Chem. Res. Toxicol. 2010, 23, 55–65
55
Passive Dosing for Producing Defined and Constant Exposure of Hydrophobic Organic Compounds during in Vitro Toxicity Tests Kilian E. C. Smith,*,† Gertie J. Oostingh,‡ and Philipp Mayer† Department of EnVironmental Chemistry and Microbiology, National EnVironmental Research Institute, Aarhus UniVersity, FrederiksborgVej 399, P.O. Box 358, 4000 Roskilde, Denmark, and Department of Molecular Biology, UniVersity of Salzburg, Hellbrunnerstrasse 34, 5020 Salzburg, Austria ReceiVed August 8, 2009
Toxicity testing of hydrophobic organic compounds (HOCs) in plastic cell culture plates is problematic due to compound losses through volatilization and sorption to the wells and culture medium constituents. This leads to poorly defined exposure and reduced test sensitivity. Passive dosing can overcome these problems by the continual partitioning of HOCs from a dominating reservoir loaded in a biologically inert polymer such as silicone, providing defined and constant freely dissolved concentrations and also eliminating spiking with cosolvents. This study aimed to select a suitable passive dosing format for in vitro tests in multiwell plates and characterize its performance at 37 °C. Silicone O-rings were the most suitable format; they were both practical and demonstrated excellent passive dosing performance. (1) The rings were loaded by partitioning from a methanol solution containing polycyclic aromatic hydrocarbons (PAHs) (log KOW, 3.33-6.43) that served as model compounds, followed by removal of the methanol with water. This resulted in highly reproducible HOC concentrations in the silicone O-rings. (2) The release of PAHs into aqueous solutions was rapid and reproducible, with equilibrium partitioning being reached within hours. (3) The buffering capacity of the O-rings was sufficient to maintain stable concentrations over more than 72 h. The O-rings were then applied to test a range of PAHs at their aqueous solubility in an array of established in vitro cell culture assays with human cells and cell lines. These included the formation of reactive oxygen species, induction of the IL-8 cytokine promoter, and secretion of MCP-1 by the cells. The biological responses depended on the melting point of the individual PAHs and their maximum chemical activities (amax). Only those PAHs with the highest amax stimulated the formation of reactive oxygen species and MCP-1 secretion, while they inhibited the induction of the IL-8 cytokine promoter. 1. Introduction The advantages of in vitro tests, including their ethical acceptance, suitability for test automation, high reproducibility, and relatively low costs, have led to an increased use of such tests for the toxicity testing of hydrophobic organic compounds (HOCs) (1-4). For cell culture experiments, the vessel of choice is the cell culture plate, which is available in various architectures and manufactured from different plastics. There is, however, an incompatibility between those properties of cell culture plates necessary for their successful application in in vitro tests and the proper testing of HOCs. On the one hand, the plates should be biologically inert and allow for gas exchange. On the other hand, these same characteristics can lead to unacceptable losses of HOCs by sorption to the plastic material and volatilisation (5, 6). Furthermore, experimental conditions are often not optimal with regard to HOC losses. In vitro assays with mammalian cells are conducted at elevated temperatures (normally 37 °C), exacerbating volatile losses. Many test chemicals can bind to protein or lipid in the culture medium, leading to their reduced availability (7, 8). These issues make it difficult to control and thus to properly define the target dose in in vitro tests, which poses challenges in interpreting * To whom correspondence should be addressed. Tel: +45 4630 1248. Fax: +45 4630 1114. E-mail:
[email protected]. † Aarhus University. ‡ University of Salzburg.
the toxicity data, comparing data from different studies, and extrapolating in vitro data to in vivo observations (8-10). To overcome these limitations, efforts could be directed toward using cell culture plates from more HOC-compatible materials such as glass, with the downsides of increased costs and a less optimal cell adherence as compared to cell culturetreated plastic. However, the need for gas exchange and thus of volatile loss remains, as does the sorption of HOC to the medium constituents. Alternatively, one might accept that sorption and volatilization take place and better define exposure during the test by analytical measurements or modeling approaches (7, 8). In an estrogenicity reporter gene assay, Heringa et al. (7) successfully used nondepletion solid-phase microextraction (nd-SPME) to sample the freely dissolved compound concentrations as a proxy for target dose. However, if the time required for sampling is longer than the test duration or if compound loss is rapid, it remains difficult to adequately define exposure. Similarly, modeling the partitioning to lipids and proteins in the culture medium and the target cells to derive available concentrations does not take into account the compoundspecific losses during the test, which result in different decreases in exposure of the test substances (6, 8). Another strategy is to maintain constant exposure concentrations by a resupply of test substance during the test. Flowthrough systems are often used to maintain constant exposure concentrations in tests with aquatic organisms, and chemostats are well-established for studies with bacteria, yeast, and unicel-
10.1021/tx900274j CCC: $40.75 2010 American Chemical Society Published on Web 11/20/2009
56
Chem. Res. Toxicol., Vol. 23, No. 1, 2010
lular algae. Unfortunately, it is not practical to operate in vitro toxicity tests in a flow-through mode. Alternatively, the test substance might be supplied by passive dosing, also known as partitioning-driven administration or partition-controlled delivery (11-15). Passive dosing involves the partitioning of HOC(s) into the test medium from a dominating reservoir loaded in a biologically inert polymer phase such as silicone (11-14, 16). The continual partitioning compensates for compound losses, resulting in constant freely dissolved HOC exposure concentrations. Furthermore, by optimizing the architecture of the polymer phase, it can be ensured that equilibrium partitioning prevails. This simplifies the application, since with knowledge of the polymer:water partitioning of the test chemical, freely dissolved concentrations are known a priori without the need for analytical confirmation. A further important advantage of passive dosing is that target cells are not exposed to cosolvents, which can themselves be toxic to cells or modify the toxicity of the test chemical. Passive dosing is highly suited to the small size and high replication of in vitro tests but requires a new passive dosing format that physically fits the dimensions of cell culture plates and is compatible with the standard operation of in vitro tests. This requires: (1) operation under sterile conditions, (2) the possibility for cell attachment at the well bottom prior to exposure, (3) the ability to perform plate reader measurements during exposure, and (4) eventual analytical confirmation of exposure during or after the test. Silicone was chosen as the polymer material because of its proven biocompatibility, excellent partitioning properties (17, 18), and low mass transfer resistance for hydrophobic chemicals (14, 19). Silicone is commercially available in many shapes, and various silicone formats were initially considered for passive dosing in the cell culture plates, including coated glass fibres (20), silicone tubing (21), silicone rods (22), thin silicone coatings (12), thin silicone films (14), silicone cast into the plate wells (13), and silicone oil (23). For in vitro cell culture experiments, the most suitable and flexible format in meeting all of the above requirements was the silicone O-ring, manufactured out of food-grade silicone. They are easy to introduce and remove from the wells and are cheaply available in large quantities in a highly standardized form. This study systematically investigated the applicability of passive dosing using silicone O-rings for the routine toxicity testing of HOCs in in vitro tests using commercially available plastic cell culture plates. Particular emphasis was placed on choosing a flexible and practical passive dosing format, developing a simple and reproducible HOC polymer-loading protocol, and characterizing the performance of the system under in vitro experimental conditions. The passive dosing performance of the silicone O-rings was tested with a range of PAHs. These were selected as model compounds since they possess a wide range of physicochemical properties and thus undergo different loss processes and can also be easily measured. Finally, the applicability of this passive dosing format was demonstrated in several different in vitro cellular assays using different cell types. These included the formation of reactive oxygen species (ROS) in peripheral blood mononuclear cells (PBMCs), interleukin (IL)-8 promoter induction in human bronchial epithelial (A549) cells, and the secretion of monocyte chemotactic protein (MCP)-1 by human monocyte (THP-1) cells.
2. Experimental Procedures Caution: Some of the listed PAHs are harmful, and adequate measures should be taken during handling to eliminate exposure and to ensure their proper disposal.
Smith et al. 2.1. Chemicals and Materials. Food-grade silicone O-rings with an outer diameter of 14.4 mm, inner diameter of 9.6 mm, mass of 231 mg (C.V. 1%, n ) 10), and a calculated volume of 0.171 mL were used for passive dosing in the cell culture plates (order no. ORS-0096-24., Altec, Cornwall, United Kingdom). Costar 24-well flat bottom cell culture-treated polystyrene plates (Corning Inc., Corning, NY) were used as supplied. The following PAHs were selected as model hydrophobic compounds: naphthalene (99%, Sigma, Germany), acenaphthene (99%, Sigma), fluorene (99%, Aldrich, Germany), phenanthrene (99.5%, Aldrich), anthracene (99%, Acros, Belgium), fluoranthene (99%, Aldrich), pyrene (>99% Fluka, Germany), benz(a)anthracene (99%, Aldrich), chrysene (99% Cerilliant, TX), benzo(a)pyrene (98%, Cerilliant), and dibenzo(a,h)anthracene (97%, Aldrich). Ethylacetate (p.a. grade) and methanol (HPLC grade) were used as solvents (Merck, Darmstadt, Germany). Milli-Q water was used (Super Q treated, Millipore, MA). The cell culture medium consisted of RPMI 1640 medium supplemented with 10% v/v fetal calf serum (FCS), penicillin (end concentration, 100 U mL-1), streptomycin (end concentration, 100 µg mL-1), and L-glutamine (end concentration, 2 mM) as recommended by the distributor. All cell culture medium reagents were obtained from PAA Laboratories (Pasching, Austria). Ficoll-Pague Plus (GE Healthcare, Vienna, Austria) was used for the density centrifugation. Carboxy-DCFDA was obtained from Molecular Probes (Carlsbad, CA). MCP-1-specific antibodies were obtained from Immunotools (Friesoythe, Germany). The CellTiterBlue test was obtained from Promega (Madison, WI). 2.2. Cleaning and Loading of the O-Rings. 2.2.1. Initial Cleaning. O-Rings were cleaned by soaking once overnight in ethylacetate, followed by three washes, each time overnight, in methanol. Any methanol adhering to the O-rings was removed by three overnight washes with Milli-Q water. Cleaned O-rings were stored until use in a sealed bottle in Milli-Q water. Before the PAH loading step, they were dried by wiping the surfaces using lintfree tissue. 2.2.2. O-Ring Loading with PAHs. The silicone for passive dosing can be loaded with HOCs by solvent evaporation (11, 12, 16, 22). However, this can lead to evaporative losses or the crystallization of the test substance at the polymer-air interface (11). Therefore, in the present study, another loading approach was applied based on partitioning from a methanol standard solution (24). Methanol was used as the loading solvent since PAHs are sufficiently soluble in it, it causes limited swelling of the silicone as compared to other nonpolar solvents, and it can be quantitatively removed from the silicone by rinsing with water. In this way, no solvent comes into contact with the biological material during the toxicity test. O-Rings were loaded with PAHs by partitioning at 21 °C from methanol for a minimum period of 72 h. For some experiments, rings were loaded with individual PAHs to the saturation level using methanol suspensions of the respective PAH. The PAH crystals in the methanol suspension then served to maintain the saturation level in the methanol (13, 25). For other experiments, rings were loaded to below saturation levels using PAH solutions in methanol. After loading, the O-rings were then removed from the loading solution, and the surfaces were thoroughly wiped using lint-free tissue to remove any adhering suspension. Finally, they were rinsed three times for 1 h each with a small volume of Milli-Q water to remove any residual methanol. 2.3. Sampling and Analysis. 2.3.1. Water and Medium Samples. For the water samples, 50 µL of water was mixed with 50 µL of methanol and analyzed directly by HPLC. For the cell culture medium samples, 50 µL of medium was added to 950 µL of methanol. This resulted in precipitation of the medium constituents, and an aliquot of the clear supernatant was taken for HPLC analysis. To test that no PAHs remained sorbed to the precipitate, a methanol stock solution with known PAH concentrations was diluted with medium as above. The concentrations in the supernatant were measured and compared to the known values, taking the dilution into consideration. Measured concentrations in the spiked medium samples were not reduced by sorption to the precipitate (mean concentration, 97% of nominal value; range, 94-99%).
PassiVe Dosing To Control Exposure during in Vitro Tests 2.3.2. O-Ring Samples. O-Rings were removed from the experimental set-ups as required, and the surfaces were thoroughly wiped using lint-free tissue. PAHs in the silicone were extracted by adding 10 mL of methanol (20 mL for the PAH release experiment), leaving to stand for 24 h and analyzing an aliquot of the methanol using HPLC. This extraction was quantitative; a second extraction using 5 mL of fresh methanol yielded less than 1.5% of the first extraction. 2.3.3. HPLC Analysis. PAH analysis was by HPLC with fluorescence detection (Agilent 1100 HPLC equipped with a G1321A FLD operated at Ex, 260 nm, and Em, 350, 420, 440, and 500 nm). A volume of 30 µL sample was injected at 28 °C, and the PAHs were separated on a CP-Ecospher 4 PAH column (Varian Inc., Palo Alto, CA), operated at a flow rate of 0.5 mL min-1. Methanol and water were used as the mobile phase: 50% methanol between t ) 0 and 2 min, linear gradient from 50 to 75% methanol between t ) 2 and 7 min, linear gradient from 75 to 100% methanol between t ) 7 and 35 min, and 100% methanol from 35 to 50 min. PAH concentrations in the samples were quantified using a nine-point external standard calibration curve. Signal integration was done using the HP Chemstation software (B.03.01, Agilent Technologies, Palo Alto, CA). 2.4. Characterization of Passive Dosing Performance. 2.4.1. Kinetics and Reproducibility of O-Ring Loading with PAHs. A methanol loading solution containing the 11 PAHs listed above was prepared, each at a concentration of approximately 45 mg L-1. Three batches, each of 12 cleaned and dried O-rings, were added to separate 60 mL volumes of the loading solution. These were kept static at 21 °C, and after 0, 0.5, 1, 2, 4, 8, 16, 32, 64, and 128 h, one O-ring was removed from each loading batch (n ) 3). The O-ring-sorbed PAHs were immediately extracted and analyzed as detailed above. To discount any permanent binding of PAHs to the silicone or losses during storage, O-rings were loaded for 128 h and then removed, cleaned, and stored at 21 °C for either 30 or 60 days in sealed 10 mL autosampler vials (each n ) 3). These were extracted and analyzed as above. O-Rings that had only been through the cleaning procedure above were extracted as blanks, with measured PAHs below the instrument detection limits. 2.4.2. Release into Water or Medium in Cell Culture Plates. Separate batches of O-rings were each loaded from methanol suspensions of naphthalene, phenanthrene, anthracene, fluoranthene, pyrene, and benzo(a)pyrene as detailed above. Sufficient wells for the destructive sampling of triplicate wells per time point were filled with 1 mL of either Milli-Q water or cell culture medium, and the cell culture plates were placed in an oven at 37 °C for 1 h to allow them to come to temperature. Using a separate cell culture plate for each PAH, a single loaded O-ring was added to each well, and the plates were placed in an oven at 37 °C on an orbital shaker operating at 200 rpm. At 0, 0.33, 0.66, 1.33, 2.66, 5.33, 12, 24, 48, and 72 h, samples (n ) 3) of either water or medium were taken from separate wells and analyzed. To study total PAH losses during passive dosing, the concentrations of PAHs in the O-rings (n ) 3) were determined immediately after loading and after 22 and 72 h of passive dosing in the cell culture plates. To study PAH losses from the water or medium after termination of passive dosing, after 22 h of dosing, O-rings were removed from a number of wells. Samples of water or medium were then taken after 2, 4, 6, 8, 12, and 24 h and analyzed. The above experiment was also repeated under static conditions at 37 °C using phenanthrene as an example PAH, to study the release and loss kinetics in water in the absence of additional mixing. 2.4.3. Equilibrium Water Concentrations at 37 °C. Separate batches of O-rings were each loaded from methanol suspensions of naphthalene, phenanthrene, anthracene, fluoranthene, pyrene, or benzo(a)pyrene as above. A further batch of O-rings was loaded using the methanol solution containing the 11 PAHs already described. The loaded O-rings were added to 10 mL (naphthalene, 1 mL) of Milli-Q water in glass vials closed with Teflon-lined lids and placed in an oven at 37 °C for 72 h, and a sample of water taken and analyzed. These concentrations represent the equilibrium partitioning concentrations in water at the temperature of the in
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 57 vitro tests and were used as a reference to compare the measured concentrations in the plastic well plates. Finally, the O-rings were extracted and analyzed as described above. 2.5. Application of Passive Dosing in in Vitro Cell Culture Assays. The feasibility and practicality of using the silicone O-ring passive dosing format for in vitro testing was investigated using three different cell culture assays measuring the formation of ROS by PBMCs, IL-8 promoter induction in A549 cells, and the secretion of MCP-1 by THP-1 cells, respectively. These assays have already been established and validated in our laboratory (26-28). They use different cell types, including primary cells, cell lines, suspensions of cells, and adhering cells, thus allowing the versatility of the passive dosing method to be studied. Initially, the viability of all cell types in the presence of PAHs at their aqueous solubilities using passive dosing with the O-rings was determined up to a total incubation time of 72 h. This was done using the CellTiterBlue test according to the manufacturer’s instructions. Subsequently, the end points in the different assays were measured upon exposure of the cells to individual PAH compounds at their aqueous solubilities using passive dosing. Testing at aqueous solubility is equivalent to testing at the compound maximum chemical activity (amax). The equilibrium partitioning of PAHs into cells and their membranes is a spontaneous process driven by the partial molar free energy and can be quantified by the compound chemical activity (29). Equilibrium between the cell and the surrounding medium is reached when they have equal chemical activities. Batches of O-rings were each loaded to saturation with one of the 10 PAHs listed above [dibenzo(a,h)anthracene not tested] using methanol suspensions of the respective compound. Loading the O-ring silicone to saturation results in aqueous solubility concentrations in the assays. Control O-rings were treated in exactly the same manner except that pure methanol was used during the loading step. Thereafter, the loaded and control O-rings were shipped to the University of Salzburg, Austria, where the cell culture assays were carried out. During initial testing, it was observed that crosscontamination of adjacent wells by the lighter PAHs occurred. Therefore, the different PAHs were each tested on a separate cell culture plate. 2.5.1. Formation of ROS by PBMCs. This is a short assay, that is, a 2 h incubation of the cells in the presence of the test substance. Therefore, the cell culture medium was pre-equilibrated with the O-rings before adding the cells to start the assay. A single O-ring was placed in each well with 900 µL of cell culture medium and incubated overnight at 37 °C and 5% CO2 to allow PAH equilibrium to be attained with the medium. Following standard procedures, PBMCs were isolated from the buffy coats of healthy blood donors (kindly provided in an anonymous manner by the blood services of the general hospital in Salzburg) by density centrifugation using Ficoll-Pague Plus. Samples were added to cell culture medium to a final concentration of 10 million cells mL-1, and 100 µL of this freshly prepared cell suspension was added to each well containing the pre-equilibrated medium. The cells were then incubated with the O-rings still present for 2 h at 37 °C and 5% CO2. After this, the formation of ROS was determined via the production of fluorescent dye (28), with the carboxy-DCFDA assay being performed according to the manufacturer’s instructions. Briefly, a stock solution of 10 mM carboxy-DCFDA was prepared in DMSO and stored in the dark at -20 °C. From this, a second stock solution with a concentration of 1 mM was prepared in phosphate-buffered saline (PBS) immediately prior to use, and 10 µL was added to each well. The plates were kept static at 37 °C and 5% CO2 for an additional hour, before the fluorescence of 10,000 cells was analyzed using flow cytometry on a Becton Dickinson FACSCanto. 2.5.2. IL-8 Promoter Induction in A549 Cells. This reporter gene assay with a panel of stably transfected cell lines has been previously used for testing PAH-induced immunomodulatory effects (26). To investigate the applicability of passive dosing with silicone O-rings in this assay, stably IL-8 promoter transfected A549 cells were plated out at a density of 5 × 104 cells mL-1 in 24-well plates and left at 37 °C and 5% CO2 overnight to allow the cells to adhere
58
Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Smith et al.
and obtain their normal morphology. To each well containing the cells, one O-ring loaded to saturation with a single PAH was added, and the plates were incubated at 37 °C and 5% CO2 for 24 h. Thereafter, the luciferase assay was performed by analyzing the relative light units (RLU) using a TECAN plate reader as previously described (26). 2.5.3. Secretion of MCP-1 by THP-1 Cells. Another established assay to determine immunomodulatory effects of chemicals is the detection of cytokines secreted by cells in culture. Recent observations in our laboratory have shown that the secretion of MCP-1 is a reliable marker for the analysis of such effects (27). THP-1 cells were plated out at a density of 1 × 106 cells mL-1 in 24-well plates, and one O-ring loaded to saturation with a single PAH was immediately added to each well. These were then incubated for 24 h at 37 °C and 5% CO2. After this, the secretion of MCP-1 by THP-1 cells was determined using MCP-1-specific antibodies according to the manufacturer’s instructions and described in ref 27. Experiments performed on a single day were done in duplicate; that is, each PAH tested on a separate well plate with two wells each dosed with a single PAH-loaded O-ring. Experiments were also repeated at least once on a different day. After assay completion, exposure was confirmed by equilibrating the O-rings with pure water, giving measurements of the freely dissolved concentrations at the end of the test. The O-rings were rinsed with water and cleaned using lint-free tissue, before adding 1 mL of Milli-Q water and leaving overnight in closed autosampler vials at 37 °C. The equilibrium freely dissolved concentrations were then measured. Loading the O-rings to saturation gives aqueous concentrations close to solubility when passive dosing; therefore, the measurements were compared to PAH aqueous solubilities from the literature (30). 2.6. Data Treatment. The O-ring loading data were analyzed using a one-compartment model (eq 1), where t (h) is time, Csilicone(t) (mg L-1) is the silicone concentration at time t, Csilicone(eq) (mg L-1) is the equilibrium silicone concentration, and kloading (h-1) is the loading rate constant.
Csilicone(t) ) Csilicone(eq)(1 - exp-kloadingt)
(1)
The release of PAHs into water or medium was analyzed in the same way (eq 2), where Cwater(t) (mg L-1) is the water concentration at time t, Cwater(eq) (mg L-1) is the equilibrium water concentration, and krelease (h-1) is the release rate constant. In the experiments with cell culture medium, the total medium concentrations, Cmedium (mg L-1), were used in place of Cwater.
Cwater(t) ) Cwater(eq)(1 - exp-kreleaset)
(2)
Data were fitted by the least-squares method using Graphpad Prizm 5 (San Diego, CA), giving estimates of the appropriate rate constant and equilibrium concentrations. As a quantitative measure of the time to reach equilibrium, the time to reach 95% steady state was used (eq 3). Here, t95%ss (h) is the time to reach 95% steady state, and k is either the loading or the release rate constant as specified above.
t95%ss )
3 k
(3)
The equilibrium PAH distribution between methanol and silicone is described using the dimensionless partitioning ratio KMeOH:silicone (eq 4). Methanol and silicone PAH concentrations (mg L-1) were from the O-ring loading experiment described above.
KMeOH:silicone )
CMeOH Csilicone(eq)
(4)
The equilibrium PAH distribution between silicone and water is described using the dimensionless partitioning ratio Ksilicone:water (eq
5). Silicone and water PAH concentrations (mg L-1) were from the 37 °C equilibrium experiments described above.
Ksilicone:water )
Csilicone(eq) Cwater(eq)
(5)
The distribution between the methanol and the water, KMeOH:water, was calculated by multiplying the above KMeOH:silicone and Ksilicone:water (eq 6).
KMeOH:water ) KMeOH:silicone × Ksilicone:water
(6)
Because of the miscibility of methanol with water, KMeOH:water is not directly measurable and thus calculated. They are useful to link the concentrations in the methanol loading solution to the freely dissolved concentration in the test. Passive dosing was applied in the present study to control freely dissolved concentrations at the aqueous solubility limit of the PAHs, which is equivalent to the testing at their amax (13). The amax of a PAH is numerically identical to its fugacity ratio and can be estimated at the temperature of interest (T, K) from its reported melting point (Tm, K), assuming the entropy of melting to be 56 J mol-1 K-1, that is, Walden’s rule (13, 31):
amax ) exp[(6.8)(1 - Tm /T)]
(7)
It has been reported that narcosis or baseline toxicity of organic chemicals is related to their chemical activity (32-34) and that high melting point chemicals cannot provide a high enough amax to exert baseline toxicity (13, 35). The observed responses in the in vitro tests were therefore also plotted against the melting point temperatures as well as to the amax of the PAHs.
3. Results and Discussion 3.1. O-Ring Loading. The loading of PAHs from methanol into the silicone rings was completed within less than 3 h for all 11 tested PAHs (Table 1). Figure 1 shows how the concentrations of three representative PAHs increased in the O-rings during the loading. The time to attain equilibrium increased by a factor of approximately two from naphthalene to dibenzo(a,h)anthracene, in line with increasing PAH molecular weight. After equilibrium was reached, the concentrations remained highly constant over a period of up to 128 h (5 days). The relative standard deviations for the final equilibrium silicone concentrations (t ) 4-128 h, n ) 18) were less than 5% for all PAHs [range, 2.2% for acenaphthene to 4.9% for benzo(a)pyrene]. This attests to both the precision of the loading protocol and a homogeneous silicone matrix without regions with slower sorption kinetics. When loaded O-rings were stored for periods of 30 and 60 days, the silicone concentrations were the same as those measured immediately after loading (data not shown). This indicates that during prolonged storage, permanent binding to the silicone matrix or losses did not occur. This has practical implications; a large batch of rings can be prepared in one effort and stored for subsequent use. Moreover, several in vitro experiments can be carried out over a prolonged period of time using a batch of O-rings loaded at the same time and with exactly the same characteristics. 3.2. PAH Release into Water and Medium. The success of passive dosing in producing defined and constant exposure is dependent on (i) the release from the silicone being sufficiently fast to compensate system losses so that equilibrium partitioning prevails and (ii) the capacity of the silicone HOC reservoir being sufficiently large to buffer any losses. 3.2.1. Release. Figure 2A,B shows the release from the O-rings into Milli-Q water of naphthalene and phenanthrene,
PassiVe Dosing To Control Exposure during in Vitro Tests
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 59
Table 1. PAH Loading and Release Kinetics of Silicone O-Ringsa loading from methanol
naphthalene acenaphthene flourene phenanthrene anthracene fluoranthene pyrene benz(a)anthracene chrysene benzo(a)pyrene dibenzo(a,h)anthracene
rate constant (h-1) 2.38 1.80 1.89 1.66 1.82 1.41 1.35 1.38 1.37 1.32 1.10
release into water
SE
time to 95% steady state (h)
SE
0.098 0.059 0.074 0.073 0.081 0.060 0.074 0.066 0.069 0.085 0.064
1.26 1.67 1.59 1.81 1.65 2.13 2.22 2.17 2.20 2.28 2.74
0.052 0.055 0.063 0.080 0.074 0.091 0.122 0.103 0.111 0.147 0.161
rate constant (h-1)
SE
time to 95% steady state (h)
4.77b
0.402
0.63
2.20 2.44 1.80 1.83
0.138 0.233 0.141 0.170
0.35c
0.071
release into medium
SE
rate constant (h-1)
SE
time to 95% steady state (h)
0.053
3.55b
SE
0.223
0.85
0.053
1.37 1.23 1.66 1.64
0.086 0.117 0.131 0.152
1.32 2.30 0.93 1.04
0.108 0.367 0.075 0.076
2.27 1.31 3.22 2.89
0.186 0.209 0.259 0.213
8.69
1.798
0.57
0.035
5.28
0.324
a The loading of 11 PAHs from excess methanol solution containing the individual compounds at 45 mg/L was under static conditions at 21°C. The release of six PAHs first loaded to amax was into 1 mL of Milli-Q water or medium in 24-well cell culture plates under shaking at 200 rpm at 37°C. The time to 95% steady state was calculated as 3/rate constant. b Data after 3 h was not included because of the decrease in silicone concentrations. c The final 72 h sampling point not included as well above aqueous solubility.
Figure 1. Loading of the silicone O-rings with three representative PAHs from a methanol solution containing each compound at a concentration of 45 mg/L. Each point represents the mean concentration plus standard deviation from triplicate O-rings. The dashed line shows the fit of the one-compartment model used for calculating the rate constants.
respectively. These were representative of the behavior of the other PAHs, with release being sufficiently rapid that maximum concentrations were reached within less than 3 hours for all compounds except benzo(a)pyrene (see Table 1). The release of phenanthrene under static conditions was similarly rapid; maximum concentrations were reached in just over 3 hours (t95%ss ) 3.38 h). This is sufficiently rapid for in vitro assays, and if required, shaking can be used to further enhance the release kinetics. When using cell culture medium, significant sorption of hydrophobic test compounds to the medium constituents occurs, which can lead to altered release kinetics. On the one hand, a greater mass of chemical needs to partition into the medium; on the other hand, the medium constituents can lead to an enhanced diffusive transport into the medium (14, 19, 36, 37). Because both processes depend variably on the compound properties, release of the six PAHs into the RPMI cell culture medium was also measured. As a representative example, the release of phenanthrene into medium is shown in Figure 2C, with the release rate constants and time to reach equilibrium summarized for all compounds in Table 1. As compared to water, release into medium was marginally slower, and stable medium concentrations were still reached within about 2-3 hours for all compounds, with the exception of benzo(a)pyrene, which had a t95%ss of 5.28 h. The release of benzo(a)pyrene into Milli-Q water required almost 9 h to attain stable concentrations (Table 1). The water concentrations of benzo(a)pyrene increased throughout the test to around three times the independently measured equilibrium partitioning concentrations. Measurements above the equilibrium
Figure 2. Release and loss kinetics of (A) naphthalene into 1 mL of water, (B) phenanthrene into 1 mL of water, and (C) phenanthrene into 1 mL of cell culture medium, contained in 24-well plates and maintained at 200 rpm and 37 °C. Each point represents the mean concentration plus standard deviation from three replicate wells. In the graphs with water, the bold dashed line shows the independently measured equilibrium partitioning concentrations in water at 37 °C.
level are likely due to association of benzo(a)pyrene with an additional sorbing phase, perhaps leached from the wall material. This does, however, illustrate an important benefit of passive dosing. Because passive dosing is based on partitioning, the freely dissolved concentrations determining exposure will be
60
Chem. Res. Toxicol., Vol. 23, No. 1, 2010
equal to the equilibrium partitioning levels, despite the observed higher total aqueous concentrations. 3.2.2. Equilibrium Partitioning. The concentrations from the separate equilibrium water concentration experiment were compared to the final stable aqueous concentrations measured in the cell culture plates. Importantly, this allows equilibrium partitioning to be distinguished from steady state, both of which are characterized by stable aqueous concentrations. A close correspondence between the two indicates that equilibrium partitioning applies. Final stable water concentrations in the wells were estimated from the fitting of the data (n ) 30) with eq 2. For naphthalene, only data up to 3 h were used (n ) 15), since concentrations decreased after this time due to volatilization or other losses. For benzo(a)pyrene, the 72 h sample was not included (n ) 27), which markedly exceeded the water solubility of benzo(a)pyrene. Final stable water concentrations in the water release experiment ranged from 78% for anthracene to 136% for benzo(a)pyrene of the independently measured equilibrium water concentrations in the vials. Therefore, for all PAHs, equilibrium partitioning between the freely dissolved and the O-ring silicone PAHs was approached. This has the important advantage that the freely dissolved concentrations can be reliably predicted a priori based on knowledge of the silicone to water partitioning behavior of the test compounds. 3.2.3. Stability of the Equilibrium Concentrations. The equilibrium water and medium concentrations were maintained for the whole 72 h duration of the test for all compounds except naphthalene (data not shown). For phenanthrene, anthracene, fluoranthene, and pyrene, the coefficients of variation for the water concentrations after attainment of equilibrium, between 4 and 72 h (n ) 18), were less than 5%. Given that these measurements were from different wells, each dosed with a separate O-ring, this demonstrates the precision of the passive dosing procedure for establishing constant exposure concentrations. For benzo(a)pyrene, the coefficient of variation was slightly higher at 18% due to the sorption artifact. Naphthalene was anomalous, characterized by a rapid release into water with the concentrations reaching 85% of equilibrium, indicating that equilibrium partitioning initially applied (Figure 2A). After about 3 h, a reduction in the water concentrations was observed. This is due to the high volatility of this compound at 37 °C, leading to rapid losses exceeding the buffering capacity of the rings. This was corroborated by the measured parallel decrease in the O-ring silicone concentrations (Figure S1 of the Supporting Information). For the remaining PAHs, the O-ring silicone concentrations remained unchanged throughout the 72 h (Figure S1 of the Supporting Information). 3.2.4. Losses in the Absence of Passive Dosing. The loss of naphthalene and phenanthrene from wells with water after removal of the O-rings was rapid as shown in Figure 2. For naphthalene and phenanthrene, the loss was substantial, with over 70% of the compound being lost from the solution within 24 h. The losses of the remaining PAHs from water ranged from 74 (anthracene) to 56% [pyrene and benzo(a)pyrene]. Note that higher losses can be expected when HOCs are introduced using solvent, because additional sorption to the plastic well plates can be expected in situations without a pre-exposure period. Also shown is the more moderate loss of phenanthrene from medium after removal of the O-rings (Figure 2C). PAH losses from medium ranged from 62% (naphthalene) to insignificant [fluoranthene, pyrene, and benzo(a)pyrene], which can be explained by PAH-equilibrated medium constituents functioning as an additional reservoir and buffering HOC losses. In practical terms, pre-equilibrating the medium prior to adding biological
Smith et al.
Figure 3. Freely dissolved fractions of the total PAH in cell culture medium at 37 °C are shown in panel A. These were calculated using the equilibrium PAH medium and water concentrations from the release experiments. Confirmation of exposure using duplicate O-rings from the IL-8 promoter assay as an example is shown in panel B. After completion of the assay, the O-rings were cleaned and equilibrated with 1 mL of pure water at 37 °C, and the PAH water concentrations were measured. Chrysene freely dissolved concentrations were below detection limits. The mean concentrations together with their standard deviations are plotted against the PAH solubilities taken from the literature (30).
material provides an efficient way of buffering the freely dissolved phase when, for example, adding the cells. This was done for the ROS cellular assay below. 3.2.5. Freely Dissolved Fraction in the Cell Culture Medium. Cell culture medium constituents can significantly sorb hydrophobic organic chemicals (36, 37). Freely dissolved fractions in the cell culture medium were thus determined as the ratio between measured concentrations in pure water (Cfree) and culture medium (Ctotal), both after equilibration with the silicone O-rings. These are plotted in Figure 3A against the logarithm of the octanol:water partitioning ratio (KOW) calculated using the SPARC online calculator (38). Freely dissolved fractions ranged from 30% for naphthalene down to about 0.1% for benzo(a)pyrene. This is in agreement with other studies investigating the binding of similar HOCs in solutions with high protein and lipid contents (36, 37) and highlights one of the most important features of passive dosing. Because the freely dissolved PAH concentrations are controlled by equilibrium partitioning from the silicone, these are unchanged by the presence of sorption in the cell culture medium. Furthermore, they can be predicted in advance based on knowledge of the silicone:water or methanol:water partitioning. In contrast, when spiking PAHs via a cosolvent, only between 0.1 and 30% of
PassiVe Dosing To Control Exposure during in Vitro Tests
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 61
Table 2. PAH Equilibrium Partitioning Ratios for Methanol:Silicone (KMeOH:silicone, L L-1), Silicone:Water (Ksilicone:water; L L-1), and Methanol:Water (KMeOH:water; L L-1) silicone:water methanol:silicone
naphthalene acenaphthene flourene phenanthrene anthracene fluoranthene pyrene benz(a)anthracene chrysene benzo(a)pyrene dibenzo(a,h)anthracene
KMeOH:silicone (L L-1)
SE
2.78 2.06 2.75 3.68 3.47 3.97 3.53 4.56 4.70 4.47 5.80
0.047 0.037 0.049 0.071 0.061 0.080 0.069 0.083 0.086 0.089 0.117
below saturation
Log Ksilicone:water KMeOH:silicone (L L-1) 0.44 0.31 0.44 0.57 0.54 0.60 0.55 0.66 0.67 0.65 0.76
711 2565 2820 3302 6320
SE 8.3 57.2 88.9 32.9 109.7
the total concentrations would be in the freely dissolved form. This has two implications. First, without detailed knowledge of the partitioning to and the amount of medium constituents, the effective exposure concentrations are unknown. Second, because only a small fraction of the more hydrophobic compounds in fact is freely dissolved, this leads to a significant reduction in the apparent test sensitivity. This lack of sensitivity has been observed in in vitro assays as compared to in vivo tests (8, 10, 39). The above results can be summarized as follows: • Equilibrium partitioning concentrations were attained within hours for water and cell culture medium. • The concentrations remained highly stable for over 72 h (apart from naphthalene). • Freely dissolved concentrations at equilibrium are proportional to the concentrations in the loading solution (KMeOH:water) and the silicone O-rings (Ksilicone:water) and can be varied between saturation (for limit testing) down to lower levels (for dose-response testing). The O-rings can be applied to in vitro tests in various ways. For toxicity assays lasting more than a few hours, the O-rings can simply be added to the wells at the beginning of the test, and exposure concentrations will then build up at the beginning of the test. This strategy would generally require agitation of the plates, which speeds up the release from the O-rings. In many situations, it might be better to establish exposure concentrations before adding the biological material, and this can be done by placing the O-rings into wells containing medium a day before the actual test. If confluent cell layers are required, the medium can be pre-equilibrated by passive dosing in separate vessels and then added to the wells containing the pregrown cells at the same time as the loaded O-rings. For more volatile test substances, such as naphthalene, the decrease in the silicone concentrations can be overcome by simply replacing the O-rings at suitable intervals. 3.2.6. Partitioning between Methanol-Silicone-Water. On the basis of knowledge of the partitioning behavior of the test substances, the exposure concentrations can be calculated. Therefore, the data from the loading and release experiments were used to calculate methanol:silicone (KMeOH:silicone), silicone: water (Ksilicone:water), and methanol:water (KMeOH:water) partitioning ratios using eqs 4-6 (Table 2). The Ksilicone:water values at 37 °C increased linearly with PAH hydrophobicity (log KOW ) 0.834 log Ksilicone:water - 0.027; r2 ) 0.982), as has also been observed previously for silicone from other manufacturers (40, 41). The Ksilicone:water values for rings loaded to below saturation and at saturation showed a very close correspondence (Table 2), which confirms that loading from a saturated methanol solution was
at saturation
Log Ksilicone:water Ksilicone:water (L L-1) 2.85 3.41 3.45 3.52 3.80
SE
methanol:water
Log KMeOH:water Ksilicone:water (L L-1)
704
15.9
2.85
5155 7240 15986 17167
176.9 300.3 413.2 434.2
3.71 3.86 4.20 4.23
1954 5288 7746 18983 25133 63459 60607
168002
7591.6
5.23
974851
SE
Log KMeOH:water
55.0 151.8 280.9 748.0 1133.1 2076.5 1939.7
3.29 3.72 3.89 4.28 4.40 4.80 4.78
48230.7
5.99
governed by simple partitioning. For the calculated methanol to water partitioning ratios at 37 °C, KMeOH:water increased linearly with PAH hydrophobicity (log KOW ) 0.940 log KMeOH:water + 0.038; r2 ) 0.978), and they are useful for calculating the equilibrium freely dissolved concentrations in the tests from known methanol loading concentrations (nominal or measured). This might circumvent the need to analyze the often very low freely dissolved concentrations, which can be hard to measure in small volume samples. 3.3. In Vitro Cell Culture Assays. To investigate the applicability of passive dosing for use in in vitro cell culture assays, three different assays were chosen that have already been established and validated in our laboratory (26-28). Viability measurements of the different cell types in the presence of the 10 PAHs at their aqueous solubility when using passive dosing showed no cytotoxicity for any of the PAHs in any the different cell types (data not shown). Therefore, the effects observed below on ROS formation, IL-8 promoter induction, or MCP-1 secretion were not due to cytotoxicity. Exposure was confirmed at test completion by equilibrating the O-rings with a small volume (
