 1996). Moreover, the kin and kout were theoretically independent of the different initial exposure concentrations during the experiments (Rubach et al., 2010). Despite the heterogeneity, patterns in terrestrial animals such as earthworms, birds, and mammals were similar (Hendriks et al., 2001). Chemicals with high Kow had high kin. The kin of a chemical in fish could positively correlated with the Kow of the chemical, and then the chemical was partitioned through the gills into the blood of the fish and eventually affected the toxicity of chemicals to fish (Yang and Sun, 1977). The variation in kin among chemicals was likely related with various factors including uptake pathways, and specific receptor (Ivey et al., 2016; Molinaro, 2016). Moreover, Kow might lead to large differences in kin among chemicals, because the lipophilicity of chemicals highly determines the bioconcentration in organisms (Verhaar et al., 1999). Moreover, the kin varied within a large range, supporting conclusions that other physiochemical factors of chemicals, such as particle size and hydrophobicity, also determined uptake of chemicals (Nel et al., 2006; Zhang et al., 2015). The observed differences in the kout of chemicals were possibly attributed to differences in metabolism pathways and specific enzyme activity and kinetics, leading to different detoxification processes (Eaton et al., 2008), but other physiochemical factors such as Kow and molecular mass were also of importance. In addition, high kout values were very likely related to high physiological activity of organic compounds leading to a high metabolic turnover. As shown in Fig. 2, BCFs derived from both models agreed well with ranges published in literatures (Ali et al., 2018; El-Amrani et al., 2012; Hakk et al., 2009; Wu et al., 2018; Zarco-Fernandez et al., 2016; Zhao et al., 2018). The investigated chemicals rank in their BCFs from 5

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

limited Kow values (Chessells et al., 1992), bioavailability and biotransformation, short exposure duration not to reach effective equilibrium (Connell and Hawker, 1988), and combinations of factors above (Zhu et al., 2015). We thus suggest to consider these two models for the prediction of lipid:water partition coefficients for chemicals with a log Kow value above 4 as Kow has been shown to be a good model parameter for lipid-water partitioning in numerous studies (Hansch et al., 1989; Seydel and Schaper, 1981). The high Kow of organic compounds resulted in the high BCF, indicating high accumulation of chemicals in zebrafish (Figs. 3 and 4). Similarly, the higher molecular mass of an organic compound was, the higher the BCF of an organic compound was. Sorption coefficients or environmental risk assessments are usually based on the Kow of the individual chemicals. This is a reasonable assumption if the affinity of a chemical for 1-octanol can reflect its affinity to dissolved organic carbon (DOC) or lipids in organisms. Kow is readily available for investigated chemicals. BCFs of organic compounds were largely dependent of Kow. There was linear correlation between the Kow and BCFs of organic chemicals in some fish (Dover, 1995). Bioaccumulation in the fish (G. ufinis) was roughly correlated with Kow of 49 pesticides (r = −0.76). there was a good fitting (r = -0.93) between Kow and BCF of seven organic toxicants such as PCB in rainbow trout (Ou et al., 2018). In addition, BCFs of eight chemicals in rainbow trout muscle complied a straight-line relationship with Kow (r = 0.948) due to testing similar structure organic toxicants such as 1,1,2,2-tetrachloroethylene and carbon tetrachloride (Nichols et al., 2015; Tanoue et al., 2015). The lower correlation coefficient between the BCFs of organic compounds in zebrafish and Kow (r = 0.506) in this study is obtained due to using organic compounds with a wide range of structures including insecticides, fungicides, herbicides, and polychlorinated and brominated flame retardants. Altogether, this indicates that the Kow and molecular weight influence TK process and subsequent TD process. Consistently, TK differences led to the chemical toxicity variability in liver enlargement of male rats (Gomis et al., 2018).

unexpected extremely high values for organic compounds to the lowest values for nano-metal oxides. Except for nano-metal oxides, the surprisingly large span in the bioconcentration of organic compounds in zebrafish might be reduced when normalized to the available lipid information (Pery et al., 2014). For Metribuzin, Benzocaine, Clofibric acid, and naphthalene the kin and kout-derived BCFs were more than 1 order of magnitude higher than the measured BCFs. The low rates of uptake and elimination predicted for these compounds in two compartments (with high estimated k12 and k21 values in Table 1) indicated that at infinite equilibrium time, the compounds presented in the metabolized and detoxified compartments might have a great influence on internal concentrations and measured BCF values. For other compounds, measured BCFs and kin and kout –derived BCFs were in good agreement. BCF of BPA derived from one compartment model agreed well with value published in literatures, indicating BPA seem to follow monophasic metabolism. BCF of nano-CuO derived from two compartment model slightly agreed well with value published in literatures, indicating nano-CuO seem to follow biphasic metabolism. However, according to above results, the two compartment model simulations agreed better with BPA concentrations in fish than one compartment. Also, the one compartment model simulations agreed better with nanoCuO concentrations in fish than two-compartment. This discrepancy can be attributed to that BCFs were calculated in fish after short exposure duration not to reach effective equilibrium and resulted in underestimation of actual BCF of BPA (Arnot and Gobas, 2006). For some chemicals, measured BCFs which reflect a steady state or equilibrium condition, generally achieved over long-term exposures may overestimate tissue levels in fish that may be exposed only for a short period of time. 45% of measured BCF values in literatures are subject to at least one major source of uncertainty and that measure errors generally result in an underestimation of actual BCF values (Arnot and Gobas, 2006). Kinetic BCFs determined as the ratio kin/kout from the one-compartment TK model better reflect a steady state or equilibrium condition during short period exposure. Possibly, kinetic BCFs in fish should be considered a benchmark for equilibrium portioning theory, which substitutes for measured BCFs from long-term exposures (Mackay, 2014). As biotransformation seems to be of great importance, it may be expected that residues of chemicals in fish probably are linked to the bioavailable fraction of chemicals in whole body. This conclusion would require more validation studies, with respect to uptake pathways, metabolism mechanisms, and dose dependency of TK process.

3.2.2. Nano metal oxides For nano metal oxides, one compartment derived kin and logBCF both had a negative relation with particle size in linear regressions (R2 > 0.5and p < 0.05, Fig. 5 and Table S2), but kin was positively correlated with Zeta potential in linear regressions (R2 > 0.5and p < 0.05, Fig. 5 and Table S2). However, kout positively correlated with particle size and specific surface area, respectively, in linear regressions (R2 > 0.5and p < 0.05, Fig. 6), and kout had a negative relation with Zeta potential in linear regressions (R2 = 0.47 and p < 0.05, Fig. 6). This indicated that particle size of nano metal oxides was a major factor that affected influx (kin) and efflux (kout) of nanoparticles. There was no significant correlation between t95 and particle size, specific surface area, and Zeta potential (p = 0.30, Fig. 6). Consistently, two compartment derived kin and logBCF both had a negative relation with particle size in linear regressions (R2 > 0.5and p < 0.05, Fig. S55), but kin and logBCF both had no significant correlation with specific surface area and Zeta potential, respectively, Fig. S55). In contrast, kout negatively correlated with specific area, but positively correlated with Zeta potential in linear regressions (R2 > 0.5and p < 0.05, Fig. S56). There was a significantly positive correlation between t95 and specific surface area (R2 > 0.5and p < 0.05, Fig. S56). In addition, k12/k21 was positively correlated with specific surface area, rather than particle size and Zeta potential (Fig. S56). Those complicated phenomenon indicated that particle size of nano metal oxides was a major factor that affected influx (kin) and efflux (kout) of nanoparticles. Structural properties of nanoparticles such as particle size and surface area can directly affect biological effects (Navarro et al., 2008). In this study, the main reason for the mild difference in the accumulation of two crystalline nano-TiO2 in zebrafish may be that the structural

3.2. Factors influencing toxicokinetics 3.2.1. Organic compounds As was shown in Fig. 3 and Table S7, one compartment derived kin had a positive relationship with log BCF, obviously attributed to BCF = kin/kout. kin and Log BCF showed a significant correlation with log Kow and molecular mass of organic compounds in linear regressions, respectively. Also, two compartment derived kin and BCF both positively correlated with log Kow and molecular mass (Fig. 4). For one and two compartment model, there was no significant correlation between kout and t95 and log Kow and molecular mass (p = 0.30, Figs. S53 and S54). The metabolism probability k12/k21 did not correlate with increasing Kow and molecular mass according to a common slope. BCF and first-order kin for persistent organic pollutants in aquatic organisms mostly present positive relationships with Kow (Tong et al., 2017). For the log Kow values below 4 and molecular mass below 300, BCFs of chemicals overlapped confidence intervals in the linear regression. In contrast, for chemicals with log Kow values exceeding 4 complied with the linear regression. For certain chemicals, BCFs overlapping confidence intervals in the linear regression, which was due to various factors including reduced membrane permeability related to molecular size (Dimitrov et al., 2003), decreased lipid solubility and 6

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

Fig. 3. Linear regressions of one-compartment TK model derived kin and logBCF individually against log Kow (left panel) and molecular mass (right panel) of organic compounds (n = 17) with equations, determination coefficient R2, and significance level p values given (α = 0.05), respectively.

accumulation was probably particle size-dependent, indicating that there was a filter-feeder in fish (Adams et al., 2006). Similarly, it was previously found that nanoparticle with size from 0.9 to 18,000 μm3 accumulated by D. magna (Koch and Peters, 1987). Other studies

characteristics of the two are different, because A-TiO2 had smaller particle size and larger specific surface area than R–TiO2 and thus led to its lower accumulation in zebrafish. Compared with the control, fish could accumulate all the tested NMs (Figs. S1–S13). Nevertheless, this

Fig. 4. Linear regressions of two-compartment TK model derived kin and log BCF individually against log Kow (left panel) and molecular mass (right panel) of organic compounds (n = 17) with equations, determination coefficient R2, and significance level p values given (α = 0.05), respectively. 7

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

Fig. 5. Linear regressions of one compartment TK model derived kin and BCF individually and particle size (nm), specific surface area (m2·g−1), and Zeta potential (mV) of nano metal oxides (n = 8) with determination coefficient R2, and significance level p values given (α = 0.05).

suggested that particles less than 50 μm in diameter were more easily accumulated by D. magna (Hund-Rinke and Simon, 2006). However, D. magna either prevented large particles from entering the filter chamber or excluded them through the post-abdominal claw (Koch and Peters, 1987). In addition, larger particles easily aggregated into flocculent

masses which tended to decrease accumulation by D. magna, and thus mitigated the toxicity. The reason why kout positively correlated with particle size (Fig. S56) might be that larger particles could be retained in the gut and easily excreted from organisms, as better described by one

Fig. 6. Linear regressions of one compartment TK model derived kout and t95 individually and particle size (nm), specific surface area (m2·g−1), and Zeta potential (mV) of nano metal oxides (n = 8) with determination coefficient R2, and significance level p values given (α = 0.05). 8

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

have implications for the risk assessment of time-varying exposures. These chemicals with slow elimination obviously require long time for complete elimination and potential recovery, or further exposure will have enhanced effects (Ashauer et al., 2007a, 2007b). Future research should reduce those uncertainties and identified biological factors in organisms that determine uptake and elimination, in relation to sensitivity of the investigated species, not presented in this study. Given chemicals lack of a specific TK, our results provide an excellent case for comparing and predicting the TK of different compounds in a certain species by TK model. The uptake of EDCs and PAHs is estimated by Kow and molecular mass. Moreover, the log Kow has also been proposed to test how long time a chemical reaches the incipient LC50 in acute toxicity tests (Baas et al., 2010). Given that the Kow correlates with the kin and BCF, by using linear relationships between TK parameters and log Kow, it is possible to make predictions on the TK process of other chemicals according to the log Kow of the chemical (Baas et al., 2010). The uptake of nanoparticles is estimated by particle size. We believe that our results are useful cases of such theoretical estimates and prediction and also make a starting point for TK extrapolation for untested chemicals.

compartment TK model (Zhang et al., 2015). We tried to explain the positive correlation between metabolism probability k12/k21 and specific surface area (Fig. S56). Maybe smaller nanoparticles have fewer atoms, which translate to fewer electron energy levels. Thus, electron energy bands become important influence factor of NMs in biota. For nanoparticles, size and surface charging are fundamental characteristics because they initiate predominant control over reactivity and transportation (Liu et al., 2008). According to our results, size and surface charging both determined the types and distributions of surface reactive sites capable of nanoparticle uptake, distribution, transformations and even the toxicity of nanoparticles. On account of the high proportion of surface atoms, the small variations in specific surface areas would determine the fate and reactivity of NMs in biota. We suggested that the evolution of particle size, surface area, and Zeta potential, and thus the uptake, distribution, metabolism, and excretion of effective particle diameters, should be predicted with high accuracy in modeling (Mullaugh and Luther, 2011). 3.3. Uncertainty The method presented to evaluate TK in various chemicals has several limitations. Firstly, differences in experimental design of chemicals increase incomparability due to chemical differences. The sensitivity differences between larvae and adults of zebrafish were not accounted for, although it is well known that sensitivity differences exist between life stages and might increase uncertainty on the explanatory power of kin. Thus, BCF variability can be attributed to organism size, e.g. embryo, larvae and adult zebrafish for chemicals from different studies. Organism body size has previously been identified as an influencing factor in the bioaccumulation of organic chemicals (Hendriks et al., 2001). In accumulation and elimination experiments, only survivors in organisms were sampled and uptake and elimination kinetics were addressed rather than toxic effects of chemicals. Sublethal exposure concentrations used in this study affected uptake and elimination, because all chemicals were exposed to less than their respective EC50s. Metal and metal oxide NPs were measured as total concentration in the zebrafish studies. However, metal and metal oxide NPs can dissolve into ions while we did not consider actually NPs or dissolved ions, and thus substantially impacted TK modeling. Another critically important aspect is that BCF values for NPs may differ from those for regular chemicals in that absorption across the gut tract cannot be assumed (Petersen et al., 2018; Petersen and Henry, 2011). Instead, the NP could predominately be in the organisms' gut tracts, and thus whether one compartment (gut) or two compartment (gut and other organs) was being modeled was somewhat determining BCF values.

4. Conclusions In this study, we can conclude that EDCs, PAHs and metal oxide NMs can be accumulated and cleared in zebrafish, which can be fitted well by one and two compartment TK model. kin, BCF and kout observed in some nano metal oxides were significantly slower than organic chemicals For organic compounds, a significant correlation was found between the kin or BCF by the fish and the octanol-water partition coefficient (Kow) of a chemical. A significant correlation was also found between the kin or BCF by fish and molecular mass. A significant negative correlation between the kin or BCF and particle size of nanoparticles and a significant positive correlation between kout and particle size indicated that NMs particle size did matter in biological influx and efflux processes. It would therefore appear that the TK of untested chemicals by a fish may be extrapolated from known chemical properties, and even the TD process of untested chemicals to a fish may be differed from known TK process. Declaration of competing interest The authors declare no competing financial interest. Acknowledgments: This study was supported by the National Natural Science Foundation of China (41877498), Tianjin Natural Science Foundation (193714040616) and the National Water Pollution Control and Treatment Science and Technology Major Project (2017ZX07301002). We also thank the editor and two reviewers for providing critical suggestions and comments that helped to improve the manuscript greatly.

3.4. Implications for extrapolations Different factors determined the TK variation in organic compounds and nano-metal oxides exposure to zebrafish (Danio Rerio). Kow and molecular mass correlated the TK variation in organic compounds. Particle size determined the TK variation in nano-metal oxides. Altogether, this indicates that toxicity differences in zebrafish can dominantly be explained by differences in TK, and remaining differences can be explained by mechanisms of toxic action. The most potential application of TK variation for ERA depends on parameters BCFs, t95, and other risk indicators. At present, ERA guidelines consider BCFs as initiated values for environmental risk assessment of chemicals with bioconcentration. For example, (e.g., BCF > 1 L/g for chemicals in a fish was assumed to be degradable. t95 is another important value for ERA, which is derived from the TK process, or the time that the fish eliminate 95% of the accumulated chemical into clean water (i.e., the time fish need to reach the steady state). This value depends on the kout. t95 defines the minimum interval time required for the organisms to recover in repeated exposure. Therefore, the high elimination abilities

Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ecoenv.2019.109810. This document includes accumulation data of organic compounds and nano-metal oxides in larvae and adult zebrafish used in this paper. Ethical approval This article does not contain any studies with animals performed by any of the authors. 9

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

References

Koch, P.C., Peters, R.H., 1987. Suppression of adult copulatory behaviors following LiClinduced aversive contingencies in juvenile male rats. Dev. Psychobiol. 20, 603–611. Kuehnert, A., et al., 2013. The internal concentration of organic substances in fish embryosA toxicokinetic approach. Environ. Toxicol. Chem. 32, 1819–1827. Lamon, L., et al., 2019. Physiologically based mathematical models of nanomaterials for regulatory toxicology: a review. Comput. Toxicol. (Amsterdam Neth.) 9, 133–142. Lee, B.G., et al., 1998. Uptake and loss kinetics of Cd, Cr and Zn in the bivalves Potamocorbula amurensis and Macoma balthica: effects of size and salinity. Mar. Ecol. Prog. Ser. 175, 177–189. Lee, J.H., et al., 2002. Prediction of time-dependent PAH toxicity in Hyalella azteca using a damage assessment model. Environ. Sci. Technol. 36, 3131–3138. Liu, J., et al., 2008. The non-oxidative dissolution of galena nanocrystals: insights into mineral dissolution rates as a function of grain size, shape, and aggregation state. Geochem. Cosmochim. Acta 72, 5984–5996. Liu, Q., et al., 2019. Uptake kinetics, accumulation, and long-distance transport of organophosphate esters in plants: impacts of chemical and plant properties. Environ. Sci. Technol. 53, 4940–4947. Ma, Y., et al., 2018. Sex dependent effects of silver nanoparticles on the zebrafish gut microbiota. Environ. Sci. Nano. 5, 740–751. Mackay, D., 1982. Correlation of bioconcentration factors. Environ. Sci. Technol. 16, 274–278. Mackay, D., et al., 2014. Relationships between exposure and dose in aquatic toxicity tests for organic chemicals. Environ. Toxicol. Chem. 33, 2038–2046. Maltby, L., et al., 2005. Insecticide species sensitivity distributions: importance of test species selection and relevance to aquatic ecosystems. Environ. Toxicol. Chem. 24, 379–388. Molinaro, A., 2016. The witch weed is able to detect a wide range of chemicals from plants through its sensitive and specific receptors: the strigolactone story updated. Chembiochem 17, 129–131. Mullaugh, K.M., Luther III, G.W., 2011. Growth kinetics and long-term stability of CdS nanoparticles in aqueous solution under ambient conditions. J. Nanoparticle Res. 13, 393–404. Navarro, E., et al., 2008. Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi. Ecotoxicology 17, 372–386. Nel, A., et al., 2006. Toxic potential of materials at the nanolevel. Science 311, 622–627. Nichols, J.W., et al., 2015. Observed and modeled effects of pH on bioconcentration of diphenhydramine, a weakly basic pharmaceutical, in fathead minnows. Environ. Toxicol. Chem. 34, 1425–1435. Ou, W., et al., 2018. Development of chicken and fish muscle protein - water partition coefficients predictive models for ionogenic and neutral organic chemicals. Ecotoxicol. Environ. Saf. 157, 128–133. Pery, A.R.R., et al., 2011. Predicting in vivo gene expression in macrophages after exposure to benzo(a)pyrene based on in vitro assays and toxicokinetic/toxicodynamic models. Toxicol. Lett. 201, 8–14. Pery, A.R.R., et al., 2014. A physiologically based toxicokinetic model for the zebrafish Danio rerio. Environ. Sci. Technol. 48, 781–790. Petersen, B., et al., 2018. Ionic structure around polarizable metal nanoparticles in aqueous electrolytes. Soft Matter 14, 4053–4063. Petersen, E.J., Henry, T.B., 2011. Ecotoxicity of fullerenes and carbon nanotubes: a critical review of evidence for nano-size effects. Biotechnol. Nanotechnol. Risk Assess.: Minding Manag. Potential Threats Around US 1079, 103–119. Pittinger, C.A., Pecquet, A.M., 2018. Review of historical aquatic toxicity and bioconcentration data for the brominated flame retardant tetrabromobisphenol A (TBBPA): effects to fish, invertebrates, algae, and microbial communities. Environ. Sci. Pollut. Control Ser. 25, 14361–14372. Rapaport, R.A., Eisenreich, S.J., 1984. Chromatographic determination of octanol-water partition coefficients (Kow's) for 58 PCB polychlorinated biphenyl congeners. Environ. Sci. Technol. 18, 163–170. Reeves, P.G., et al., 2007. Determination of selenium bioavailability from wheat mill fractions in rats by using the slope-ratio assay and a modified Torula yeast-based diet. J. Agric. Food Chem. 55, 516–522. Rubach, M.N., et al., 2010. Toxicokinetic variation IN 15 freshwater arthropod species exposed to the insecticide chlorpyrifos. Environ. Toxicol. Chem. 29, 2225–2234. Sanz-Landaluze, J., et al., 2015. Zebrafish (Danio rerio) eleutheroembryo-based procedure for assessing bioaccumulation. Environ. Sci. Technol. 49, 1860–1869. Scholz, S., 2012. Special Issue: Zebrafish Teratogenesis. Reprod. Toxicol. 33, 127. Segner, H., 2009. Zebrafish (Danio rerio) as a model organism for investigating endocrine disruption. Comp. Biochem. Physiol. C Toxicol. Pharmacol. 149, 187–195. Seydel, J.K., Schaper, K.J., 1981. Quantitative structure-pharmacokinetic relationships and drug design. Pharmacol. Ther. 15, 131–182. Sousa, C.A., et al., 2019. Metal(loid) oxide (Al2O3, Mn3O4, SiO2 and SnO2) nanoparticles cause cytotoxicity in yeast via intracellular generation of reactive oxygen species. Appl. Microbiol. Biotechnol. Stadnicka-Michalak, J., et al., 2014. Measured and modeled toxicokinetics in cultured fish cells and application to in vitro - in vivo toxicity extrapolation. PLoS One 9. Swietaszczyk, C., Jodal, L., 2019. Derivation and presentation of formulas for drug concentrations in two-, three- and four-compartment pharmacokinetic models. J. Pharmacol. Toxicol. 100, 106621. Tan, Q.-G., Wang, W.-X., 2011. Acute toxicity of cadmium in Daphnia magna under different calcium and pH conditions: importance of influx rate. Environ. Sci. Technol. 45, 1970–1976. Tan, Q.-G., Wang, W.-X., 2012. Two-compartment toxicokinetic-toxicodynamic model to predict metal toxicity in Daphnia magna. Environ. Sci. Technol. 46, 9709–9715. Tanoue, R., et al., 2015. Uptake and tissue distribution of pharmaceuticals and personal care Products in wild fish from treated-wastewater-impacted streams. Environ. Sci. Technol. 49, 11649–11658.

Adams, L.K., et al., 2006. Comparative toxicity of nano-scale TiO2, SiO2 and ZnO water suspensions. Water Sci. Technol. 54, 327–334. Ali, U., et al., 2018. Accounting for water levels and black carbon-inclusive sedimentwater partitioning of organochlorines in Lesser Himalaya, Pakistan using two-carbon model. Environ. Sci. Pollut. Control Ser. 25, 24653–24667. Arnot, J.A., Gobas, F.A.P.C., 2006. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14, 257–297. Ashauer, R., et al., 2006. Predicting effects on aquatic organisms from fluctuating or pulsed exposure to pesticides. Environ. Toxicol. Chem. 25, 1899–1912. Ashauer, R., et al., 2007a. Modeling combined effects of pulsed exposure to carbaryl and chlorpyrifos on Gammarus pulex. Environ. Sci. Technol. 41, 5535–5541. Ashauer, R., et al., 2007b. New ecotoxicological model to simulate survival of aquatic invertebrates after exposure to fluctuating and sequential pulses of pesticides. Environ. Sci. Technol. 41, 1480–1486. Baas, J., et al., 2010. Understanding toxicity as processes in time. Sci. Total Environ. 408, 3735–3739. Baas, J., et al., 2015. A simple mechanistic model to interpret the effects of narcotics. SAR QSAR Environ. Res. 26, 165–180. Babic, S., et al., 2018. Assessment of river sediment toxicity: combining empirical zebrafish embryotoxicity testing with in silico toxicity characterization. Sci. Total Environ. 643, 435–450. Baird, D.J., Van den Brink, P.J., 2007. Using biological traits to predict species sensitivity to toxic substances. Ecotoxicol. Environ. Saf. 67, 296–301. Brock, T.C.M., et al., 2010. Introduction to the Guidance on Linking Aquatic Exposure and Effects in the Risk Assessment for Plant Protection Products. Linking Aquatic Exposure and Effects: Risk Assessment of Pesticides. pp. 15–24. Brox, S., et al., 2014. A quantitative HPLC-MS/MS method for studying internal concentrations and toxicokinetics of 34 polar analytes in zebrafish (Danio rerio) embryos. Anal. Bioanal. Chem. 406, 4831–4840. Chen, J., et al., 2017. Bioconcentration and metabolism of emodin in zebrafish eleutheroembryos. Front. Pharmacol. 8. Chessells, M., et al., 1992. Influence of solubility in lipid on bioconcentration of hydrophobic compounds. Ecotoxicol. Environ. Saf. 23, 260–273. Choi, J.-S., et al., 2018. Quasi-QSAR for predicting the cell viability of human lung and skin cells exposed to different metal oxide nanomaterials. Chemosphere 217, 243–249. Connell, D.W., Hawker, D.W., 1988. Use of polynomial expressions to describe the bioconcentration of hydrophobic chemicals by fish. Ecotoxicol. Environ. Saf. 16, 242–257. Cosnefroy, A., et al., 2012. Selective activation of Zebrafish Estrogen Receptor subtypes by chemicals by using stable reporter Gene Assay developed in a Zebrafish liver cell line. Toxicol. Sci. 125, 439–449. Dimitrov, S.D., et al., 2003. Bioconcentration potential predictions based on molecular attributes - an early warning approach for chemicals found in humans, birds, fish and wildlife. QSAR Comb. Sci. 22, 58–68. Dover, G., 1995. Molecular coevolution and the esterase enigma. Trends Ecol. Evol. 10, 286–287. Du, W., et al., 2019. Comparison study of zinc nanoparticles and zinc sulphate on wheat growth: from toxicity and zinc biofortification. Chemosphere 227, 109–116. Eaton, D.L., et al., 2008. Review of the toxicology of chlorpyrifos with an emphasis on human exposure and neurodevelopment. Crit. Rev. Toxicol. 38, 1–125. El-Amrani, S., et al., 2012. Bioconcentration of pesticides in Zebrafish eleutheroembryos (Danio rerio). Sci. Total Environ. 425, 184–190. Gao, Y., et al., 2016. Application of biotic ligand and toxicokinetic-toxicodynamic modeling to predict the accumulation and toxicity of metal mixtures to zebrafish larvae. Environ. Pollut. 213, 16–29. Gao, Y., et al., 2015. Prediction of acute toxicity of cadmium and lead to zebrafish larvae by using a refined toxicokinetic-toxicodynamic model. Aquat. Toxicol. 169, 37–45. Gomis, M.I., et al., 2018. Comparing the toxic potency in vivo of long-chain perfluoroalkyl acids and fluorinated alternatives. Environ. Int. 113, 1–9. Grech, A., et al., 2017. Toxicokinetic models and related tools in environmental risk assessment of chemicals. Sci. Total Environ. 578, 1–15. Griffitt, R.J., et al., 2009. Comparison of molecular and histological changes in zebrafish gills exposed to metallic nanoparticles. Toxicol. Sci. 107, 404–415. Hakk, H., et al., 2009. Absorption, distribution, metabolism and excretion (ADME) study with 2,2',4,4',5,6'-hexabromodiphenyl ether (BDE-154) in male Sprague-Dawley rats. Xenobiotica 39, 46–56. Hansch, C., et al., 1989. Toward a quantitative comparative toxicology of organic compounds. Crit. Rev. Toxicol. 19, 185–226. Hendriks, A.J., et al., 2001. The power of size. 1. Rate constants and equilibrium ratios for accumulation of organic substances related to octanol-water partition ratio and species weight. Environ. Toxicol. Chem. 20, 1399–1420. Hou, W.-C., et al., 2013. Biological accumulation of engineered nanomaterials: a review of current knowledge. Environ. Sci. Process. Impacts 15, 103–122. Hund-Rinke, K., Simon, M., 2006. Ecotoxic effect of photocatalytic active nanoparticles TiO2 on algae and daphnids. Environ. Sci. Pollut. Control Ser. 13, 225–232. Ivey, C.E., et al., 2016. Application of a Hybrid Chemical Transport-Receptor Model to Develop Region-specific Source Profiles for PM2.5 Sources and to Assess Source Impact Changes in the United States. Klok, C., et al., 2014. Estimating the impact of petroleum substances on survival in early life stages of cod (Gadus morhua) using the Dynamic Energy Budget theory. Mar. Environ. Res. 101, 60–68.

10

Ecotoxicology and Environmental Safety 186 (2019) 109810

Y. Gao, et al.

Wang, S., et al., 2019. Toxicity of some prevalent organic chemicals to tadpoles and comparison with toxicity to fish based on mode of toxic action. Ecotoxicol. Environ. Saf. 167, 138–145. Wang, W.-X., et al., 1996. Kinetic determinations of trace element bioaccumulation in the mussel Mytilus edulis. Mar. Ecol. Prog. Ser. 140, 91–113. Wang, W.X., Dei, R.C.H., 1999. Factors affecting trace element uptake in the black mussel Septifer virgatus. Mar. Ecol. Prog. Ser. 186, 161–172. Wu, Y., et al., 2018. Uptake and elimination of emerging polyfluoroalkyl substance F-53B in zebrafish larvae: response of oxidative stress biomarkers. Chemosphere 215, 182–188. Yang, C.F., Sun, Y.P., 1977. Partition distribution of insecticides as a critical factor affecting their rates of absorption from water and relative toxicities to fish. Arch. Environ. Contam. Toxicol. 6, 325–335. Yang, Y.-F., et al., 2017. Toxicity-based toxicokinetic/toxicodynamic assessment of bioaccumulation and nanotoxicity of zerovalent iron nanoparticles in Caenorhabditis elegans. Int. J. Nanomed. 12, 4607–4621. Zarco-Fernandez, S., et al., 2016. Bioconcentration of ionic cadmium and cadmium selenide quantum dots in zebrafish larvae. Chemosphere 148, 328–335. Zhang, Y., et al., 2015. Accumulation and elimination of iron oxide nanomaterials in zebrafish (Danio rerio) upon chronic aqueous exposure. J. Environ. Sci. 30, 223–230. Zhao, H., et al., 2018. Effect of copper on the accumulation and elimination kinetics of fluoroquinolones in the zebrafish (Dario rerio). Ecotoxicol. Environ. Saf. 156, 135–140. Zhou, W., et al., 2005. Estimation of n-octanol/water partition coefficients (K-ow) of all PCB congeners by density functional theory. J. Mol. Struct. Theochem. 755, 137–145. Zhu, C., et al., 2015. Bioconcentration and trophic transfer of polychlorinated biphenyls and polychlorinated dibenzo-p-dioxins and dibenzofurans in aquatic animals from an e-waste dismantling area in East China. Environ. Sci. Process. Impacts 17, 693–699. Zhu, X., et al., 2009. Acute toxicities of six manufactured nanomaterial suspensions to Daphnia magna. J. Nanoparticle Res. 11, 67–75.

Tiwari, P.K., et al., 2019. Liquid assisted pulsed laser ablation synthesized copper oxide nanoparticles (CuO-NPs) and their differential impact on rice seedlings. Ecotoxicol. Environ. Saf. 176, 321–329. Tokarz, J., et al., 2013. Zebrafish and steroids: What do we know and what do we need to know? J. Steroid Biochem. 137, 165–173. Tong, L., et al., 2017. Modification of polychlorinated phenols and evaluation of their toxicity, biodegradation and bioconcentration using three-dimensional quantitative structure-activity relationship models. J. Mol. Graph. Model. 71, 1–12. Vaal, M.A., et al., 2000. Variation in sensitivity of aquatic species to toxicants: practical consequences for effect assessment of chemical substances. Environ. Manag. 25, 415–423. Van Hattum, B., Montanes, J.F.C., 1999. Toxicokinetics and bioconcentration of polycyclic aromatic hydrocarbons in freshwater isopods. Environ. Sci. Technol. 33, 2409–2417. Venquiaruti Escarrone, A.L., et al., 2016. Uptake, tissue distribution and depuration of triclosan in the guppy Poecilia vivipara acclimated to freshwater. Sci. Total Environ. 560, 218–224. Verhaar, H.J.M., et al., 1999. Modeling the bioconcentration of organic compounds by fish: a novel approach. Environ. Sci. Technol. 33, 4069–4072. Vosges, M., et al., 2010. 17 Alpha-Ethinylestradiol disrupts the ontogeny of the forebrain GnRH system and the expression of brain aromatase during early development of zebrafish. Aquat. Toxicol. 99, 479–491. Wallace, W.G., et al., 2003. Subcellular compartmentalization of Cd and Zn in two bivalves. I. Significance of metal-sensitive fractions (MSF) and biologically detoxified metal (BDM). Mar. Ecol. Prog. Ser. 249, 183–197. Wallace, W.G., Luoma, S.N., 2003. Subcellular compartmentalization of Cd and Zn in two bivalves. II. Significance of trophically available metal (TAM). Mar. Ecol. Prog. Ser. 257, 125–137. Wang, H., et al., 2013. Acute toxicity, respiratory reaction, and sensitivity of three cyprinid fish species caused by exposure to four heavy metals. PLoS One 8.

11