844
Chem. Res. Toxicol. 2005, 18, 844-854
Chemistry-Toxicity Relationships for the Effects of Diand Trihydroxybenzenes to Tetrahymena pyriformis Aynur O. Aptula,†,‡ David W. Roberts,*,† Mark T. D. Cronin,† and T. Wayne Schultz§ School of Pharmacy and Chemistry, Liverpool John Moores University, Byrom Street, Liverpool, L3 3AF, England, and College of Veterinary Medicine, The University of Tennessee, 2407 River Drive, Knoxville, Tennessee 37996-4543 Received December 2, 2004
This paper presents a mechanistic analysis of aquatic toxicity data, quantified as pIGC50 assessed in the 40 h Tetrahymena pyriformis population growth impairment assay, for 40 polyhydroxybenzene derivatives. The toxicity trends of these phenolic compounds have been shown to be consistent with mechanistic organic chemistry principles. Thus, it is shown that the compounds can be grouped into two chemical mechanism of action domains, according to whether they can be oxidized to electrophilic quinones or quinone methides. Compounds in which the hydroxy groups are oriented meta, but not ortho or para, to one another cannot be oxidized to electrophilic quinones or quinone methides and act as polar narcotics. Their toxicities are found to be well-correlated with hydrophobicity (modeled by log D): pIGC50 ) 0.83 ((0.04) log D - 1.27 ((0.09): n ) 10, r2 (adj) ) 0.981, q2 ) 0.974, s ) 0.15, and F ) 460. Compounds with hydroxy groups oriented ortho or para to one another are more toxic than predicted by this equation, and the toxicity trends within this group of compounds are rationalized in terms of the electrophilic chemistry of their oxidation products. A quantitative correlation is demonstrated between toxicity and electrophilicity of the oxidation products, as modeled by the activation energy index (AEI), a new molecular orbital parameter derived from the computed highest occupied molecular orbital (HOMO) and HOMO-1 orbital energies of the electrophiles and the intermediates for Michael addition of n-butylamine: pIGC50 (adj) ) -0.49 ((0.06) AEI + 6.85 ((0.69): n ) 18, r2 (adj) ) 0.810, q2 ) 0.774, s ) 0.24, and F ) 73. Outliers to these quantitative structure-activity relationships (QSARs) are easily rationalized in terms of their chemistry (tetrabromocatechol, 4,6-dinitro-1,2,3-trihydroxybenzene, and 2,3,4-trihydroxybenzophenone) or in a demonstrable deficiency in the descriptor (the methyl-substituted hydroquinones, for which the AEI parameter as defined here fails to model the electron donation effects of the methyl groups). The AEI parameter is a mechanism-based molecular orbital parameter new to QSAR and, on the basis of the present findings, it shows promise for further applications. However, some deficiencies have been identified with it, particularly with regard to modeling the electronic effects of methyl (and presumably other alkyl) groups, and there is scope to refine the concept so as to deal with these deficiencies.
Introduction Hydroxylated aromatic compounds, or phenols, are used in many industries and consumer products, and many are naturally occurring. Hence, they have become widely distributed in nature (1-3). Because of their prevalence in the environment and their likelihood of eliciting often unknown toxic effects, there is much interest in determining their potential hazard. Consequently, they have been the subject of many toxicological studies (e.g., 4-6), covering a variety of toxic end points representing both environmental and human health effects. While considerable information exists regarding the toxicity and fate of simple phenols, there are still considerable data and knowledge gaps. This is a motiva* To whom correspondence should be addressed. Tel: + 44 151 334 8926. Fax: + 44 151 231 2170. E-mail:
[email protected]. † Liverpool John Moores University. ‡ Current address: SEAC, Unilever Colworth, Sharnbrook, Bedfordshire, MK44 1LQ, England. § The University of Tennessee.
tion for the rational testing and modeling of effects (7) and eventually the formation of quantitative structureactivity relationships (QSARs). For the successful use of QSARs to predict toxicological and fate effects, especially in a regulatory setting, there is a growing realization that the “applicability domain” of the model must be defined (8). The definition of the applicability domain includes the chemical properties, structural features, and biological effects of the training set of compounds and, in our view, is most reliable when based on the understanding of the underlying chemical mechanisms of the toxic end point. The term “phenolic compounds” defines a “structural domain” in which all compounds have at least one aromatic OH group, but it does not define a mechanistic domain or a mode of action domain. With regard to acute ecotoxicity, there is no one single mechanism or even mode of toxic action characteristic of the phenols. These compounds are associated with a range of possible mechanisms, the nature of which depend on the number and substitution pattern of the hydroxyl groups and on
10.1021/tx049666n CCC: $30.25 © 2005 American Chemical Society Published on Web 04/08/2005
Toxicological Chemistry of Phenolic Compounds
the presence or absence of other substituents on the aromatic ring. In addition, it is possible that certain mechanisms may even be species specific and relate to the metabolism and distribution of the compounds in vivo. The mechanisms of acute toxicity of phenols have been relatively well-studied and range from narcotic (unspecific) to electrophilic (specific) effects, which may or may not be reliant on metabolic activation (4). Chemically inert phenols (e.g., substituted by alkyl or halogen groups) exhibit toxicity by membrane perturbation, a mechanism referred to as polar narcosis (9-11). Phenols that interfere reversibly with mitochondrial membranes and the build up of the proton gradient are termed uncouplers of oxidative phosphorylation (12). More difficult to define, and less well-characterized biologically, are the phenols that may be oxidized (either abiotically or enzymatically) to electrophilic species of various types. An example is 4-aminophenol, which in a fish toxicity study of 110 phenols was found to be considerably more toxic than predicted by the equation for polar narcosis, its excess toxicity being attributed to a proelectrophile mechanism involving a p-quinoid metabolite (13). Electrophiles, often referred to as reactive toxicants, are associated with irreversible interaction with biological macromolecules. Some phenols may be directly electrophilic, and others may become active through tautomeric keto forms. In some cases, the toxicity is driven from the chemical properties of (an)other substituent(s) rather than from the phenolic group. In terms of QSAR development, the modeling and prediction of polar narcosis and respiratory uncoupling are well-established and relatively straightforward. Such toxicity is well-modeled by hydrophobicity-dependent relationships (14, 15). However, the toxicity of compounds that may act either directly or indirectly as electrophiles is much more difficult to model (16). In particular, the modeling of the toxicity of polyhydroxylated compounds is regarded as posing a difficult challenge (5, 16). Aromatic compounds with two or more hydroxy groups positioned on the ring in an ortho or para configuration to each other are potentially capable of being oxidized to reactive species, which can act as electrophiles by Michael type nucleophilic addition reactions (6). Michael type electrophilicity (hereafter the term Michael is used to indicate Michael type) is often associated with toxicological activity, such as skin sensitization potential (17) and enhanced (relative to nonelectrophilic compounds) acute aquatic toxicity (18). In terms of acute toxicity, these compounds have a potency greater than that associated with nonspecific effects. Despite some knowledge of the chemistry and reactivity of these compounds, there are currently no satisfactory QSAR models for their toxicity. In addition, there has been no rational approach to predict their toxicity using computational chemistry developed from a mechanistic basis. The aim of this study was to consider the chemistrytoxicity relationships for the acute aquatic toxicity of polyhydroxylated benzenes (those with two or more hydroxyl groups). A particular emphasis of this study was the consideration of novel computational descriptors to assess potential reactivity.
Materials and Methods Biological Data. Toxicity data were taken from the literature (4, 5) except for the five compounds, for which new
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 845 measurements were made as follows: 2,3-dihydroxybenzaldehyde, 2,3,4-trihydroxybenzophenone, 4,6-dinitro-1,2,3-trihydroxybenzene, 2,5-dihydroxybenzaldehyde, and 2,5-dichlorohydroquinone. The data were obtained from the population growth impairment test with the common ciliate Tetrahymena pyriformis (strain GL-C), which was conducted following the protocol described by Schultz (19). This 40 h static assay uses population density quantified spectrophotometrically at 540 nm as its quantified end point. The test protocol permits 8-9 cell cycles in controls. Following range finding, each chemical was tested in three replicate tests. Two controls were used to provide a measure of the acceptability of the test by indicating the suitability of the medium and test conditions as well as a basis for interpreting data from other treatments. The first control had no test material and was inoculated with T. pyriformis. The other, a blank, had neither test material nor inoculum. Each test replicate consisted at minimum of eight different concentrations of each R,β-unsaturated chemical with duplicate samples at each concentration. Only replicates with control-absorbency values of >0.60 but of 95%. In all cases, no further purification was undertaken. Calculation of Molecular Descriptors. Initial threedimensional (3D) molecular geometries were generated using the TSAR for Windows (ver 3.3) molecular spreadsheet (Accelrys Ltd., Oxford, England). The CORINA conformational analysis software converted SMILES strings into 3D structures using a pseudo-force field minimization technique (pseudo-molecular mechanics) (20). All of the resultant geometries were then used as an input for the final gas phase geometry optimization. Stereoelectronic parameters, which have been used, were calculated with MOPAC 93 (21) in the gas phase using the AM1 all-valence electron, semiempirical Hamiltonian. The quantum chemical descriptors included the energies of the highest and the subjacent occupied molecular orbital (EHOMO and EHOMO-1, respectively). Logarithms of the octanol/water partition coefficient (log P), compound acidity (pKa), and hydrophobicity, corrected for ionization (log D) at pH 7.4 were calculated using the ACD/Labs software (ACD/Labs software, 1995, Advanced Chemistry Development Inc., Toronto, Canada). Log P values were all in the following ranges: compounds modeled by eq 1 (Table 1), 0.064.27; compounds modeled by eq 3 (Table 2), -1.10-2.47. Statistical Analysis. The MINITAB statistical software ver. 13.1 (MINITAB Inc., State College, PA) was used, and leastsquares multiple linear regression analysis was performed by the stepwise addition of descriptors. The final model was chosen on the basis of the squared correlation coefficient (r2), leaveone-out cross-validated regression coefficient (rCV2 or q2) calculated by the cross-validation (PRESS) method, standard error of the regression (s), the Fisher’s criterion (F), and standard error of regression coefficients.
Results and Discussion Toxicity data for 39 polyhydroxylated benzene derivatives plus one monoester of a dihydroxybenzene were collected. Among these, 12 compounds (including the monoesterified dihydroxybenzene) had their hydroxy groups only in the meta configuration. Initial mechanistic chemistry considerations suggested that these compounds could be modeled separately. Meta Orientation of Hydroxy Groups. meta-Dihydroxy benzenes and 1,3,5-trihydroxybenzene derivatives
846
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Aptula et al.
Table 1. Resorcinols and Their Toxicities, Observed and Calculated from Eq 1 (Polar Narcosis) pIGC50 obs (mmol/L)
calcd (mmol/L)
0.05 0.54 0.76 1.22 1.62 2.37 2.74 3.35 3.02 3.88
-1.26 -0.87 -0.65 -0.39 0.13 0.97 1.06 1.31 1.37 1.80
-1.23 -0.83 -0.64 -0.26 0.07 0.69 0.99 1.50 1.22 1.94
4.88
-0.98
0.66
-2.08
9.12
3.02
1.11
1.23
no.
CAS
name
log P
pKa
1 2 3 4 5 6 7 8 9 10
108-73-6 49650-88-6 108-46-3 504-15-4 95-88-5 137-19-9 2437-49-2 500-66-3 131-56-6 136-77-6 outlier 601-89-8 derivative 136-36-7
1,3,5-trihydroxybenzene 2-(2-hydroxyethyl)resorcinol resorcinol 5-methylresorcinol 4-chlororesorcinol 4,6-dichlororesorcinol 2,4,6-tribromoresorcinol 5-pentylresorcinol 2,4-dihydroxybenzophenone 4-hexylresorcinol
0.06 0.54 0.76 1.22 1.67 2.58 4.27 3.35 3.17 3.88
9.06 9.50 9.45 9.56 8.24 7.55 5.83 9.58 7.72 10.03
2-nitroresorcinol
1.49
resorcinol monobenzoate
3.03
11 12
log D
Table 2. Experimental Toxicity to T. pyriformis of Chemicals Treated as Proelectrophiles, and Calculated Descriptorsa pIGC50 no.
CAS
name
obs (mmol/L)
adj (mmol/L)
AEI
log P
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
488-17-5 452-86-8 120-80-9 2138-22-9 3316-09-4 24677-78-9 87-66-1 533-73-3 1143-72-2 3264-71-9 123-31-9 95-71-6 608-43-5 700-13-0 527-18-4 615-67-8 583-69-7 87-87-6 771-63-1 1079-21-6 1194-98-5 824-69-1 1198-55-6 488-47-1 601-89-8
3-methylcatechol 4-methylcatechol catechol 4-chlorocatechol 4-nitrocatechol 2,3-dihydroxybenzaldehyde 1,2,3-trihydroxybenzene 1,2,4-trihydroxybenzene 2,3,4-trihydroxybenzophenoneb 4,6-dinitro-1,2,3-trihydroxybenzeneb hydroquinone methylhydroquinoneb 2,3-dimethylhydroquinoneb trimethylhydroquinoneb tetramethylhydroquinoneb chlorohydroquinone bromohydroquinone tetrachlorohydroquinone tetrafluorohydroquinone phenylhydroquinone 2,5-dihydroxybenzaldehyde 2,5-dichlorohydroquinone tetrachlorocatechol tetrabromocatecholb 2-nitroresorcinol
0.28 0.37 0.75 1.06 1.17 1.03 0.85 0.44 0.88 1.18 0.47 1.86 1.41 1.34 1.28 1.26 1.68 2.11 1.84 2.01 0.71 1.81 1.70 0.98 0.66
0.28 0.37 0.45 1.06 1.17 1.03 0.85 0.44 0.88 0.88 -0.13 1.86 1.11 1.34 0.68 1.26 1.68 1.51 1.24 2.01 0.71 1.51 1.40 0.68 0.66
12.618 12.806 13.051 12.017 11.832 11.882 12.146 13.176 11.062c 9.787c 14.111 11.413c 11.133c 11.122c 10.793c 12.686 11.105 11.018 11.332 10.269 12.879 10.229 11.152 10.959 12.029
0 0 -0.59 -0.40 -1.08 -1.10 -0.83 0.03 0.05 -1.08 0.27 0.04 0.53 1.22 1.71 0.46 0.62 1.39 0.34 2.47 -0.24 0.65 0.71 1.37 -0.43
a Log P and AEI values are for the corresponding electrophiles (quinones or quinone methides). b Not included in regression analysis for eq 3 (see text). c AEI values are for the nonionized enolic forms.
cannot be oxidized to quinones. Thus, unless they have other reactive or activating substituents, they are expected to act as polar narcotics. This implies a relationship of toxicity with hydrophobicity. Table 1 shows the toxicities to T. pyriformis (pIGC50), log P, pKa, and log D values for all 11 such compounds. Because some of the compounds have pKa values below the test pH, log D rather than log P is the appropriate hydrophobicity parameter to use. Regression analysis reveals a strong correlation between pIGC50 and log D, with one positive outlier (eq 1; Figure 1).
pIGC50 ) 0.83 ((0.04) log D - 1.27 ((0.09)
(1)
where n ) 10, r2 ) 0.983, r2 (adj) ) 0.981, q2 ) 0.974, s ) 0.15, and F ) 460. Table 1 lists the pIGC50 values calculated from eq 1 for all of the compounds, including the strongly positive outlier, 2-nitroresorcinol (see below). Also shown in Table 1 are data for resorcinol monobenzoate, which has only one free hydroxyl group. Phenolic esters of this type are often quite reactive as hard
Figure 1. Plot of toxicity vs hydrophobicity for resorcinols.
electrophilic acylating agents (resorcininol monobenzoate would be a benzoylating agent), and because of this reactivity, they can act as skin sensitizers [phenyl benzoate is a strong skin sensitizer in guinea pigs (22)]. However, the toxicity of resorcinol monoebenzoate is
Toxicological Chemistry of Phenolic Compounds
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 847
Scheme 1. Michael Electrophilicity of 2-Nitroresorcinol
Scheme 3. Michael and Vinylic Substitution Reactions of para-Quinones
Scheme 2. Redox Reactions of Hydroquinones
Scheme 4. Michael and Vinylic Substitution Reactions of ortho-Quinones
very well-predicted by eq 1, which may suggest that either hard nucleophilic groups are not well-represented in T. pyriformis or their reactivity is suppressed under the conditions of the test protocol. 2-Nitroresorcinol. We propose that this compound acts as a Michael acceptor electrophile by a concerted tautomerisation-nucleophilic addition reaction (Scheme 1). We considered an alternative stepwise reaction pathway, also shown in Scheme 1, whereby 2-nitroresorcinol is assumed to be in equilibrium with its keto/aci tautomer, which reacts with the nucleophile. To assess this possibility, we estimated the equilibrium constant for the tautomerisation from the total occupied orbital energies of 2-nitroresorcinol and the keto/aci tautomer. The latter is estimated to be 77.4 kJ/mol higher in energy than the former, from which the equilibrium constant at 25 °C is estimated as 2.8 × 10-14. Clearly, the equilibrium concentration of the keto/aci tautomer is several orders of magnitude too low for the stepwise pathway to be a realistic possibility. The concerted mechanism of Scheme 1 is consistent with the results of a spectroscopic study by Kovacs et al. (23), which shows that both hydroxyl hydrogens of 2-nitroresorcinol are strongly hydrogen bonded to the oxygens of the nitro group, with a hydrogen bond length of 1.7 Å. Thus, only slight alterations in bond angles and bond lengths are required for the proposed concerted reaction pathway. Ortho and Para Orientation of Hydroxyl Groups. Toxicity data for monocyclic aromatics with ortho or para orientation of hydroxyl groups are shown in Table 2, together with their associated physicochemical parameters. Before proceeding to quantitative analysis of the data, we need to consider the mechanistic reaction chemistry of these compounds and particularly of their oxidation products. 1. para-Dihydroxybenzenes (Hydroquinones). para-Dihydroxybenzenes (hydroquinones) can undergo one-electron oxidation to semiquinones (Scheme 2), these being radical anions, which can undergo one-electron reduction back to the hydroquinones (redox recycling) (24,
25). The semiquinones can be oxidized further to paraquinones (benzoquinones), or the benzoquinones can be formed directly in a two-electron transfer reaction (Scheme 2). Benzoquinones can act, to a greater or lesser extent depending on the substituents, as Michael acceptor electrophiles (Scheme 3) (24, 25). 2. ortho-Dihydroxybenzenes (Catechols). For these compounds, one-electron oxidation to ortho- semiquinones is possible but less well-established than for hydroquinones. However, catechols can be oxidized abiotically and enzymatically to ortho-quinones, which can act, to a greater or lesser extent depending on the presence or absence of other substituents, as Michael acceptor electrophiles (Scheme 4) (24, 25). ortho-Quinones are much more reactive as Michael acceptors than the isomeric para-quinones (24). This is because the COCO group is more electronegative than the CO group. For either type of quinone, Michael reactivity depends on the substituents: electron-donating substituents deactivate, while electron-attracting substituents activate. In light of these principles, it is interesting to consider the selection of T. pyriformis toxicity data shown in Table 3, viewed from the perspective of the working hypothesis that ortho- and para-dihydroxybenzenes act via their electrophilic oxidation products. Some of the toxicity (pIGC50) values in Table 3 have been adjusted, where there are planes of symmetry (other than the plane of the molecule) such that more than one equally reactive Michael electrophilic center can result from oxidation of the parent compound. For example, in the ortho-quinone derived from catechol, a nucleophile can attack equally well at positions 3 and 5. Thus, the concentration of potentially reactive sites is twice the catechol concentration. Consequently, to correct for this fact, the experi-
848
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Aptula et al.
Table 3. T. pyriformis Toxicity Data for Selected Catechols and Hydroquinones
Scheme 5. ortho-Quinone Methide from Methylhydroquinone
pIGC50 no.
name
obs (mmol/L)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
3-methylcatechol catechol 4-nitrocatechol hydroquinone methylhydroquinone 2,3-dimehylhydroquinone trimethylhydroquinone tetramethylhydroquinone bromohydroquinonea phenylhydroquinone methoxyhydroquinoneb 2,3-dicyanohydroquinoneb 1,2,4-trihydroxybenzene 1,2,3-trihydroxybenzene
0.280 0.758 1.170 0.473 1.850 1.410 1.340 1.280 1.680 2.010 2.200 -0.441 0.439 0.850c
adj (mmol/L) 0.280 0.457 1.170 -0.129 1.850 1.110 1.340 0.679 1.680 2.010 2.200 -0.441 0.439 0.850
a Assumed to react by substitution of halogen: The first step is the same as for Michael addition, but the intermediate reacts further by expulsion of the halogen. b Not included in Table 2 (see text). c Note that although 1,2,3-trihydroxybenzene has a plane of symmetry, its pIGC50 value does not require adjustment, since it gives rise to only one Michael electrophilic center, the 5-position.
mental pIGC50 should be adjusted by -log2. Similarly, the experimental pIGC50 value for hydroquinone should be adjusted by -log4. In line with the greater activating effect of the COCO group, catechol is more toxic than hydroquinone. The pattern of toxicity of the substituted catechols is entirely consistent with what would be expected on the basis of the electronic effects of the substituents. However, this is not so for the substituted hydroquinones; in particular, the effects on toxicity of methyl substitution and methoxy substitution are contrary to their effects on Michael reactivity of the para-quinones. This raises the questions as to why one Me or OMe group increases the toxicity of hydroquinone so drastically, whereas further Me groups reduce the toxicity. Our interpretation is that these groups all introduce alternative reaction possibilities. A methyl substituent in hydroquinone provides an alternative two-electron oxidation possibility, oxidation, initially to a hydroxy-substituted ortho-quinone methide as shown in Scheme 5. para-Quinone methides are highly reactive [more so than quinones (24)] Michael acceptor electrophiles whose role in toxicology of many phenols has been quite extensively investigated (26-29). paraQuinone methides with an enolic hydroxyl group ortho to the carbonyl group can be formed in vivo and in vitro by base-catalyzed tautomerisation of 4-alkyl- or 4alkenyl-substituted ortho-quinones resulting from oxidation of 4-alk(en)yl catechols (26, 27). By analogy, it is possible that the ortho-quinone methides proposed here are formed via tautomerization of initially formed paraquinones, although for simplicity these are not shown in Scheme 5. ortho-Quinone methides, presumably because they are normally less stable, are less well-known than para-quinone methides but have been generated and trapped as short-lived intermediates (30-33). However, in the ortho-quinone methides derived from methyl hydroquinone and related compounds, the enolic hydroxyl group meta to the exocyclic methylene group has a resonance stabilizing effect. Furthermore, when this hydroxyl group is present, it can tautomerize to the probably more stable, yet still strongly Michael electrophilic, alicyclic R-methylene-ketone (Scheme 5). The
absence of these possibilities for resonance stabilization and tautomerization would explain why monohydroxy analogues such as ortho-cresol are well-modeled by polar narcotic toxicity QSARs without electrophilicity parameters. Thus, the introduction of a methyl group in the hydroquinone structure creates the possibility of quinone methide and R-methylene-ketone formation, with the consequence that methyl hydroquinone is much more toxic than hydroquinone. Although the R-methyleneketone form is probably the more stable, the hydroxyquinone methide is formed first and is probably the more reactive as a Michael electrophile. Further methyl groups, by their electron-donating effects, make the quinone methide and the R-methylene-ketone less electrophilic, to an extent depending on the position of the methyl group relative to the exocyclic methylene group. An “ortho” methyl group, as in the oxidation product of 2,3dimethylhydroquinone, has a direct electron-donating effect to the reaction center, destabilizing the transition state. The effects of “meta” and “para” methyl groups, as in the oxidation product of trimethyl hydroquinone, are indirect and smaller: By their electron-donating effects, they reduce the electronegativity of the carbonyl group, which is “ortho” to the exocyclic methylene group, so that the stabilizing effect of this carbonyl group on the transition state is lessened. Therefore, the reactivity of the Michael electrophiles derived from methyl-substituted hydroquinones should follow the order: methyl > trimethyl > 2,3-dimethyl > tetramethyl. This descending order of reactivity matches the observed order of toxicity (Scheme 5 and Table 3). When methoxyhydroquinone is oxidized to the quinone (Scheme 6), the methoxy group becomes an enol ether and can act as an SN2 electrophile methylating agent.
Toxicological Chemistry of Phenolic Compounds Scheme 6. SN2 Reactivity of para-Quinone from Methoxy Hydroqinone
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 849 Scheme 8. Strongly Acidic Hydroxyquinones from 2,3,4-Trihydroxybenzophenone and 4,6-Dinitro-1,2,3-trihydroxybenzene
Scheme 7. Quinones Derived from Trihydroxybenzenes
Enol ethers are not usually reactive as SN2 electrophiles, since a simple enolate ion is too basic to be a good leaving group (34). However, in this case, the leaving group is the anion of a vinylogous carboxylic acid (a relatively strong acid due to the electronic effect of the other carbonyl group). This explains why methoxyhydroquinone is so much more toxic than hydroquinone. Phenylhydroquinone was found to be much more toxic than hydroquinone. The simplest reason is that the phenyl group can delocalize the nominal negative charge on the carbon atom to which it is bonded, in the intermediate formed in the first step of the Michael addition reaction. As a result, phenylbenzoquinone is a more reactive Michael acceptor than benzoquinone. A suitable explanation can also be found for the low toxicity of 2,3-dicyanohydroquinone. The simplest answer in this case is that 2,3dicyanohydroquinone is probably completely ionized and hence almost completely nonbioavailable at the pH of the test system. 3. Trihydroxybenzenes (Scheme 7). 1,2,3- and 1,2,4-Trihydroxybenzenes can be oxidized to hydroxyquinones. 1,2,3-Trihydroxybenzene can only give a hydroxy ortho-quinone, but 1,2,4-trihydroxybenzene can, on paper, give either a hydroxy ortho-quinone or a hydroxy para-quinone. The latter is the more likely in view of the greater stability of para-quinones relative to orthoquinones. If we consider these trihydroxybenzenes as catechols or hydroquinones with an extra hydroxy group, we can predict how the additional hydroxy group affects the electrophilicity of the corresponding quinones, relative to the ortho-quinone derived from catechol. For 1,2,3-trihydroxybenzene, the corresponding hydroxyquinone (see Scheme 7) is a Michael acceptor with the double bond activated by a COCOC(OH)d group. If the enolic hydroxyl group tautomerizes, which we consider unlikely, the activating group will be COCOCO. Either of these is more electronegative than a COCO
group. Consequently, the quinone from 1,2,3-trihydroxybenzene should be more electrophilic than the orthoquinone from catechol, and on that basis, 1,2,3-trihydroxybenzene should be more toxic than catechol. For 1,2,4-trihydroxybenzene, the hydroxy para-quinone shown in Scheme 7 is the most likely oxidation product. The activating group is COC(OH)d, which is more electronegative than the CO group of the para-quinone derived from hydroquinone but less electronegative than the COCOC(OH)d group, and its inductive effect will be correspondingly weaker. On that basis, 1,2,4-trihydroxybenzene should be less toxic than 1,2,3-trihydroxybenzene but more toxic than hydroquinone. The values for toxicity to T. pyriformis shown in Table 3 for these two trihydroxybenzenes are in agreement with these arguments. The two substituted 1,2,3-trihydroxybenzenes in the data set (see Table 2) also require further consideration. 4,6-Dinitro-1,2,3-trihydroxybenzene might be expected to be very much more toxic than 1,2,3-trihydroxybenzene due to the activating effects of the two nitro groups. However, the difference between the toxicity values is relatively small. This is probably because the enolic hydroxyl group in the ortho-quinone oxidation product is highly acidic and almost completely ionized at physiological pH, its calculated pKa being 2.3 ( 0.2. The anion will be less Michael reactive than the neutral compound (Scheme 8). A similar argument applies for 2,3,4-trihydroxybenzophenone (calculated pKa 4.5 ( 1.6), which is only marginally more toxic than 1,2,3-trihydroxybenzene despite the presence of the strongly electronegative benzoyl group. The hydroxyquinone resulting from oxidation of 2,3,4-trihydroxybenzophenone has three possible tautomers (plus the tetraketo tautomer). All of these tautomers will be highly acidic and give the same highly delocalized anion when they ionize. Again, the anion will be less Michael reactive than the neutral compound. Quantitative Modeling. From the above discussion, it is clear that the toxicity of di- and trihydroxybenzene derivatives with ortho or para orientation of hydroxyl groups can be related to the Michael reactivity (Michael electrophilicity) of their corresponding quinone, quinone methide, or R-methylene-ketone derivatives. Therefore, to develop a quantitative model in the form of a mechanism-based QSAR, a suitable parameter is required to represent the Michael electrophilicity of these derivatives.
850
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Michael reactions proceed via a metastable intermediate (I) with carbanion character, and it is reasonable to take this intermediate as a good approximation to the transition state. Thus, Michael electrophilicity, which can be expressed as the activation energy for the reaction of an electrophile (E) with a nucleophile, can be modeled by the energy differences between the occupied molecular orbitals of I and electrophile, E. Rather than using the total energy difference between all correlated pairs of occupied orbitals, which, although in principle more rigorous, risks accumulation of errors, we adopted the approach of considering only those orbitals that we expected to change most as E is converted and for which the energy changes are most dependent on the position and nature of substituents. Our first assumption is that the σ-orbital framework of the molecule does not change significantly on going from electrophile, E, to intermediate, I. Of course, there is a substantial σ-energy change corresponding to the new σ-bond to the nucleophile, but we assume that this energy change does not vary significantly between different Michael electrophiles. Therefore, we focused on the energy changes of the π-orbitals in the transformation from electrophile E to intermediate I. Conjugated organic molecules can be safely assumed to have their bonding framework decomposed into σ- and π-components, where the highest energy occupied MOs and the lowest energy unoccupied MOs have predominantly π-character. Thus, for the simplest Michael electrophiles such as R,βunsaturated aldehydes and ketones, there are only two occupied π-orbitals, and we assume that their energies are represented by EHOMO and EHOMO-1 as calculated by the MOPAC quantum chemical package. For the quinones and quinone methides discussed in the present work, there are at least four occupied π-orbitals. In a recent paper (35), this approach was adopted to derive an activation energy index (AEI) for electrophilic isothiazol-3-ones, which have at least three occupied π-orbitals (four if a lone pair from the sulfur atom is assumed to have π-character). Defining the AEI as ∆EHOMO + ∆EHOMO-1 (i.e., ignoring lower energy πorbitals), it was found that the reaction chemistry and skin sensitization properties were well-rationalized in terms of AEI values. There is an implicit assumption in the above definition of AEI, which is that as the structure of E is changed, most of the variation in energy differences between E and I is reflected in ∆EHOMO and ∆EHOMO-1. In other words, for the π-orbitals below HOMO-1, although the energies change on going from E to I, the energy changes do not vary greatly as the structure of E changes. For a preliminary assessment of whether the above definition of AEI is adequate for modeling the present toxicity data, we considered the four compounds containing hydroxyl groups, with ortho or para orientation, and no other substituents. These compounds are catechol, hydroquinone, 1,2,3-trihydoxybenzene, and 1,2,4-trihydroxybenzene. AEI values were calculated for the corresponding quinones, assumed to react with n-butylamine to form the intermediate I, using the definition AEI ) ∆EHOMO + ∆EHOMO-1. Although we considered thiol nucleophiles likely to be very important in the toxic mechanism, we chose butylamine because the modeling is easier and on the grounds that reactivities toward BuNH2 and thiolate ions should be correlated by an LFER of the form log k (RS-) ) a log k (BuNH2) + b. The toxicities
Aptula et al.
Figure 2. Plot of toxicity vs AEI for catechol, hydroquinone, 1,2,3-trihydroxybenzene, and 1,2,4-trihydroxybenzene.
(pIGC50) and AEI values are shown in Table 2, and a plot of pIGC50 vs AEI is shown in Figure 2. Although this is only a four-point correlation, the high linearity suggests that ∆EHOMO + ∆EHOMO-1 may be an adequate definition of AEI for modeling toxicity of di- and trihydroxybenzene derivatives with ortho or para orientation of hydroxyl groups. The regression equation is
pIGC50 (adj) ) -0.50 ((0.04) AEI + 6.93 ((0.57) (2) where n ) 4, r2 ) 0.985, r2 (adj) ) 0.978, q2 ) 0.887, s ) 0.06, and F ) 133. AEI values were therefore calculated for all of the compounds listed in Table 2. Examination of the AEI values listed in Table 2 reveals the expected general trend for decreasing AEI (indicating higher reactivity) with introduction of electron-withdrawing groups. However, comparing methyl-, 2,3-dimethyl-, trimethyl-, and tetramethylhydroquinone, it can be seen that AEI decreases with increasing methyl group substitution, contrary to what would be expected (and what is found in reactivity trends for other methyl-substituted Michael electrophiles) given the electron-donating effects of methyl groups. A similar trend, contrary to expectation, is seen in the AEI values for catechol and its 3-methyl and 4-methyl derivatives. It appears that the AEI as currently defined does not adequately model the effects of methyl substitution. For this reason, the methyl-substituted hydroquinones were omitted from the regression analysis presented below. With catechol and its two methyl derivatives, the differences in AEI values are quite small; therefore, these compounds were not omitted from the regression analysis. In some cases, the oxidation product has two or more different potential reaction sites. Our approach with such cases was to calculate AEI values for each of the possibilities and to choose the one with the lowest AEI (the lower the AEI value, the lower the activation energy and the greater the reactivity). Scheme 9 shows all of the quinone structures and their Michael intermediates for reaction with butylamine, which were used to derive the AEI values listed in Table 2. Regression analysis for all of the compounds in Table 2, except those otherwise indicated, and including 2-nitroresorcinol, gave the QSAR:
pIGC50 (adj) ) -0.49 ((0.06) AEI + 6.85 ((0.69) (3) where n ) 18, r2) 0.821, r2 (adj) ) 0.810, q2 ) 0.774, s ) 0.24, and F ) 73. Tetrabromocatechol is a negative outlier
Toxicological Chemistry of Phenolic Compounds
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 851
Scheme 9. Quinone and Michael Intermediate Structures Used for AEI Calculations
to this equation. It is the most hydrophobic of the compounds listed in Table 2, and its toxicity is better predicted by the polar narcosis eq 1, although still underpredicted. Its log D value is 3.03 (5), and substituting this value in eq 1 gives a calculated pIGC50 value of 1.24. The experimental value of 0.98 is below the predicted value by slightly less than two standard deviations. Its toxicity is probably affected by low bioavailability. The compounds covered by eq 3 are all significantly more toxic than predicted by the polar narcosis QSAR of eq 1. This is illustrated by the most hydrophobic com-
pound [phenylhydroquinone, log P ) 2.47, log D ) 2.43 (4)] and the least reactive compound [hydroquinone, AEI ) 14.111, log P ) 0.27, log D ) 0.59 (4)], for which we have calculated excess toxicity values as pIGC50 (obs, adj) minus pIGC50 (eq 1): phenylhydroquinone, excess toxicity 1.26; hydroquinone, excess toxicity 0.65. Equation 3 is of high statistical quality. In particular, the low value of s (standard deviation of residuals) denotes close fit to the regression line. Chlorohydroquinone is the least well-fitted to the regression line (residual + 0.57), and its removal would improve the r2 value, but because we see no mechanistic justification, we have not removed it. It will be recalled that 2-nitroresorcinol was identified as a positive outlier to the polar narcosis eq 1, and this was attributed to its proposed direct electrophilic reactivity. 2-Nitroresorcinol is the only compound, in the whole data set, which is proposed to act as an electrophile in its own right rather than as a proelectrophile, and it fits the regression line for eq 3 very well, as indicated in Figure 3. The addition of log P of the electrophiles as a second parameter did not lead to a regression equation better than eq 3:
pIGC50 (adj) ) -0.46 ((0.07) AEI + 0.07 ((0.08) log P + 6.46 ((0.81) (4) Figure 3. Plot of toxicity vs AEI for 18 compounds listed in Table 2.
where n ) 18, r2 ) 0.831, r2 (adj) ) 0.818, q2 ) 0.768, s
852
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
) 0.25, and F ) 37. The standard deviation on the log P coefficient is larger than the coefficient itself, indicating nonsignificance of the log P term. Similarly, log P of the parent compounds was found to be nonsignificant (equation not shown). We interpret this in the light of a pharmacokinetic model published some years ago (36), as indicating that the toxicity-determining step, assumed here to be reaction of the electrophile with a biological nucleophile, takes place in a nonlipid environment.
Discussion This study has assessed structure-activity and QSARs for polyhydroxylated benzenes. The study has focused upon mechanistic considerations and potential electrophilic reactivity. From the results, a number of conclusions can be drawn. First, we note the very good linearity and statistical fit of the polar narcosis plot for the resorcinols (eq 1). When a less than perfect fit to a linear QSAR is obtained, the reason may be inadequately precise biological data, inadequate descriptors (e.g., poor experimental log P values, failure of a log P calculation method to adequately represent structural feature in some of the compounds), or more than one mechanism of action. It is often assumed that biological data are intrinsically less reliable or consistent than physicochemical experimental data, but there is in reality no underlying reason that this should be so. In the present case, the compounds represented in the regression equation are all closely related, so the log P and pKa values used to estimate log D are likely to be self-consistent across the data set, and we can be confident of only one mechanism of action. The very good statistical fit of eq 1 (r2 ) 0.983, s ) 0.15) indicates that the biological data must also be very selfconsistent. Statistical fits of this quality have been found on several other occasions (37-39) for fish toxicity but not previously with T. pyriformis. We can infer from the present findings that when the statistical fit of a QSAR based on T. pyriformis is inferior to that of eq 1, unless there is good reason to expect increased variability in the toxicity data, the reason is likely to be either that more than one mechanism of action is represented in the data set or that the physicochemical descriptors used do not correlate perfectly with the in vivo physicochemical processes they are intended to model. Repeatability of T. pyriformis toxicity values has been examined (40). It was observed that relatively poor repeatability was common among the chemicals believed to elicit toxicity after either abiotic or biotic transformation, in particular, derivatives with multiple electron-releasing substituents such as hydroxy groups. On this basis, the fact that the polar narcosis QSAR of eq 1 for the resorcinols is statistically better than the QSAR of eq 3 for the proelectrophilic compounds may at least in part be due to the inherent greater variability of the latter toxicity data. Equation 3, relating toxicity to the reactivity parameter AEI, is not improved by the addition of the hydrophobicity parameter log P. Several other groups of electrophiles and proelectrophiles similarly show dependence of toxicity on reactivity but not on hydrophobicity (36, 41). On the basis of a pharmacokinetic mathematical model for the toxicity of electrophiles and proelectrophiles published some years ago (36), this indicates that the site of action of the Michael acceptors electrophiles discussed
Aptula et al.
here is in an aqueous environment in the interior of the cell, rather than in a membrane. This does not of course imply that membrane-associated proteins are in general not targets for electrophiles in toxic processes. For example, in skin sensitization, the RAI model is based on membrane proteins being the target (42). However, in the present case, as in the fish studies referred to in refs 36 and 41, the results are consistent with aqueous targets dominating in producing toxicity and are difficult to explain on the basis of membrane-bound targets. A tendency has previously been identified (36, 43) whereby with increasing log P, the toxicity of electrophiles and proelectrophiles converges on a narcosis mechanism, in terms of both the physiological effects observed and the fit to the narcosis model. It arises because some compounds can have several toxicities, e.g., a dihydroxybenzene (ortho or para orientation of OH groups) with not too high log P will have a general narcosis toxicity, a polar narcosis toxicity, and a proelectrophile toxicity. If tested on its own, the proelectrophile EC50 is the first EC50 to be reached and the dihydroxybenzene is classed as a proelectrophile toxicant. If tested in a mixture with polar narcotics, its own polar narcotic contribution may lead to the polar narcotic EC50 of the mixture being reached before the proelectrophile toxicity of the dihydroxybenzene. In this situation, the dihydroxybenzene would display its polar narcotic toxicity. Similarly, if tested in a mixture with general narcotics, the general narcotic toxicity of the dihydroxybenzene will contribute to the general narcotic toxicity of the mixture and the dihydroxybenzene will display its general narcotic toxicity. If the dihydroxybenzene is incrementally modified by making it increasingly more hydrophobic, the point will come when its narcotic EC50 is lower than its proelectrophile EC50. Now it would be classed as a narcotic (but its proelectrophilic toxicity would still able to contribute in mixtures with other proelectrophiles or electrophiles). However, in our data set, the log P range does not extend high enough for any of the proelectrophiles studied to have narcotic EC50 < proelectrophile EC50. The QSAR denoted by eq 3 relates the toxicity of the polyhydroxybenzenes to the physicochemical properties not of the parent compounds but of their electrophilic oxidation products, with one exception, 2-nitroresorcinol. This compound is treated as a direct-acting electrophile. This implies that the oxidation step is faster than the toxicity-producing reaction of the electrophile with biological nucleophiles. Other metabolic transformations of the parent compounds and their derived electrophiles, which could on paper lead to detoxification, do not appear to be significant as compared to the oxidation and the electrophilic reaction: If they were, the correlation with AEI would not apply unless these competing metabolic transformations were also correlated with AEI, an assumption for which there would be no justification.
Conclusions In the present work, the toxicity trends of the various polyhydroxybenzenes have been shown to be consistent with mechanistic organic chemistry principles, based on the concept that 1,3-dihydroxybenzenes (resorcinols), unless special activating substituents are present, act by a physical chemistry mechanism as polar narcotics while compounds with ortho or para orientation of hydroxy
Toxicological Chemistry of Phenolic Compounds
groups act as pro-Michael acceptor electrophiles. Using appropriately defined mechanistic descriptors, log D for the resorcinols and AEI for the pro-Michael acceptor electrophiles, two QSARs, each based on a single descriptor, have been derived. Outliers to these QSARs are easily rationalized in terms of their chemistry (tetrabromocatechol, 4,6-dinitro-1,2,3-trihydroxybenzene, and 2,3,4-trihydroxybenzophenone) or in a demonstrable deficiency in the descriptor (the methyl-substituted hydroquinones, for which the AEI parameter as defined here fails to model the electron donation effects of the methyl groups). Some of the rationalizations are based on predictive application of organic reaction mechanistic principles and could usefully be tested by chemical experiments as part of an experimental program to extend the knowledge base on reaction chemistry of phenolic compounds, which would be highly desirable. Our aim in this study was not to produce QSARs to be used for predictive purposes but rather to use QSAR methodology for the investigation of mechanism, in an analogous manner to the use of LFER plots in organic reaction mechanism studies. However, the QSARs having been developed, it is appropriate to comment on their predictive value. The log D-based polar narcosis QSAR of eq 1, for compounds with meta orientation of OH groups, is expected to be well-predictive for other compounds with meta orientation of OH groups (and probably for many phenols with a single aromatic OH group), although there will always be special cases such as 2-nitroresorcinol, which this QSAR will underpredict. A “rule” for detecting such cases is to look for electronattracting unsaturated groups, which give the opportunity for quinonoid tautomers. For para and ortho orientation of OH groups, unless the compound is very hydrophobic, we should always expect proelectrophilic toxicity, and the polar narcosis QSAR will always underpredict. We can be confident about this in view of how well our mechanism-based QSAR using AEI fits the data. However, we cannot at present be so confident in how well the AEI-based QSAR will predict other compounds. We have already found some, in this paper, which are underpredicted and which we can postrationalize, but such compounds may not always be recognizable in advance. The AEI parameter is a mechanism-based molecular orbital parameter new to QSAR. The high statistical quality of eq 3 encourages us to apply the concept in further QSAR studies. However, we have identified some deficiencies with it, particularly with regard to modeling the electronic effects of methyl (and presumably other alkyl) groups, and we plan further work to refine the concept so as to deal with these deficiencies. It is important to note that the AEI is the difference in combined HOMO and HOMO-1 energies between the electrophile and the intermediate, thus representing the activation energy for reaction. This may be contrasted to an index used by Ghiotto et al. and Barone et al. (44, 45), defined as ∆ΕH,H-1, the difference between the HOMO and HOMO-1 energies for the compound studied. There is no theoretical reason that this index should be applicable to the electrophilic toxicity, and indeed, we found no significance between ∆ΕH,H-1 and toxicity for the compounds studied here. Recently, many of the polyhydroxybenzenes discussed here have been the subject of a QSAR study by some of the present authors, based on the proposal of a free
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 853
radical mechanism of action (5). In that study, multivariate analysis was applied to obtain QSARs based on two descriptors, one electronic (bond dissociation energy or absolute hardness) and one hydrophobic (log D) but with several statistical outliers not all of which could readily be rationalized in chemical mechanistic terms. The results of the present work point to two mechanisms of action, one hydrophobicity-dependent (modeled by eq 1) and one reactivity-dependent (modeled by eq 3). Equations 1 and 3 are each based on a single descriptor, and the outliers are all simply rationalized mechanistically. Although the greater simplicity of the present QSARs does not of itself prove that the mechanism of action is proelectrophilic rather than free radical, the proelectrophile mechanism is simpler, better able to rationalize the outliers, and is well-precedented for substituted catechols and hydroquinones in other toxic endpoints, particularly skin sensitization. Consequently, the proelectrophile mechanism is our currently preferred model.
Acknowledgment. A.O.A. acknowledges receipt of a Fellowship from the Leverhulme Trust.
References (1) Gilner, M., Moore, G. A., and Cederberg, H. (1994) Hydroquinone. The WHO Environmental Series, Vol. 157, pp 1-178, WHO Publications, Albany, NY. (2) Moulton, M. P., and Schultz, T. W. (1986) Comparisons of several structure-toxicity relationships for chlorophenols. Aquat. Toxicol. 8, 121-128. (3) Reynolds, G. W., Epstein, W. L., and Rodriguez, E. (1986) Urushiol contact allergy from plants in the family hydrophyllaceae. Contact Dermatitis 14, 39-48. (4) Schultz, T. W., Sinks, G. D., and Cronin, M. T. D. (1997) Identification of mechanisms of toxic action of phenols to Tetrahymena pyriformis from molecular descriptors. In Quantitative Structure-Activity Relationships in Environmental Sciences-VII (Chen, F., and Schu¨u¨rmann, G., Eds.) pp 329-342, SETAC Press, Pensacola, FL. (5) Netzeva, T. I., Aptula, A. O., Chaudary, S. H., Duffy, J. D., Schultz, T. W., Schu¨u¨rmann, G., and Cronin, M. T. D. (2003) Structure-activity relationships for the toxicity of substituted poly-hydroxylated benzenes to Tetrahymena pyriformis: Influence of free radical formation. QSAR Comb. Sci. 22, 575-582. (6) Roberts, D. W., and Benezra, C. (1993) Quantitative structureactivity relationships for skin sensitisation potential of urushiol analogues. Contact Dermatitis 29, 78-83. (7) Cronin, M. T. D. (2004) Predicting chemical toxicity and fate in humans and the environmentsAn introduction. In Predicting Chemical Toxicity and Fate (Cronin, M. T. D., and Livingstone, D. J., Eds.) pp 3-13, CRC Press, Boca Raton, FL. (8) Jaworska, J. S., Comber, M., Auer, C., and van Leeuwen, C. J. (2003) Summary of a workshop on regulatory acceptance of (Q)SARs for human health and environmental endpoints. Environ. Health Perspect. 111, 1358-1360. (9) Bradbury, S. P., Carlson, R. W., and Henry, T. R. (1989) Polar narcosis in aquatic organisms. In Aquatic Toxicology and Hazard Assessment, 12th ed. (Cowgill, U. M., and Williams, L. R., Eds.) pp 59-73, ASTM STP 1027, American Society for Testing and Materials, Philadelphia, PA. (10) Schultz, T. W., Holcombe, G. W., and Phipps, G. L. (1986) Relationships of quantitative structure-activity to comparative toxicity of selected phenols in the Pimephales promelas and Tetrahymena pyriformis test systems. Ecotoxicol. Environ. Saf. 12, 146-153. (11) Roberts, D. W., and Costello, J. F. (2003) Mechanisms of action for general and polar narcosis: A difference in dimension. QSAR Comb. Sci. 22, 226-233. (12) Terada, H. (1990) Uncouplers of oxidative phosphorylation. Environ. Health Perspect. 87, 213-218. (13) Lipnick, R. L., Bickings, C. K., Johnson, D. E., and Eastmond, D. A. (1986) Comparison of QSAR Predictions with Fish Toxicity Screening Data for 110 Phenols, pp 153-176, American Society for Testing and Materials, ASTM STP 891, 1985, Philadelphia, PA.
854
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
(14) Bearden, A. P., and Schultz, T. W. (1997) Relationships of structure-activity for Tetrahymena and comparative toxicity with Pimephales: A mechanism of action approach. Environ. Toxicol. Chem. 16, 1311-1317. (15) Schultz, T. W., Cronin, M. T. D., Walker, J. D., and Aptula, A. O. (2003) Quantitative structure-activity relationships (QSARs) in toxicology: A historical perspective. J. Mol. Struct. (THEOCHEM) 622, 1-22. (16) Cronin, M. T. D., Aptula, A. O., Duffy, J. C., Netzeva, T. I., Rowe, P. H., Valkova, I. V., and Schultz, T. W. (2002) Comparative assessment of methods to develop QSARs for the prediction of the toxicity of phenols to Tetrahymena pyriformis. Chemosphere 49, 1201-1221. (17) Barratt, M. D., Basketter, D. J., and Roberts, D. W. (1997) Structure-activity relationships for contact hypersensitivity. In Allergic Contact Dermatitis. The Molecular Basis (Lepoittevin, J.P., Basketter, D. A., Goosens, A., and Karlberg, A.-T., Eds.) pp 129-154, Springer, Heidelberg. (18) Schultz, T. W., Netzeva, T. I., Roberts, D. W., and Cronin, M. T. D. (2005) Structure-toxicity relationships for the effects to Tetrahymena pyriformis of aliphatic, carbonyl-containing, R,βunsaturated chemicals. Chem. Res. Toxicol. 18, 330-341. (19) Schultz, T. W. (1997) TETRATOX: Tetrahymena pyriformis population growth impairment endpoint: A surrogate for fish lethality. Toxicol. Methods 7, 289-309. (20) Sadowski, J., and Gasteiger, J. (1993) From atoms and bonds to three-dimensional atomic coordinates: Automatic model builders. Chem. Rev. 93, 2567-2581. (21) Dewar, M. J. S, Zoebisch, E. G., Healy E. F., and Stewart, J. J. P. (1985) AM1: A new general purpose quantum mechanical molecular model. J. Am. Chem. Soc. 107, 3902-3909. (22) Barratt, M. D., Basketter, D. A., and Roberts, D. W. (1994) Skin sensitization structure activity relationships for phenyl benzoates. Toxicol. in Vitro 8, 823-826. (23) Kovacs, A., Keresztury, G., and Izvekov, V. (2000) Intramolecular hydrogen-bonding in 2-nitroresorcinol. A combined FT-IR, FTRaman and computational study. Chem. Phys. 253, 193-204. (24) Monks, T. J., Hanzlik, R. P., Cohen, G. M., Ross, D., and Graham, D. G. (1992) Quinone chemistry and toxicity. Toxicol. Appl. Pharmacol. 112, 2-16. (25) Bolton, J. L., Trush, M. A., Penning, T. M., Dryhurst, G., and Monks, T. J. (2000) Role of quinones in toxicology. Chem. Res. Toxicol. 13, 135-160. (26) Iverson, S. L., Hu, L. Q., Vukomanovic, V., and Bolton, J. L. (1995) The influence of the p-alkylsubstituent on the isomerization of o-quinones to p-quinone methidessPotential activation mechanism for catechols. Chem. Res. Toxicol. 8, 537-544. (27) Bolton, J. L., Wu, H. M., and Hu, L. Q. (1996) Mechanism of isomerization of 4-propyl-o-quinone to its tautomeric p-quinone methide. Chem. Res. Toxicol. 9, 109-113. (28) Bolton, J. L., Turnipseed, S. B., and Thompson, J. A. (1997) Influence of quinone methide reactivity on the alkylation of thiol and amino groups in proteins: Studies utilizing amino acid peptide models. Chem-Biol. Interact. 107, 185-200. (29) Yu, L., Liu, H., Li, W., Zhang, F., Luckie, C., van Breemen, R. B., Thatcher, G. R. J., and Bolton, J. L. (2004) Oxidation of raloxifene to quinoids: Potential toxic pathways via a diquinone methide and o-quinones. Chem. Res. Toxicol. 17, 879-888. (30) Gautam, D. R., Litinas, K. E., Fylaktakidou, K. C., and Nicolaides, D. N. (2003) Reactions of o-quinones with alpha-methyl- (or methylene) susbtituted phosphorus ylides. Synthesis of benzo[b]furan derivatives. J. Heterocycl. Chem. 40, 399-404.
Aptula et al. (31) Leo, E. A., Delgado, J., Domingo, L. R., Espinaos, A., Miranda, M. A., and Tormos, R. (2003) Photogeneration of o-quinone methides from o-cycloalkenylphenols. J. Org. Chem. 68, 96439647. (32) Selenski, C., Mejorado, L. H., and Pettus, T. R. R. (2004) Diastereoselective [4+2] reactions of o-quinone methides with a chiral enol ether: Asymmetric synthesis of (+)-R-momosifoliol. Synlett 6, 1101-1103. (33) Yadav, J. S., Reddy, B. V. S., Sadashiv, K., and Padmavani, B. (2004) [4+2] Cycloaddition of ortho-quinone methides promoted by ionic liquids: An efficient and mild protocol for the synthesis of tetrahydropyranobenzopyrans. Adv. Synth. Catal. 346, 607610. (34) Smith, M. B., and March, J. (2001) March’s Advanced Organic Chemistry, 5th ed., pp 465-468, Wiley, New York. (35) Aptula, A. O., Roberts, D. W., and Cronin, M. T. D. (2004) From experiment to theory: Molecular orbital parameters to interpret the skin sensitisation potential of 5-chloro-2-methylisothiazol-3one (MCI) and 2-methylisothiazol-3-one (MI). Chem. Res. Toxicol. 18, 324-329. (36) Roberts, D. W. (1989) Acute lethal toxicity quantitative structureactivity relationships for electrophiles and pro-electrophiles: Mechanistic and toxicokinetic principles. In Aquatic Toxicology and Environmental Fate: Eleventh Volume, ASTMSTP1007 (Suter, G. W., II, and Lewis, M. A., Eds.) pp 490-506, American Society for Testing and Materials, Philadelphia, PA. (37) Ko¨nemann, H. (1981) Quantitative structure-activity relationships in fish toxicity studies Part 1: Relationship for the 50 industrial pollutants. Toxicology 19, 209-221. (38) Veith, G. D., Call, D. J., and Brooke, L. T. (1983) Structure toxicity relationships for the fathead minnow, Pimephales promelas: Narcotic industrial chemicals. Can. J. Fish. Aquat. Sci. 40, 743748. (39) Roberts, D. W. (2004) Environmental risk assessment of surfactants: Quantitative structure-activity relationships for aquatic toxicity. In Handbook of Detergents: Part B, Environmental Impact (Zoller, U., Ed.) Chapter 10, pp 271-298, Dekker, New York. (40) Seward, J. R., Sinks, G. D., and Schultz, T. W. (2001) Reproducibility of toxicity across mode of toxic action in the Tetrahymena population growth impairment assay. Aquat. Toxicol. 53, 33-47. (41) Hermens, J., Busser, F., Leeuwangh, P., and Musch, A. (1985) Quantitative correlation studies between the acute lethal toxicity of 15 organic halides to the guppy (Poecilia reticulata) and chemical reactivity towards 4-nitrobenzylpyridine. Toxicol. Environ. Chem. 9, 219-236. (42) Roberts, D. W., and Williams, D. L. (1982) The derivation of quantitative correlations between skin sensitisation and physicochemical parameters for alkylating agents and their application to experimental data for sultones. J. Theor. Biol. 99, 807-825. (43) Lipnick, R. L., Watson, K. R., and Strausz, A. K. (1987) A QSAR study of the acute toxicity of some industrial chemicals to goldfish: Narcosis, electrophile and pro-electrophile mechanisms. Xenobiotica 17, 1011-1025. (44) Ghiotto, R. C. T., Lavarda, F. C., and Ferreira, F. J. B. (2004) Antioxidant activity of flavonols. Int. J. Quantum Chem. 97, 949952. (45) Barone, P. M., Camillo, V. B., Jr., and Galva˜o, D. S. (1996) Theoretical approach to identify carcinogenic activity of polycyclic aromatic hydrocarbons. Phys. Rev. Lett. 77, 1186-1189.
TX049666N
