Bioaccumulation of tetrabromobisphenol A in a laboratory-based fish–water system based on selective magnetic molecularly imprinted solid-phase extraction

Science of the Total Environment 650 (2019) 1356–1362

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Bioaccumulation of tetrabromobisphenol A in a laboratory-based fish– water system based on selective magnetic molecularly imprinted solid-phase extraction Liqin Hu a,1, Tingting Zhou b,1, Dan Luo a, Jingwen Feng a, Yun Tao a, Yusun Zhou b, Surong Mei a,⁎ a State Key Laboratory of Environment Health (Incubation), Key Laboratory of Environment and Health, Ministry of Education, Key Laboratory of Environment and Health (Wuhan), Ministry of Environmental Protection, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, #13 Hangkong Road, Wuhan, Hubei 430030, China b Department of Clinical Laboratory, The Affiliated Hospital of Qingdao University, Qingdao, Shandong 266003, China

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• MSPE • A laboratory-based fish–water system • BCF of TBBPA and exposure time

a r t i c l e

i n f o

Article history: Received 13 June 2018 Received in revised form 1 September 2018 Accepted 1 September 2018 Available online 03 September 2018 Editor: P Holden Keywords: Tetrabromobisphenol A Bioconcentration factor Magnetic solid-phase extraction Fish–water system

a b s t r a c t Recently, magnetic solid-phase extraction (MSPE) using magnetic molecularly imprinted polymers (MMIPs), which is a simple process with excellent selectivity, has attracted much attention for the determination of environmental pollutants. In this study, MMIPs were used as an adsorbent to establish a selective MSPE method coupled with high-performance liquid chromatography using ultraviolet detection (HPLC-UV) for the determination of tetrabromobisphenol A (TBBPA) levels in water and fish samples. The samples were collected from a laboratory-based fish–water system after 0, 2, 5, 8, 11, 20, 30, and 50 days. We found that the concentrations of TBBPA in the sample group spiked with TBBPA decreased in the water samples over time and increased in the fish samples from 2 to 30 days, then finally decreased. The calculated bioconcentration factor (BCF) increased over time, reaching 33.98 L/kg after 50 days exposure to TBBPA. Linear and exponential kinetic models were applied to fit the correlation between BCF and exposure time, and the constant of the time-dependent BCF (Ku) ranged from 0.0364 to 1.5250 L/kg per day with a corresponding R2 of 0.6786 to 0.9985. Simplified mathematical models to evaluate the transfer characteristics of TBBPA in a laboratory-based fish–water system have been developed. © 2018 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, #13 Hangkong Road, Wuhan 430030, Hubei, China. E-mail address: [email protected] (S. Mei). 1 These authors equally contributed to this work.

https://doi.org/10.1016/j.scitotenv.2018.09.002 0048-9697/© 2018 Elsevier B.V. All rights reserved.

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

1. Introduction Tetrabromobisphenol A (TBBPA) is one of the most common brominated flame retardant (BFR) compounds, covering over 60% of the total BFR market (Liu et al., 2017). Approximately 80% of TBBPA is employed in reactive application and the remaining 20% is accounted for by additive application. Both reactive- and additive-treated products result in the release of TBBPA into the environment (Pollock et al., 2017). For example, TBBPA has been detected in dust, wastewater, soil, and sediment, as well as in organisms (Abafe and Martincigh, 2016; Inthavong et al., 2017; Kim et al., 2016; Lee et al., 2015). Nonoccupational TBBPA exposure in humans is mainly via the consumption of fish products because of the persistence and bioaccumulation of TBBPA (Gu et al., 2017). Therefore, analysis of the distribution of TBBPA in fish products and investigation of its bioaccumulation are of particular importance for human health. The concentrations of TBBPA in environmental water and aquatic organisms have been investigated and found to be variable (Ashizuka et al., 2008; Gong et al., 2017; Lee et al., 2015; Suzuki and Hasegawa, 2006). TBBPA has been detected at levels of not detectable (nd) through 1.8 μg/L in seawater samples derived from coastal areas of Qingdao in Northern China (Yin et al., 2011). In addition, the level of TBBPA in 29 kinds of fish samples from three different regions of Japan was determined and ranged from 0.01 to 0.11 ng/g wet weight (Ashizuka et al., 2008). However, these investigations only evaluated the distribution of TBBPA in environmental water or aquatic organisms; a unified assessment metric for the bioaccumulation of TBBPA was not determined. Gu et al. (2017) utilized the bioconcentration factor (BCF) to assess the bioaccumulation in bivalves and mussels of TBBPA in aqueous media and found TBBPA concentrations in seawater and bivalves ranging from nd to 2.79 ng/L and from nd to 158 ng/g lipid weight, respectively. Moreover, the average values of BCF of TBBPA for oysters and mussels were calculated to be 3.19 × 104 (±1.06 × 105) and 3.67 × 104 (±1.06 × 105) L/kg, respectively. Generally, chemicals are defined as bioaccumulative if log BCF N 3.3 or the substituted log Kow (the logarithm of the octanol–water partition coefficient) is N4.5 (Grisoni et al., 2015). Thus, TBBPA was concluded to be bioaccumulative based on this marine field study (Gu et al., 2017). In contrast, another study has drawn a negative conclusion (Hardy, 2004). The BCF is a fundamental property of a substance in terms of its likelihood of concentrating in organisms and is used as a bioaccumulation hazard assessment metric in many regulatory contexts (Gissi et al., 2015; Lombardo et al., 2010). To avoid unnecessary animal testing and overcome the problem of limited experimental data (Mansouri et al., 2012), different models have been applied to predict the BCF of various chemicals, although no specific model simultaneously satisfies the requirements for regression and classification of chemicals (Gissi et al., 2015). Quantitative structure–activity relationship (QSAR) models, as recommended by the European Commission, are used in the Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) procedure, in which the majority of QSAR models are based on log Kow, which can be used to determine the BCF because of the good correlation between BCF and log Kow. However, it is challenging to use QSAR models for log Kow N 9, and, even when using complex QSAR models, only slight improvement to the predictive accuracy is achieved (Grisoni et al., 2015). Considering the complexity of the current model and the contradictory experimental BCF data for TBBPA, the overarching goal of our study was to evaluate the BCF of TBBPA based on our experimental data and develop a simplified mathematical model to evaluate the transfer characteristics of TBBPA in a laboratory-based fish–water system. Among the different kinds of analytical methods in TBBPA detection, chromatography and chromatography–mass spectroscopy methods are commonly used because of their separation efficiency and low detection limit. However, samples require special handling before chromatography or mass spectroscopy analysis. The most widely used sample

1357

preparation technology is solid-phase extraction (SPE). The adsorbents utilized in SPE for TBBPA include Oasis MAX (Chu and Letcher, 2013), LC-Si (Jian et al., 2016), HLB (Yang et al., 2014), and molecularly imprinted polymers (MIPs) (Shen et al., 2013; Yin et al., 2012). MIPs have drawn significant attention because of their excellent selectivity toward targets and their highly selective binding cavities (Li et al., 2015; Madikizela et al., 2018). However, only Shen et al. (2013) and Yin et al. (2012) focus on the utility of MIPs in SPE for TBBPA. In those studies, traditional SPE columns involving time-consuming extraction processes were used. Subsequently, magnetic particles were introduced into the preparation of MIPs, and the obtained magnetic molecularly imprinted polymers (MMIPs) can be separated rapidly with the help of a magnet (Karimi et al., 2016; Li et al., 2016; Zhou et al., 2015). However, no studies have been performed on the determination of TBBPA via magnetic solid-phase extraction (MSPE) with MMIPs as the adsorbent. In our work, a selective method involving MMIPs for SPE coupled with HPLC-UV for TBBPA determination was developed and applied to detect TBBPA in water and fish samples in a laboratory-based fish– water system. Finally, the BCF was utilized to assess the bioaccumulation of TBBPA in fish and simplified mathematical models were used to investigate the transfer characteristics of TBBPA from water to fish. 2. Experimental 2.1. Chemicals and reagents TBBPA (99%) was obtained from Sigma-Aldrich, St. Louis, MO, USA. Tetrabromobisphenol A-bis(dibromopropyl ether) (TBBME, 98%), bisphenol A (BPA, 99%), tetrachlorobisphenol A (TCBPA, 98%), 4,4sulphonyl-bis-(2,6-dibromophenol) (TBBPS, 98%), hexafluorobisphenol A (BPAF, 98%), and tetrabromobisphenol A diallyl ether (TBBDE, 99%) were purchased from Meryer Chemical Technology Co., Ltd., Shanghai, China. Tetraethylorthosilicate (TEOS), 3-aminopropyl triethoxysilane (APTES), and polyacrylic acid (PAA) were purchased from Sigma (St. Louis, MO, USA). HPLC-grade methanol and acetonitrile were obtained from Merck, Beijing, China. Ammonium hydroxide (NH3·H2O) was procured from Sinopharm Group Chemical Reagent Co. Ltd., Shanghai, China. Acetic acid (HAc), potassium hydroxide (KOH), and potassium dihydrogen phosphate were supplied by J&K Chemical Co, Shanghai, China. 2.2. Instruments X-ray diffraction (XRD) patterns of MMIPs were recorded using a PANalytical B.V. X'Pert PRO X-ray diffractometer (Holland). The magnetic characteristics of MMIPs were determined using a LakeShore 7400 vibrating sample magnetometer (USA). Chromatographic analysis was performed on an Agilent 1260 HPLC apparatus equipped with an ultraviolet-diode array detector (UV-DAD), and the detection wavelength of the spectra was 212 nm. A Waters Sunfire C18 column (150 × 4.6 mm, 5 mm) was used for the chromatographic separation, with a column temperature of 25 °C. The mobile phase of the separation was methanol/water (85:15, v/v), with an injection volume of 20 μL and a flow rate of 1.0 mL/min. The KS 103B shaking table was purchased from IKA, Germany, and the N2 concentrator was obtained from Anpu Co., China. 2.3. MMIPs synthesis The preparation of the MMIPs was carried out as in our previous study (Zhou et al., 2016). Briefly, 0.5 mM TBBPA was dissolved in ethanol (80 mL). Then, well-dispersed Fe3O4 nanoparticles (NPs; 0.2 g) in water (20 mL) were added to the mixture. Next, 1 mL of NH3∙H2O was added, and the solution was ultrasonicated for 6 min. Furthermore, while the mixture was mechanically stirred at 200 rpm in a water

1358

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

bath at 30 °C, 2 mM APTES was added and prepolymerized for 5 min with the template TBBPA. Finally, 2.5 mM TEOS was added, and the polymerization was performed for 16 h. The synthesized MMIPs were separated from the suspension by a magnet. After twice washing with a methanol/0.05 M KOH solution (2:1, v/ v), the polymers were washed with methanol four times. Then, the product was dried in a vacuum oven at 55 °C for 12 h. Magnetic nonmolecularly imprinted polymers (MNIPs) were synthesized without a template during polymerization via an identical procedure to that used for the MMIPs. 2.4. Construction of the fish–water system and sample collection A schematic diagram of the construction of the fish–water system is shown in Fig. S1 (in the Supplementary Material). Fish, fish tanks (40 × 30 × 30 cm), and aerators were purchased from a bird and flower market in Wuhan, China, and the type of fish selected for the experiment was koi carp (Cyprinus carpio). Before use, the fish tanks and aerators were disinfected with KMnO4 (0.02%, w/v). Then, 10 L of water was placed into one fish tank, which was placed in the full sun for several days before the test. Then, 14 fish were placed into each of two tanks: one tank was spiked with 100 μg/L TBBPA and the other was not spiked. After 0, 2, 5, 8, 11, 20, 30, and 50 days, we collected a fish and 50 mL water from the constructed systems and set these samples aside at −20 °C until further analysis.

2.8. Data modelling The BCF was used to assess the capability of bioaccumulation of chemicals, which was calculated based on the ratio of the chemical concentration in organisms to the chemical concentration of the surrounding medium at steady state (Fernandez and Gardinali, 2016). BC F compound ¼

C organism ðng=kgÞ C environment ðng=LÞ

ð1Þ

However, this verified model for demonstrating the relationship between bioaccumulation and exposure time is limited. In previous investigation of the bioaccumulation kinetics and BCF of chlorinated pesticides, the first-order linear regression model was applied to analyze the bioaccumulation of TBBPA (Satyanarayan and Ramakant, 2004). Meanwhile, the exponential model, which had been used for predicting the BCF of soil pesticide residues (Hwang et al., 2018) and assessing the bioaccumulation of a commonly used antifouling biocide for Thalassia (Fernandez and Gardinali, 2016), was selected to predict the transfer characteristics of TBBPA. Because of the uptake process of TBBPA from water to fish, the linear and exponential mathematical models in our study were used as follows (Fernandez and Gardinali, 2016; Satyanarayan and Ramakant, 2004). Considering the elimination process of chemical in organisms (Garcia-Galan et al., 2017; Silva et al., 2015), the polynomial–linear and polynomial–exponential models were developed below:

2.5. Procedure for magnetic solid-phase extraction (MSPE)

Linear model : BCF ðtÞ ¼ BCF 0 þ K u t

ð2Þ

The MSPE process for fish samples included activation, loading, washing, and elution. First, 100 mg of MMIPs in a 50-mL bottle was activated successively with 5 mL methanol and methanol/water (1:1, v/v). For loading, the activated MMIPs were added to the prepared water samples and extracted solutions of the fish samples, and the mixtures were shaken for 30 min at 25 °C. After the loading solutions had been separated with the help of an external magnet, 5 mL water/acetonitrile (1:9, v/v) was used to remove the impurities nonspecifically adsorbed onto the MMIPs. When the washing solution was discarded with the help of the magnet, the MMIPs were eluted with 5 mL methanol/acetic acid (4:1, v/v) solution for 30 min. Finally, the eluent was collected and concentrated at 25 °C under N2 evaporation and reconstituted in 200 μL methanol/water (1:1, v/v) for HPLC-UV analysis.

Polynomial–linear model : BCF ðtÞ ¼ BCF 0 þ K u t þ K e t2

ð3Þ

Exponential model : BC F ðtÞ ¼ BCF 0  eK u t

ð4Þ

Polynomial–exponential model : BCF ðtÞ ¼ BCF 0  eK u tþK e t

3. Results and discussion

Water and fish samples were collected from the fish–water system when the feeding times at 0, 2, 5, 8, 11, 20, 30, and 50 days. For the fish samples, the fish were ground into homogenates, which were placed in 50 mL centrifuge tubes. Then, 5 mL acetonitrile was added and the mixtures were shaken for 5 min to extract the TBBPA. The supernatants were collected via centrifugation. After drying with nitrogen, 40 mL of methanol/water solution (1:1, v/v) was added to the centrifuge tubes, and then the mixtures were used as the loading samples. For water samples, 20 mL water sample was mixed with an equal volume of methanol for further loading.

3.1. Characterization and evaluation of MMIPs

Before constructing the fish–water system, water and fish samples were analyzed using the MSPE-HPLC-UV method, and no TBBPA was found. Thus, calibration curves were established using water and fish as matrices and six different concentrations of TBBPA were added to the water (5.0–1000 μg/L) and fish samples (50–2000 ng/g). The recoveries of spiked samples were used to validate the accuracy and precision of the method. In this study, the spiked concentrations were 5, 10, and 50 μg/L in water samples and 100, 500, and 1000 ng/g in fish samples.

ð5Þ

here, BCF(t) is defined as the ratio of the TBBPA concentration in fish at the time of sampling (ng/kg) to the TBBPA concentration in water at the time of sampling (ng/L). BCF0 and Ku are fitting parameters representing the fish uptake factor at the time of placement in the water and the linear or exponential change constant of time-dependent BCF (d−1), respectively. In addition, Ke represents the linear or exponential change constant of the square of the time-dependent BCF (d−2).

2.6. Sample preparation

2.7. Validation of the MSPE-HPLC-UV method

2

As shown in Fig. 1A, the three prepared magnetic materials had similar XRD patterns, and reflections corresponding to the cubic inverse spinel structure of Fe3O4 at [220], [311], [400], [422], [511], and [440] were observed at 29.9°, 35.3°, 42.9°, 53.3°, 56.9°, and 62.5°, respectively. The positions and relative intensities of the diffraction peaks were in good agreement with those from the JCPDS card (16-0629) for Fe3O4, which indicates that the crystallinity of Fe3O4 remained after modification with the MMIPs and MNIPs. In Fig. 1B, the saturated magnetization (Ms) of Fe3O4 was 60.343 emu/g, which was higher than that of the MMIPs (30.494 emu/g) and MNIPs (27.185 emu/g). The reason for the decrease in Ms in the MMIPs and MNIPs was the coated MIPs and nonmolecularly imprinted polymers (NIPs) on the Fe3O4 nanoparticles, respectively. The prepared MMIPs and MNIPs exhibited excellent superparamagnetic properties and could be separated from the suspensions in 2 min by the magnet. To investigate the adsorption characteristics of the prepared MMIPs further, kinetic adsorption and selectivity adsorption experiments were conducted in this work, and results of which are in the Supplementary Material. As shown in Fig. 2A, the adsorption time to achieve

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

1359

Fig. 1. XRD spectra (A) and hysteresis loops (B) of the Fe3O4 NPs, MMIPs, and MNIPs.

equilibrium for the MMIPs (200 min) was longer than that for the MNIPs (20 min), which may be attributed to the specific target recognition effect on the tailored stereo cavities and binding sites. Therefore, more time is needed for the template to orient itself to fit into the imprinted cavities of the MMIPs (Xie et al., 2015). Meanwhile, the selectivity of the MMIPs toward TBBPA and its analogs was investigated. Fig. 2B shows the adsorption characteristics of MMIPs and MNIPs toward TBBPA and its analogs. The imprinting factor (IF) is defined as the ratio of the adsorption capacity for TBBPA and its analogs of MMIPs to that of MNIPs. As the value of IF increased, the selective recognition ability of MMIPs also increased. Thus, as illustrated in Fig. 2B, the MMIPs showed excellent selective recognition ability for TBBPA. 3.2. Validation of the MSPE-HPLC-UV method Under the optimal MSPE conditions (shown in the Supplementary Material, Fig. S3), the extraction recoveries of TBBPA in the water and fish samples were investigated, with spiked concentrations of TBBPA in the water samples of 5, 10, and 50 μg/L and in the fish samples of 100, 500, and 1000 ng/g. The extraction recoveries were 73.26–83.25% with root squared deviation (RSD) values of 9.23–12.53% for the water

samples and 69.13–81.57% with RSD values of 7.64–9.62% for the fish samples (in the Supplementary Material, Table S2), which suggests that the developed MSPE procedure was suitable for the pretreatment of water and fish samples. We prepared a calibration curve based on the matrix match to the water and fish samples. The TBBPA in the water samples exhibited a satisfactory linearity, with a correlation coefficient (R2) of 0.9868 in the range of 5.0–1000.0 μg/L, and the limit of detection (LOD) and limit of quantification (LOQ) were 1.0 μg/L and 5.0 μg/L, respectively. Meanwhile, the TBBPA in the fish samples also obtained a good linearity, ranging from 50 to 2000 ng/g (R2 = 0.9766), and the LOD and LOQ were 15.2 and 50.0 ng/g, respectively. The above-mentioned LOD and LOQ were defined as 3:1 and 10:1 signal-to-noise (S/N) ratios, respectively. Thus, using the developed method, the level of TBBPA in real samples was determined. The recoveries of the samples spiked with different concentrations of TBBPA were used to evaluate the accuracy and precision of the developed method. The recoveries of the spiked water and fish samples were 89.68–100.05% and 98.81–112.16%, respectively, with corresponding RSDs of b5% (Table 1). Compared with the chromatogram of TBBPA extracted by MNIPs in the fish samples, the MMIPs provided higher chromatographic peaks and a more stable baseline (Fig. 3). These results suggest that the

Fig. 2. Kinetic adsorption curves of MMIPs and MNIPs for TBBPA (0.05 mg/mL) (A) and adsorption capacity and IF of MMIPs for TBBPA and its structural analogs (1.0 mg/mL) (B).

1360

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

MMIPs as MSPE adsorbents had a strong ability to absorb TBBPA and resist matrix effects from complex samples. 3.3. Bioaccumulation of TBBPA in a laboratory-based fish–water system The developed MSPE-HPLC-UV method was applied to determine the concentrations of TBBPA in water and fish samples collected from the laboratory-based fish–water system. TBBPA was not determined in water and fish samples of the blank group but was detected in the experimental group samples spiked with 100 μg/L TBBPA in water. As shown in Table 2, the concentrations of TBBPA gradually decreased over time in the water samples, whereas they increased from days 2 to 30 in the fish samples, and then decreased. Furthermore, the BCF, which was calculated based on the ratio of the chemical concentration in organisms to the chemical concentration of the surrounding medium at steady state, increased over time in our experiments. After 50 days of exposure, the BCF of TBBPA in fish was 33.98 L/kg; thus, TBBPA was defined as non-bioaccumulative (Grisoni et al., 2015; Hardy, 2004). This result is in accordance with those of other studies assessing the BCF of TBBPA (Springborn Life Sciences, 1989a, 1989b; Hardy, 2004). Although, log kow of TBBPA was predicted as 7.2 based on the EPI Suite KOWWIN model (U.S. EPA, 2000), the log kow value of 5.9 was finally selected for calculating log BCF because a reasonable and reliable experimental result is a priority over a theoretical value from software of the EPI Suite KOWWIN model (Pittinger and Pecquet, 2018). Considering a log kow of TBBPA equal to 5.9, the recommended QSAR model was applied to predict the BCF. An empirical linear equation of the QSAR model is as follows (Grisoni et al., 2015): logBCF ¼ a∙ logkow þ b

ð6Þ

where a and b represent empirical constants, for which Veith's linear equation (Veith et al., 1979) and Mackay's linear equation (Mackay, 1982) were used: the former with a = 0.85, b = −0.70, and the latter with a = 1, b = −1.32. These values of a and b were introduced in Eq. (6), as the calculated log BCF values of 4.315 and 4.58, respectively. In addition, a bilinear equation developed by Bintein and co-authors was applied to predict the value of logBCF, with the modified equation expressed as follows (Bintein et al., 1993): logBCF ¼ 0:910  logBCF−1:975  log6:8  10−7  logBCF þ 1−0:786

ð7Þ

the calculated logBCF was 4.5830 (Grisoni et al., 2015). Finally, we used the EPI Suite BCFBAF Meylan model (U.S. EPA, 2000) to predict the logBCF of TBBPA. According to the BCFBAF method, the logBCF was calculated as 3.5598, based on the equation below: logBCF ¼ 0:6598 logkow −0:333 þ

X

ð8Þ

correction value

the correction value is related to particular structural fragments, but it does not contribute to the logBCF of TBBPA. The logBCF values calculated

Table 1 Spiked recoveries of TBBPA in water and fish samples analyzed using the MSPE-HPLC-UV method. Samples

Spiked concentration (μg/L or ng/g)a

Spiked recoveries (%, n = 3)

RSD (%, n = 3)

Water

5 10 50 100 500 1000

89.68 100.05 96.72 112.16 102.92 98.81

0.53 3.41 4.52 3.62 4.43 2.57

Fish

a The unit of spiked concentration is micrograms per liter and nanograms per gram in water and fish samples, respectively.

Fig. 3. Typical chromatograms of a TBBPA standard solution (100 ng/mL) (a) and spiked fish sample (100 ng/g) extracted with MMIPs (b) and MNIPs (c).

above were all N3.3. Thus, TBBPA was theoretically defined as bioaccumulative. In conclusion, the theoretical value of logBCF is contradictory to our experimental result. Obviously, the theoretical value of logBCF is determined using the value of logkow setting in the software, so more attention should be paid to the logBCF provided by models. Meanwhile, good agreement existed between our contradictory result and that of a related study assessing bioaccumulation of TBBPA in fish (Hardy, 2004). Both studies compared BCFs measured in the laboratory and predicted by models, and the estimated values of BCF were predictive of bioaccumulation, but the BCFs determined in the laboratory fish studies were b500 L/kg. A possible reason for the differing results derives from the ease of metabolism and elimination of TBBPA in fish (Hardy, 2004), thus bioaccumulation was not represented as a characteristic feature of the TBBPA measured herein. To a certain extent, it showed that it is necessary to introduce the elimination constant (Ke) into simple mathematical models for analyzing the transfer characteristics of TBBPA in laboratory-based fish–water systems. 3.4. Simplified mathematical models The transfer characteristics of TBBPA from water to fish is still unknown. Because the bioaccumulation of environmental pollutants among different species and environmental media, including aquatic macrophytes, fish, shrimp, water, and soils, had been investigated in previous studies, linear and exponential models were selected in our study for analyzing the transfer characteristics of TBBPA. The fitting results of this analysis are shown in Table 3 and Fig. 4. Meanwhile, polynomial–linear and polynomial–exponential models were developed based on our experimental results, and Ke was introduced owning to the elimination process of chemical in organisms (Garcia-Galan et al., 2017; Silva et al., 2015). Therefore, four mathematical models were applied to fit our experimental data, in which Ku of TBBPA ranged from 0.0364 to 1.5250 L/kg per day with corresponding R2 values from 0.6786 to 0.9985. In particular, the polynomial–linear and Table 2 Concentrations of TBBPA in water (spiked with 100 μg/L) and fish samples from the water–fish system detected by the developed MSPE-HPLC-UV method. Sample (Days)

Water (μg/L)

Fish (ng/g)

BCF (L/kg)

0 2 5 8 11 20 30 50

98.75 88.89 77.31 61.16 49.55 44.96 37.61 32.24

– 62.32 143.01 226.37 263.16 756.38 1226.06 1095.67

– 0.70 1.85 3.70 5.31 16.82 32.59 33.98

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

1361

Table 3 Different time-dependent BCF prediction models and parameters. Models Liner model Polynomial linear model Exponential model Polynomial exponential model

Formula

BCF0 (L/kg)

Ku (L/kg/day)

Ke (L/kg/day2)

R2

BCF(t) = BCF0 + Kut BCF(t) = BCF0 + Kut + Ket2 BCF(t) = BCF0 × eKut BCF(t) = BCF0 × eKut+Ket2

−0.8961 −5.8329 6.1871 0.9411

0.8034 1.5250 0.0364 0.1887

– −0.0140 – −0.0023

0.8753 0.9210 0.6786 0.9985

polynomial–exponential models (Fig. 4) fitted the experimental data better than the linear and exponential models. It should be noted that Ke could not be appropriately defined in the polynomial–linear or exponential model, which probably represents the elimination pathways of fecal egestion, metabolic biotransformation, or growth dilution based on some related studies assessing the bioaccumulation process of environmental pollutants (Garcia-Galan et al., 2017; Hwang et al., 2018). 4. Conclusion We successfully established a MSPE method for the extraction of TBBPA from water and fish samples using MMIPs as the adsorbent and applied the developed MSPE-HPLC-UV method to determine the TBBPA levels in a laboratory-based fish–water system. Moreover, a polynomial–exponential model was developed to evaluate the transfer characteristics of TBBPA, which showed a correlation of 0.9985 between

BCF and exposure time. In the future, field-based studies should be conducted to confirm the simplified mathematical models developed in this study. Acknowledgments This work was supported by the National Key R&D Program of China (2017YFC0212003), the National Natural Science Foundation of China (No. 21577043), and the Open Project of Key Laboratory of Environment and Health, Ministry of Education (2017GWKFJJ02). We also appreciate the Analytical and Testing Center of Huazhong University of Science and Technology for the XRD pattern analysis. Competing financial interests The authors declare no competing financial interests.

Fig. 4. Different time-dependent BCF prediction models fitting the experimental data from a laboratory-based fish–water system: linear model (A), polynomial–linear model (B), exponential model (C), and polynomial–exponential model (D).

1362

L. Hu et al. / Science of the Total Environment 650 (2019) 1356–1362

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2018.09.002. References Abafe, O.A., Martincigh, B.S., 2016. Determination and human exposure assessment of polybrominated diphenyl ethers and tetrabromobisphenol A in indoor dust in South Africa. Environ. Sci. Pollut. Res. 23, 7038–7049. Ashizuka, Y., Nakagawa, R., Hori, T., Yasutake, D., Tobiishi, K., Sasaki, K., 2008. Determination of brominated flame retardants and brominated dioxins in fish collected from three regions of Japan. Mol. Nutr. Food Res. 52, 273–283. Bintein, S., Devillers, J., Karcher, W., 1993. Nonlinear dependence of fish bioconcentration on n-octanol/water partition coefficient. SAR QSAR Environ. Res. 1, 29–39. Chu, S.G., Letcher, R.J., 2013. Halogenated phenolic compound determination in plasma and serum by solid-phase extraction, dansylation derivatization and liquid chromatography-positive electrospray ionization-tandem quadrupole mass spectrometry. J. Chromatogr. A 1320, 111–117. Fernandez, M.V., Gardinali, P.R., 2016. Risk assessment of triazine herbicides in surface waters and bioaccumulation of Irgarol and M1 by submerged aquatic vegetation in Southeast Florida. Sci. Total Environ. 541, 1556–1571. Garcia-Galan, M.J., Sordet, M., Bulete, A., Garric, J., Vulliet, E., 2017. Evaluation of the influence of surfactants in the bioaccumulation kinetics of sulfamethoxazole and oxazepam in benthic invertebrates. Sci. Total Environ. 592, 554–564. Gissi, A., Lombardo, A., Roncaglioni, A., Gadaleta, D., Mangiatordi, G.F., Nicolotti, O., et al., 2015. Evaluation and comparison of benchmark QSAR models to predict a relevant REACH endpoint: the bioconcentration factor (BCF). Environ. Res. 137, 398–409. Gong, W.J., Zhu, L.Y., Jiang, T.T., Han, C., 2017. The occurrence and spatial-temporal distribution of tetrabromobisphenol A in the coastal intertidal zone of Qingdao in China, with a focus on toxicity assessment by biological monitoring. Chemosphere 185, 462–467. Grisoni, F., Consonni, V., Villa, S., Vighi, M., Todeschini, R., 2015. QSAR models for bioconcentration: is the increase in the complexity justified by more accurate predictions? Chemosphere 127, 171–179. Gu, S.Y., Ekpeghere, K.I., Kim, H.Y., Lee, I.S., Kim, D.H., Choo, G., et al., 2017. Brominated flame retardants in marine environment focused on aquaculture area: occurrence, source and bioaccumulation. Sci. Total Environ. 601, 1182–1191. Hardy, M.L., 2004. A comparison of the fish bioconcentration factors for brominated flame retardants with their nonbrominated analogues. Environ. Toxicol. Chem. 23, 656–661. Hwang, J.I., Zimmerman, A.R., Kim, J.E., 2018. Bioconcentration factor-based management of soil pesticide residues: endosulfan uptake by carrot and potato plants. Sci. Total Environ. 627, 514–522. Inthavong, C., Hommet, F., Bordet, F., Rigourd, V., Guerin, T., Dragacci, S., 2017. Simultaneous liquid chromatography-tandem mass spectrometry analysis of brominated flame retardants (tetrabromobisphenol A and hexabromocyclododecane diastereoisomers) in French breast milk. Chemosphere 186, 762–769. Jian, L.I., Wang, Y.F., Zhou, X.Q., Shi, Z.X., 2016. Determination of 18 brominated flame retardants in dairy product by gel permeation chromatography-solid phase extraction coupled to gas chromatography-mass spectrometry and high performance liquid chromatography-tandem mass spectrometry. Chin. J. Anal. Chem. 44, 1742–1747. Karimi, M.A., Ranjbar, M., Akbarpoor, M., 2016. Preparation of magnetic molecularly imprinted polymer nanoparticles for selective adsorption and separation of βestradiol. J. Clust. Sci. 27, 1067–1080. Kim, U.J., Lee, I.S., Oh, J.E., 2016. Occurrence, removal and release characteristics of dissolved brominated flame retardants and their potential metabolites in various kinds of wastewater. Environ. Pollut. 218, 551–557. Lee, I.S., Kang, H.H., Kim, U.J., Oh, J.E., 2015. Brominated flame retardants in Korean river sediments, including changes in polybrominated diphenyl ether concentrations between 2006 and 2009. Chemosphere 126, 18–24. Li, B., Zhou, Y., Wu, W., Liu, M., Mei, S., Zhou, Y., et al., 2015. Highly selective and sensitive determination of dopamine by the novel molecularly imprinted poly(nicotinamide)/ CuO nanoparticles modified electrode. Biosens. Bioelectron. 67, 121–128. Li, C., Ma, X., Zhang, X., Rui, W., Yuan, C., Li, Z., 2016. Magnetic molecularly imprinted polymer nanoparticles-based solid-phase extraction coupled with gas chromatography-

mass spectrometry for selective determination of trace di-(2-ethylhexyl) phthalate in water samples. Anal. Bioanal. Chem. 408, 1–8. Liu, A.F., Shen, Z.S., Tian, Y., Shi, R.G., Liu, Y., Zhao, Z.S., et al., 2017. Thin-layer chromatography coupled with high performance liquid chromatography for determining tetrabromobisphenol A/S and their derivatives in soils. J. Chromatogr. A 1526, 151–156. Lombardo, A., Roncaglioni, A., Boriani, E., Milan, C., Benfenati, E., 2010. Assessment and validation of the CAESAR predictive model for bioconcentration factor (BCF) in fish. Chem. Cent. J. 4 (Suppl. 1), S1–S11. Mackay, D., 1982. Correlation of bioconcentration factors. Environ. Sci. Technol. 16, 274–278. Madikizela, L.M., Tavengwa, N.T., Chimuka, L., 2018. Applications of molecularly imprinted polymers for solid-phase extraction of non-steroidal anti-inflammatory drugs and analgesics from environmental waters and biological samples. J. Pharm. Biomed. Anal. 147, 624–633. Mansouri, K., Consonni, V., Durjava, M.K., Kolar, B., Oberg, T., Todeschini, R., 2012. Assessing bioaccumulation of polybrominated diphenyl ethers for aquatic species by QSAR modeling. Chemosphere 89, 433–444. 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. Res. 25, 14361–14372. Pollock, T., Mantella, L., Reali, V., Decatanzaro, D., 2017. Influence of tetrabromobisphenol A, with or without concurrent triclosan, upon bisphenol A and estradiol concentrations in mice. Environ. Health Perspect. 125 (087014-1–087014-10). Satyanarayan, S., Ramakant, 2004. Bioaccumulation kinetics and bioconcentration factor of chlorinated pesticides in tissues of Puntius ticto (Ham.). J. Environ. Sci. Health B 39, 321–332. Shen, W., Xu, G., Wei, F., Yang, J., Cai, Z., Hu, Q., 2013. Preparation and application of imprinted polymer for tetrabromobisphenol A using tetrachlorobisphenol A as the dummy template. Anal. Methods 5, 5208–5214. Silva, L.J.G., Pereira, A., Meisel, L.M., Lino, C.M., Pena, A., 2015. Reviewing the serotonin reuptake inhibitors (SSRIs) footprint in the aquatic biota: uptake, bioaccumulation and ecotoxicology. Environ. Pollut. 197, 127–143. Springborn Life Sciences, 1989a. Bioconcentration and Elimination of Residues by Fathead Minnows (Pimephales promelas) Exposed to Tetrabromobisphenol A. Springborn Life Sciences (Inc Report #89-3-2952, Study #1199-1287-6105-141). Springborn Life Sciences, 1989b. Bioconcentration and Elimination of Residues by Eastern Oysters (Crassostrea virginica) Exposed to Tetrabromobisphenol A. Springborn Life Sciences (Inc Report #89-1-2918, Study #1199-0788-6106-142). Suzuki, S., Hasegawa, A., 2006. Determination of hexabromocyclododecane diastereoisomers and tetrabromobisphenol A in water and sediment by liquid chromatography/mass spectrometry. Anal. Sci. 22, 469–474. U.S. EPA, 2000. Estimation Program Interface (EPI) Suite, Version 1.68–2000. Veith, G.D., Defoe, D.L., Bergstedt, B.V., 1979. Measuring and estimating the bioconcentration factor of chemicals in fish. J. Fish. Res. Board Can. 36, 1040–1048. Xie, L.W., Guo, J.F., Zhang, Y.P., Hu, Y.C., You, Q.P., Shi, S.Y., 2015. Novel molecular imprinted polymers over magnetic mesoporous silica microspheres for selective and efficient determination of protocatechuic acid in Syzygium aromaticum. Food Chem. 178, 18–25. Yang, Y., Lu, L., Zhang, J., Yang, Y., Wu, Y., Shao, B., 2014. Simultaneous determination of seven bisphenols in environmental water and solid samples by liquid chromatography-electrospray tandem mass spectrometry. J. Chromatogr. A 1328, 26–34. Yin, J., Meng, Z., Zhu, Y., Song, M., Wang, H., 2011. Dummy molecularly imprinted polymer for selective screening of trace bisphenols in river water. Anal. Methods 3, 173–180. Yin, Y.M., Chen, Y.P., Wang, X.F., Liu, Y., Liu, H.L., Xie, M.X., 2012. Dummy molecularly imprinted polymers on silica particles for selective solid-phase extraction of tetrabromobisphenol A from water samples. J. Chromatogr. A 1220, 7–13. Zhou, Y., Zhou, T., Jin, H., Jing, T., Song, B., Zhou, Y., et al., 2015. Rapid and selective extraction of multiple macrolide antibiotics in foodstuff samples based on magnetic molecularly imprinted polymers. Talanta 137, 1–10. Zhou, T., Feng, Y., Zhou, L., Tao, Y., Luo, D., Jing, T., et al., 2016. Selective and sensitive detection of tetrabromobisphenol-A in water samples by molecularly imprinted electrochemical sensor. Sensors Actuators B Chem. 236, 153–162.