Environ. Sci. Technol. 2002, 36, 869-876
Ozonation of Chrysene: Evaluation of Byproduct Mixtures and Identification of Toxic Constituent S. L. LUSTER-TEASLEY,† J. J. YAO,† H. H. HERNER,† J. E. TROSKO,‡ AND S . J . M A S T E N * ,† Department of Civil and Environmental Engineering and Department of Pediatric Medicine and Human Development, Michigan State University, East Lansing, Michigan 48824
The effects of chrysene and the ozonated byproducts on in vitro gap junctional intercellular communication (GJIC) were evaluated using the scrape loading/ dye transfer (SL/ DT) technique. A 1 mM solution of chrysene was ozonated at dosages of 1.75, 3, 4.25, and 5 mol O3/mol chrysene (Chr). The early ozonation mixture, 1.75 mol O3/mol Chr, exhibited greater inhibition to GJIC than chrysene and irreversible damage to cells leading to cell death. To determine the compounds potentially responsible for the increase in toxicity, the byproducts formed upon treatment with 1.44 mol O3/mol Chr were separated into 14 fractions using RPHPLC. The major compounds identified in the fractions were 2-(2′-formyl) phenyl-1-naphthaldehyde, 2-(2′formyl) phenyl-1-naphthoic acid, and 2-2-carboxyphenyl-1-naphthoic acid. 2-(2′-Formyl) phenyl-1-naphthaldehyde was determined to be the compound causing GJIC inhibition in sample fractions and byproduct mixtures.
Introduction Polycyclic aromatic hydrocarbons (PAH) are an important class of environmental carcinogens found in many petroleum hazardous waste sites. PAH compounds at these sites are often present as components of nonaqueous phase liquids (NAPLs) such as coal tars, creosotes, and petroleum distillates (1). Unfortunately, due to years of improper disposal, PAH contamination has become a major environmental concern in both soil and groundwater. These compounds present a remediation challenge because they are highly recalcitrant, insoluble in water, and tend to accumulate on solid surfaces (2). The U.S. Environmental Protection Agency (EPA) lists 16 PAHs as priority pollutants and eight as carcinogens or potential carcinogens causing skin, liver, and/or lung cancer in humans. The need for effective methods to remediate PAH compounds has resulted in a considerable amount of research investigating the use of ozone for the destruction of these compounds (3-8). In-situ treatment technologies using ozone have also shown promise in remediating both contaminated groundwater and soil (8-15). While ozone is effective at oxidizing specific PAH contaminants, ozonation results in the formation of numerous PAH byproduct derivatives (5). The potential production of ozonation byproducts must be considered when evaluating the efficacy * Corresponding author phone: (517)353-8539; fax: (517)355-0250; e-mail:
[email protected]. † Department of Civil and Environmental Engineering. ‡ Department of Pediatric Medicine and Human Development. 10.1021/es011090q CCC: $22.00 Published on Web 01/17/2002
2002 American Chemical Society
of remediation efforts. Remediation goals are often impaired by a lack of information on the byproducts produced, on individual byproduct toxicity, and on the interactive toxicity of complex mixtures remaining following treatment efforts. Acknowledgment of these limitations has led to the development and implementation of toxicity reduction, identification, and evaluation (TRE/TIE) programs (16). The primary focus of risk assessment for PAH contaminated sites is to evaluate the carcinogenic risk. Predictions and models for risk assessments are then based on empirical or epidemiological data, suspected mechanisms of action, and extrapolation of human effects from laboratory animal studies (17-19). Considerable research is available for genotoxicity (mutagenicity) and on cytotoxicity (16, 20-28). The first stage of cancer requires an irreversible genotoxic/ mutagenic; however, the second stage of cancer development involves tumor promotion, which would result from an epigenetic event (1, 17). Some PAHs have been proven to be epigenetically toxic by being able to disrupt gap junctional intercellular communication (GJIC) (1, 29-32). The GJIC assay is a nongenotoxic assay that measures the ability for cells to transfer information, low molecular weight molecules, and small regulatory and macromolecular substances through the cytoplasm of one cell to the next cell through channels called gap junctions. GJIC controls cell homeostasis, tumor promotion, and cell synchronization. Only a few studies have investigated epigenetic toxicity of ozonation byproducts using the GJIC assay (1, 32-35). Byproducts formed during aqueous and soil ozonation could present a potential risk for continued exposure to toxicants despite the removal of the original toxic PAHs. Therefore, it is important to understand the toxicity of these mixtures and to identify the toxic constituents if we are to truly assess the efficacy of ozonation for remediation. In this study, chrysene was selected as the target PAH. Chrysene is identified as a weak carcinogen that promotes skin, liver, and lung carcinomas (36). It is found in smoke and soot from burning coal, gasoline, and organic matter. The general population may be exposed to chrysene in anthropogenic combustion emissions, from dust particles carried to water, soil, and crops, cigarette smoke, and coal/fire cooked food. Exposure to chrysene and other PAHs may also occur through skin contact with products containing creosote-treated wood, asphalt roads, or coal tar. Copeland et al. (3) proposed that chrysene reacts with ozone at the 5,6 or 11,12 bonds resulting in 48% yield of 2-2-carboxyphenyl-1-naphthoic acid. 2-2Carboxyphenyl-1-naphthoic acid, together with a compound thought to be a lactone, was produced when chrysene was ozonated and treated with hydrogen peroxide in acetic acid (5). Rodd (4) presented findings where the oxidation of chrysene with sodium dichromate in acetic acid yields chrysene-5, 6-quinone, and 5,6-dihydro-5,6-dihydroxide chrysene. To determine the epigenetic toxicity of chrysene and the resultant ozonation byproducts, chrysene was ozonated, and the toxicity of the mixture of byproducts produced was evaluated using a nongenotoxic assay measuring gap junctional intercellular communication (GJIC) in rat liver epithelial cells. The ozonated mixtures that demonstrated an increased toxicity were fractionated using reverse phase HPLC (RP-HLPC). These fractions were then tested for their ability to inhibit cellular communication using the GJIC assay and to identify the toxic constituents. The complete study consisted of dose response, time response, time recovery, and cytotoxicity for the ozonated compounds. VOL. 36, NO. 5, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
869
Experimental Section Chemicals. Chrysene (98% purity, Aldrich Chemicals) was dissolved in acetonitrile (ACN) (99.8% purity, EM Science, Gibbstown, NJ) and adjusted to a pH 3-4 using acidified deionized water to make a 1 mM solution. The low solubility of chrysene in pure water (6 µg/L) required using an acetonitrile/water (ACN/H2O) mixture for ozonation. Acetonitrile was selected as the solvent because it has a low reactivity with ozone (t1/2 g 18 years at pH 7 and [O3] ) 20.8 mM). A 10% water concentration is sufficient to act as a participating solvent in ozonolysis (37). Since, at higher aqueous concentrations, chrysene precipitated, a 90%/10% (v/v) (ACN/H2O) solution was used in all experiments. The molecular weight of chrysene (MW 228) was used to calculate the stoichiometric ratios for ozone dosages and concentrations. The ozonated samples were identified by the ozone dosage applied using the units mole ozone/mol Chr (mol O3/mol Chr). The ozonated samples ranged from 1.75 to 5 mol O3/mol Chr. For the toxicology studies, dried ozonated samples were dissolved in ACN, and the units were reported as µM as Chr. The molecular weight of chrysene was used as the basis to calculate the concentrations because the exact molecular weight of the ozonated mixtures was unknown. The µM as Chr can be converted to mg/L ozonated sample by multiplying the concentration by the molecular weight of chrysene and dividing by 1000. Ozone Treatment. A semibatch system was used to ozonate the solution (37). Ozone was generated in dried oxygen using a corona discharge ozone generator (Polymetrics Model T-408, San Jose, CA). The flow of ozone into the reactor was regulated at 200 mL/min using a Sidetrack mass flow controller (Sierra Instruments Inc., Monterey, CA). The tubing (1/8” i.d.), connectors, and valves were constructed of Teflon or 316 stainless steel. The concentrations of ozone in the influent and effluent gas streams were measured spectrophotometrically at 258 nm using an UVvisible light spectrophotometer (Model 1201, Shimadzu Scientific Instruments, Japan). The absorbance values for ozone were converted to concentration units using a molar absorptivity coefficient for ozone of 3000 M-1 cm-1 (37). After the solutions were ozonated for the desired time (i.e., ozone concentration), the solution was flushed with helium to purge residual ozone and to terminate ozone reactions. The sample was removed, and 2 g of Na2S2O3 was added as a free radical quenching agent. Sodium nitrite, commonly used to quench ozone, was not used so as to avoid subsequent radical reactions. The ozonated samples were then kept in the dark and shaken at 250 rpm overnight in 250 mL glass bottles. Solid Na2S2O3 was filtered from the solution. The sample was then rotary evaporated (Buchi WaterBath with Rotavapor, Brinkman, Westbury, NY) to recover the solid byproduct. The dry sample was stored in a freezer at 13 °C until it was resuspended in pure acetonitrile for reverse phase HPLC analysis and toxicity studies. RP-HPLC Analysis. A Gilson HPLC unit (Worthington, OH), UV detector (Model 116), and an Altima C18 reverse phase column (5 µM) with dimensions 4.6 × 250 mm were used for HPLC Analysis (Alltech Co., Deerfield, IL). The effluent was monitored at two wavelengths (225 and 260 nm). Pure ACN (99.8% purity) and HPLC grade water were used as the mobile phases. The linear gradient used for the mobile phase was 25%/75% ACN/H2O at the time of injection and increased to 90%/10% ACN/H2O over 15 min. The mobile phase was then held at 90%/10% ACN/H2O for 3 min and then linearly decreased to 25%/75% ACN/H2O over 2 min and held for an additional 5 min at this ratio. The total run time was 25 min. Sample Purification. Samples were resuspended in ACN just prior to fractionation. Fractionation of the solid byprod870
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 5, 2002
uct was performed using a Gilson HPLC unit (Worthington, OH) equipped with two pumps (Model 303), manometric module (Model 802b), dynamic mixer (Model 811), UV detector (Model 116), sample injector (Model 231) and diluter (Model 401), coupled with Altima C18 column (5 µM, Alltech Co., Deerfield, IL). A 100 µL injection loop was used, and the eluent was monitored at 240 nm because this wavelength provided the best resolution of peaks. The mobile phase flowrate was 1 mL/min. The HPLC linear gradient method discussed in the HPLC analysis section above was used to fractionate the sample. The fractionated samples were then rotary evaporated until complete dryness and stored in a freezer. Identification. Fractionated samples were evaporated under helium, and the residue was dried over P2O5 in vacuo for 8 h. The completely dried sample was derivatized by silylation using bis-trimethylsilyl/trifluoracetamide (BSTFA) and 1% of trimethylchlorosilane (TMCS) (Regis Technologies, Inc., Morton Grove, IL) at 100 °C for 1 h to convert all substituted -OH and -COOH groups to their respective volatile TMS-ether (-OSiMe3) or TMS-ester (-CO2SiMe3) derivatives. GC/MS was performed using a JEOL AX-505H double-focusing mass spectrometer coupled with a HewlettPackard 5890J GC (Norwalk, CT). A DB5MS (30 m length × 0.32 mm i.d. × 0.25 µM film thickness) fused silica capillary column (J&W Scientific, Rancho Cordova, CA) was employed for GC separation. A splitless injector was used with a column head pressure of 10 psi using helium as the carrier gas, producing a flow rate of ca. 1 mL/min. The initial column temperature was held for 2 min at 100 °C, ramped at 20 C/min to 220 °C, then ramped at 5 °C/min to 280 °C, and finally ramped at 20 °C/min to 300 °C. The mass spectrometer was operated in electron impact mode. Mass calibration of the spectrometer was performed using perfluorkerosine. Toxicology Studies. Dried ozonated byproduct and fractions were dissolved in acetonitrile (99.8% purity, EM Science, Gibbstown, NJ). Acetonitrile (ACN) was selected as the solvent because it has little effect on GJIC at a final ACN concentration up to 1.75% in cell culture medium (33). All experiments were conducted at an ACN concentration of 1% or less in cell culture medium. Chrysene was treated at stoichiometric ratios of 1.75, 3, 4.25, and 5 mol O3/mol Chr, and 14 fractionated samples from a 1.44 mol O3/mol Chr sample were analyzed for toxicity. Cell Culture. WB-F344 rat epithelial cell lines were obtained from Dr. J. W. Grisham and M. S. Tsao of the University of North Carolina (Chapel Hill, NC). Cells were cultured in 25 mL of D medium (Formula No. 78-5470EG, GIBCO Laboratories, Grand Island, NY) containing 5% fetal bovine serum (FBS) (GIBCO Laboratories, Grand Island, NY) and 0.2% gentamycin. The cells were incubated at 37 °C in a humidified atmosphere containing 5% CO2 and 95% air. The cells were grown in 150 mm plastic flasks, and the culture was split and new medium was added every other day. WBF344 are a diploid, nontumorigenic cell line on which a large database of toxicity/cancer information for other in vivo tested chemicals can be used to compare results (38). Bioassay for GJIC. Bioassays were conducted in 35 mm2 Petri dishes with confluent cultures grown for 2 days in 2 mL of D medium supplemented with 5% fetal bovine serum. The procedure for the scrape loading/dye transfer (SL/DT) technique was adapted from the method used by El-Fouly et al. (39) and is described in detail by Herner et al. (35). All tests were run in triplicate and at noncytotoxic levels determined by the neutral red uptake assay kit (Sigma Chemical Co., St. Louis, MO). Three photographs were taken for each concentration tested, and the distance the dye traveled perpendicular to the scrape was measured. All photographs were taken within 1 h of experiment completion.
FIGURE 1. RP-HPLC profile of chrysene and ozonation byproduct mixtures: (A) chrysene; (B) 0.41 mol O3/mol Chr; (C) 1.75 mol O3/mol Chr; (D) 3 mol O3/mol Chr; (E) 5 mol O3/mol Chr. Under fluorescent light, the Lucifer yellow dye will fluoresce to indicate the distance the dye travels from the scrape. This distance was measured and compared to a control group of cells that were exposed to acetonitrile only (vehicle controls) but assayed using the identical SL/DT method. For each picture, measurements were taken every 1 cm for a total of 10 cm. A total of 30 measurements (10 from each photo) were averaged together to obtain a representative fraction of control (f). The results were reported as an average ( standard deviation determined at the 95% confidence interval (C.I.). GJIC was assessed by the decrease in communication of the cells exposed to the toxicant compared to the vehicle control group (acetonitrile only). Complete communication is identified as an f value of 1.0 or 100% communication seen in the control. GJIC values less than an f value 0.5 fractional value (50%) indicate a significant decrease because the cells are communicating at half or less than half of normal communication levels. GJIC values between 0.0 and 0.3 are considered to correspond to no intercellular communication. These interpretations for GJIC are consistent with Upham et al. (32) and that of Herner et al. (35). Controls were run for each experiment to standardize the measurement for normal cell-cell communication at the time of the experiment. Using
the t-test it was determined that the mean f values obtained from the control cells without solvent, and the vehicle controls do not differ at a 95% confidence interval (95% C.I.). Statistical analyses using the two-tailed t-test and F-test at 95% C.I. were used to compare within treatment and between treatment variations in the mean f values. Bioassay for Cytotoxicity. Cytotoxicity was tested using the neutral red uptake assay according to the method of Borenfreund and Puerner (40). WB-F344 cells were grown using the same method as the cells used for the GJIC assay. It is necessary to use noncytotoxic levels of toxicants in GJIC studies, since if levels used are cytotoxic, then differentiation between decreased communication due to cell death cannot be distinguished from decreased intercellular communication due to blockage of gap junctions. Therefore, to accurately measure decreased GJIC activity, only noncytotoxic concentrations were used.
Results RP-HPLC of Whole Byproduct Mixtures. Figure 1 presents the RP-HPLC chromatograms for chrysene and the byproducts for stoichiometric ratios of 0.41, 1.75, 3, and 5 mol O3/ mol Chr. Pure chrysene had a retention time of 19-20 min. VOL. 36, NO. 5, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
871
FIGURE 2. Dose response resulting from contact with ozonation byproduct mixtures. Cells exposed for 30 min. b Chrysene, 41.75 mol O3/mol Chr, 0 3 mol O3/mol Chr, [ 4.25 mol O3/mol Chr, 3 5 mol O3/mol Chr. A GJIC (f) value 1 is indicative of total communication, while an f e 0.3 is indicative of complete inhibition of communication. The mean f values for 1.75 mol O3/mol Chr were statistically different from controls and chrysene at the 95% confidence level. The 3 mol O3/mol Chr mean f values did not differ from those of chrysene for all doses tested. Both 4.25 and 5 mol O3/mol Chr means did not differ from the vehicle controls at concentrations less than 50 µM. The F-test indicated that there is no reason to believe that the precision in the standard deviations differ. As the ozone concentration increased, chrysene quickly degraded, and the byproducts formed increased in polarity. The production of polar compounds was evident with the detection of compounds with retention times of 2-3 min. As the mobile phase was linearly increased to 90%/10% ACN/ H2O, the less polar, more hydrophobic organic compounds were separated and eluted from the column. The concentrations of the compounds having retention times of 3-16 min increased with increasing ozone concentration. The intensity of the peaks that eluted after 18.6 min decreased with increasing ozonation.
GJIC Evaluation Dose Response. For dose response experiments, cells were exposed to varying doses of a toxicant for 30 min and then assayed to determine GJIC levels. Figure 2 is a comparison of the dose response curves for the mixtures tested. The results of the dose experiments show that chrysene moderately inhibits gap junctional communication. Due to the low solubility of chrysene, concentrations greater than 50 µM as Chr could not be tested. Ozonated samples exhibited higher solubility than chrysene with the samples becoming more readily dissolved in ACN with increasing ozone dose. Therefore, concentrations greater than 50 µM as Chr for the ozonated samples could be evaluated for GJIC inhibition. Inhibition of GJIC was greatest for the 1.75 mol O3/mol Chr sample, while the f values for the 3 mol O3/mol as Chr were similar to that obtained with chrysene. At this ozone concentration, the chrysene peak had completely disappeared and only the byproduct peaks remained. The mixtures generated using greater ozone dosages (i.e., 4.25 and 5 mol O3/mol Chr) have f > 0.9 at concentrations less than 50 µM 872
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 5, 2002
as Chr, indicating minimal or no inhibition of cellular communication. Inhibition for the 5 mol O3/mol Chr was not seen until concentrations of 180-240 µM as Chr were reached, where f ) 0.95 at 180 µM as Chr and f ) 0.6 at 240 µM as Chr concentration. Error bars are not included in Figures 2 and 3b for the sake of readability; however, all differences are consistent and reproducible with the differences for f values no greater than ( 0.06. Determining Toxic Compounds and Byproduct Identification. To determine the compounds potentially responsible for the increase in toxicity, a 1.44 mol O3/mol Chr sample was separated into 14 fractions (F1-F14) using RP-HPLC (Figure 3a). The earlier fractions (F1-F6) were separated so that little to no impurities were detected. During the fractionation process, it was observed that as the ACN concentration increased, the difficulty in obtaining pure fractionated samples also increased. Fraction 7 (F7) was still very pure; the unidentified impurities accounted for only a very small percentage. While the remaining fractions (F8F14) were impure, the major compounds in each fraction were identified. The impurities in the fractionated samples could not be identified as the peaks associated with these compounds were either were too small (i.e., very small quantity) or were small compared to that of the major compounds (
