Environ. Sci. Technol. 2001, 35, 4097-4102
Partition Controlled Delivery of Hydrophobic Substances in Toxicity Tests Using Poly(dimethylsiloxane) (PDMS) Films R . S T E P H E N B R O W N , * ,† PARVEEN AKHTAR,† JOHAN ÅKERMAN,‡ LAURA HAMPEL,† IGOR S. KOZIN,† LEEN A. VILLERIUS,§ AND H A N S J . C . K L A M E R * ,§ Fiber-Optic Environmental Sensors Group, School of Environmental Studies, Queen’s University, Kingston, Ontario, Canada, K7L 3N6, AquaSense, P.O. Box 95125, 1090 HC Amsterdam, The Netherlands, and National Institute for Coastal and Marine Management/RIKZ, Ministry of Transport and Public Works, P.O. Box 207, 9750 AE Haren, The Netherlands
Interpretation of toxicity test results may be hampered when doubt exists about the actual exposure concentration. Processes that are responsible for differences between the nominal and the actual concentration in aqueous test systems may include sorption, precipitation, volatilization, chemical and biological degradation, and uptake into biological or test tissue. In this study, the use of a poly(dimethylsiloxane) (PDMS) film containing the test compound is introduced as a versatile technique for partition controlled delivery of hydrophobic compounds to aqueous toxicity tests. Two methods developed produced preloaded films, having toxicant added to the PDMS prepolymer solution before film deposition and curing, and postloaded films, which are created by the addition of toxicant in a solvent to an alreadypolymerized PDMS film. Preloaded films were generally more easily prepared, may better accommodate larger molecules, and have a higher capacity than postloaded films. Postloaded films provided film-solution partition coefficients with higher precision and allowed for the use of films from stock and thus for a more portable technique. Chemical analysis showed that equilibrium between films and the aqueous solution was established within 7-10 min and was maintained for a suite of aromatic compounds (log Kow ranging from 2.8 to 6.1). The reliability of the film technique was demonstrated by application to the Microtox bacterial toxicity tests of solutions of polycyclic aromatic hydrocarbons (PAHs).
Introduction Many aqueous tests (bioassays) are conducted using nominal toxicant concentrations. When testing compounds that are water-soluble and nonvolatile, the dose may be expressed in * Corresponding authors phone: (613)533-2655; fax: (613)5336669; e-mail:
[email protected] (Brown) and e-mail:
[email protected] (Klamer). † Queen’s University. ‡ AquaSense. § National Institute for Coastal and Marine Management/RIKZ, Ministry of Transport and Public Works. 10.1021/es010708t CCC: $20.00 Published on Web 09/08/2001
2001 American Chemical Society
FIGURE 1. Visualization of a test vial containing contaminant-loaded PDMS film. Partitioning from the film compensates for loss processes thus controlling the truly dissolved concentration. (actual) concentration units, and interpretation of doseresponse curves is generally straightforward (1). When compounds are sparingly soluble (hydrophobic) and/or volatile, however, there may be significant differences between nominal and actual concentrations. The dominant mechanism for contaminant uptake by aquatic organisms will vary between organisms and compounds, either directly from the aqueous phase via absorption through membranes, by uptake from a particle phase at the respiratory surface, or by ingestion with food (2-4). This will depend on various factors, including compound solubility and hydrophobicity. For subcellular components or microorganisms such as those used in aqueous toxicity tests, however, the freely dissolved aqueous concentration of test compounds is the primary contributor to the intensity of the response (5). Nevertheless, toxicity tests with effect concentrations up to several orders of magnitude above the aqueous solubility of the test compounds have been published (6-8). In these tests, difficulties may arise in maintaining known and dissolved concentrations, especially since the solutions must either be supersaturated or actually be suspensions. As the actual dissolved concentration will deviate from the nominal concentration (calculated as total mass added/total volume), the toxicity of the compounds or samples under investigation will be both underestimated and unreliable. The “need” to have excess contaminant to see effects in some test procedures is explained by many loss mechanisms occurring after addition to the aqueous solution (see Figure 1). A nominal concentration greater than the solubility must initially be added to compensate for the losses, i.e., the excess material is acting as a “reservoir” to keep delivering the dissolved compound (8). Obtaining relevant dose-response curves and effect concentration values would be dramatically aided by finding an alternate delivery scheme, whereby the toxicant would be present at meaningful and systematically varied concentrations throughout toxicity testing. Our approach is to use partition of toxicants from polymer films to do just that. Partition of contaminant from a polymer film will not generate a toxic response for compounds where effect concentrations are above solubility, even after accounting for losses, since partition cannot produce concentrations above the solubility. Where toxic responses are observed, effect concentrations will be known since a controlled concentration below the aqueous solubility for the compound is produced and maintained by partition from a film reservoir. Effect concentration values are expected to be below those VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4097
previously reported using nominal concentrations in cases where significant contaminant losses occurred in the conventional test. Partition Controlled Delivery. Using a partition-controlled (i.e. equilibrium) delivery procedure of toxicant from a reservoir was recently reported (9). In this example, the solid phase was a C-18 extraction disk, where the C-18 film is on particles trapped in a fluoropolymer mesh (i.e. an Empore disk). A solution of the contaminant(s) in hexane was added to the C-18 material, and the solvent was allowed to evaporate. Stable concentrations up to aqueous solubility could be established for nonpolar halogenated aromatic compounds. Preliminary algae toxicity tests were apparently successfully completed; however, problems with release of nondissolved toxicants from the C-18 disks complicated the evaluation of the test results (9). In this paper, we present an alternative method for toxicant delivery wherein a poly(dimethylsiloxane) (PDMS, silicone rubber) polymer film containing the analyte compound is applied to the sample vessel, and the analyte desorbs from the film into the solution (Figure 1). Two different methods of film/analyte preparation are discussed. Preloaded films have toxicant added to the PDMS prepolymer solution in hexane before film deposition, thus the toxicant is incorporated into the film as it polymerizes. Postloaded films, on the other hand, are created by the addition of toxicant in solution to an already-polymerized PDMS film. Parameters that determine the success of the method include the capacity of the films to compensate for losses, rapidly establishing equilibrium, and applicability to mixtures as well as pure compounds. Film/water partition coefficients have been determined for a variety of compounds, including polycyclic aromatic hydrocarbons (PAHs), both as pure analytes as well as in mixtures. The kinetics of dissolution of analytes from the film were studied as a function of temperature and time. Tests were carried out in glass cuvettes and polystyrene microplates to anticipate usage of the procedure in different test configurations. Microtox broadspectrum bacterial toxicity tests were run with contaminant solutions prepared using both preloaded and postloaded films. Response data are presented and compared with the literature to demonstrate the shift to lower effect concentrations for compounds that have reported EC50 values above their aqueous solubility.
Materials and Methods Chemicals and Solvents. PAHs naphthalene (Nap), fluorene (Flu), anthracene (Ant), phenanthrene (Phe), fluoranthene (Fla), pyrene (Pyr), and benzo[a]pyrene (Bap) were obtained from different suppliers at 99+% purity. 1-Chloroanthracene, 2-aminoanthracene, 9-anthracenecarbonitrile, and 9-methanolanthracene were obtained from Sigma-Aldrich (Breda, The Netherlands) at maximal available purities; a 16 PAH standard was obtained from Supelco (#48743, Mississauga, Canada). HPLC grade hexane, methanol, and acetonitrile and spectroscopic grade ethanol were obtained from Fisher Scientific, (Ottawa, Canada) and J. T. Baker (Deventer, The Netherlands). Test water was double-distilled water or prepared using a Waters Milli-Q system. PDMS prepolymer (“Silicone” single component preparation, GE Silicones, Waterford, NY) was purchased from a local hardware store. Stock solutions were prepared either in methanol, ethanol, or hexane. Film Preparation. 96-Well white polystyrene microplates (350 µL/well, Perkin-Elmer Life Science/Wallac) and 4-mL glass cuvettes (Azur Environmental) were used for film deposition. A 6 mg/mL poly(dimethylsiloxane) (PDMS) solution was prepared by dissolving PDMS prepolymer in hexane. Preloaded films were prepared by adding a known concentration of compound in hexane to the PDMS solution 4098
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
before film formation. Films were prepared by depositing from 100 µL to 800 µL of PDMS solution into the cuvette or microplate well and allowing the hexane to evaporate and then the film to cure for at least 3 h. Volumes greater than 300 µL in microplate wells were prepared by consecutive addition of multiple aliquots of 200 µL each. Postloaded films were prepared by placing toxicant dissolved in hexane over a cured blank film. The film swelled due to the presence of the solvent, which subsequently evaporated leaving the toxicant compound in the film. For both types of films, the maximum amount of toxicant which could be added was determined either by a visible loss of film integrity or the presence of particles in the equilibrated solution. Postloaded films were noted to have lower total capacity than preloaded films. Analytical Procedures. PAHs were analyzed using fluorescence detection on three different systems. Analysis of individual PAHs from microplate-tests used a fluorescence spectrometer (QuantaMaster QM1, Photon Technology International, London, ON, Canada). Excitation and emission slits were set to 2 nm band-pass for all measurements. Synchronous Fluorescence Scanning (SFS) was used, where excitation and emission wavelengths were scanned simultaneously with a fixed offset. Samples were diluted 50:50 (v:v) with ethanol, and Ant and Phe were detected with offset values of 122 and 53 nm, respectively. Additionally, PAHs in mixtures from microplate tests were analyzed using a Varian HPLC system (consisting of a ProStar 240 quaternary pump, 430 autosampler, 360 Fluorescence detector (FLD), 330 Photodiode Array detector, and a 150 × 4.6 mm HewlettPackard-Zorbax C18 column). PAHs from cuvette tests were analyzed using a Hewlett-Packard HPLC system (consisting of a HP1050 system equipped with a Waters SymmetryShield 150 × 3.9 mm C18 column operating at 22 °C, coupled to a HP 1050 diode-array and HP 1046A fluorescence detector). Depending on the compound(s) present, either isocratic or gradient elution was used. Desorption of Compounds from Films. Test compounds in films were equilibrated against water or Microtox diluent on an orbital shaker (700 or 900 rpm) in the dark, at ambient temperature, or at 15 °C (WTB Binder KB115 incubator, Fisher Scientific, The Netherlands), respectively. Determination of Partition Kinetics and Film Stability. The time required to establish equilibrium between film and aqueous Phe and Ant concentrations was investigated at different temperatures, using films in microplates and glass cuvettes. Preloaded and postloaded films were prepared. Water or diluent was added, and the test vials were shaken at 700 or 900 rpm in the dark. Solution concentrations were determined, either at constant temperature (15 °C) or at intervals of 5 °C between 5 °C and 35 °C. Approximately 500 µL aliquots were removed at regular intervals and mixed with 500 µL of methanol; 50 µL of internal standard (9-methanolanthracene) was added with the Beeline3 automatic pipettor (Hook & Tucker Instruments, New Addington, U.K.). All analyses were carried out in triplicate. Determination of Film:Solution Partition Coefficient Kfs. Films were prepared with different test compound loadings. After equilibration, approximately 250 µL (plate) or 500 µL (cuvette) aliquots were removed and mixed with 500 µL of methanol; 50 µL internal standard (9-methanolanthracene) was added using the automatic pipettor for total aqueous volume determination. After careful removal of residual water, the film was exhaustively extracted with several aliquots of ethanol; the film’s volume was calculated using a PDMS density of 0.96 g/mL. Compound concentrations were determined (SFS or HPLC/FLD), and log Kfs was calculated from the slope of the linear part of the CS(olution) vs CF(ilm) curve. All analyses were carried out at least in duplicate.
Compensation for Substantial Analyte Loss. A PDMS “bead” was added to an equilibrated film/water Ant system and to an aqueous Ant control solution, to simulate loss processes. Dissolved Ant concentrations were determined as above, at regular intervals, and the capacity of the film to compensate the decrease in dissolved Ant was evaluated. Microtox Tests. Bacterial toxicity tests were run with slight adaptations to the standard protocols (Azur Environmental). In brief, films were prepared with different loadings and equilibrated against 500 µL of Microtox diluent (a readyto-use saline solution). Preincubated bacteria were added, and Microtox test responses were recorded after 15 and 30 min incubation at 15 °C under continuous agitation at 900 rpm. The percent response with respect to a blank was calculated using the Microtox software; EC50 values were calculated using a sigmoidal dose-response curve fitting algorithm (GraphPad Prism version 3.00 for Windows, GraphPad Software, San Diego, CA).
Results and Discussion Partition Controlled Delivery. The key parameters in partition controlled delivery of compounds to solution are the time to reach equilibrium, the equilibrium concentration achieved, and the stability of the equilibrium concentration. The time to reach equilibrium should be a few minutes or less so that solutions may be prepared and used in a convenient manner. More importantly, rapid equilibration also indicates the ability of the film to compensate dynamically for losses of compound from solution. The solution concentration, after equilibrium has been reached, should be predictable based on the concentration of the compound in the film and should be stable at this value over the course of the toxicity test. Characterizing Partition Kinetics. As was previously pointed out by Mayer et al. (9-11), partitioning of compounds from a reservoir to the aqueous phase is greatly aided by a high surface-to-volume ratio and by introducing turbulent conditions. Preparing a 5 µL film in Microtox vials yielded an S/V ratio of 5.3 × 104 m2/m3, which is comparable to the S/V ratio of 7.1 × 104 m2/m3 calculated by Mayer et al. (11) for 15 µm PDMS coating on glass-fiber (SPME). Films prepared in microplate wells have a slightly less favorable S/V ratio of 3.3 × 104 m2/m3. Turbulence can be introduced by using an orbital shaker to minimize equilibration times. In initial experiments with Bap and 2-aminofluorene, it was shown that agitation greatly facilitated rapid equilibration from our films (data not shown). Agitation was optimized by determining the time needed to reach equilibrium for Phe and Ant in both glass cuvettes and polystyrene 96 well microplates. Film release rate constant k and time to steady-state tss (the time when Ct ≈ 0.95Ceq) were calculated using the following expressions
Ct ) Ceq ‚ (1 - e-kt)
(1)
tss ) 3/k
(2)
and
where Ct and Ceq are the dissolved concentrations at time ) t and at equilibrium, respectively. Optimal shaking speeds were determined to be 900 rpm for cuvettes and 700 rpm for microplatessthe latter system requiring a reduced speed to prevent spillage. As is shown in Figure 2, stable Ant concentrations were reached in times of 7-10 min for cuvettes and 30 min for microplates, which is similar to the time required to reach 90% of the final dissolved concentration (t90%) reported by Mayer et al. (9, 10). Preloaded and postloaded films with identical Ant concentrations produced
FIGURE 2. Release of Ant from preloaded PDMS film in glass cuvettes and polystyrene microplates at optimized shaking velocities. Dissolved concentrations are normalized. Time to steady-state tss is defined as 3/k (refer to eqs 1 and 2). Symbols represent averaged values ( σ of individual samples (n ) 3). (Curves drawn to aid the eye).
FIGURE 3. Phe equilibrium concentrations in solution versus film for postloaded and preloaded films. Kfs was determined by calculation of the slope of the initial linear part of the curve. Symbols represent average values ( σ (n ) 3). The dashed line depicts the reported aqueous solubility at 25 °C (23); the shaded area depicts the estimated range of Phe solubility in Microtox diluent at 15 °C. (Curves drawn to aid the eye). identical equilibrium times. It is apparent from Figure 2 that fast desorption from the film in cuvettes leads to the common phenomenon of initial overshoot of the dissolved concentration, followed by relaxation toward equilibrium concentration. Solution concentrations in cuvettes were monitored for several films and compounds for up to 17 days, and no significant change in concentration was observed (data not shown). Partition Coefficients and Controlled Concentrations. The determination of the film:solution partition coefficient (Kfs) is straightforward: Kfs is calculated as the slope of the initial linear section of a plot of Cs versus Cf (see, e.g., Figure 3 for Phe). In our case, we define two separate Kfs terms to distinguish films partitioning in pure water (film:water partition coefficient, Kfw) from those partitioning in a toxicity test diluent solution (film:diluent partition coefficient, Kfd) but continue to use the Kfs term when referring to the two partition constants collectively. Table 1 gives average Kfw and Kfd values (n ) 3) and standard deviations for all compounds tested, indicating excellent precision in this measurement. Literature Kfs values are included for comparison. Kfd values were consistently lower than Kfw values, which is contrary to the trend expected based on salting out effects, though comparison with literature values indicates that this difference may be within the absolute error in the measurements. The results in Figure 3 demonstrate the ability of films prepared with different Phe concentrations to control the dissolved Phe concentration in Microtox diluent to a maximum of 4.1 µM. This maximum should correspond to VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4099
TABLE 1. Partition Coefficients for Each Tested Compoundk compound
abbrev
CAS no.
log Kowa
Sw [µM]
log Kfw log Kfd log Kfd log KPDMS:water postloadedb (SD) postloaded (SD) preloaded (SD) lit.
naphthalene fluorene phenanthrene
Nap Flu Phe
91-20-3 86-73-7 85-01-8
3.30 4.08 4.49
242c 11.4c 6.17c
3.96 (0.03) 4.03 (0.04) 4.20 (0.04)
anthracene
Ant
120-12-7
4.49
0.25c
fluoranthene
Fla
206-44-0
4.95
1.19c
pyrene benzo[a]pyrene acridine acridone 1-methylanthracene 1-chloroanthracene 1-aminopyrene 2-aminofluorene 9-anthracenecarbonitrile
Pyr Bap
129-00-0 50-32-8 260-94-6 578-95-0 610-48-0 4985-70-0 1606-67-3 153-78-6 1210-12-4
4.95 6.12 3.41 5.28g 4.99 5.20 3.72 2.85 3.92
0.65c 0.007g 214h i 1.28j i 2.63j 48.4j 2.01j
a
3.72 (0.01) 3.96 (0.01)
3.71 (0.08) 4.14 (0.06)
4.43 (0.11) 4.48 (0.10)f 4.54 (0.03)
4.20 (0.03)
4.11 (0.06)
4.36 (0.02)
4.53 (0.04)
4.61 (0.03)
4.44 (0.05) 5.47 (0.09)
3.17 (0.03) 4.78 (0.03) 3.27 (n ) 1) 2.82 (0.03)
4.31 (0.08) 4.36 (0.16) 3.58 (0.20)
3.02d 3.71d 3.48 (0.09),e 3.96d 3.98d 4.26 (0.20),e 4.71d 4.86d 5.39d i i i i i i i
b
Calculated using the ClogP algorithm (Biobyte, Claremont, U.S.A.). Determined in mixture experiments. c Selected by Mackay et al. (12). Reference 13. e Reference 14. f From an individual test. g Reference 15. h Reference 17. i Not available. j Calculated from ref 16. k Kfw and Kfd are film:water and film:diluent partition coefficients, respectively (values between brackets are standard deviations; n ) 3 for postloaded values; n ) 5-11 for preloaded values). Kow is the octanol:water partition coefficient. Sw is the aqueous solubility [µM]. d
the solubility of Phe in this aqueous medium. For comparison, aqueous solubility data from the literature are included in Table 1. The reduced solubility of Phe in Microtox diluent compared to pure water may be partly explained by the “salting out” effect. Assuming that Microtox diluent is a 2% NaCl solution, the total molar salt concentration equals 0.34 M. Using the Setschenov equation and the NaCl salting constant from May et al. (18), the aqueous solubility of Phe is reduced by 24% to 4.9 µM. Additionally, the published solubility data is for 25 °C, while the partitioning studies with diluent were performed at 15 °C. This temperature decrease should cause a further reduction in solubility. An approximation of the change in solubility as a function of temperature can be determined using the approach reported by Banerjee (19). In this paper, eq 3 is presented to relate solubility in water (Sw) to temperature (T)
log Sw ) 1.9 - 0.89 log γ - 2.93(mp - T)/(T + 273) (3) where γ is the activity coefficient of the solute as determined by the UNIFAC calculation (20) and mp is the melting point of the compound (in °C). This can be rearranged to calculate the difference in solubility between two temperatures as
log S1 - log S2 ) 2.93(mp - T2)/(T2 + 273) 2.93(mp - T1)/(T1 + 273) (4) where subscripts indicate respective temperatures (in Celsius). Note that this assumes the value for γ to be constant over the temperature range studied. While some change in γ is expected, Banerjee (19) notes that this is smaller than the error in solubility predictions over the 10 °C to 70 °C temperature range. Using this equation, the solubility of Phe (mp 99.5 °C) at 15 °C is calculated to be 75% of the value at 25 °C, and a similar calculation for Ant (mp 217.5 °C) gives 68%. Combining these estimates with the results of the salting out effect, it may be concluded that the film is able to deliver aqueous concentrations up to solubility. Similar results were obtained in experiments with BaP. Using film concentrations up to 0.02 M resulted in dissolved concentrations leveling off at 8.2 × 10-9 M, which compares favorably with the reported solubility of 7.2 × 10-9 M ( 1.35 × 10-9 M (15). A dissimilarity in partition controlled concentrations in microplates compared with cuvettes, which proved to be a general phenomenon, is observed in Figure 3. Irreversible 4100
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
absorption of contaminant from solution into the polystyrene microplate matrix reduced the actual contaminant concentration in the film, producing lower solution concentrations and higher apparent Kfs values. This had a more pronounced effect at lower film concentrations. Most of this loss was prevented by using larger film volumes, producing films that while curing also covered the walls of the microplate wells, though apparent Kfs values were still higher than for glass cuvettes. Kinetic studies with different film volumes indicated that tss remained the same, proving that the desorption rate was independent of the film volume. Capability of Films To Compensate for Substantial Compound Loss from Solution. Addition of a PDMS bead to the cuvette solutions acted as a model for toxicant loss through processes such as precipitation, adsorption, biological and/or chemical degradation, and uptake into biological or test tissue. After addition of the bead to a simple Ant solution (7 × 10-8 M) in a glass vial, the solution concentration decreased dramatically (unstirred system, 80% loss over 100 min), while control experiments without a bead added showed no significant loss. After addition of a bead to a vial where Ant was delivered from a PDMS film, there was no measurable reduction of the solution concentration. From this it was concluded that, greatly facilitated by the fast equilibration (Figure 2), desorption from the film was able to compensate for loss by absorption into the bead. A mass-balance approach may be used to estimate the expected impact of losses on the film concentration Cf and hence the partition controlled concentration. The total material in the film may be calculated as Cf × Vf, while the material lost on equilibration with solution is Cs × Vs. Since Cf ) Cs × Kfs, it can be stated that Cf will be essentially constant if Vf × Kfs is much greater than Vs. When losses in addition to equilibration with the solution are added, the same approach may be used where total material lost is compared to the Cf × Vf factor. If greater film capacity is needed for a particular experiment, without a change in the partition controlled concentration, the film volume may simply be increased with no change in Cf. Partitioning of PAHs from Films After Storage. An important parameter of both preloaded and postloaded films is their stability over time. The possibility of using films from stock (stored dry) would make the technique more versatile and portable. To test stability, films preloaded with Ant in microplates and unloaded films in cuvettes were stored for
TABLE 2. Toxicity of PAHs in the Microtox Testd compound
microtox EC50 lit. [µM]
Sw [µM]
microtox EC50 (SD) this study [µM]
fluorene phenanthrene anthracene acridine
19.4a 0.28a 1874a 38.9c
11.4 6.45 0.24 214
3.27 (0.01) 0.44 (0.01) NRb 196 (6.2)
a Reference 7. b No response. c TerraTox 2000, TerraBase Inc. d EC 50 values are calculated after 15 min incubation at 15 °C. Values between brackets are the standard deviations (n ) 6 for Flu, Phe, and Ant; n ) 3 for acridone). Sw is the aqueous solubility [µM].
FIGURE 4. Dissolved concentrations of Phe and Ant after equilibration from aged postloaded cuvettes and preloaded microplates, respectively. The horizontal axis gives the age of the film at which partitioning was started. Symbols represent average concentrations ( σ (n ) 3). Shaded area gives average normalized Cs ( 2σ. (Curves drawn to aid the eye). several days in the dark at ambient temperature. At regular intervals, unloaded films were postloaded with Phe. After equilibrating preloaded and postloaded films with water and diluent, respectively, dissolved concentrations were determined. With one outlier after 5 days storage (microplate), all dissolved concentrations were within a range of 2σ (Figure 4), with solutions prepared from postloaded films displaying a smaller range in dissolved concentrations. Comparison of Kfs with Kow Values. In Table 1, average log Kfs values are listed along with the octanol-water partition coefficients (log Kow), calculated using the ClogP algorithm (Biobyte Corp., Claremont, U.S.A.), with the exception of 9-(10H)-acridone, for which log Kow is from Bleeker et al. (17). Film:diluent and film:water partition coefficients are denoted as Kfd and Kfw, respectively. A plot of log Kfs vs log Kow gave a slope of 0.77 ( 0.06 with an intercept of 0.63 (r2 ) 0.933). This slope is lower than the near-unit slope reported for partitioning of nonpolar halogenated compounds (log Kow between 2.5 and 5) between a C18-phase and water (recalculated from Mayer et al. (9, 10), using slow-stirring data from De Maagd et al. (15) and de Bruijn et al. (21) only, instead of the mixed-origin data used by Mayer (10)). Selecting preload Kfs data improved the regression and resulted in a slightly steeper slope (r2 ) 0.990; slope is 0.84 ( 0.04); replacing ClogP data with slow-stirring log Kow data (when available (15, 21)) further improved the slope, however, with a small sacrifice toward the correlation (r2 ) 0.979; slope is 0.90 ( 0.06). The difference in slope of the postloaded film data suggests a different mode of interaction between the test molecules and PDMS, compared to their interactions with octanol or C18. Since PDMS is a highly cross-linked polymer, it is expected that partition of compounds will be influenced by steric interactions with the polymer network. This would impede partition for larger molecules compared with smaller molecules, which results in a systematic decrease in partitioning as Kow increases, since molecular size increases in close correlation with Kow for the compounds studied in this work. Partitioning from a Film Loaded with PAH Mixtures. The performance of the films was further evaluated by partitioning PAHs from mixtures. In one experiment, the films were postloaded with an equimolar mixture of Nap, Flu, Phe, Ant, Fla, and Pyr (data not shown). After equilibrating with water, Kfs values for Ant and Pyr were identical, within error, to those obtained in single-compound tests (Table 1). In a separate experiment, the Phe concentration in the film was fixed at 6.0 × 10-2 M, while the Ant concentration varied between 5 × 10-4 M and 8 × 10-3 M in a six-step dilution
FIGURE 5. Microtox dose-response curves of Flu, Phe, and Ant and of the supposedly equitoxic mixture (bottom). Symbols represent average responses ( σ (n ) 6). A fit of Ant data to the sigmoidal curve did not converge. series. The equilibrium Ant concentrations in solution were greater than for the single-compound experiment. Presumably, PDMS has a finite capacity for total added compounds, and in this case the excess Phe reduced the available capacity, which then caused desorption of more Ant molecules into the solution (i.e. a lower apparent Kfs). Further experiments using mixtures, including extracts of contaminated sediments, are currently being conducted, and results will be published separately. Microtox Broad-Spectrum Bacterial Toxicity Test. The Microtox bacterial toxicity was originally developed by Microbics Corp. (now Azur Environmental, Carlsbad, CA). This bioassay uses the photoluminescent bacterium Vibrio fisheri. Toxicity is based upon reduction in light-emission relative to a control (blank). With the exception of Nap and Phe, literature EC50 values of PAHs are all above the aqueous solubility (6, 7) (Table 2). The reported EC50 of Flu is close to and possibly within error of the reported solubility (7); the reported toxicity of Ant is without any doubt incorrect. Microtox tests were run in 1:1 (v:v) dilution series with the highest concentrations as close to the reported aqueous solubility as possible (Figure 5 and Table 2). PAH (Flu, Ant, Phe) concentrations in the film and in solution were determined. For comparison, PAH concentrations in parallel dilution series in the absence of bacteria were recorded; they showed a slightly lower but negligible difference in the dissolved concentrations. After EC50 calculation and assuming concentration-addition, equitoxic Flu and Phe mixtures were prepared accordingly and tested (Figure 5 and Table 2). While the EC50 value of Phe was close to the reported value, EC50 of Flu was actually below the solubility. Some response was found in the Ant tests, but signal reductions were highly irregular making it impossible to calculate an EC50. Highly precise EC50 values for Flu and Phe were determined; the coefficients of variation were less than 2.5% (n ) 6), which is much more precise than the literature values (95% confidence intervals for Flu and Phe are 13-28 µM and VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4101
2.0-4.4 µM, respectively (7)). This precision significantly adds to the applicability of the film procedure. The lack of toxic response of Ant in comparison with its isomer Phe may be explained by the high melting point of Ant (217.5 °C vs 99.5 °C of Phe). Assuming that the mechanism of toxicity of PAHs in the Microtox test is (nonpolar) narcosis, the observation by Mayer et al. (9, 10) that compounds with high melting points cannot exert baseline toxicity as single species is supported. With EC50 values for Flu, Phe, and the “equitoxic” mixture of 3.27 µM, 0.44 µM, and 1.64 µM, respectively, the additivity index of the binary mixture is 0.87. Using the additivity indices of Chen and Chiou (22) for nonreactive toxicants having either parallel or nonparallel dose-response curves, the combined toxicity of this binary mixture of Flu and Phe in the Microtox test may be classified as simple addition. This conclusion could not be reached properly using the literature data, as the reported EC50 value of Flu is above its solubility. Future studies will be carried out to resolve the possible contribution of other PAHs (e.g., Ant) in mixture toxicity. Further Remarks. Summarizing the performance of postloaded vs preloaded films, the preloaded films are more easily prepared, may better accommodate larger molecules, and have a higher capacity than postloaded films. The postloaded films, on the other hand, provide Kfs values with generally higher precision, will not chemically react with functional groups of contaminant molecules, and, importantly, allow for the use of films from stock and thus make the technique portable. In principle, both techniques are applicable to small-scale toxicity tests of sparingly soluble chemicals as pure compounds and in mixtures. Toxicity tests that need a larger volume of the aqueous phase can easily be accommodated by increasing the film volume (Brown et al.; results of Japanese Medaka egg early life-stage tests to be published elsewhere). If the film method were to be used in testing a variety of samples, then the relative toxicity between them would be closer related to bioavailability and thus better reflect differences in toxicity in a real exposure situation.
Acknowledgments This project was financially supported by the Dutch Ministry of Transport and Public Works, National Institute for Coastal and Marine Management/RIKZ, project WONS*TOX. Comments of Drs. Karin Legierse, Gert-Jan de Maagd, and Philipp Mayer were greatly appreciated.
Literature Cited (1) Aquatic Toxicity Testing of Sparingly Soluble, Volatile and Unstable Substances; Monograph No. 26; European Center for
4102
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
(2) (3) (4) (5)
(6) (7) (8) (9) (10) (11)
(12)
(13) (14) (15)
(16) (17) (18)
(19) (20) (21) (22)
Ecotoxicology and Toxicology of Chemicals (ECETOC): Brussels, 1996. Opperhuizen, A.; van de Velde, E. W.; Gobas, F. A. P. C.; Liem, D. A. K.; van der Steen, J. M. D. Chemosphere 1985, 14, 1871. Anliker, R.; Moser, P.; Poppinger, D. Chemosphere 1988, 17, 1631. Lien, G. J.; McKim, J. M. Aquat. Toxicol. 1993, 27, 15. Vaes, W. H. J.; Ramos, E. U.; Hamwijk, C.; Holsteijn, I van; Blaauboer, B. J.; Seinen, W.; Verhaar, H. J. M.; Hermens, J. L. M. Chem. Res. Toxicol. 1997, 10, 1067. Kaiser, K. L. E.; Ribo, J. M. Toxicity Assess. 1988, 3, 195. Jacobs, M. W.; Coates, J. A.; Delfino, J. J.; Bitton, G.; Davis, W. M.; Gracia, K. L. Arch. Environ. Contam. Toxicol. 1993, 24, 461. Klamer, H. J. C.; Villerius, L. A.; Roelsma, J.; de Maagd, P. G-J.; Opperhuizen, A. Environ. Toxicol. Chem. 1997, 16, 857. Mayer, P.; Wernsing, J.; Tolls, J.; De Maagd, P. G. J.; Sijm, D. T. H. M. Environ. Sci. Technol. 1999, 33, 2284. Mayer, P. Ph.-D. Dissertation, Utrecht University, The Netherlands, 2000. Mayer, P.; Vaes, W. H. J.; Wijnker, F.; Legierse, K. C. H. M.; Kraaij, R. H.; Tolls, J.; Hermens, J. L. M. Environ. Sci. Technol. 2000, 34, 5177. Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated handbook of physicochemical and environmental fate for organic chemicals. Part III; Lewis Publishers: Chelsea, U.S.A., 1992. Doong, R.; Chang, S. 2000 Anal. Chem. 72, 3647. Mayer, P.; Vaes, W. H. J.; Hermens, J. L. M. Anal. Chem. 2000, 72, 459. Maagd, P. G. J. De; ten Hulscher, D. Th. E. M.; van den Heuvel, H.; Opperhuizen, A.; Sijm, D. T. H. M. Environ. Toxicol. Chem. 1998, 17, 251. Meylan, W. M.; Howard, P. H.; Boethling, R. S. 1996. Environ. Toxicol. Chem. 15, 100. Bleeker, E. A. J.; van der Geest, H. G.; Klamer, H. J. C.; de Voogt, P.; Wind, E.; Kraak, M. H. S. Polyc. Arom. Comput. 1999, 13, 191. May, W. E. The solubility behavior of polycyclic aromatic hydrocarbons. In Petroleum in the Marine Environment; Advances in Chemistry Series No. 185; American Chemical Society: Washington, DC, 1980; Chapter 7. Banerjee, S. Water Res. 1996, 30, 2222. Hansen, H. K.; Rasmussen, P.; Fredenslund, A.; Schiller, M.; Gmehling, J. Ind. Eng. Chem. Res. 1991, 30, 2352. de Bruijn, J.; Busser, F.; Seinen, W.; Hermens, J. L. M. Environ. Toxicol. Chem. 1989, 8, 499. Chen, C.-Y.; Chiou, Y.-S. Environ. Toxicol. Water Qual. 1995, 10, 97.
Received for review March 6, 2001. Revised manuscript received March 25, 2001. Accepted June 28, 2001. ES010708T
