1936
Chem. Res. Toxicol. 2010, 23, 1936–1946
Acute and Chronic Toxicity toward the Bacteria Vibrio fischeri of Organic Narcotics and Epoxides: Structural Alerts for Epoxide Excess Toxicity Ulrike Blaschke,†,‡ Albrecht Paschke,† Ines Rensch,§ and Gerrit Schu¨u¨rmann*,†,‡ UFZ Department of Ecological Chemistry, Helmholtz Centre for EnVironmental Research, Permoserstrasse 15, 04318 Leipzig, Germany, Institute for Organic Chemistry, Technical UniVersity Bergakademie Freiberg, Leipziger Strasse 29, 09596 Freiberg, Germany, and UFZ Personnel Department, Helmholtz Centre for EnVironmental Research, Permoserstrasse 15, 04318 Leipzig, Germany ReceiVed August 31, 2010
The acute and chronic bacterial toxicity of 34 organic compounds comprising 19 baseline narcotics and 15 epoxides has been determined with regard to 30-min bioluminescence and 24-h growth inhibition in terms of EC50 (effective concentration 50%) values employing Vibrio fischeri. For the narcotics, linear regression of log EC50 on log Kow (octanol/water partition coefficient) yields r2 (squared correlation coefficient) and rms (root-mean-square error) values of 0.95 and 0.44 (30-min), and 0.94 and 0.34 (24h), respectively. Employing the resultant baseline narcosis models, toxicity enhancement (Te) values were derived as a ratio of narcosis-predicted over experimental EC50 for the epoxides. For seven aliphatic epoxides, log Te was below 1 in both assays, indicating narcosis-range toxicity with regard to 30-min bioluminescence and 24-h growth inhibition. Concerning eight nonaliphatic epoxides, log Te values up to 2.4 were observed, reflecting excess toxicity through an enhanced electrophilic reactivity of the compounds. Here, however, the intercorrelation between both assays was very low (r2 ) 0.09). The results are discussed in terms of electronic substituent effects activating an SN2-type epoxide reaction with nucleophilic protein sites and side-chain activation offering alternative electrophile-nucleophile reaction routes at side-chain sites, leading to respective structural alerts as indicators of excess toxicity. Surprisingly, 30-min bioluminescence appears to be slightly more sensitive to chemical stress than 24-h growth, which holds both for baseline narcotics and for most of the epoxides. This is also reflected by effective narcosis doses 50%, ED50, of 7.1 mmol/kg (30-min) and 7.7 mmol/kg (24-h) estimated from narcosis theory. Keeping in mind the different end points (bioluminescence vs growth) involved, this finding demonstrates that chronic toxicity is not always more sensitive than acute toxicity, calling for analyses with regard to further respective cases and associated mechanistic causes. Introduction Epoxides are electrophilic three-membered cyclic ethers with a substantial ring strain and can undergo ring-opening through a reaction with nucleophiles. The mutagenicity and carcinogenicity of epoxides have been reviewed thoroughly in previous reports (1). In addition to DNA reactivity, epoxides may also bind covalently to nucleophilic protein sites, both of which is in competition with enzymatic phase-I biotransformation through epoxide hydrolase (EH) forming 1,2-diols (2). Various unsaturated aliphatic and aromatic xenobiotics are subject to cytochrome P450 oxidation to yield epoxides (3), which may subsequently be detoxified through EH-mediated hydrolysis. The electrophilic reactivity of epoxides may also lead to skin sensitization (4), and in this context, conjugated dienes may act as prohaptens, leading to allergic contact dermatitis following P450-mediated oxidation to epoxide metabolites (5). From the mechanistic viewpoint, the covalent attack on epoxides by nucleophilic DNA and protein sites proceeds * Corresponding author. Tel: +49-341-235-1262. Fax: +49-341-2351785. E-mail:
[email protected]. † UFZ Department of Ecological Chemistry, Helmholtz Centre for Environmental Research. ‡ Technical University Bergakademie Freiberg. § UFZ Personnel Department, Helmholtz Centre for Environmental Research.
Scheme 1. SN2-Type Ring-Opening of Epoxides through a Reaction with a Nucleophile NuH
through an SN2-type ring-opening, resulting in a chemical modification of the endogenous biomolecule. The general reaction is shown in Scheme 1, with NuH representing the nucleophile. In case of reactive side chains such as R,βunsaturated carbonyl moieties, additional pathways of reactive toxicity may play a role, as will be outlined below in more detail. Concerning aquatic toxicology, only few studies have explored the potential hazard associated with epoxides. In early investigations, the acute fish toxicity of 12 oxirane derivatives was determined and explained in terms of both hydrophobicity quantified as log Kow (octanol/water partition coefficient) and chemical reactivity toward 4-(4-nitrobenzyl)pyridine (6), and epoxides have been subject to further analyses of electrophilic toxicity mechanisms (7). Moreover, the reactivity toward the glutathione (GSH) of four epoxides could partly explain their potency for algal growth inhibition (8). These findings suggest that the electrophilic reactivity of epoxides may play an important role for their aquatic toxicity, as was shown recently
10.1021/tx100298w 2010 American Chemical Society Published on Web 11/04/2010
Acute and Chronic Bacterial Toxicity of Epoxides
to be the case for R,β-unsaturated aldehydes, esters, and ketones (9) employing a kinetic GSH chemoassay (10). The marine Gram-negative bacterium Vibrio fischeri has been widely used as the test organism for screening the aquatic toxicity of contaminants, and in this respect have shown correlations with the toxicological responses of various other aquatic species (11). In the context of REACH (12), Vibrio fischeri has been discussed as an in vitro tool to support the human toxicology screening of contaminants (13), and recently, a flash assay variant of the 30-min bioluminescence has been used to investigate nanoparticle toxicity (14). In the present investigation, the acute and chronic aquatic toxicities of 15 aliphatic and nonaliphatic epoxides were determined employing two different bioassays with the bacterium Vibrio fischeri. Short-term toxicity was characterized through EC50 values quantifying the compound concentration yielding 50% inhibition of bioluminescence after 30 min of exposure. For the long-term toxicity, a 24-h growth inhibition assay was used, yielding respective EC50 values. To characterize the impact of reactivity on toxicity, baseline narcosis models for both bioassays were derived through additional analyses of 19 organic narcotics, quantifying the toxicity enhancement (Te) as a ratio of narcosis-predicted over experimental EC50 values (7, 15). The results reveal distinct relationships between electronic substituent effects and the reactive component to toxicity as modeled through Te, leading to structural alerts for the predictive screening of the bacterial excess toxicity of epoxides. Surprisingly, 30-min bioluminescence inhibition appears to be slightly more sensitive to both baseline narcotics and reactive toxicants than 24-h growth inhibition. Possible reasons include long-term adaptation to chemical stress as well as the difference in end points, calling for future comparative investigations of bioluminescence and growth inhibition with identical exposure regimes.
Materials and Methods Test Compounds. The test set comprises 34 organic compounds, covering 19 baseline narcotics (following the classification of ref 15) and 15 epoxides. The narcotics ethylene glycol, methanol, acetone, ethanol, 2-butanone, 1-propanol, 1-butanol, dichloromethane, chloroform, 1-hexanol, 1-heptanol, 1-octanol, and p-xylene were purchased from Merck (Darmstadt, Germany), and 2-methylpropanol, methyl-t-butyl ether, 1,2,4-trichlorobenzene, diphenyl ether, and 4-bromodiphenyl ether from Flucka (Buchs, Switzerland). With regard to the epoxides, 1,2-propylene oxide, 2,3-epoxypropyl iso propyl ether, glycidyl methacrylate, 2,2dimethyloxirane, 1,2-epoxybutane, n-butyl glycidyl ether, styrene oxide, and 2,3-epoxypropyl phenyl ether were obtained from Merck, and 1,2-epoxyhexane, 1,2-epoxyoctane, 1,2-epoxydodecane, ethyl 2,3-epoxy-3-phenylpropionate, chalcone epoxide, and oct-7-enyloxirane were from Alfa Aesar (Karlsruhe, Germany). All test compounds were used in at least “p.a.” purity, and their log Kow (octanol/water partition coefficient) values as well as logarithmic Henry’s law constants and water solubilities were calculated with EPISuite (16). Test Organism. The toxicity tests were conducted with the bioluminescent bacterium Vibrio fischeri strain NRRL B-11177 from the German Resource Centre for Biological Material (DSMZ, Braunschweig, Germany). Vibrio fischeri was originally classified as Photobacterium fischeri and later reclassified as AliiVibrio fischeri, after Urbanczyk et al. (17) had found new relationships in the family Vibrionaceae. Short-Term Bioassay (30-min Bioluminescence). The shortterm measurement of the bioluminescence of Vibrio fischeri cells in a luminometer (Hach Lange, Du¨sseldorf, Germany) was performed according to the manufacturer’s standard procedure (derived from the ISO 11348-1:2007 directive) at 15 °C. Employing
Chem. Res. Toxicol., Vol. 23, No. 12, 2010 1937 defrosted and regenerated bacteria, EC50 (effective concentration 50%) values were derived as compound concentrations yielding 50% inhibition of the bioluminescence (see below). For frosting the bacteria, cells were diluted and brought to the exponential growth phase by incubation at room temperature. Subsequently, the bacteria were concentrated by centrifugation and frozen in special sugar solution to be stored. Exposure to the test chemical was maintained for 30 min at 15 °C, followed by measurement of the maximum emission of the bioluminescence at 490 nm. The aqueous test samples had a volume of 1 mL and contained 2% (w/v) NaCl. Each test compound was run in two to four replicates (depending on how well the results from the individual runs agreed), with one blank control and nine concentrations per run. For volatile chemicals (see also below), the standard amounts of liquid were increased to a 3-fold volume as compared to the ISO standard instructions, employing shorter cuvettes with specifically manufactured caps to minimize the headspace. Long-Term Bioassay (24-h Growth). The long-term measurement concerns growth inhibition in terms of corresponding EC50 (concentration yielding 50% inhibition of growth) values at room temperature, and was also performed with the bacterium Vibrio fischeri. Cells were grown at room temperature and shaken at 75 rpm (Swip SM 25, Bu¨hler, Germany) in a medium (according to ISO 11348-1) consisting of 0.5 M NaCl, 44.2 mM NaH2PO4, 12.1 mM K2HPO4, 0.83 mM MgSO4, 3.8 mM (NH4)2HPO4, 41 mM glycerol, 5 g/L peptone, and 0.5 g/L yeast extract, respectively. At least once a week, 20 µL of the cultures were diluted with 5 mL of media. Moreover, the cultures were plated at least every half year to control their bioluminescence and for biological contamination to exclude a genetic drift. For the toxicity tests, cultures in the exponential growth phase, diluted 1:200 overnight, were used. Chemicals with a good water solubility (Sw g 1 mol/L) were diluted in 2% (w/v) salt solution. Less soluble compounds (Sw < 1 mol/L) were dissolved in dimethyl sulfoxide (DMSO) and then diluted in the aqueous solution. The DMSO concentration in the bioassay was kept below 0.1% (w/v) to avoid effects on bacterial growth. A blank control with 2% salt solution and when necessary an additional DMSO control was always available. Each compound was measured again at least twice in at least five different concentrations. The standard assay was made under sterile conditions when the culture media were mixed with cells from an overnight culture to a cell concentration of 104 cells/mL. After filling 5 mL of the mixture onto 100 mL glass flasks with ground-in glass stoppers, the test chemical was added in different concentrations. If possible, the amount of test chemical was not more than 5 µL. Otherwise, the amount increased until the dose-response curve could be measured. For volatile chemicals (see below), the standard assay was modified by mixing 5 mL of culture medium and the test chemical in a specifically manufactured glass flask. These flasks were similar to flasks of the nonvolatile assay but were closed with a cap and septum (22 mm silicone/PTFE septum, with 3.2 mm EPE, VWR, Germany). After overnight equilibration, cells of an overnight culture were injected through the septum to a cell concentration of 104 cells/mL. The subsequent handling followed the one for nonvolatile substances. The cell concentration was measured after inoculation and after 24 h of incubation with a CASY cell counter (Model TTC, Innovatis, Germany). For the measurement, 50 µL suspensions were transferred to 10 mL Casyton in a Casy-cup. For determining the 24-h EC50 value, the initial cell concentration was subtracted from the final concentration to yield the relative inhibition. EC50 Determination. For both the 30-min bioluminescence test and the 24-h growth inhibition test, the measured values were converted to percentage inhibition data. These latter data were used for generating concentration-response curves according to the fourparameter logistic function as implemented in SigmaPlot 11 (Systat Software Inc., Chicago, USA). For each compound, the EC50 value
1938
Chem. Res. Toxicol., Vol. 23, No. 12, 2010
Blaschke et al.
resulting from the concentration-response curve was calculated from the following equation:
y ) min +
(max - min) x -Hillslope 1+ EC50
( )
(1)
In eq 1, x denotes an exposure concentration value used for setting up the concentration-response curve, y the associated relative inhibition (expressed as the percentage number usually between 0 and 100), min and max the minimum and maximum relative inhibition values resulting from curve fitting, respectively, Hillslope the slope parameter of this Hill concentration-response model, and EC50 (as 4-th parameter) the EC50 value resulting from the curve fit. For volatile chemicals (see below), the test compound concentration was corrected for theoretical loss due to volatilization before generating the concentration-response curve. Excess Toxicity. First, the subset of narcotics was used to calibrate a regression equation for log EC50 vs log Kow (log EC50 ) a · log Kow + b with regression parameters a and b representing the baseline narcosis dependence on log Kow for the two bioassays (30-min bioluminescence inhibition and 24-h growth inhibition). Second, a comparison of experimental EC50 and the associated EC50 calculated from baseline narcosis as introduced by Lipnick (7) yielded a quantification of the toxicity enhancement Te in terms of
Te )
Results and Discussion
EC50(calc.) EC50(exp.)
(2)
Note that Te was originally termed “excess toxicity” (7). Because Te is defined as a (dimensionless) toxicity ratio and as such is not a toxicity itself, we prefer the term “toxicity enhancement” (which is also compatible with the abbreviation Te). Accordingly, the toxicity in terms of EC50 or LC50 associated with Te > 1 is called excess toxicity, and respective compounds are called excess-toxic. Equation 2 was applied to the 15 epoxides, and a threshold of log Te ) 1 was used to discriminate between narcotic-level (observed log Te < 1) and excess-toxic (observed log Te g 1) compounds. The threshold was chosen to account for expected statistical scatter within the narcosis level, assuming that log Te values 5% were classified as volatile and otherwise as nonvolatile. For the former,
Narcosis-level Toxicity toward Vibrio fischeri. In Tables 1 and 2, the Vibrio fischeri response toward 19 baseline narcotics is summarized in terms of log EC50 values and associated Hill slopes (characterizing the steepness of the concentration-response curves) for the 30-min bioluminescence and the 24-h growth inhibition assays, respectively. As can be seen from the Tables, log Kow ranges from -1.20 to 4.94, and a similar variation over six log units is observed for the bioluminescence EC50 (from 1.07 M to 1.01·10-6 M for methanol and 1-dodecanol, respectively, see Table 1). Interestingly, the 24-h growth inhibition EC50 varies by only 5 orders of magnitude (from 1.14 M to 2.21·10-5 M for ethylene glycol and 1-dodecanol, respectively, Table 2), indicating a correspondingly smaller sensitivity toward the variation in log Kow of organic contaminants. Note further that with regard to bioluminescence, a reproducible EC50 value could not be determined for 1-heptanol. Linear regression of log EC50 on log Kow led to
log EC50(lum)[M] ) -1.01((0.06)log Kow - 0.85((0.15) n ) 17, r2 ) 0.95,
(4)
rms ) 0.44, q2cv ) 0.93, rmscv ) 0.52, F1,15 ) 346 for 30-min bioluminescence inhibition (indicated by “lum”) and to
log EC50(growth)[M] ) -0.77((0.05)log Kow - 0.81((0.13) n ) 16, r2 ) 0.94,
(5)
2 rms ) 0.37, qcV ) 0.93, rmscv ) 0.42, F1,15 ) 351
for 24-h growth inhibition. For deriving eq 4, methyl t-butyl ether (MTBE) had been removed as a significant outlier, suggesting a different mode of action of this compound. The associated data distributions of log EC50 vs log Kow are shown in Figure 1. The above-mentioned larger sensitivity of luminescence EC50 toward variation in hydrophobicity translates into
Acute and Chronic Bacterial Toxicity of Epoxides
Chem. Res. Toxicol., Vol. 23, No. 12, 2010 1939
Table 1. Set of 19 Organic Baseline Narcotics with Predicted Log Kow Values, Experimental EC50 Values Characterizing 30-min Bioluminescence Inhibition of the Bacteria Vibrio fischeri, and Hill Slopes of the Associated Concentration-Response Curvesa name
CAS number
log Kow
EC50 [M]
(() ∆ EC50 [M]
log EC50 [M]
Hill slope
ethylene glycol methanol acetone ethanol 2-butanone 1-propanol 2-methylpropanol 1-butanol methyl-t-butyl etherb dichloromethane chloroform 1-hexanol 1-heptanolc 1-octanol p-xylene 1,2,4-trichlorobenzene diphenyl ether 4-bromodiphenyl ether 1-dodecanol
107-21-1 67-56-1 67-64-1 64-17-5 78-93-3 71-23-8 78-83-1 71-36-3 1634-04-4 75-09-2 67-66-3 111-27-3 111-70-6 111-87-5 106-42-3 120-82-1 101-84-8 101-55-3 112-53-8
-1.20 -0.63 -0.24 -0.14 0.26 0.35 0.77 0.84 1.43 1.34 1.52 1.82 2.31 2.81 3.09 3.93 4.05 4.94 4.77
1.70 1.07 1.12 × 10-1 5.60 × 10-1 3.21 × 10-2 1.17 × 10-1 1.87 × 10-2 2.82 × 10-2 3.00 × 10-4 4.53 × 10-2 4.50 × 10-3 6.00 × 10-4 n.d. 2.35 × 10-5 5.52 × 10-5 2.03 × 10-5 2.06 × 10-5 7.20 × 10-6 1.01 × 10-6
3.64 × 10-1 8.67 × 10-2 6.20 × 10-3 3.36 × 10-2 3.40 × 10-3 1.33 × 10-2 1.30 × 10-3 2.70 × 10-3 6.25 × 10-5 5.40 × 10-3 1.10 × 10-3 4.36 × 10-5 3.60 × 10-6 1.83 × 10-6 8.52 × 10-6 5.97 × 10-6 8.20 × 10-6 4.19 × 10-6 1.51 × 10-7
0.23 0.03 -0.95 -0.25 -1.49 -0.93 -1.73 -1.55 -3.52 -1.34 -2.35 -3.22 n.d. -4.63 -4.26 -4.69 -4.69 -5.14 -6.00
1.3 2.1 1.9 2.5 1.8 1.8 1.3 1.4 0.6 2.2 1.5 1.1 n.d. 1.2 1.1 1.4 0.7 0.8 0.8
a Logarithmic octanol/water partition coefficients, log Kow, were predicted using EPISuite (16). EC50 [mol/L] denotes the effective concentration yielding 50% bioluminescence inhibition after 30 min of exposure, ∆EC50 represents the associated standard error, Hill slope refers to eq 1, and n.d. denotes not determined. b The 30-min bacterial toxicity of methyl-t-butyl ether (MTBE) deviates significantly both from the baseline narcosis regression equation (eq 4) as well as from the intercorrelation between 30-min bioluminescence and 24-h growth inhibition (r2 ) 0.91 including MTBE, r2 ) 0.99 without MTBE; see text), suggesting a mode of action different from narcosis. c A reproducible EC50 could not be determined.
Table 2. Set of 19 Organic Baseline Narcotics with Experimental EC50 Values Characterizing the 24-h Growth Inhibition of the Bacteria Vibrio fischeri, and Hill Slopes of the Associated Concentration-Response Curvesa name
EC50 [M]
(() ∆EC50 [M]
log EC50 [M]
Hill slope
ethylene glycol methanol acetone ethanol 2-butanone 1-propanolb 2-methylpropanolb 1-butanol methyl-t-butyl ether dichloromethane chloroform 1-hexanol 1-heptanol 1-octanol p-xyleneb 1,2,4-trichlorobenzene diphenyl ether 4-bromodiphenyl ether 1-dodecanol
1.14 6.79 × 10-1 1.94 × 10-1 2.80 × 10-1 5.70 × 10-2 n.d. n.d. 5.46 × 10-2 6.47 × 10-2 1.83 × 10-2 1.92 × 10-2 1.60 × 10-3 7.00 × 10-4 2.00 × 10-4 n.d. 2.00 × 10-4 2.00 × 10-4 5.43 × 10-5 2.21 × 10-5
1.37 × 10-1 3.55 × 10-2 2.63 × 10-2 1.12 × 10-2 4.40 × 10-3 n.d. n.d. 3.30 × 10-3 9.30 × 10-3 5.00 × 10-4 1.90 × 10-3 3.00 × 10-4 8.74 × 10-5 9.75 × 10-5 n.d. 7.11 × 10-5 2.43 × 10-5 1.45 × 10-5 6.75 × 10-6
0.06 -0.17 -0.71 -0.55 -1.24 n.d. n.d. -1.26 -1.19 -1.74 -1.72 -2.80 -3.15 -3.70 n.d. -3.70 -3.70 -4.27 -4.65
2.7 5.7 2.7 6.2 3.4 n.d. n.d. 3.0 2.3 6.8 3.9 1.4 2.2 1.2 n.d. 1.5 3.0 0.8 0.8
a For CAS numbers and log Kow values, see Table 1. EC50 [mol/L] denotes the effective concentration yielding 50% growth inhibition after 24 h of exposure, ∆EC50 represents the associated standard error, Hill slope refers to eq 1, and n.d. denotes not determined. b Only the 30-min bioluminescence inhibition EC50 values were determined (see Table 1).
a correspondingly steeper negative slope (-1.01 vs -0.77), as can also be seen clearly from the Figure. The overall correlation between 30-min and 24-h log EC50 is r2 ) 0.91 (n ) 15 due to missing values for some compounds as listed in Tables 1 and 2) and increases to r2 ) 0.99 (n ) 14) when MTBE is omitted, again suggesting a non-narcotic mode of action of this compound with regard to 30-min bioluminescence inhibition. A further interesting issue is how the two assays compare with regard to their absolute sensitivities. Surprisingly, the 30min bioluminescence assay is more sensitive toward narcosislevel toxicants with log Kow > 0 than the 24-h growth inhibition (see Tables 1 and 2, and Figure 1). In other words, for the presently analyzed set of baseline narcotics, the chronic toxicity
Figure 1. Baseline narcosis log EC50 [mol/L] vs log Kow for 30-min bioluminescence (9) and 24-h growth (O) inhibition, using the data listed in Tables 1 and 2. The solid lines represent the linear regression predictions according to eqs 4 (n ) 17, r2 ) 0.95, rms ) 0.44; slope ) -1.01) and 5 (n ) 16, r2 ) 0.94, rms ) 0.37; slope ) -0.77) for the 30-min and 24-h end point, respectively.
toward Vibrio fischeri turns out to be smaller than the acute (or short-term) toxicity. Because of the above-mentioned difference in slope (relative sensitivity) between the two regression lines, this difference in absolute sensitivity increases with increasing log Kow (provided log Kow > 0). This finding demonstrates that the short-term performance of organisms can indeed be more sensitive to (chemical) stress than their long-term performance, keeping in mind that the 30-min end point (bioluminescence inhibition) differs substantially from the 24-h end point (growth inhibition). A further difference between these two end points concerns the steepness of the concentration-response curve as quantified through the Hill slope (defined through eq 1). As can be seen from Tables 1 and 2, the concentration-response curves associated with the 24-h growth inhibition are steeper in most cases than the short-term bioluminescence counterparts. Moreover, the growth inhibition Hill slopes vary more substantially,
1940
Chem. Res. Toxicol., Vol. 23, No. 12, 2010
Blaschke et al.
Scheme 2. Chemical Structures of the 15 Epoxides Listed in Tables 3 and 4a
a
1,2-Propylene oxide (A1), 2,2-dimethyloxirane (A2), 1,2-epoxybutane (A3), 1,2-epoxyhexane (B1), 1,2-epoxyoctane (B2), oct-7-enyloxirane (B3), 1,2epoxydodecane (C1), isopropyl glycidyl ether (C2), glycidyl methacrylate (C3), n-butyl glycidyl ether (D1), styrene oxide (D2), glycidyl phenyl ether (D3), ethyl 2,3-epoxy-3-phenylpropionate (E1), chalcone epoxide (E2), and allyl glycidyl ether (E3).
ranging from 1.2 (1-octanol) to 6.8 (dichloromethane) as compared to the bioluminescence inhibition slopes from 0.7 (diphenyl ether) to 2.5 (ethanol). As noted some time ago (18), the narcotic dose ED50 corresponding to narcosis-level exposure EC50 values can be estimated from narcosis regression relationships through
ED50 ) fL·10b
(6)
where fL denotes the fractional lipid content, and b the intercept of the narcosis regression equation. Assuming fL ) 0.05 and an organismic density of 1 g/cm3, b ) -0.85 from eq 4 yields ED50 ) 7.1 mmol/kg (0.05·10-0.85 ) 7.1·10-3 mol/L ≈ 7.1 mmol/kg) for 30-min bioluminescence, and with b ) -0.81 from eq 4, ED50 ) 7.7 mmol/kg is obtained for 24-h growth. Interestingly, these QSAR-derived narcotic doses that reflect the body burden associated with baseline narcosis (19) are in the LD50 (lethal dose 50%) range observed for the fish species guppy and fathead minnow (20). Moreover, the slightly larger (calculated) narcotic dose associated with 24-h growth inhibition indicates again its slightly smaller absolute sensitivity toward baseline narcotics as compared to that with 30-min bioluminescence inhibition. Recently, Vighi et al. (21) have published a 15-min bioluminescence narcosis regression equation with a similar slope (-0.94) but a substantially smaller negative intercept (-0.39) that was derived from 23 compounds with a log Kow range from -1.22 to 5.66 (r2 ) 0.92 and rms ) 0.512, calculated from the data of ref 21). Note that the intercept would correspond (according to eq 6) to a narcotic dose of 20 mmol/kg that appears to be unrealistically large. A possible explanation could be that the shorter exposure time may have been insufficient to ensure thermodynamic equilibrium. However, it was also reported that the differences
of the results from 5, 15, and 30 min of exposure have been negligible (21). Thus, further investigation may be needed to clarify these discrepancies that translate into a substantial difference in absolute sensitivity of the short-term bioluminescence assay by a factor of almost three. Epoxide Toxicity toward Vibrio fischeri. For seven epoxides with aliphatic side chains (including oct-7-enyloxirane with an isolated terminal double bond remote from the epoxide ring) and eight epoxides with side chains containing unsaturated moieties or heteroatoms or both (see Scheme 2), the 30-min and 24-h bacterial toxicities are summarized in Tables 3 and 4, respectively. Here, log Kow ranges from 0.37 (1,2-propylene oxide) to 4.79 (1,2-epoxydodecane), and bioluminescence EC50 from 1.35·10-1 M (log EC50 -0.87, 1,2-propylene oxide) to 1.35·10-6 M (log EC50 -5.87, chalcone epoxide; see Table 3). Again, the 24-h EC50 variation is smaller, covering only 2.5 orders of magnitude (log EC50 [M] from -1.88 to -4.33, Table 4). The Hill slopes (eq 1) quantifying the steepness of the concentration-response curves range from 0.8 to 2.0 in the case of the 30-min bioluminescence inhibition, and from 1.7 to 12.9 with regard to the 24-h growth inhibition. It follows that the latter assay shows steeper concentration-response curves with a larger slope variation for both narcosis level and reactive toxicity (see the Hill slope columns in Tables 1-4). As outlined in the Introduction, epoxides are electrophilic at the ring carbons with a regioselective preference for reactions at the unsubstituted ring carbon if available, and thus may undergo an SN2-type reaction (where the leaving group remains in the molecule) with nucleophilic groups of proteins and the DNA (see Scheme 1 above). While protein attack would be expected to result in excess toxicity, covalent binding at the DNA represents a mutagenic event possibly leading to genotoxicity. As can be seen from Table 3, all aliphatic epoxides
Acute and Chronic Bacterial Toxicity of Epoxides
Chem. Res. Toxicol., Vol. 23, No. 12, 2010 1941
Table 3. Set of 15 Epoxides with Predicted Log Kow Values, Experimental 30-min EC50 Values Characterizing Bioluminescence Inhibition of the Bacteria Vibrio fischeri, Logarithmic Toxicity Enhancement (Log Te) Values Regarding Excess Toxicity, Hill Slopes of Associated Concentration-Response Curves, and Literature Data on Mutagenicitya name
CAS number
log Kow
EC50 [M]
(() ∆EC50 [M]
log EC50 [M]
log Te
Hill slope
mutagenicity
Aliphatic Epoxides 1,2-propylene oxide 2,2-dimethyloxirane 1,2-epoxybutane 1,2-epoxyhexane 1,2-epoxyoctane oct-7-enyloxirane 1,2-epoxydodecane
75-56-9 558-30-5 106-88-7 1436-34-6 2984-50-1 85721-25-1 2855-19-8
0.37 0.83 0.86 1.85 2.83 3.67 4.79
1.35 × 10-1 1.08 × 10-2 9.90 × 10-3 1.30 × 10-3 1.00 × 10-4 3.09 × 10-5 6.51 × 10-6
1.26 × 10-2 2.20 × 10-3 8.00 × 10-4 4.00 × 10-4 3.50 × 10-5 3.80 × 10-6 1.47 × 10-6
-0.87 -1.97 -2.00 -2.89 -4.00 -4.51 -5.19
-0.35 0.28 0.29 0.17 0.30 -0.04 -0.49
2.0 1.0 1.4 0.8 0.9 1.1 1.0
+d +c +d +b -b +b,e -d
Nonaliphatic Epoxides allyl glycidyl ether isopropyl glycidyl ether glycidyl methacrylate n-butyl glycidyl ether styrene oxide glycidyl phenyl ether ethyl 2,3-epoxy-3-phenylpropionate chalcone epoxide
106-92-3 4016-14-2 106-91-2 2426-08-6 96-09-3 122-60-1 121-39-1 5411-12-1
0.45 0.52 0.81 1.08 1.59 1.61 2.55 3.36
4.00 × 10-4 3.01 × 10-2 6.00 × 10-4 2.30 × 10-3 1.52 × 10-5 8.06 × 10-5 9.57 × 10-6 1.35 × 10-6
9.65 × 10-5 1.30 × 10-3 6.34 × 10-5 3.00 × 10-4 1.20 × 10-6 6.69 × 10-6 1.14 × 10-6 7.03 × 10-8
-3.40 -1.52 -3.22 -2.64 -4.82 -4.09 -5.02 -5.87
2.09 0.15 1.56 0.70 2.36 1.62 1.60 1.63
0.8 1.7 1.2 1.1 0.8 0.8 0.9 1.1
+c +c +c +c +c +c -d n.a.
a Logarithmic octanol/water partition coefficients, log Kow, were predicted using EPISuite (16). EC50 [mol/L] denotes the effective concentration yielding 50% bioluminescence inhibition after 30 min of exposure, ∆EC50 represents the associated standard error, log Te quantifies the toxicity enhancement relative to baseline narcosis (eq 2), Hill slope refers to eq 1, (+) and (-) indicate mutagenic and not mutagenic, respectively, according to the literature, and n.a. denotes not available. b Vonderhude et al. 1991 (25). c Canter et al. 1986 (26). d National Toxicology Program (NTP), Salmonella TA100 (27). e The result for 1,2-epoxy-7-decene was used as the surrogate for oct-7-enyloxirane.
Table 4. Set of 15 Epoxides with Experimental 24-h EC50 Values Characterizing Growth Inhibition of the Bacteria Vibrio fischeri, Logarithmic Toxicity Enhancement (Log Te) Values Regarding Excess Toxicity, and Hill Slopes of the Associated Concentration-Response Curvesa EC50 [M]
(() ∆EC50[M]
log EC50[M]
log Te
Hill slope
Aliphatic Epoxides 1,2-propylene oxide 2,2-dimethyloxirane 1,2-epoxybutane 1,2-epoxyhexane 1,2-epoxyoctane oct-7-enyloxirane 1,2-epoxydodecane
name
1.33 × 10-2 2.28 × 10-2 1.67 × 10-2 4.70 × 10-3 6.00 × 10-4 2.00 × 10-4 >2.6 × 10-4 b
7.00 × 10-4 1.40 × 10-3 1.20 × 10-3 3.00 × 10-4 1.00 × 10-4 8.19 × 10-6 n.d.
-1.88 -1.64 -1.78 -2.33 -3.22 -3.70 >-3.59
0.78 0.19 0.30 0.09 0.22 0.05 1 for at least one of the two assays. These results indicate that epoxide activation through unsaturated or polar
1942
Chem. Res. Toxicol., Vol. 23, No. 12, 2010
Figure 2. Short-term (30-min) bioluminescence inhibition of aliphatic (9) and nonaliphatic (b) epoxides toward the bacteria Vibrio fischeri shown as log EC50 [mol/L] vs log Kow, using the toxicity data listed in Table 3. The solid line represents baseline narcosis (eq 4), and the broken line the threshold log Te ) 1 (toxicity enhancement, see eq 2) for discriminating between narcosis-level and excess toxicity.
Figure 3. Long-term (24-h) growth inhibition of aliphatic (9) and nonaliphatic (b) epoxides toward the bacteria Vibrio fischeri shown as log EC50 [mol/L] vs log Kow, using the toxicity data listed in Table 4. The solid line represents baseline narcosis (eq 5) and the broken line the threshold log Te ) 1 (toxicity enhancement, see eq 2) for discriminating between narcosis-level and excess toxicity.
side chains as well as the presence of reactive side chains may result in significant excess toxicities, as will be discussed in more detail in the next section. In Figures 2 and 3, epoxide log EC50 is plotted against log Kow for the short-term and longterm assays, respectively, showing also the lines representing baseline narcosis and the threshold log Te ) 1 for discriminating between narcosis level and significant excess toxicity. Coming back to the study of Deneer et al. (6), large Te values from 692 to 5880 (log Te 2.84-3.77) were reported for three heteroatom-activated epoxides, the overall largest Te of 6275 (log Te 3.80) for 1,3-butadienediepoxide where the two oxirane rings are subject to mutual negative inductive effects enhancing their electrophilicity (see the next section), and a still substantial Te value of 72 (log Te 1.86) for styrene oxide (the only activated epoxide that is also subject of the present investigation). Even for 1,2,7,8-diepoxyoctane, the fish toxicity Te was 149 (log Te 2.17) (6), although the two epoxide units are linked through an aliphatic C4 chain that appears to be too long for transmitting a significant inductive electronic effect.
Blaschke et al.
For the latter, a possible explanation could be that two oxirane rings in one molecule simply offer an enhanced probability for electrophile-nucleophile reactions. In this context, it is also possible that the intramolecular distance between the two epoxide rings imposed by their aliphatic C4 link might be particularly suited for approaching two different nucleophilic sites of proteins, including the possibility of cross-linking between different proteins and a resultant severe impairment of their physiological functioning. This hypothesis could be checked by testing aliphatic terminal diepoxides with different chain lengths, and comparing the resultant Te values with the ones for corresponding monoepoxides (keeping in mind the expected monotonic decrease in Te with increasing Kow for the latter). Apart from this issue, the qualitative patterns observed for Vibrio fischeri and Poecilia reticulata for activated epoxides are similar insofar as for both species, significant excess toxicities are observed. By contrast, the species sensitivity appears to differ for aliphatic (nonactivated) epoxides in the lower Kow range, where excess toxicity was observed only for fish (6) as outlined above. Excess Toxicity of Activated Epoxides and Epoxides with Reactive Side Chains. The subset of eight nonaliphatic epoxides contains four glycidyl ethers, one of which has an allylic moiety, and one a phenyl moiety (structures C2, D1, D3, and E3 in Scheme 2). The associated 24-h growth inhibition log Te values are 0.92 (isopropyl glycidyl ether), 1.42 (n-butyl glycidyl ether), 2.14 (allyl glycidyl ether), and 1.94 (glycidyl phenyl ether), respectively, indicating significant excess toxicities for this type of activated epoxide (with 0.92 being very close to the pragmatically selected threshold of log Te ) 1). The latter can be rationalized by considering the electronattracting inductive effect of the ether oxygen, reducing the epoxide electron density and thus increasing the electrophilicity at the epoxide carbons. However, the oxygen inductive effect cannot explain the substantial Te difference between the isopropyl and allyl derivatives that have very similar log Kow values and the same number of carbons. Here, the glycidyl moiety can be understood as electron-attracting substituent of the allylic double bond, the latter of which thus becomes polarized with a reduced electron density. Accordingly, nucleophilic addition at the β-carbon can take place, offering an alternative route to the SN2-type reaction of the epoxide moiety as shown in Scheme 3. As a third possible route, P450-mediated monooxygenation through insertion of an oxygen atom between the methylene carbon in allylic position and one of its hydrogen atoms yields a hemiacetal that in turn may decompose into an alcohol and acrolein as an electrophilic metabolite (see the right pathway in Scheme 3). Note that a respective P450-mediated generation of acrolein has been discussed earlier as an explanation for the excess toxicity of pentaerythritol triallyl ether (7). Moreover, the relatively high 30-min and 24-h log Te values of glycidyl phenyl ether (1.62, 1.94; see Tables 3 and 4) are striking, considering the fact that this compound is significantly more hydrophobic than the other three glycidyl ethers. In this case, two possible causes can be considered. First, mesomeric stabilization of the electronic ground state invokes a polar resonance form that increases the electron-attracting inductive effect on the epoxide unit (see Scheme 4, top). Second, the methylene group linking the oxirane and aromatic ether moieties is as such attached to two electron-attracting groups that possibly enable a side-chain SN2 pathway with the phenolate ion as the leaving group (keeping in mind that PhO- is a much weaker
Acute and Chronic Bacterial Toxicity of Epoxides
Chem. Res. Toxicol., Vol. 23, No. 12, 2010 1943
Scheme 3. Electrophilic Reaction Pathways of Allyl Glycidyl Ethera
a Left: SN2-type oxirane ring-opening. Middle: nucleophilic addition (AN) at terminal allylic carbon. Right: P450-mediated oxygenation to a hemiacetal that decomposes into an alcohol and the electrophile acrolein.
Scheme 4. Electrophilic Reaction Pathways of Glycidyl Phenyl Ethera
a Top: mesomeric enhancement of the negative inductive effect of the phenoxy moiety. Bottom left: SN2-type oxirane ring-opening. Bottom right: SN2 reaction at the methylene carbon with phenoxide as the leaving group.
Scheme 5. SN2-Type Ring-Opening of Styrene Oxide, Facilitated through Tautomeric Stabilization of the Intermediate
base and thus a much better leaving group than aliphatic RO-; see Scheme 4). Accordingly, the four glycidyl ethers already show two different structural causes for the excess toxicity of epoxide derivatives: On the one hand, activation takes place through substituents with a significant electron-attracting inductive effect (possibly enhanced by mesomeric stabilization of the nonaliphatic ether side chain). On the other hand, a side chain may offer an alternative route of electrophilic reactivity, leading to an overall increased probability for attacking nucleophilic sites of proteins or the DNA. For styrene oxide (structure D2 in Scheme 2), our 30-min log Te value of 1.59 is similar to the one observed for guppy (1.86, see above). Note further that its guppy log Te is close to the one of 1,2-propylene oxide (1.93) (6), suggesting a stronger reactivity component to toxicity of the former because of its significantly larger hydrophobicity (log Kow 1.59 vs 0.37). Besides considering the (slight) negative inductive effect of the
phenyl ring, this aromatic substituent might enable a tautomeric stabilization of the anionic intermediate as depicted in Scheme 5, which is not possible for aliphatic side chains. However, the 24-h Vibrio fischeri growth inhibition of styrene oxide is still in the narcosis range (log Te 0.81), suggesting a more efficient elimination from the bacteria during long-term exposure. The excess toxicity of glycidyl methacrylate (structure C3 in Scheme 2) can be traced back to the presence of an additional Michael-type reaction pathway, involving the unsaturated β-carbon of the methacrylate unit as an electrophilic reaction site (see Scheme 6, top). In both ethyl 2,3-epoxy-3-phenylproprionate and chalcone epoxide (E1 and E2 in Scheme 2), the oxiran ring is attached to a phenyl group and a carbonyl group, both of which activate the epoxide reactivity through negative inductive effects, apparently overcompensating its reduced steric availability (because now both epoxide carbons carry substituents). For chalcone epoxide, the Schiff base formation with endogeneous amines such as peptide side-chain amino acid
1944
Chem. Res. Toxicol., Vol. 23, No. 12, 2010
Blaschke et al.
Scheme 6. Reaction Mechanisms of the Toxicity of Epoxides with Reactive Side Chains in Addition to the SN2-Type Oxirane Ring-Openinga
a Top: 1,4 Michael-type addition of glycidyl methacrylate, with subsequent keto-enol tautomerism yielding the final ketone. Bottom: Schiff base formation of chalcone epoxide with endogeneous amines.
lysine would offer an alternative electrophilic reaction pathway (see Scheme 6, bottom). With both compounds, however, the 24-h Te is significantly smaller than the 30-min Te (Tables 3 and 4). Structural Alerts. The possible mechanistic reasons of the observed excess toxicity as discussed above can be summarized in terms of structural alerts for discriminating between narcosislevel and excess toxicity. First, epoxides carrying only aliphatic side chains R are not sufficiently reactive in the 30-min and 24-h Vibrio fischeri assays for exerting significant Te values, and thus yield narcosis-level toxicity. This contrasts with the observed excess toxicity of small-chain (low hydrophobic) epoxides toward guppy (6), which should be kept in mind when employing bacteria assays to screen organic electrophiles for aquatic excess toxicity. Second, heteroatom-containing side chains Z activate the epoxide unit through negative inductive effects, which may result in substantial excess toxicities. In the present study, Z was restricted to -CH2-O-R but is likely to cover more generally electron-attracting side chains. Further examples already tested with fish (6) are glycidol (-CH2-OH) and 1,3-butadienediepoxide (-epoxide). Third, epoxide conjugation to moieties capable of mesomeric stabilization of the reaction intermediate may also lead to significant Te values, with styrene oxide (-phenyl) being the only corresponding example of the present test set (see Scheme 5). Fourth, electrophilic side-chain reactivity offers an alternative pathway that alone or in competition with SN2-type epoxide ring-opening may lead to significant excess toxicities. In addition to the presently analyzed examples shown in Schemes 3, 4, and 6, a side-chain SN2 reaction at the methylene carbon is likely to cause the excess fish toxicity observed (6) for epichlorohydrine (-CH2-Cl) and epibromohydrine (-CH2-Br). In Scheme 7, the resulting epoxide structural patterns associated with narcosis-level (a) and excess toxicity (b) are summarized, with Z comprising both -I substituents and reactive side chains. Protein Toxicity versus Mutagenicity. Epoxides are known for their potency to exert mutagenicity. According to a statistical analysis of a large database, 63% of the compounds containing an oxirane ring were mutagenic (22). More specifically, structure-activity studies of aliphatic and nonaliphatic epoxides have revealed the following general trends: 1,2-disubstitution is associated with reduction or loss of mutagenicity as compared to monosubstituted epoxides, mutagenicity decreases with increasing alkyl chain length, terminal epoxides appear to be strongly mutagenic than nonterminal counterparts, and electro-
Scheme 7. Structural Alerts Indicating Narcosis Level (a) and Excess Toxicity (b) of Epoxides Toward the Bacteria Vibrio fischeria
a R denotes an aliphatic side chain, Ph the phenyl moiety, and Z an electron-withdrawing substituent exerting a negative inductive (-I) effect or a reactive side chain offering an additional electrophilic reaction pathway. Examples of Z include -CH2-O-R, -CH2-OH, -CH2-X (X ) halogen), and -epoxide extrapolated from fish toxicity studies (6).
negative substituents increase the mutagenic potential significantly (23, 24). Because these structure-mutagenicity relationships show some similarity with the present findings concerning excess toxicity, it appeared interesting to compare the log Te values for the epoxides under current analysis with data on mutagenicity from the literature (25-27). In particular, one might speculate that besides protein toxicity, mutagenicity could be a further cause contributing to 24-h growth inhibition, taking into account the fact that under the bioassay conditions, Vibrio fischeri has about 15 reproduction cycles within 24 h. As can be seen from Table 3, however, there is no correlation between mutagenicity (last column of the table) and 30-min log Te, and a comparison with 24-h log Te of Table 4 shows a corresponding lack of correlation for the growth inhibition assay. Among the seven nonaliphatic epoxides with mutagenicity information (25-27), six are mutagenic, but of these, isopropyl glycidyl ether and n-butyl glycidyl ether have 30-min log Te values
