Passive Dosing of Polycyclic Aromatic Hydrocarbon (PAH) Mixtures to

Mar 11, 2013 - ... Presow , Roger van Egmond , Jon A. Arnot , Donald Mackay , Philipp Mayer ... Martin Holmstrup , Christian Damgaard , and Philipp Ma...
1 downloads 0 Views 640KB Size
Download PDF
Article pubs.acs.org/est

Passive Dosing of Polycyclic Aromatic Hydrocarbon (PAH) Mixtures to Terrestrial Springtails: Linking Mixture Toxicity to Chemical Activities, Equilibrium Lipid Concentrations, and Toxic Units Stine N. Schmidt,*,† Martin Holmstrup,‡ Kilian E. C. Smith,† and Philipp Mayer† †

Department of Environmental Science, Aarhus University, Frederiksborgvej 399, PO Box 358, DK-4000 Roskilde, Denmark Department of Bioscience, Aarhus University, Vejlsøvej 25, PO Box 314, DK-8600 Silkeborg, Denmark

‡

S Supporting Information *

ABSTRACT: A 7-day mixture toxicity experiment with the terrestrial springtail Folsomia candida was conducted, and the effects were linked to three different mixture exposure parameters. Passive dosing from silicone was applied to tightly control exposure levels and compositions of 12 mixture treatments, containing the polycyclic aromatic hydrocarbons (PAHs) naphthalene, phenanthrene, and pyrene. Springtail lethality was then linked to sum chemical activities (∑a), sum equilibrium lipid concentrations (∑Clipid eq.), and sum toxic units (∑TU). In each case, the effects of all 12 mixture treatments could be fitted to one sigmoidal exposure−response relationship. The effective lethal chemical activity (La50) of 0.027 was well within the expected range for baseline toxicity of 0.01−0.1. Linking the effects to the lipid-based exposure parameter yielded an effective lethal concentration (LClipid eq 50) of 133 mmol kg−1 lipid in good correspondence with the lethal membrane burden for baseline toxicity (40−160 mmol kg−1 lipid). Finally, the effective lethal toxic unit (LTU50) of 1.20 was rather close to the expected value of 1. Altogether, passive dosing provided tightly controlled mixture exposure in terms of both level and composition, while ∑a, ∑Clipid eq., and ∑TU allowed baseline toxicity to be linked to mixture exposure.

■

INTRODUCTION Organisms in the environment are exposed to complex mixtures of hydrophobic organic contaminants (HOCs), whereas toxicity of these contaminants is mainly tested on an individual compound basis.1−4 The first part of the present study therefore addresses the challenges of controlling both level and composition of HOC mixture exposure in toxicity experiments. In traditional laboratory experiments, where the test compound(s) are added via an organic solvent, the effective exposure often decreases due to e.g., sorption, uptake, (bio)transformation, and evaporation.5,6 The impact of these processes often varies considerably among mixture constituents,7 and test organisms are then exposed to mixtures where both levels and compositions change in time. Recently, various passive dosing formats have been developed to tightly control the exposure in laboratory experiments.8−13 In passive dosing, a silicone polymer is loaded with the HOC(s) and subsequently applied to establish and maintain the exposure by continuous equilibrium partitioning. In this way, the method provides well-defined and constant exposure during the experiment. Passive dosing has been used for the toxicity testing of individual HOCs,10,12−14 and more recently also of mixtures of fixed composition.11,15−17 The present study goes one step further by exploring the ability of © 2013 American Chemical Society

passive dosing to control mixture level and composition. Specifically, the terrestrial springtail Folsomia candida was exposed to four mixtures containing naphthalene, phenanthrene, and pyrene. Each of these polycyclic aromatic hydrocarbon (PAH) mixtures was tested at three exposure levels, resulting in 12 mixture treatments (Table 1). Based on previous results with passive dosing, it was expected that the method would be able to control both levels and compositions of PAH mixtures during the 7-day toxicity experiment with F. candida. Landrum and co-workers specified two fundamental requirements in mixture (eco)toxicology, namely establishing dose− response relationships and determining the causative agent(s).18 One challenge for meeting these requirements is that the toxicity of an HOC is poorly related to its total concentration in soil.5,19 Another challenge is to identify exposure parameters that allow a meaningful way of adding the exposure to the mixture constituents in accordance with the Special Issue: Rene Schwarzenbach Tribute Received: Revised: Accepted: Published: 7020

November 22, 2012 February 8, 2013 March 10, 2013 March 11, 2013 dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

Table 1. Exposure Compositions and Estimated Exposure Levels of the Twelve Mixture Treatments mixturea PHE & PYR

NAPH & PYR

NAPH & PHE

NAPH, PHE & PYR

nomenclature

dilutionb

Σac (at 21 °C)

ΣClipid eq.d mmol kg−1

ΣTUe

1.1 1.2 1.3 2.1 2.2 2.3 3.1 3.2 3.3 4.1 4.2 4.3

4 8 16 4 8 16 4 8 16 4 8 16

0.047 0.023 0.012 0.065 0.032 0.016 0.090 0.045 0.022 0.101 0.050 0.025 0.012−0.101

213 107 53 355 178 89 428 214 107 498 249 125 53−498

1.78 0.88 0.45 3.02 1.58 0.79 4.20 2.17 1.08 4.35 2.29 1.09 0.45−4.35

range a

PHE is phenanthrene, PYR is pyrene, and NAPH is naphthalene. bDilution factor relative to saturation. cSum chemical activities, calculated from eq 2. dSum equilibrium lipid concentrations, calculated from eq 5. eSum toxic units, calculated from eq 7.

established “concentration addition” concept for mixture toxicity.1,20 In the second part of the present study, three exposure parameters were therefore explored for their ability to link baseline toxicity to mixture exposure: (1) F. candida lethality was linked to sum chemical activities (∑a, Table 2). The chemical activity (denoted a) quantifies the energetic level of an HOC relative to a reference state.19 The subcooled liquid of the HOC (a = 1) serves as reference state, and chemical activity is then defined between 0 and 1. The toxicity of individual PAHs is closely linked to their chemical activity,14 and baseline toxicity has been reported to initiate within the chemical activity range of 0.01 to 0.1.10,12,14 Therefore, it was hypothesized that PAH mixture toxicity is linked to ∑a and will initiate in the chemical activity range 0.01−0.1. (2) Lipid membranes are considered the site of toxic action for baseline toxicants,21,22 and the sum lipid concentration (∑Clipid) may thus be a suitable exposure parameter for these compounds. It is difficult to directly determine ∑Clipid, but it can be estimated by adding ratios of chemical activity and activity coefficient in lipid23 or products of medium concentration and lipid to medium partition ratio for each mixture constituent. Both approaches are based on an equilibrium assumption, meaning that they yield sum equilibrium partitioning concentrations in lipids (∑Clipid eq., Table 2).24 The toxicity of individual PAHs is closely linked to their lipid concentration,14 and baseline toxicity has been reported to initiate within the concentration range of 40−160 mmol kg−1 lipid.21 Therefore, it was hypothesized that PAH mixture toxicity is linked to ∑Clipid eq. and will initiate in the concentration range 40−160 mmol kg−1 lipid. (3) The toxic unit (TU) approach is well established in (eco)toxicological research,1 with the TU of a mixture constituent calculated as the ratio between exposure concentration and effective concentration, e.g. LC50 (Table 2).1,25,26 The novelty in the present study was to calculate TU from freely dissolved rather than total concentrations, and it was hypothesized that PAH mixture toxicity is closely linked to ∑TU, and furthermore that 50% lethality would require a ∑TU of approximately 1.

Springtail lethality was plotted against ∑a, ∑Clipid eq., and ∑TU, respectively, and in each case one sigmoidal exposure− response curve was fitted to all data. The descriptive capability of each exposure parameter was evaluated based on the quality of the fit (expressed by r2-values), while the predictive capability was evaluated by testing the specific hypotheses given above.

■

EXPERIMENTAL SECTION Materials and Chemicals. Medical-grade polydimethylsiloxane (PDMS) silicone was prepared using the MDX4-4210 kit from Factor II, Inc. (USA). The 10-mL glass vials and airtight screw caps (with Teflon-coated septa) were from Mikrolab Aarhus A/S, Denmark. The test compounds were naphthalene (99%, Aldrich, Germany), phenanthrene (≥99.5%, Aldrich, USA), and pyrene (≥99.0%, Sigma, Switzerland). Ethanol (96%, Kemetyl A/S, Denmark), Milli-Q water (Super Q treated, Millipore, USA), methanol (99.9%, Merck, Germany), and lint-free tissue (Assistent, Germany) were used as described below. (1). Controlling Mixture Exposure. Passive dosing was used to tightly control the levels and compositions of PAH mixtures in an acute toxicity experiment with F. candida. The passive dosing vials were made, loaded, and cleaned as has been described in detail earlier,14 and the procedures are therefore only briefly outlined below. In short, 500 mg (±1%) silicone was cast and cured in the bottom of a 10-mL glass vial, and the cured PDMS was subsequently cleaned with ethanol and MilliQ water. A saturated methanol stock solution was made for each of the three PAHs, and saturation was ensured by the presence of PAH crystals. Four mixtures were prepared by mixing an equal volume of saturated and crystal free supernatant containing phenanthrene and pyrene (mixture 1), naphthalene and pyrene (mixture 2), naphthalene and phenanthrene (mixture 3), and all three PAHs (mixture 4). Loading solutions were then prepared by dilution with methanol until each of the four mixtures was diluted 4-, 8-, and 16-fold relative to saturation, resulting in a total of 12 mixture treatments (see Table 1 for an overview of the mixture treatments and their nomenclature). Loading of the silicone was done by equilibrium partitioning of the PAHs between 1000 μL of loading solution and PDMS in closed passive dosing vials for at least 48 h. The depletions of the methanol loading solutions were 15.7%, 12.5%, and 12.9% for 7021

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

solubilities at 20 °C28 and the appropriate dilution factors. Three outlier concentrations were identified using Grubbs’ test29 and excluded from the exposure confirmation data set. One naphthalene measurement was significantly below its estimated value (p < 0.01, mixture treatment 4.3, Table 1), and two pyrene measurements significantly exceeded their estimated values (p < 0.01, mixture treatments 1.1 and 4.3, Table 1). The observed lethality in these three treatments was similar to observations in other replicates and were not identified as outliers, which suggests deviations of the measured concentrations to be caused by postexperimental steps. The low naphthalene concentration was likely the result of evaporative losses during the exposure confirmation, while the higher pyrene concentrations likely resulted from pyrene sorbing to small amounts of dust or biological material, leading to an overestimation of the freely dissolved concentrations. No PAHs were detected in the 10 controls.14 (2). Linking Toxicity to Mixture Exposure. Springtail lethality was determined at day 7 of the mixture toxicity experiment, and these results were used to explore the applicability of sum chemical activities (∑a), sum equilibrium lipid concentrations (∑Clipid eq.), and sum toxic units (∑TU) as exposure parameters for mixtures. Springtail lethality was fitted with a sigmoidal exposure−response function with variable slope using least-squares regression by the GraphPad Prism 5.0 software (GraphPad Software, Inc., USA).

naphthalene, phenanthrene, and pyrene, respectively (see Supporting Information (SI) for the calculations of depletion), and all estimated exposure levels, i.e., sum chemical activities and sum equilibrium lipid concentrations, were corrected for these minor depletions. Each mixture treatment was done in five replicates, and controls were loaded using pure methanol. After loading, the spent loading solution was removed and the silicone was cleaned four times with Milli-Q water to quantitatively remove the methanol and then dried with lintfree tissue. The PAHs were pre-equilibrated between the silicone and air in closed passive dosing vials overnight to ensure the appropriate exposure levels from the beginning of the experiment. Naphthalene, phenanthrene, and pyrene equilibrate between PDMS and water within hours in the passive dosing system,12 and diffusion coefficients in air are 3−4 orders of magnitude higher than in water for these PAHs.27 In this way, equilibration overnight was sufficient to ensure equilibrium in the system. Then, a small droplet of Milli-Q water (approximately 2 μL) was placed at the silicone surface to ensure sufficient humidity during the experiment. Loading of the passive dosing vials was done at room temperature (approximately 21 °C). The mixture toxicity experiment was carried out in parallel with a bioconcentration experiment,14 and a total of 10 controls was included in the two experiments. Only one springtail did not survive in the 10 controls (average survival 99%), demonstrating that the passive dosing format, loading, and cleaning procedures did not cause lethality. Age synchronized F. candida were cultured and maintained as has been described earlier,10,14 and the springtails had an average fresh weight of 141 μg and an average dry weight of 58 μg (39−42 days old, average weight of 500 individuals). To begin the experiment, ten springtails (45−48 days old) were transferred to each passive dosing vial. The springtails could move freely within the vial, and most springtails were observed on the PDMS surface. The uptake of PAHs via water was negligible due to the tiny water volume, a limited water uptake, and low aqueous concentrations. This leaves exposure through air and direct contact with silicone as the two routes of PAH uptake in the passive dosing system. Springtail survival was examined daily for 7 days (Figure S1); the springtails were transferred to Petri dishes with a clean and moistened charcoal/plaster-of-Paris blend and characterized as living when able to walk in a coordinated manner, if necessary, after gentle stimulation with a fine brush. At day 7, the springtails recovered for at least 2 h in Petri dishes before examination as above. The experiment was conducted at 20 °C with a 12:12 h light/dark photoperiod. The toxicity data from one replicate of mixture treatment 1.1 (Table 1) was lost since the glass vial tipped over, but the exposure confirmation results could still be included in the data set (see below). After ending the experiment, the exposure was confirmed analytically in order to investigate the ability of the passive dosing system to tightly control the levels and compositions of the PAH mixtures during the 7 days, and the procedure has been described in detail earlier.14 In short, 1000 μL of Milli-Q water was equilibrated with the cleaned silicone, and 500 μL of equilibrated water was mixed with 500 μL of methanol before PAH analysis by HPLC with multiband fluorescence detection and quantification by external calibration.14 In this way, the exposure level of each mixture constituent in the 12 mixture treatments was measured (Cfree final, n = 135), and these were compared to estimated levels as calculated from water

Springtail lethality =

100 (log LC50 − C)hillslope

1 + 10

(1)

∑a was calculated for each of the 12 mixture treatments (Table 1) by simply adding the chemical activity (a) of the mixture constituents. Σa = a PAH1 + a PAH2( + a PAH3)

(2)

The chemical activities of the individual PAHs (a, dimensionless) were calculated from their maximum chemical activity (amax) and the appropriate dilution factor. The amax of each PAH was estimated from its melting temperature (Tm, K) and the experimental temperature (T, K) according to Yalkowsky et al.,30 assuming the entropy of melting to be 56 J mol−1 K−1 (i.e., Walden’s rule). amax = e

⎛ T ⎞ 6.8 × ⎜1 − m ⎟ T ⎠ ⎝

(3)

This amax is 0.255 for naphthalene, 0.165 for phenanthrene, and 0.049 for pyrene at 21 °C using the melting temperatures from Lide.31 Percent springtail lethality was fitted as a function of ∑a, as described above. The measured freely dissolved concentrations from the exposure confirmation (Cfree (final)) were translated into chemical activities using subcooled liquid solubilities (SL), which were estimated from water solubilities (SS) at 20 °C28 and amax values at 21 °C. a=

Cfree S → SL = S SL amax

(4)

These chemical activities were used for calculating the mixture compositions at day 7. ∑Clipid eq. was calculated for each of the 12 mixture treatments (Table 1). 7022

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

ΣC lipid eq. = C lipid eq.PAH1 + C lipid eq.PAH2( + C lipid eq.PAH3) (5)

The approach here was to calculate and add the equilibrium partitioning concentration of all mixture constituents in lipid (Clipid eq.). For convenience, the equilibrium partitioning concentration in lipid is termed the equilibrium lipid concentration throughout this article. The Clipid eq. (mol L−1) of each PAH was estimated from its chemical activity (a, dimensionless) and a medium- and compound-specific activity coefficient (γlipid, L mol−1).23 a C lipid eq. = γlipid (6) The maximum equilibrium lipid concentration is 1352, 653, and 323 mmol kg−1 lipid for naphthalene, phenanthrene, and pyrene, respectively, when using the maximum chemical activities given above, the activity coefficients in Mayer et al.,23 and a lipid density of 0.9 kg L−1 lipid.32 Percent springtail lethality was fitted as a function of ΣClipid eq., as described above. ∑TU was calculated for each of the 12 mixture treatments (Table 1).

∑ TU = TUPAH

1

+ TUPAH2( + TUPAH3)

(7)

The toxic unit (TU, dimensionless) of each PAH was calculated from its measured freely dissolved concentration (Cfree (final), μg L−1) and its effective lethal concentration in water (LCfree 50, μg L−1);

TU =

Cfree LCfree50

(8)

The LCfree 50 values when applying passive dosing to F. candida were previously determined to be 2387, 176, and 95 μg L−1 for naphthalene, phenanthrene, and pyrene, respectively.14 Percent springtail lethality was fitted as a function of ΣTU, as described above. The effective chemical activity (La50), effective equilibrium lipid concentration (LClipid eq 50), and effective toxic unit (LTU50) causing 50% springtail lethality were also determined.

Figure 1. (A) Exposure confirmation of naphthalene (NAPH), phenanthrene (PHE), and pyrene (PYR) at the end of the mixture toxicity experiment. The freely dissolved concentration (Cfree final) of each mixture constituent was measured in the passive dosing vials after ending the experiment and compared to estimated concentrations (Cestimated). Each of the 135 measurements is displayed by a circle, and the results group together according to PAH and dilution level; e.g., the measurements of naphthalene at a dilution level of 4 (n = 15, mixture treatments 2.1, 3.1, and 4.1, Table 1) are seen in the upper right corner of the figure, whereas the measurements of pyrene at a dilution level of 16 (n = 15, mixture treatments 1.3, 2.3, and 4.3, Table 1) are seen in the lower left corner of the figure. A 1:1 line is included for visual reference, and three outliers of 135 measurements are displayed in red. (B) The measured Cfree final of each mixture constituent plotted in a bar diagram (n = 132), with the average Cfree final value indicated above each bar (μg L−1). Each bar represents the average ± standard deviation of 5 replicates, and * denotes 4 replicates. For the nomenclature of the mixture treatments please consult Table 1.

■

RESULTS AND DISCUSSION (1). Controlling Mixture Exposure. New experimental methods are needed for controlling exposure in mixture toxicity experiments, and the performance of passive dosing for tightly controlling both the exposure levels and compositions of mixtures was investigated. Exposure Level. Freely dissolved PAH concentrations were measured after ending the experiment and plotted against the estimated concentrations (Figures 1A and S2). The measured Cfree final values were generally in good agreement with the corresponding estimates with average differences of 12.7 ± 3.4% for naphthalene and 25.3 ± 7.4% for pyrene, while measured phenanthrene concentrations were on average 43.5 ± 4.2% above the estimated values, partly in agreement with previous results.14 These moderate differences include the uncertainties related to (1) the estimated levels, (2) the analytical measurements, and (3) the ability of the passive dosing system to control exposure. The analytical measurements further indicated that mixture exposure levels were kept constant during the experiment, which is in accordance with previous results.15−17 The measured concentrations were plotted in bar diagrams (Figure 1B). Exposure levels were

precisely controlled between replicates (n = 5, e.g., naphthalene in mixture treatment 2.1) with average relative standard deviations (RSDs) between replicates of 1.3% for naphthalene, 1.2% for phenanthrene, and 1.7% for pyrene (n = 9 × 5, Figure 1B). More importantly, exposure levels were also precisely controlled between mixture treatments (n = 15, e.g., naphthalene in mixtures treatments 2.1, 3.1, and 4.1) with 7023

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

average RSDs between mixture treatments of 2.7%, 1.6%, and 2.6% for naphthalene, phenanthrene, and pyrene, respectively (n = 3 × 15, Figure 1B). These results demonstrate that the exposure control of one PAH was unaffected by the presence of other PAHs, which is a prerequisite for the independent control of mixture levels and compositions. Finally, the exposure in a dilution series (n = 15, e.g. naphthalene in mixture treatments 2.1−2.3) was well produced and controlled by passive dosing (Figure 1B). Overall, the phase partitioning of PAHs from the PDMS allowed a precise exposure control with RSD values for final measurements of Cfree in the low percentage range in accordance with recent research applying passive dosing as an analytical speciation technique.33 Exposure Composition. Controlling mixture composition evidently results from the tightly controlled exposure level of each individual mixture constituent. A comparison of actual to estimated compositions revealed good agreement, indicating that the composition of each mixture treatment was tightly controlled and kept constant during the 7-day toxicity experiment (Table S1). Furthermore, the results were highly reproducible in the dilution series of a mixture (e.g., mixture treatments 1.1−1.3, Table S1). The difference between the measured and estimated phenanthrene levels (see above) was to some extent reflected in the mixture compositions with phenanthrene constituting 2−8% more than estimated, while the reverse was the case for naphthalene, which constituted 2− 8% less than estimated (Table S1). Pyrene was within 2% of the estimated values. Altogether, passive dosing tightly controlled exposure levels and compositions of the 12 mixture treatments during the 7-day toxicity experiment, as was expected and in line with previous results.16 (2). Linking Toxicity to Mixture Exposure. In the following, the springtail lethality caused by the 12 mixture treatments is linked to exposure expressed as sum chemical activities (Σa), sum equilibrium lipid concentrations (ΣClipid eq.), and sum toxic units (ΣTU). Sum Chemical Activities. ∑a was chosen as an exposure parameter since differences in chemical activity drive partitioning of HOCs into target membranes, including both diffusive uptake and internal distribution34,35 (Table 2). Additionally, springtail lethality caused by the 12 mixture treatments was expected to be related to ∑a since the baseline toxicity of individual HOCs has been reported to initiate within a rather narrow chemical activity range of 0.01−0.1,19,36 and baseline toxic mixtures are expected to follow the “concentration addition” concept.1 The effects for all 12 mixture treatments were linked to Σa by one sigmoidal exposure− response function (Figure 2, Table S2). The descriptive capability when linking springtail lethality to Σa after 7 days exposure was high (r2 = 0.89) and in line with previous studies with aquatic invertebrates.15−17 Furthermore, the predictive capability was high, since springtail lethality initiated well within the hypothesized range of chemical activity for baseline toxicity of 0.01−0.1 (Figure 2, Table 2). The effective lethal chemical activity, La50, of 0.027 (95% confidence interval 0.026−0.028) was in line with results obtained for the individual PAHs having La50 values of 0.019 (naphthalene), 0.026 (phenanthrene), and 0.052 (pyrene, Figure 2).14 These results indicate that the baseline toxicity of individual PAHs and PAH mixtures is comparable when expressing exposure in terms of (Σ)a. Based on these and previous results,10,12,14,19,36,37 Σa is therefore expected to be a useful exposure parameter when describing and predicting the baseline toxicity of HOC mixtures, but

Table 2. Three Exposure Parameters for Mixture Toxicity ∑aa basis

prediction assumptions input

basis

prediction assumptions input

basis prediction assumption input

Chemical activity is a multimedia parameter.19 Differences in chemical activity drive partitioning of HOCs into target membranes.34,35 Baseline toxicity initiates within a narrow chemical activity range.10,12,14,19,36 Baseline mixture toxicity initiates at ∑a = 0.01−0.1. Equilibrium partitioning and concentration addition. Melting temperature, experimental temperature, and dilution factor relative to saturation or freely dissolved concentration and subcooled liquid solubility (eqs 2−4). ∑Clipid eq.b Lipid membranes are presumed site of toxic action for baseline toxicants.21,22 Differences in HOC partitioning and activity coefficients in lipid are taken into account.3,22 Baseline toxicity initiates within a narrow concentration range.21 Baseline mixture toxicity initiates at ∑Clipid eq. = 40−160 mmol kg−1 lipid. Equilibrium partitioning and concentration addition. Chemical activity and activity coefficient in lipid or medium concentration and lipid to medium partition ratio (eqs 5 and 6). ∑TUc An established concept in (eco)toxicological research. Differences in HOC partitioning, activity coefficients in membrane lipid, and intrinsic toxicity are taken into account. 50% mixture toxicity requires a ∑TU of approximately 1. Concentration addition. Exposure concentration and effective concentration (eqs 7 and 8).

a Sum chemical activities, calculated from eq 2. bSum equilibrium lipid concentrations, calculated from eq 5. cSum toxic units, calculated from eq 7. HOC is short for hydrophobic organic contaminant.

further research is needed with other organic mixtures, other test organisms, and other exposure routes in order to determine the full potential, applicability domain, and limitations of this approach. Sum Equilibrium Lipid Concentrations. ΣClipid eq. was chosen as exposure parameter since lipid membranes are believed to be the site of toxic action for baseline toxicants.21,22 In this way, differences in HOC partitioning plus differences in activity coefficients in lipid are taken into account3,22 (Table 2). The effects for all 12 mixture treatments were linked to ΣClipid eq. by one sigmoidal exposure−response function as described above (Figure 3A, Table S2). Evidently, Clipid eq. is an equilibrium partitioning estimate that only equals the actual concentration in lipid in the case of equilibrium, whereas in the case of under-equilibration the actual concentration in lipid is lower than the calculated Clipid eq.. Naphthalene and anthracene (the not acutely toxic isomer of phenanthrene) were determined to be at or near equilibrium in F. candida within 7 days when applying the passive dosing system, while pyrene was found to be somewhat under-equilibrated.14 Nevertheless, springtail lethality was closely linked to ΣClipid eq. (r2 = 0.90) confirming the high descriptive capability when linking effects to a lipid-based exposure parameter.3 The similarity in the descriptive capabilities of Σa and ΣClipid eq. is explainable by activity coefficients in animal lipid being rather similar for PAHs, only deviating a factor of approximately 2.8 between 11 PAHs and 1.6 between naphthalene, phenanthrene, and pyrene.23 Baseline toxicity was also well predicted by ∑Clipid eq., although lethality initiated at the high end of the expected concentration range of 40−160 mmol kg−1 lipid (LClipid eq 50 = 133 mmol kg−1 lipid, 95% confidence interval 125−140 mmol 7024

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

Figure 2. Lethality to Folsomia candida after 7 days exposure to 12 mixture treatments as a function of sum chemical activities. Error bars represent the standard error of the mean (n = 5), and the shaded area is the chemical activity range 0.01−0.1. As a reference, the springtail lethality caused by naphthalene (NAPH), phenanthrene (PHE), and pyrene (PYR) is shown as a function of chemical activities.14

expected value of 1.1 This small deviation from unity is likely due to toxicity observations for individual PAHs and their mixtures originating from different experiments with different batches of F. candida. Parallel experiments with individual HOCs and their mixtures would be needed to determine the LTU50 even more accurately and to reveal small degrees of synergism or antagonism. However, this was not the purpose of the present study. The results of the present study are in good agreement with the three PAHs acting by the same toxic mechanism of action, and they further support a high predictive capability of ∑TU (Figure 3B). The main advantage of ∑TU is that it neither assumes equilibrium partitioning nor a certain toxic mechanism of action. However, the main disadvantage of ∑TU is that it requires effective concentrations that accurately describe the toxicity of the individual mixture constituents under the experimental conditions of the mixture experiment. In some cases, it can be difficult if not impossible to generate such effective concentrations.17 Altogether, the descriptive capabilities of the three exposure parameters were high and similar despite the fact that Σa (r2 = 0.89), ΣClipid eq. (r2 = 0.90), and ΣTU (r2 = 0.93) gradually take differences in more processes/properties into account, i.e., differences in HOC partitioning, activity coefficients in lipid, and intrinsic toxicities. Likewise, the predictive capabilities of the three exposure parameters were high for the tested PAH mixtures. These findings are in accordance with and support established models for the prediction of mixture toxicity based on equilibrium partitioning and concentration addition, such as the target lipid model developed by Di Toro and co-workers3,38 and the sigma PAH model developed by Swartz and coworkers.39 The present study demonstrates that ∑a and ∑Clipid eq. offer new possibilities for expressing mixture exposure, which can be directly related to mixture effects. None of these exposure parameters requires toxicity data for individual mixture constituents as input parameters, whereas they do require new experimental or analytical approaches to either control or measure chemical activities or freely dissolved concentrations. Σa was the easiest exposure parameter to estimate since the only input parameters are the melting

Figure 3. Lethality to Folsomia candida after 7 days exposure to 12 mixture treatments as a function of (A) sum equilibrium lipid concentrations and (B) sum toxic units. Error bars represent the standard error of the mean (n = 5), and the shaded area is the concentration range 40−160 mmol kg−1 lipid. The symbols for the different mixture treatments are explained in the legend in Figure 2.

kg−1 lipid, Figure 3A), in part due to the under-equilibration of pyrene.14 Sum Toxic Units. The ΣTU approach takes differences in HOC partitioning, activity coefficients in membrane lipid, plus intrinsic toxicity into account, and springtail lethality was therefore linked to ΣTU. Lethality was closely linked to ΣTU (r2 = 0.93, Figure 3B, Table S2) confirming the high descriptive capability of this exposure parameter. The effective lethal toxic unit (LTU50) was 1.20 (95% confidence interval 1.15−1.25), which is statistically different from but still rather close to the 7025

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

and Pia C. H. Petersen for technical support in the laboratory. This research project was financially supported by the European Commission (MODELPROBE, 213161; OSIRIS, COGE-037017) and The Danish Council for Independent Research (Contract 10-084579). Additionally, S.N.S. was supported by the PhD research program STAiR.

temperatures of the mixture constituents, the experimental temperature, and the dilution factors relative to saturation. For calculating ΣClipid eq. more input values were required as either activity coefficients in lipid or lipid to medium partition ratios should be available in addition to exposure levels (Table 2). There is a growing interest in relating effects not only to welldefined exposure as measured by, for example, solid phase micro extraction (SPME), but even to use the polymeric coating of the SPME fiber as a kind of partitioning surrogate for biological lipids.40−42 The applied PDMS silicone can be regarded as a biomimetic phase in line with that used for SPME, and springtail lethality was therefore linked to sum PDMS concentrations (∑CPDMS, Figure S3, Table S2). In the present study, PDMS concentrations were calculated from previously measured PAH concentrations in saturated PDMS10 and the respective dilution factor, and then corrected for the depletion in the present loading solutions. The effective lethal PDMS concentration (LCPDMS 50 ) was 9.8 mM (95% confidence interval 8.9−10.7) and in line with a previous value of 8.7 mM determined in a toxicity experiment with 10 PAHs.10 The wider perspective of such polymer-based effect concentrations is the possibility of applying equilibrium sampling into polymers for the prediction of the baseline toxic potential.41,42 Springtail lethality was not as closely linked to ∑CPDMS (r2 = 0.78) as to ∑a, ∑Clipid eq., and ∑TU, which suggests that PDMS might not be the optimal biomimetic phase as has also been suggested in earlier studies.43 In summary, passive dosing was found to be a practical and high performing method for conducting mixture toxicity experiments while tightly controlling both exposure levels and compositions of mixtures, and ∑a, ∑Clipid eq., and ∑TU provided three possibilities for linking mixture toxicity to mixture exposure with high descriptive and predictive capabilities.

■

■

ASSOCIATED CONTENT

S Supporting Information *

Detailed calculation of PAH depletion in methanol loading solutions, three figures, and two tables. Figure S1 shows springtail lethality caused by the 12 mixture treatments as a function of sum chemical activities and exposure time. Figure S2 shows the exposure confirmation of naphthalene, phenanthrene, and pyrene on linear scales. Figure S3 shows springtail lethality caused by the 12 mixture treatments as a function of sum PDMS concentrations. Table S1 shows the exposure compositions and levels of the 12 mixtures treatments. Table S2 shows the exposure−response equations and 95% confidence interval for each of the four exposure parameters. This information is available free of charge via the Internet at http://pubs.acs.org.

■

REFERENCES

(1) Backhaus, T.; Faust, M. Predictive environmental risk assessment of chemical mixtures: A conceptual framework. Environ. Sci. Technol. 2012, 46 (5), 2564−2573. (2) Schwarzenbach, R. P.; Escher, B. I.; Fenner, K.; Hofstetter, T. B.; Johnson, C. A.; von Gunten, U.; Wehrli, B. The challenge of micropollutants in aquatic systems. Science 2006, 313 (5790), 1072− 1077. (3) Di Toro, D. M.; McGrath, J. A. Technical basis for narcotic chemicals and polycyclic aromatic hydrocarbon criteria. II. Mixtures and sediments. Environ. Toxicol. Chem. 2000, 19 (8), 1971−1982. (4) Syberg, K.; Jensen, T. S.; Cedergreen, N.; Rank, J. On the use of mixture toxicity assessment in REACH and the Water Framework Directive: A Review. Hum. Ecol. Risk Assess. 2009, 15 (6), 1257−1272. (5) Alexander, M. Aging, bioavailability, and overestimation of risk from environmental pollutants. Environ. Sci. Technol. 2000, 34 (20), 4259−4265. (6) Hurdzan, C. M.; Lanno, R. P. Determining exposure dose in soil: The effect of modifying factors on chlorinated benzene toxicity to earthworms. Chemosphere 2009, 76 (7), 946−951. (7) Schreiber, R.; Altenburger, R.; Paschke, A.; Kuster, E. How to deal with lipophilic and volatile organic substances in microtiter plate assays. Environ. Toxicol. Chem. 2008, 27 (8), 1676−1682. (8) Mayer, P.; Wernsing, J.; Tolls, J.; de Maagd, P. G. J.; Sijm, D. Establishing and controlling dissolved concentrations of hydrophobic organics by partitioning from a solid phase. Environ. Sci. Technol. 1999, 33 (13), 2284−2290. (9) Brown, R. S.; Akhtar, P.; Akerman, J.; Hampel, L.; Kozin, I. S.; Villerius, L. A.; Klamer, H. J. C. Partition controlled delivery of hydrophobic substances in toxicity tests using poly(dimethylsiloxane) (PDMS) films. Environ. Sci. Technol. 2001, 35 (20), 4097−4102. (10) Mayer, P.; Holmstrup, M. Passive dosing of soil invertebrates with polycyclic aromatic hydrocarbons: Limited chemical activity explains toxicity cutoff. Environ. Sci. Technol. 2008, 42 (19), 7516− 7521. (11) Bandow, N.; Altenburger, R.; Lubcke-von Varel, U.; Paschke, A.; Streck, G.; Brack, W. Partitioning-based dosing: An approach to include bioavailability in the effect-directed analysis of contaminated sediment samples. Environ. Sci. Technol. 2009, 43 (10), 3891−3896. (12) Smith, K. E. C.; Dom, N.; Blust, R.; Mayer, P. Controlling and maintaining exposure of hydrophobic organic compounds in aquatic toxicity tests by passive dosing. Aquat. Toxicol. 2010, 98 (1), 15−24. (13) Kwon, H.-C.; Roh, J.-Y.; Lim, D.; Choi, J.; Kwon, J.-H. Maintaining the constant exposure condition for an acute Caenorhabditis elegans mortality test using passive dosing. Environ. Health Toxicol. 2011, 26, e2011015. (14) Schmidt, S. N.; Smith, K. E. C.; Holmstrup, M.; Mayer, P. Uptake and toxicity of polycyclic aromatic hydrocarbons in terrestrial springtails − Studying bioconcentration kinetics and linking toxicity to chemical activity. Environ. Toxicol. Chem. 2013, 32 (2), 361−369. (15) Engraff, M.; Solere, C.; Smith, K. E. C.; Mayer, P.; Dahllof, I. Aquatic toxicity of PAHs and PAH mixtures at saturation to benthic amphipods: Linking toxic effects to chemical activity. Aquat. Toxicol. 2011, 102 (3−4), 142−149. (16) Rojo-Nieto, E.; Smith, K. E. C.; Perales, J. A.; Mayer, P. Recreating the seawater mixture composition of HOCs in toxicity tests with Artemia f ranciscana by passive dosing. Aquat. Toxicol. 2012, 120− 121, 27−34. (17) Smith, K. E. C.; Schmidt, S. N.; Dom, N.; Blust, R.; Holmstrup, M.; Mayer, P. Baseline toxic mixtures of non-toxic chemicals:

AUTHOR INFORMATION

Corresponding Author

*Phone: +45 87158952; fax: +45 87155010; e-mail: stns@dmu. dk. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We greatly thank Margit M. Fernqvist for her guidance and assistance with passive dosing, Elin Jørgensen and Zdenek Gavor for their guidance and assistance with Folsomia candida, 7026

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027

Environmental Science & Technology

Article

“Solubility addition” increases the exposure for solid hydrophobic chemicals. Environ. Sci. Technol. 2013, 47 (4), 2026−2033. (18) Landrum, P. F.; Chapman, P. M.; Neff, J.; Page, D. S. Evaluating the aquatic toxicity of complex organic chemical mixtures: Lessons learned from polycyclic aromatic hydrocarbon and petroleum hydrocarbon case studies. Integr. Environ. Assess. Manage. 2012, 8 (2), 217−230. (19) Reichenberg, F.; Mayer, P. Two complementary sides of bioavailability: Accessibility and chemical activity of organic contaminants in sediments and soils. Environ. Toxicol. Chem. 2006, 25 (5), 1239−1245. (20) Bliss, C. I. The toxicity of poisons applied jointly. Ann. Appl. Biol. 1939, 26 (3), 585−615. (21) Van Wezel, A. P.; Opperhuizen, A. Narcosis due to environmental pollutants in aquatic organisms: Residue-based toxicity, mechanisms, and membrane burdens. Crit. Rev. Toxicol. 1995, 25 (3), 255−279. (22) Di Toro, D. M.; McGrath, J. A.; Hansen, D. J. Technical basis for narcotic chemicals and polycyclic aromatic hydrocarbon criteria. I. Water and tissue. Environ. Toxicol. Chem. 2000, 19 (8), 1951−1970. (23) Mayer, P.; Torang, L.; Glaesner, N.; Jonsson, J. A. Silicone membrane equilibrator: Measuring chemical activity of nonpolar chemicals with poly(dimethylsiloxane) microtubes immersed directly in tissue and lipids. Anal. Chem. 2009, 81 (4), 1536−1542. (24) van der Heijden, S. A.; Jonker, M. T. O. Evaluation of liposomewater partitioning for predicting bioaccumulation potential of hydrophobic organic chemicals. Environ. Sci. Technol. 2009, 43 (23), 8854−8859. (25) Sprague, J. B. Measurement of pollutant toxicity to fish. 2. Utilizing and applying bioassay results. Water Res. 1970, 4 (1), 3−32. (26) Berenbaum, M. C. The expected effect of a combination of agents - The general solution. J. Theor. Biol. 1985, 114 (3), 413−431. (27) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons, Inc.: Hoboken, New Jersey, USA, 2003. (28) Shaw, D. G. IUPAC Solubility Data Series. Hydrocarbons (C8C36) with Water and Seawater; Pergamon Press: Oxford, U.K., 1989. (29) GraphPad Software. http://graphpad.com/quickcalcs/Grubbs1. cfm. 2012 (accessed 8/14/2012). (30) Yalkowsky, S. H.; Orr, R. J.; Valvani, S. C. Solubility and partitioning. 3. The solubility of halobenzenes in water. Ind. Eng. Chem. Fundam. 1979, 18 (4), 351−353. (31) Lide, D. R. Handbook of Chemistry and Physics; CRC Press: Boca Raton, FL, USA, 2003. (32) Jahnke, A.; McLachlan, M. S.; Mayer, P. Equilibrium sampling: Partitioning of organochlorine compounds from lipids into polydimethylsiloxane. Chemosphere 2008, 73 (10), 1575−1581. (33) Gouliarmou, V.; Smith, K. E. C.; de Jonge, L. W.; Mayer, P. Measuring binding and speciation of hydrophobic organic chemicals at controlled freely dissolved concentrations and without phase separation. Anal. Chem. 2012, 84 (3), 1601−1608. (34) Di Toro, D. M.; Zarba, C. S.; Hansen, D. J.; Berry, W. J.; Swartz, R. C.; Cowan, C. E.; Pavlou, S. P.; Allen, H. E.; Thomas, N. A.; Paquin, P. R. Technical basis for establishing sediment quality criteria for nonionic organic chemicals using equilibrium partitioning. Environ. Toxicol. Chem. 1991, 10 (12), 1541−1583. (35) Mayer, P.; Reichenberg, F. Can highly hydrophobic organic substances cause aquatic baseline toxicity and can they contribute to mixture toxicity? Environ. Toxicol. Chem. 2006, 25 (10), 2639−2644. (36) Mackay, D.; Arnot, J. A.; Petkova, E. P.; Wallace, K. B.; Call, D. J.; Brooke, L. T.; Veith, G. D. The physicochemical basis of QSARs for baseline toxicity. SAR QSAR Environ. Res. 2009, 20 (3−4), 393−414. (37) Ferguson, J. The use of chemical potentials as indices of toxicity. Proc. R. Soc. B-Biol. Sci. 1939, 127 (848), 387−404. (38) Redman, A. D.; Parkerton, T. F.; McGrath, J. A.; Di Toro, D. M. PETROTOX: An aquatic toxicity model for petroleum substances. Environ. Toxicol. Chem. 2012, 31 (11), 2498−2506. (39) Swartz, R. C.; Schults, D. W.; Ozretich, R. J.; Lamberson, J. O.; Cole, F. A.; Dewitt, T. H.; Redmond, M. S.; Ferraro, S. P. Sigma-PAH

- A model to predict the toxicity of polynuclear aromatic hydrocarbon mixtures in field-collected sediments. Environ. Toxicol. Chem. 1995, 14 (11), 1977−1987. (40) Leslie, H. A.; Ter Laak, T. L.; Busser, F. J. M.; Kraak, M. H. S.; Hermens, J. L. M. Bioconcentration of organic chemicals: Is a solidphase microextraction fiber a good surrogate for biota? Environ. Sci. Technol. 2002, 36 (24), 5399−5404. (41) Leslie, H. A.; Hermens, J. L. M.; Kraak, M. H. S. Baseline toxicity of a chlorobenzene mixture and total body residues measured and estimated with solid-phase microextraction. Environ. Toxicol. Chem. 2004, 23 (8), 2017−2021. (42) Parkerton, T. F.; Stone, M. A.; Letinski, D. J. Assessing the aquatic toxicity of complex hydrocarbon mixtures using solid phase microextraction. Toxicol. Lett. 2000, 112, 273−282. (43) Verbruggen, E. M. J.; Vaes, W. H. J.; Parkerton, T. F.; Hermens, L. M. Polyacrylate coated SPME fibers as a tool to simulate body residues and target concentrations of complex organic mixtures for estimation of baseline toxicity. Environ. Sci. Technol. 2000, 34 (2), 324−331.

7027

dx.doi.org/10.1021/es3047813 | Environ. Sci. Technol. 2013, 47, 7020−7027