Fish bioconcentration studies with column‐generated analyte concentrations of highly hydrophobic organic chemicals

The performance of aqueous exposure bioconcentration fish tests according to Organisation for Economic Co‐operation and Development (OECD) guideline 305 requires the possibility of preparing stable aqueous concentrations of the test substances. For highly hydrophobic organic chemicals (HOCs; octanol–water partition coefficient [log KOW] > 5), testing via aqueous exposure may become increasingly difficult. A solid‐phase desorption dosing system was developed to generate stable concentrations of HOCs without using solubilizing agents. The system was tested with hexachlorobenzene (HCB), o‐terphenyl (oTP), polychlorinated biphenyl (PCB) 153, and dibenz[a,h]anthracene (DBA) (log KOW 5.5–7.8) in 2 flow‐through fish tests with rainbow trout (Oncorhynchus mykiss). The analysis of the test media applied during the bioconcentration factor (BCF) studies showed that stable analyte concentrations of the 4 HOCs were maintained in the test system over an uptake period of 8 wk. Bioconcentration factors (L kg−1 wet wt) were estimated for HCB (BCF 35 589), oTP (BCF 12 040), and PCB 153 (BCF 18 539) based on total water concentrations. No bioconcentration could be determined for DBA, probably because of the rapid metabolism of the test item. The solid‐phase desorption dosing system is suitable to provide stable aqueous concentrations of HOCs required to determine the bioconcentration in fish and represents a viable alternative to the use of solubilizing agents for the preparation of test solutions. Environ Toxicol Chem 2017;36:906–916. © 2016 The Authors. Environmental Toxicology and Chemistry Published by Wiley Periodicals, Inc. on behalf of SETAC.


INTRODUCTION
The regulatory chemical safety assessments of pesticides, biocides, pharmaceuticals, and other chemicals follow specific registration requirements, including the identification and scientific assessment of compounds with persistent, bioaccumulative, and toxic properties [1][2][3][4]. Bioaccumulative compounds enrich in organisms and may biomagnify in the food web. The bioconcentration factor (BCF), as determined in flowthrough fish tests according to Organisation for Economic Co-operation and Development (OECD) test 305 [5], is the standard endpoint in bioaccumulation assessment. The BCF is defined as the concentration of a test substance in fish divided by the concentration of the chemical in the surrounding medium. According to the Registration, Evaluation, Authorisation, and Restriction of Chemicals (REACH) regulation [1], a substance is defined as bioaccumulative or very bioaccumulative when the BCF in aquatic species is higher than 2000 or 5000, respectively. Flow-through fish tests require the preparation of stable, measurable, dissolved aqueous concentrations of the test substance, preventing the presence of the test substance as solid particles or as emulsions. For hydrophobic organic chemicals (HOCs), characterized by an octanol-water partition coefficient (log K OW ) > 5 and a solubility below approximately 0.01 mg/L to 0.1 mg/L, testing via aqueous exposure may become increasingly difficult. For these substances the dietary test is recommended, provided that the test is consistent with the relevant regulatory framework and risk assessment needs [5]. The dietary approach yields a biomagnification factor (BMF) rather than a BCF. The estimation of BCF values from data generated in the dietary studies seems to be complicated, as a result of the differences in the biomagnification and bioconcentration processes [6][7][8][9]. In addition, defined regulatory cut-off criteria for the biomagnification potential of chemical compounds are still not available. Under such circumstances, the estimation of BCFs for HOCs in flow-through fish tests will remain essential in the future. For HOCs with their low water solubility, this could be challenging with regard to obtaining stable water concentrations. The use of solvents and dispersants (solubilizing agents) is not generally recommended but may be acceptable to produce a suitably concentrated stock solution, for instance, of highly HOCs [5]. Stock solutions for fish BCF studies should preferably be prepared by simply mixing or agitating the test substance in the dilution water. An alternative method to achieve constant conditions in BCF testing by passive dosing has been explored by Adolfsson-Erici et al. [10]. In Adolfsson-Erici et al. [10], a polymer phase (silicone rubber) with fast diffusion kinetics was utilized to maintain defined concentrations of a mixture of hydrophobic substances in a bioconcentration test. The advantage of this approach is that any desired concentration can be maintained by changing the concentration in the polymer and the water flow rate across its surface. By matching the volume of the polymer phase to the physicochemical characteristics of the chemical of interest and the total volume of water generated, constant exposure concentrations can be maintained over a suitable period of This article includes online-only Supplemental Data. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. time, which was up to 8 d in the study of Adolfsson-Erici et al. [10]. However, the possible exposure periods using polymer phases are limited, which makes it challenging to maintain stable concentrations over a period of up to 60 d, which is the maximum uptake phase defined by OECD 305 [5]. Especially highly HOCs require extended exposure periods to allow the calculation of a steady state BCF (BCF ss ) as the ratio of the concentration in the fish and in the water at apparent steady state. The use of a solid-phase desorption dosing system may help to circumvent this problem and allow the continuous saturation of water with HOCs for fish BCF studies for several weeks without using solubilizing agents in those test solutions [11]. The principle of the solid-phase desorption dosing system is similar to the column elution method described by OECD guideline 105 [12], used to estimate the water solubility of chemicals. This method is based on the elution of a test substance with water from a microcolumn that is charged with an inert support material, previously impregnated with an excess of the chemical under investigation. The eluate, saturated with the test substance, might be further diluted to reach a suitable test concentration for BCF testing.
The aim of the present study was to evaluate the use of column-generated analyte concentrations (CGACs) for fish BCF studies with HOCs. Two flow-through fish tests with rainbow trout (Oncorhynchus mykiss) were carried out.

Loading and operation of the solid-phase desorption dosing system
Column-generated concentrations of HCB and oTP were prepared for the first BCF study. A solution of HCB or oTP in MTBE was mixed with Florisil (60-100 mesh) in a roundbottomed flask to reach a concentration of 1 mg g À1 (HCB) or Column-generated HOC concentrations in fish BCF studies Environ Toxicol Chem 36, 2017 907 5 mg g À1 (oTP). The solvent was subsequently evaporated to dryness on a rotary evaporator, and the material was additionally dried in a drying cabinet at 65 8C overnight. The treated carrier material was mixed with water to obtain a solid dough and filled into glass columns (diameter, 55 mm; height, 600 mm; end-capped with stainless-steel flanges). The bottom of the columns was covered with a glass fiber filter to separate matrix particles from the column inlet ( Figure 1). The columns were prepared following the examples presented in OECD guideline 312 [13]. Two columns were prepared with HCB, and 1 column was prepared with oTP. A constant flow of purified tap water was directed through each column from bottom to top (10 mL/min for each HCB column; 2 mL/min for the oTP column) by means of peristaltic pumps (Table 2, Figure 1). The single-outlet tubes of the 3 columns were connected to a mixing chamber where the column-generated concentrations were further diluted with purified tap water. This resulted in the test medium containing both test items that were to be continuously supplied to the experimental tank. Dilution water was purified by activated carbon filtration, aeration, and passage through a limestone column before being used for media preparation. In the second BCF study, CGACs of PCB 153 and DBA were tested. Preparation of the test medium was the same as in the first study, with minor modifications described in Table 2. Three columns loaded with DBA and 1 column with PCB 153 were used in the test. Throughout the uptake period of both studies, samples of pooled and diluted eluates obtained from the output of the mixing chamber were collected to estimate the CGACs for each test item. A flow diagram summarizing the loading and operation of the solid-phase desorption system is shown in Supplemental Data, Figure S1. The critical variables, which can be adjusted running the test system to maintain stable test concentrations within the desired limits, are explained.

Flow-through fish test
The flow-through fish test was carried out with juvenile rainbow trout (O. mykiss) with an average size at the start of the experiment of 3.3 AE 0.4 g and 3.8 AE 0.4 g for the first and the second studies, respectively ( Table 3). The animal experiment was performed in accordance with the German Animal Welfare Act under a permit from Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany (permit 84-02.04.2011.A069). Only healthy fish free of observable diseases and abnormalities were used in the studies. The experimental animals were raised in the hatchery of Fraunhofer IME from fertilized eggs that were obtained from Fischzucht Rameil. Two 100-L glass aquaria were used as experimental tanks and filled with 70 L of test medium (test group) or 70 L of water (control group). In the flow-through tanks a continuous flow of approximately 22 L/h (first study) or 21 L/h (second study) was maintained throughout the test, equivalent to 7.5 volume replacement and 7.2 volume replacement, respectively, per day ( Table 2). Tanks were stocked with 70 fish or 68 fish each during the first and second studies, respectively. In both studies, the test groups were exposed for 56 d (uptake period) and then transferred into a new aquarium where the fish were kept in clean water (depuration period), which was constantly replaced at the same rate as the previously applied test medium. The depuration periods were 28 d and 56 d during the first and second BCF studies, respectively. Animals exposed to the 2 test substances were compared with the unexposed control animals, which were kept under comparable conditions throughout the studies. The water in the test vessels was aerated via a stainlesssteel capillary. Temperature, pH, and the oxygen concentrations in the test vessels were measured daily. Observations were made throughout the test period on fish behavior and mortality. The light:dark cycle was adjusted to 12:12 h, and light intensities were measured once during the test. Rainbow trout in the control and test chambers obtained a daily ration equivalent to 1.5% of their body weight. A commercial feed for juvenile rainbow trout (Inicio Plus 0.8mm; Biomar) was used. Tanks were carefully cleaned every day 30 min to 60 min after feeding to avoid the accumulation of organic matter [14]. The total organic carbon (TOC) concentration in the water was estimated on a weekly basis.

Collection of water and fish samples
Samples of 4 fish were periodically removed from the test vessels during the uptake (days 7, 14, 21, 28, 35, 42, 49, and 56) and depuration (days 1, 2, 4, 8, 16, and 28) periods of the first BCF study. During the second BCF study, animals were collected on days 7, 14, 21, 28, 35, 42, 49, 52, and 56 (uptake period) and days 1, 2, 4, 8, 16, 32, and 56 (depuration period). The fish were randomly sampled, immediately weighed, and anesthetized in a water bath containing 150 mg/L MS 222 (Sigma Aldrich). The anesthetized fish were killed by a deep cut through the neck and immediately frozen in liquid nitrogen. At the end of the uptake and depuration periods 4 additional fish were removed from all tanks for lipid analysis. At each sampling time during the uptake phase and at other days where no sampling took place (Figure 2), adequate amounts of test water were collected from the experimental tank to analyze the test item concentrations.  Environ Toxicol Chem 36, 2017 C. Schlechtriem et al.

Chemical analysis of water samples
For the first BCF study, gas chromatography-mass spectrometry (GC-MS) analysis of water samples was carried out after liquid-liquid extraction (LLE), yielding total water concentrations, which means that no differentiation between freely dissolved and sorbed molecules was possible (further details are given in the Discussion section). For GC-MS analysis, samples were spiked with 100 mL of an internal standards solution containing HCB-13 C 6 and oTP-d 14 and were afterward acidified with 25 mL hydrochloric acid. After the addition of 12 mL MTBE, the vials were closed with screw caps and were agitated thoroughly on a horizontal shaker for approximately 15 min. To perform a rapid phase separation, the vials were centrifuged for 2 min at 1500 rpm. The MTBE phases were then transferred into 50-mL Zymark vessels using disposable Pasteur pipettes and were concentrated to approximately 0.2 mL by a stream of nitrogen using the TurboVap II concentration workstation (water bath temperature ¼ 40 8C). The concentrates were transferred into 300-mL microvials using Pasteur pipettes and were evaporated almost to dryness by a gentle stream of N 2 . Finally, the precipitates were redissolved in 200 mL n-hexane in an ultrasonic bath; 1 mL of the prepared samples was measured directly by GC-MS. Gas chromatographic separation was performed on a Varian 450-GC using a 30 m Â 0.25 mm VF-5ms column (Varian) with a 0.10-mm film thickness. Injections were made in the splitless mode (280 8C) with the GC oven at 60 8C. This was raised after 1 min to 280 8C at a rate of 20 8C min À1 and held at this temperature for 5 min. Mass spectrometric analysis was conducted using a Varian Ion-trap 240-MS. The detection limit of the method was 0.1 mg L À1 for both analytes.
In the second BCF study, GC-MS analysis of water concentrations was carried out after liquid-liquid extraction, yielding total water concentrations. Samples of 50 mL of water were spiked with 50 mL of an internal standards solution containing PCB 138 and DBA-d 14 in methanol and extracted by shaking with 10 mL of toluene. The organic phase was separated after centrifugation, dried over Na 2 SO 4 , concentrated in a Zymark vessel, transferred into 300-mL microvials, and concentrated under a nitrogen stream to an end volume of 150 mL. The analysis was performed by GC-MS in a selected ion monitoring mode. Gas chromatographic separation was performed on a GC 6890 N (Agilent) using a 30 m Â 0.25 mm Rxi-5sil MS column (Restek) with 0.50-mm film thickness. Injections were made in the splitless mode (300 8C) with the GC oven at 130 8C. After 1 min, the temperature was raised to 330 8C at a rate of 20 8C min À1 and held at this temperature for 5 min. The transfer line temperature was set at 300 8C. Mass spectrometric analysis was conducted using a MSD 5973 Network (Agilent). The detection limit of the method was 0.1 mg L À1 for both analytes.

Chemical analysis of fish samples
Extraction of fish samples: Fish were extracted by accelerated solvent extraction (Dionex ASE 350 system). Every fish was placed in a glass beaker with approximately 5 g of preextracted Hydro Matrix and homogenized with a metal spatula. The mixture was transferred to 33-mL accelerated solvent extraction cells containing cellulose filters. Each glass beaker was rinsed with 2 mL of 1:1 acetone:dichloromethane that were transferred to the respective accelerated solvent extraction cells. The fish samples were extracted using the following conditions: Acetone:dichloromethane (1:1) was used as the solvent mixture and heated within 5 min to 120 8C and kept at this temperature for 5 min. The cells were constantly flushed at a 30% rate and finally purged for 60 s. Extraction was carried out at 1500 psi. Two extraction cycles were applied, and the extracts were dried with Na 2 SO 4 . Afterward, aliquots were spiked with a Average column-generated concentration during uptake period calculated from cumulated column effluent concentrations (see Figure 2). b Flow rate of diluted cumulated column effluents into experimental tank. c Time-weighted average concentration of test items measured in experimental tank during uptake period.
Column-generated HOC concentrations in fish BCF studies Environ Toxicol Chem 36, 2017 internal standards and cleaned on SEP-PAK silica gel cartridges (2 g/12 mL; Waters) by eluting with hexane and dichloromethane/hexane 1:1. The eluates were concentrated to dryness, redissolved in hexane (first study) and toluene (second study) and measured by GC-MS. During the first study, the GC-MS analysis was performed according to the analysis of the water extracts. The GC-MS analysis of samples collected during the second study was performed according to the analysis of the water extracts but with a differing hold time of 15 min at 330 8C. The fish weight recorded at the time of sampling was used for calculation of tissue concentrations.

Lipid analysis of fish samples
The lipid content of the fish samples was determined gravimetrically by solvent extraction using propan-2-ol and cyclohexane as the organic solvent system [15,16].

Calculations
Average water concentrations. Average CGACs of HCB, oTP, PCB 153, and DBA were calculated by multiplying the concentrations measured in the pooled and diluted eluates after the mixing chamber by the dilution factors resulting from the connection of the different columns. Concentrations of the test items measured in the experimental tanks during the uptake period were used to calculate time-weighted average concentrations. First, weighted average concentrations were calculated by multiplying the average of 2 subsequently measured concentrations by the time period (h) between both measurements. All weighted average concentrations were then summed up and divided by the total time (h) of the uptake period resulting in the time-weighted average concentration.
Calculation of BCFs. The BCF ss is usually calculated as the quotient of the concentrations of the test item in the fish tissue (C f , mg/kg fish) in steady state and the corresponding average concentration of the test item in the water (C w , mg/L). The steady state is assumed to be reached if 3 successive analyses of the C f of samples taken at intervals of at least 2 d are within AE 20% of each other [5]. All concentrations of the test item measured in fish on each sampling date were subjected to Grubbs' test for outliers using SQS 2013 Ver 1.00. All test items did not reach steady-state concentrations in the fish tissue at the end of the uptake period and therefore did not allow the calculation of BCF ss estimates.
Instead, the kinetic bioconcentration factor (BCF k ) was derived from the uptake and depuration rates. The depuration rate constant (k 2 ) was calculated by fitting a 1-compartment model to the measured concentrations in fish during the depuration phase where C f(t) is the concentration in fish at sampling time (in mg/ kg) and C f(ti) is concentration in fish at the start of the depuration phase (in mg/kg). For the fitting, the concentrations were log e -transformed to allow linear regression of log-concentrations versus time. The uptake rate constant (k 1 ) was calculated by nonlinear regression analysis of the ratios of the concentration in fish (C f ; mg/kg) to the concentration in water (C w ; mg/L) against time during the uptake phase and including the depuration rate k 2 fitted before where k 1 is the uptake rate constant (L kg À1 d À1 ) and k 2 is the depuration rate constant (d À1 ). The BCF k was calculated as the quotient k 1 /k 2 . For growth correction, weight data for treatment and control groups were converted to natural logarithms and plotted against time. A parallel line analysis was performed checking for equality of slopes between uptake and depuration phases and treatments. If no statistically significant differences were found, data were pooled and an overall growth rate constant was calculated.
The calculated growth rate constant per day (k g ; d À1 ) was subtracted from the overall k 2 (d À1 ) to give the growth-corrected depuration rate constant, k 2g (d À1 ) Then, k 1 was divided by the k 2g to give the growth-corrected kinetic BCF, denoted BCF kg (Equation 4).
Both BCF k and BCF kg were normalized to a lipid content of 5%.
where BCF kgL (growth-corrected; Equation 5) is the kinetic bioconcentration factor normalized to a lipid content of 5% (L kg À1 ), L n is the mean lipid fraction at the end of the uptake phase in the treatment group (based on wet wt), and BCF k(g) (growth-corrected) is the kinetic bioconcentration factor (L kg À1 ). All calculations were done using Microsoft Office Excel 2010 for calculation of means and SigmaStat Ver 3.5 (Systat Software) for the regressions, using data for the parent compound in whole fish. For growth correction, SigmaPlot Ver 13.0 (Systat Software) was used.

Column-generated concentrations and exposure concentrations
The concentrations of dissolved HCB, oTP, PCB 153 and DBA are presented in Figure 3. Average column-generated concentrations for each test item are presented in Table 2. With 181 mg/L, the concentration of oTP was highest followed by HCB (7.1 mg/L). The more hydrophobic substances PCB 153 and DBA showed a lower water solubility leading to an average column-generated concentration of 0.704 mg/L and 0.104 mg/L, respectively. The test concentrations measured in the test system during the course of the BCF studies (uptake period) are presented in Figure 2. Time-weighted average concentrations of oTP and HCB were 453 ng/L and 390 ng/L, respectively. Lower concentrations were measured for PCB 153 (23.0 ng/L) and DBA (21.4 ng/L) ( Table 2). The concentrations of HCB, oTP, and PCB 153 in the test chambers were constantly maintained within AE20% of the time-weighted average concentration of the measured values during the uptake phase. In isolated cases, concentrations slightly exceeded the limit range for a short time. For DBA, constant concentrations were predominantly observed, with the exception of significantly increased concentrations measured in the test system on day 1 and between days 30 and 34 of the uptake period following accidental contact with 1 of the 3 columns.

Experimental conditions
During the first study, an average water temperature of 15.1 8C (14.6-15.8 8C), an average oxygen concentration of Column-generated HOC concentrations in fish BCF studies Environ Toxicol Chem 36, 2017 7.7 mg/L (5.8-8.9 mg/L), and an average pH of 8.2 (7.9-8.6) were measured. During the second BCF study, the average water temperature of 15.0 8C (14.7-15.4 8C), average oxygen concentration of 8.64 mg/L (7.8-9.7 mg/L), and average pH of 8.0 (7.8-8.2) were comparable to the first study. The concentrations of TOC measured in the water of the experimental tank collected before feeding at different days of the first study remained continuously below or close to a level of 2 mg/L. The concentrations were even below the maximum acceptable value of TOC defined in OECD 305 for dilution water. The TOC content of the dilution water used for both flow-through fish tests was close to 0. Throughout the second study, the concentration of TOC in the test vessel was between 4 mg/L and 7 mg/L, which is below the maximum tolerated TOC concentration of 10 mg/L apart from the TOC concentration contributed by carbon content of test substance and organic solvent (allowed as solubilizing agents) as specified by OECD 305 [5].

Growth performance of experimental animals
Animals used for the first BCF study had an average weight of 3.3 g at the start of the experiment. At the end of the uptake period, animals in the control and test groups reached an average weight of 11.0 g and 10.6 g, respectively. During the depuration period the weight of the animals further increased to 17.0 g (control group) and 16.9 g (test group). No significant difference in growth was observed between the 2 groups (Table 3). Animals used for the second BCF study were slightly smaller (2.6 g) compared with the control group (3.8 g) at the start of the experiment. However, differences disappeared during the uptake and depuration periods. Animals in both groups had an average weight of 42.2 g at the end of the study. No difference in the growth performance of control animals and exposed animals was observed in both studies. Similar growth rate constants of 0.0201 d À1 and 0.0226 d À1 were estimated for the first and second studies, respectively ( Table 4). The lipid content of the experimental animals was similar at the start of the first and second studies with 5% and around 6%, respectively. The lipid content increased during the uptake period of both studies to almost 6% in the first study and to 7% to 8% in the second study. Differences in lipid contents between control and exposed animals observed in both studies were not significant (Table 3).

Concentrations in fish during uptake and elimination periods
Tissue concentrations of the test items measured during the uptake and depuration periods are presented in Figure 4. At the end of the uptake period, tissue concentrations of 11.2 mg/kg and 293 mg/kg were reached for HCB and PCB 153, respectively. Tissue concentrations of oTP increased until day 36, obviously reaching an apparent steady-state concentration of 4.6 mg/kg. However, fish collected at a later time showed a high variation of tissue concentrations with a clear tendency to decreasing tissue concentrations toward the end of the uptake period. Dibenz[a,h]anthracene concentrations in fish were highly variable during exposure. Fish samples with concentrations below limits of quantification (LOQ) alternated with samples where slightly increased concentrations were found. At the end of the uptake period a sudden escalation to significantly increased tissue concentrations was observed. Fish samples collected during the entire depuration period did not contain DBA. Tissue concentrations of HCB and oTP measured at the end of the uptake period decreased during depuration by 63% and 94%, respectively. Concentrations of previously accumulated PCB 153 decreased by 76% during the depuration period.

Uptake and elimination rates
The k 2 values for HCB, oTP, and PCB 153 are presented in Figure 5. With k 2 estimates of 0.0239 d À1 and 0.0340 d À1 , the depuration of PCB 153 and HCB in rainbow trout was significantly slower compared with oTP (0.0999 d À1 ) as reflected in the elimination half-life and time required to reach 95% depuration (t 95% ; Table 4). Growth correction of the depuration rate constants was leading to significantly decreased k 2g values of 0.0139 d À1 , 0.0798 d À1 , and 0.0013 d À1 for HCB, oTP, and PCB 153, respectively ( Table 4). The k 1 values were calculated by nonlinear regression analysis of the ratios of concentration in fish and water against time during the uptake period ( Figure 6) and taking into account the k 2 fitted before. Similar uptake rates of 1210 d À1 and 1202 d À1 were estimated for HCB and oTP, respectively. Polychlorinated biphenyl 153 showed a lower uptake rate of 443 d À1 (Table 4).

BCF estimates
Kinetic bioconcentration factors were calculated for all test items based on the respective uptake and depuration rates. Kinetic bioconcentration factors were highest for HCB (35 589) followed by PCB 153 (18 539) and oTP (12 040). In some cases, lipid normalized and growth-corrected BCF k values differed significantly from the original estimates (Table 4).

DISCUSSION
The analysis of the CGACs during the BCF studies showed that the solid-phase desorption dosing system can continuously generate aqueous solutions of constant concentrations of HOCs over a period of 56 d. The dilution of the column eluates resulted in sufficient quantities of test solution allowing more than 5 volume replacements through each test chamber per day. Measured concentrations of test solutions were always above the detection limit of the analytical methods. The concentration of the test substances in the chambers could be maintained within AE20% of the mean of the measured values during the uptake phase. The preparation of test solutions using the solidphase desorption dosing system thus fulfilled all requirements specified in OECD guideline 305 [5]. However, several aspects need to be considered to ensure the proper functioning of the dosing system. The solid-phase desorption dosing system is especially suitable for HOCs (log K OW > 5), which allow a stable impregnation of the carrier material. Such test items usually show a low solubility in water below 10 mg/L. The substances tested in the present study were all characterized by a high hydrophobicity (log K OW 5.5-7.8) leading to CGACs in the range of 0.1 mg/L (DBA) to 180 mg/L (oTP), which were similar to the respective water solubility estimates described in the literature [17]. Some compounds are not suitable to be tested with the column desorption system. For instance, surface active compounds may form emulsions or micelles and tend to adsorb at water-solid interfaces and are thus not suitable to produce CGACs. Also, the use of highly volatile substances should be avoided because of the high losses of test items expected to occur during the column preparation and the desorption process, which may hamper the generation of stable eluate concentrations. Depending on the chemical properties of the test item, modifications of the dosing system may be required. The pH-dependent hydrolysis of a chemical may require the elution under adjusted pH conditions. Also, the risk of test-item breakdown caused by incident light is important. In this case, the desorption system should be covered to allow the elution of the test item under dark conditions. When biodegradable substances are tested, the source water may need to be treated to reduce dissolved organic carbon and bacterial concentration beyond the requirement for dilution water according to OECD guideline 305. The establishment of sterile conditions to avoid the biodegradation of a chemical is in principle possible but requires specific technical measures.
In the present study a Florisil matrix was used as carrier material. However, other inert materials such as silica gel, glass beads, or additional commercially available matrices could also in principle be used as carrier for the column elution method. The suitability of an adsorbing matrix should be tested before use in a flow-through study to guarantee the continuous supply of constant aqueous concentrations.
Careful investigations are necessary before the initiation of a flow-through study to estimate the right settings for the optimal dosing procedure. A laboratory pump with adjustable water flow (e.g., a peristaltic or membrane pump) is required to maintain a slow and constant water current through the column. This allows sufficient time for the test item to desorb from the matrix material into the water. In the present study, flow rates as low as 1 mL min À1 to 10 mL min À1 proved to be appropriate to obtain decent results. Regular concentration control analysis of water samples is required to confirm the proper functioning of Column-generated HOC concentrations in fish BCF studies Environ Toxicol Chem 36, 2017 913 the dosing system. Adjustments of the flow-rate settings might be required to compensate for concentration changes that may occur during continuous application over extended periods. Test solutions supplied to the test chambers are obtained by dilution of CGACs with water. The concentration of a test substance should be below its chronic effect level, or 1% of its acute asymptotic median lethal concentration; however, at the same time, it should be at least an order of magnitude above its limit of quantification in water by the analytical method used [5]. The detection limit of the analytical methods may have to be adjusted to allow the measurement of analyte concentration in the test solutions. Alternatively, the coupling of several column units might help to increase analyte concentrations in the test solutions. For flow-through tests, at least 5 volume replacements are preferred through each test chamber per day [5]. The present study has shown that sufficient volumes of test solution can be prepared by dilution of CGAC obtained from 1 or more columns.
The water temperature applied during a flow-through test needs to fulfill the requirements of the test species. Water temperatures should be in the range of 13 8C to 17 8C for rainbow trout and 20 8C to 25 8C for most of the other test species (e.g., common carp, zebra fish, or fathead minnow) suggested by OECD 305. Desorption/solubility of analytes is temperature-dependent. Therefore, the temperature of the elution water should be as constant as possible to avoid unintentional changes in analyte concentrations.
During the production of CGACs, any disturbance (hard shocks, concussions, etc.) of the columns should be avoided to guarantee the proper functioning of the solid-phase desorption
dosing system. Any such disturbance can lead to significant changes of test concentrations as shown in the present study, where increased analyte concentrations of DBA were measured in the test system between day 30 and day 34 of the uptake period following accidental contact with 1 of the 3 columns. The carrier material in the columns should not be packed too densely to avoid excessive pressure that may lead to formation of air bubbles in the column. For the same reason, elution water should be sufficiently equilibrated under laboratory conditions to allow degassing of excess air. If the carrier material in the glass column shows local compression and crack formation, the use of a carrier material with larger particles should be considered.
In the present study, flow-through tests were carried out over 56 d to prove the suitability of CGACs for bioconcentration studies with HOCs. Constant concentrations were observed for all test items over the entire exposure period, which is a substantial advantage compared with alternative dosing techniques such as flow-through passive dosing systems. In a study by Adolfsson-Erici et al. [10] stable aqueous concentrations were generated by passive dosing. Nevertheless, concentrations of more hydrophilic chemicals such as 2,3,4trichloroanisole, musk xylene, or chlorpyrifos showed a decreasing trend with time amounting to 13% to 85% after 13 d and 17 d of exposure. Passive dosing systems are simple, safe, and flexible tools that have clear advantages compared with the more elaborate process of solid-phase desorption dosing. Thus this might be preferred for the generation of stable concentrations of HOCs over shorter exposure periods lasting several days rather than several weeks.
According to OECD 305, the uptake phase should be run for 28 d unless it can be shown that steady state has been reached earlier. If steady state has not been reached by 28 d, the uptake phase can be extended until steady state is reached or for 60 d, whichever occurs first [5]. During the first BCF study the uptake phase was extended to 56 d to prove the suitability of the desorption system to generate stable test concentrations over or around the maximum uptake phase defined by the guideline.
The exposure time required to reach steady-state concentrations in fish can be derived from empirical relationships between the k 2 from fish and the K OW . Following the equations described in OECD 305, the predicted uptake phases for HCB and oTP are 22 d and 19 d, respectively, and are thus shorter than the extended uptake phase applied in the present study. For HCB, the steady state of test substance in fish (C f ) was obviously reached at the end of the uptake phase. No significant increase of C f in time among the last 3 successive analyses was observed. Temporary BCFs calculated for day 49 (31 113) and day 52 (30 750) were close to the BCF k of 35 589, which is well in accordance with the BCF estimate for HCB published by Adolfsson-Erici et al. [18]. Interestingly, the concentration in fish collected on day 56 (BCF 24 325) decreased to almost the level measured on day 42 (BCF 23 126), which would lead to a smaller steady-state BCF estimate when the last 3 successive analyses would be considered.
In contrast to HCB, tissue concentrations of oTP showed a high deviation but reached an apparent steady-state level already after approximately 2 wk of exposure. Tissue concentrations clearly decreased toward the end of the exposure period, reaching an average level that was even below the tissue concentrations measured on day 7. This might be explained by the adaptation of the biotransformation activity in fish following extended exposure to the dissolved analyte, leading to stronger analyte elimination during the bioconcentration process. Similar effects have been observed in previous studies [19].
With predicted uptake phases for PCB 153 of approximately 172 d, the exposure period of 56 d was expected to be insufficient to reach steady-state concentrations for PCB 153. Tissue concentrations still increased at the end of the uptake period, thus not allowing the estimation of a steady-state BCF. As described in OECD 305, steady state has not been reached by 60 d, the BCF is calculated using only the kinetic approach. Using the estimated uptake and elimination rates, BCF k values were calculated for HCB, oTP, and PCB 153. Growth correction and lipid normalization of BCF k estimates are required, especially if significant growth of the test animals occurs and the lipid content changes markedly during the uptake phase or depuration phase. This was the case for both studies; therefore, the BCF k values were normalized to a fish with a 5% lipid content and/or corrected for growth during the study period.
Bioconcentration factor values from the present study were calculated based on total aqueous concentrations measured by liquid-liquid extraction. With liquid-liquid extraction, both freely dissolved and sorbed test substance molecules are extracted. Because the freely dissolved concentration corresponds to the aqueous concentration that is bioavailable for organisms [20], a potential underestimation of BCF values is discussed if significant amounts of test substance molecules are bound by organic matter in the test system (e.g., from feed or feces) [21,14]. The smaller BCF value for PCB 153 compared to HCB in the present study might be influenced by such reasons. The OECD potential influence is addressed in OECD 305 where the maximum amount of organic matter in the test system is restricted to 10 mg/L TOC. Beyond this restriction, OECD 305 does not prescribe to discriminate between dissolved and bound species. Beside artificial influences [22] (e.g., from organic matter), physiological reasons are debated that may potentially lead to reduced bioconcentration of highly HOCs (e.g., restriction of membrane permeability for test substance molecules with increasing molecular size and hydrophobicity) [23][24][25]. Independent of the results obtained, all BCF estimates derived from the present study were clearly above 5000, underlining the high bioaccumulation potential of the 3 HOCs. With 35 589 (HCB) and 12 040 (oTP) the estimated BCF k s were higher compared with model predictions based on K OW [26], resulting in BCFs of 13 964 (HCB) and 9441 (oTP). In contrast, the kinetic BCF of PCB 153 was significantly lower (18 539) compared with the predicted value (851 138), showing that model predictions should be used with caution also for chemicals for which analytical challenges preclude straightforward measurements (e.g., for HOCs).
In contrast, aqueous exposure to DBA did not lead to bioconcentration in fish. Only minor concentrations or concentrations below LOQ were observed in fish collected during the uptake phase, which immediately disappeared at the onset of the elimination period. The low accumulation potential of DBA can be explained by the efficient metabolism of PAHs in fish, which has been described in previous studies [27,28].
Based on the results obtained in the present study, it can be concluded that CGACs allow the testing of HOCs in fish BCF studies, according to OECD 305, without using solubilizing agents. Bioaccumulation studies with HOCs therefore do not necessarily need to be tested in BMF studies that still do not fulfill the risk assessment requirements and thus challenge the use of experimental animals for such studies.