Environ. Sci. Technol. 2005, 39, 4961-4966
Quantitative Structure-Activity Relationship Models for Prediction of the Toxicity of Polybrominated Diphenyl Ether Congeners FIGURE 1. Chemical structure of the PBDEs molecule. YAWEI WANG,† HUANXIANG LIU,‡ CHUNYAN ZHAO,‡ HANXIA LIU,† Z O N G W E I C A I , †,§ A N D G U I B I N J I A N G * ,† State Key Laboratory of Environmental Chemistry and Ecotoxicology, Research Center for Eco-Environmental Sciences, Graduate School of Chinese Academy of Sciences, Beijing, 100085, China, Department of Chemistry, Lanzhou University, Lanzhou 730000, China, and Department of Chemistry, Hong Kong Baptist University, Kowloon Tong, Kowloon, Hong Kong SAR, China
Levels of polybrominated diphenyl ethers (PBDEs) are increasing in the environment and may cause long-term health problems in humans. The similarity in the chemical structures of PBDEs and other halogenated aromatic pollutants hints on the possibility that they might share similar toxicological effects. In this work, three-dimensional quantitative structure activity relationships (3-D-QSAR) models, using comparative molecular field analysis (CoMFA) and comparative similarity indices analysis (CoMSIA), were built based on calculated structural indices and a reported experimental toxicology index (aryl hydrocarbon receptor relative binding affinities, RBA) of 18 PBDEs congeners, to determine the factors required for the RBA of these PBDEs. After performing leave-one-out crossvalidation, satisfactory results were obtained with crossvalidation Q2 and R2 values of 0.580 and 0.995 by the CoMFA model and 0.680 and 0.982 by the CoMSIA model, respectively. The results showed clearly that the nonplanar conformations of PBDEs result in the lowest energy level and that the electrostatic index was the main factor reflecting the RBA of PBDEs. The two QSAR models were then used to predict the RBA value of 46 PBDEs for which experimental values are unavailable at present.
Introduction Polybrominated diphenyl ethers (PBDEs, Figure 1) have been extensively distributed as environmental contaminants due to their use as flame retardants for various polymers, for instance, in electronic equipment: computers and television sets, etc. (1). The global demand for PBDEs has increased since the 1970s and has been estimated to be close to 70 000 tonnes in 1999 (2). For example, the PBDE concentration in human blood and tissues has increased almost 100-fold since the 1970s, doubling every ∼5 years. Knowledge about the * Corresponding author phone: 8610-6284-9334; fax: 8610-62849179; e-mail:
[email protected]. † Graduate School of Chinese Academy of Sciences. ‡ Lanzhou University. § Hong Kong Baptist University. 10.1021/es050017n CCC: $30.25 Published on Web 05/28/2005
2005 American Chemical Society
toxicological mechanisms of PBDEs is urgently needed but insufficiently available. The levels of PBDEs in biota have shown a steady increase that parallels their historic rate of production. They are emerging as a significant class of environmental contaminants, having been detected in many environmental samples (4-9), human blood plasma (10), and human breast milk (11). The weak to moderate binding affinity of PBDE congeners to the aryl hydrocarbon receptor (AhR) and the weak induction of EROD (ethoxyresorufin-O-deethylase) activity suggest the possibility of dioxin-like behavior for some PBDE congeners (12). It has been shown that PBDEs impact the same body systems as polychlorinated biphenyls (PCBs) (13, 14), although they are thought to be less potent. But, very little is known about the toxicology of PBDEs, especially congener-specific data. The structure similarities between PBDEs and PCBs suggest that they might share similar toxicological pathways such as AhR signal transduction. Relationships between descriptors of chemical substances and their activities/toxicity make many workers want to obtain reliable quantitative structure activity relationships (QSAR), to understand toxicological activities, and to predict the toxicity of many new substances. Earlier, QSARs methods based on multiple linear regression (MLR) have been used for HAHs such as PCDFs to model toxicities and different physicochemical properties (15). But, MLR is limited because it does not consider the 3-D structure of molecules. In comparing the PBDEs with other halogenated aromatic compounds (HACs), this challenge seems to be further aggravated because each molecule’s AhR binding activity depends greatly on its chlorination/bromination sites and on the way in which its molecular backbone conformation affects the spatial locations of the chlorine/bromine atoms. Recently, the comparative molecular field analysis (CoMFA) and the comparative similarity indices analysis (CoMSIA) paradigm, 3-D-QSAR, based on the assumption that most intermolecular interactions are noncovalent and shapedependent, have been used to examine AhR binding affinities and properties of polychlorinated dibenzodioxins (PCDDs), dibenzofurans (PCDFs), and PCBs (16-19). The aim of the present work is to study the relationships between the quantitative structure indices and the toxicology index of 18 PBDE congeners (12). 3-D-QSAR models are built to determine the factors required for the toxicology of these compounds by using CoMFA and CoMSIA. The models were further used to predict the toxicology of 46 PBDE congeners for which the toxicology index is currently unknown.
Materials and Methods Data Set. A data set of 18 PBDE congeners was taken from the literature (12). The relative binding affinity (RBA) values (I) that were expressed as pI (-log I) (Table 1) for Ah receptor binding of individual PBDE congeners were in the µM range, indicating a weaker affinity than the reference toxicant TCDD. In addition, 46 PBDE congeners were selected to predict their toxicity by the models obtained. VOL. 39, NO. 13, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4961
TABLE 1. Experimental Values, Predicted Ah Receptor Relative Binding Affinities (RBA, -log I), and Residual Values of PBDE Congeners by CoMFA and CoMSIAa CoMFA no.
Br substituted site
1 2 3 4 7 8 12 13 15 16 17 18 21 25 28 30 32 40 41 47 49 53 61 63 66 67 70 71 72 75 76 77 78 84 85 86 98 99 100 104 110 117 119 120 124 125 126 127 128 152 153 154 155 157 159 167 181 182 183 186 187 191 194 209
2342,2′2,42,4′3,43,4′4,4′2,2′,32,2′,42,2′,52,3,42,3′,42,4,4′2,4,62,4′,62,2′,3,3′2,2′,3,42,2′,4,4′2,2′,4,5′2,2′,5,6′2,3,4,52,3,4′,52,3′,4,4′2,3′,4,52,3′,4′,52,3′,4′,62,3′,5,5′2,4,4′,62′,3,4,53,3′,4,4′3,3′,4,52,2′,3,3′,62,2′,3,4,4′2,2′,3,4,52,2′,3′,4,62,2′,4,4′,52,2′,4,4′,62,2′,4,6,6′2,3,3′,4′,62,3,4′,5,62,3′,4,4′,62,3′,4,5,5′2′,3,4,5,5′2′,3,4,5,6′3,3′,4,4′,53,3′,4,5,5′2,2′,3,3′,4,4′2,2′,3,5,6,6′2,2′,4,4′,5,5′2,2′,4,4′,5′,62,2′,4,4′,6,6′2,3,3′,4,4′,5′2,3,3′,4,5,5′2,3′,4,4′,5,5′2,2′,3,4,4′,5,62,2′,3,4,4′,5,6′2,2′,3,4,4′,5′,62,2′,3,4,5,6,6′2,2′,3,4′,5,5′,62,3,3′,4,4′,5′,62,2′,3,3′,4,4′,5,5′2,2′,3,3′,4,4′,5,5′,6,6′-
experimental
3.886
3.420 3.638
2.921
3.329 4.174
2.699 3.867 3.398 2.658 1.721 3.854 4.114
2.959
2.569
4.602 4.638
3.602
predicted 4.038 3.947 3.831 3.869 3.807 3.451 3.462 3.159 3.418 3.389 3.662 3.528 3.382 3.783 2.878 3.880 3.440 3.348 3.391 3.329 4.248 4.045 3.352 2.749 2.654 3.075 3.544 3.792 3.087 3.479 3.217 2.682 3.244 3.127 1.671 3.127 4.086 3.911 4.047 3.732 2.764 2.875 3.039 3.177 2.851 3.305 2.585 2.815 2.789 3.093 4.563 4.603 3.464 2.491 3.069 2.835 3.366 3.708 3.578 3.151 3.282 2.327 2.234 2.126
CoMSIA residual
0.055
0.002 -0.024
0.043
-0.077 -0.074
0.045 0.075 -0.081 -0.024 0.050 -0.057 0.067
-0.080
-0.016
0.039 0.035
0.024
predicted 4.304 4.187 3.847 4.630 4.632 3.814 4.391 3.575 3.455 4.268 3.761 4.386 4.209 4.505 2.734 4.065 3.702 4.101 4.351 3.144 4.287 4.483 3.831 3.188 2.717 3.582 4.278 3.716 3.483 3.493 3.475 2.674 4.242 3.707 1.809 3.707 4.046 3.924 4.022 4.062 2.987 3.167 3.102 3.516 3.428 3.903 2.566 3.185 3.359 3.421 4.486 4.725 3.557 2.510 3.352 2.942 3.451 3.988 3.507 3.378 3.124 2.293 2.362 1.904
residual
0.039
-0.035 -0.123
0.187
0.108 -0.113
-0.018 0.151 -0.095 -0.016 -0.088 -0.070 0.091
-0.143
0.003
0.116 -0.087
0.095
a The data came from ref 19. Values are binding affinities relative to TCDD; concentration of [3H]-TCDD was 1.0 nM. I (TCDD) ) 1.00. The relative binding affinities (RBA) values (I) are expressed as pI (-log I). Residual ) experimental - predicted, which are the experimental and predicted pI values, respectively.
Computing Model. All 3-D-QSAR studies were performed on SGI O2 workstations running Sybyl 6.9 molecular modeling software (20). The use of a reasonably low energy conformation in the alignment is a useful starting point for statistical comparisons of flexible structures within the CoMFA and CoMSIA models. The geometry of 4962
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 13, 2005
these compounds was subsequently optimized using the Tripos force field calculated from the Merck molecular force field (MMFF94) atomic charge. The method of Powell available in the Maxinin2 procedure was used for energy minimization until the gradient value was smaller than 0.05 kcal/mol Å.
FIGURE 2. 3-D view of all aligned compounds. Alignment Rules. The rules for selecting template molecule include (i) the most active compound; (ii) the lead and/or commercial compound; and (iii) the compound containing the greatest number of functional groups (21, 22). On the basis of these rules, BDE85 (2,2′,3,4,4′-bromodiphenyl ether), which is the most active compound in the data set, was selected as a template molecule for 3-D-QSAR. The conformational search was carried out using the multisearch routine in Sybyl. Then, the alignment of the molecules was carried out by flexible fitting (multifit) of atoms in the template molecule by applying force (force constant ) 20 kcal/mol Å). Figure 2 shows the aligned molecules. For CoMFA analysis, the basic assumption is that the observed biological properties can be well-understood and correlated with the suitable sampling of the steric (van der Waals) and electrostatic (Coulombic with 1/r dielectric) fields surrounding a set of ligand molecules. It maps gradual changes of interaction properties of molecules by evaluating the potential energy at regularly spaced grid points surrounding the aligned molecule from the structural- or fieldbased alignments. The grid spacing was 2.0 Å units. Electrostatic fields were calculated at each point using an sp3 carbon probe atom with a van der Waals radius of 1.52 Å and a +1.0 charge. The default cutoff, Ecut ) 30 kcal/mol, was applied to both fields, which meant that the steric and electrostatic energies greater than 30 kcal/mol were truncated to that value. Another 3-D-QSAR technique, comparative molecular similarity indices analysis (CoMSIA), can avoid some inherent deficiencies arising from the functional form of LennardJones and Coulomb potentials used in CoMFA. In CoMSIA, a distance-dependent Gaussian-type functional form has been introduced, which can avoid singularities at the atomic positions and the dramatic changes of potential energy for threshold grids in the proximity of the surface. Meanwhile, no arbitrary definition of cutoff limits is required in CoMSIA. In CoMSIA, similarity is expressed in terms of different physicochemical prosperities: steric occupancy, partial atomic charges, local hydrophobicity, and hydrogen bond donor and acceptor properties. Both CoMFA and CoMSIA regression models were built by partial least squares (PLS) regression (23) in conjunction with leave-one-out (LOO) cross-validation to measure the internal consistency and the predictive ability of the resulting QSAR models. To avoid over-fitting, the optimum number of components (N) used was chosen for the analysis of the highest cross-validated correlation coefficient (Q2), which is defined as
Q2 ) 1 -
∑(Y - Y ∑(Y - Y
2 pred) 2
mean)
To validate the derived models, the correlation coefficient (R2), which should be greater than 0.90, and Q2, which should
FIGURE 3. Predicted vs actual Ah receptor relative binding affinities (RBA, pI) of PBDEs congeners and distribution of residuals over the range of predicted pI for CoMFA. The dashed and dotted lines indicate the lower 95% prediction interval based on the linear regression model; the dashed lines indicate the upper 95% prediction interval based on the linear regression model.
FIGURE 4. CoMFA steric contour maps (stdev*coeff). Sterically favored areas (contribution level 80%) are represented by green polyhedra. Sterically disfavored areas (contribution level 20%) are represented by yellow polyhedra. be greater than 0.40, were obtained to show the selfconsistency and predictive capacity of the model.
Results and Discussion CoMFA. The alignment with MMFF94 charges showed a cross-validated Q2 ) 0.58 with six components. The noncross-validated R2 ) 0.995 and F ) 337.627 indicated that 3-D-QSAR was highly satisfactory. In the analysis, the contribution level of the electrostatic field was 68.6%, and the steric value was 31.4%. The predicted activities indices and the residue values of 18 PBDEs are listed in Table 1. In addition, the pI values of the 46 PBDE congener, for which experimental values are unknown at present, were predicted using this satisfactory 3-D-QSAR model and are included in Table 1. Figure 3 shows the plot of experimental versus predicted values of pI. The results of CoMFA studies are best viewed as 3-D colorcoded contour plots. The contour plots of the stddev*coeff values for the steric and electrostatic fields from the threefield CoMFA are presented in Figures4 and 5, respectively, using BDE85 as a reference structure. The green contours in Figure 4 represent the regions of high steric tolerance, which indicate that a bulky substituent is preferred in the position to produce higher values of pI, while yellow contours represent regions of unfavorable steric effects. The sterically favored green contours corresponded to the 2,4′,5,5′,6-position on the PBDE congeners. The yellow region near the 3,5-position is where a less bulky substituent is preferred for higher values of pI. The corresponding plot for the electrostatic field in Figure 5 is, in general, more VOL. 39, NO. 13, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4963
FIGURE 5. CoMFA electrostatic contour maps (stdev*coeff). Sterically favored areas (contribution level 80%) with positive charges are indicated by blue polyhedra. Sterically disfavored areas (contribution level 20%) with negative charges are represented by red polyhedra.
FIGURE 6. CoMSIA steric contour maps (stdev*coeff). Sterically favored areas (contribution level 80%) are represented by green polyhedra. Sterically disfavored areas (contribution level 20%) are represented by yellow polyhedra. difficult to interpret than the steric contours plots. The blue contours (3,3′,4,5-position) describe regions where a positively charged group enhances the values of pI. Negatively charged (electron-rich) red regions are found at the 2,5,6positions and other relative small regions. Politzer et al. (2426) have built a model in which negative molecular electrostatic potential values in the lateral positions and a central area of positive potential influenced the activity of PCDDs. Also, McKinney et al. (27, 28) have proposed a stacking model for the Ah receptor interaction that for PCDDs, high polarizability and low steric hindrance in the central part of the molecule are needed. The structural similarities between PBDEs and PCDDs suggest that PBDEs might activate the AhR signal transduction (29). In general, the toxicity of planar HAC is extremely sensitive to both number and position of the halogen substituents. Presently, two main types of models, electrostatic and dispersion, have been proposed for explaining the PCDDs-AhR interaction (30). The first is focused on the ligand-receptor complex, which is based on the view that the effective interaction with the receptor depends on the molecular electrostatic potential (MEP) around the ligand (31, 32). The second considers that molecular polarizability can be used to gain insight into the origins of PCDDs specific binding to their receptor protein (33). Future research on the relationships between the descriptors, which correlate with electrostatic properties and aromatic-aromatic interactions, and the toxicities of HACs (their receptor affinities), including PBDEs, is necessary for an improvement on the present level of knowledge. CoMSIA. Figures6-8 show the CoMSIA contour maps of the steric, electrostatic, and hydrophobicity fields. The value of the attenuation factor we selected is 0.15. The crossvalidated Q2 is 0.68 with six components. The non-crossvalidated R2 ) 0.982 and F ) 98.049. In this analysis, the contribution level of the electrostatic field was 78.8%, the steric value was 1.0%, and the hydrophobicity was 20.2%. 4964
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 13, 2005
FIGURE 7. CoMSIA electrostatic contour maps (stdev*coeff). Sterically favored areas (contribution level 80%) with positive charges are indicated by blue polyhedra. Sterically disfavored areas (contribution level 20%) with negative charges are represented by red polyhedra.
FIGURE 8. CoMSIA hydrophobic contour maps (stdev*coeff). Sterically favored areas (contribution level 80%) are indicated by yellow polyhedra. Sterically disfavored areas (contribution level 20%) are represented by white polyhedra.
FIGURE 9. Predicted vs actual Ah receptor relative binding affinities (RBA, pI) of PBDE congeners and distribution of residuals over the range of predicted pI for CoMSIA. The dashed and dotted lines indicate the lower 95% prediction interval based on the linear regression model; the dashed lines indicate the upper 95% prediction interval based on the linear regression model. The predicted activity indices for 64 PBDE congeners, as well as residues for the 18 PBDEs with experimental pI values, are summarized in Table 1. Figure 9 was the plot of the experimental values versus predicted values of pI. Analysis of CoMSIA steric contour maps (Figure 6) indicates that the steric favorable green region, which represents a preferred occupancy of the pocket of the acceptor, is found at the 2,5,5′,6-positions. This is similar to the result of CoMFA. The electrostatic properties were summarized in Figure 7. Positively charged groups that increase the values of pI were contoured by blue isopleths, and negatively charged groups were surrounded by red polyhedra. Blue areas were observed for the 2′,3,3′,4′,5-
positions. The red region around ring B suggests that substitutions on this ring indicate the partial electrostatic interactions of the PBDE congeners. In a CoMSIA study, hydrophobic similarity index fields are constructed, and hydrophobic contour maps are shown in Figure 8. Yellow polyhedra, which are found at the 2-position and around ring B, indicate that hydrophobic substituents are good for increasing the potency, while hydrophilic substituents are beneficial to the activity at the regions of white contours, which are found at the 3,4,4′,5-positions. Several members of the broad class of HAC produce a characteristic toxicity syndrome (34). Golas et al. (35) built a multiple parameter linear regression equation for the competitive binding EC50 of 13 PCDD congeners to human placental cytosolic aryl hydrocarbon (Ah)
pEC50 (M) ) 6.246 + 1.632π - 1.764σm0 + 1.382HB where π, σm0, and HB are the physiochemical parameters for substituent lipophilicity, meta-directing electronegativity, and hydrogen bonding capacity, respectively. The equation reflects the factors that relate with the toxicity of PCDDs, which are similar to the results of PBDEs in this work. Kodavabti et al. (36) pointed out that for the PCBs family, the coplanar congeners that bind most strongly to the AhR exhibited the smallest energy differences between their equilibrium and coplanar geometries. But, in our study, Figure 2 shows that the issue of ligand planarity is inconsequential when PBDEs bind to AhR since they are nonplanar. As compared to CoMFA, CoMSIA is a relatively new alternative molecular field analysis method. It is touted to be less affected by changes in molecular alignment and to provide more smooth and interpretable contour maps as a result of employing a Gaussian-type distance dependence with the molecular similarity indices it uses (37). In this study, steric and electrostatic contours of CoMSIA were similarly distributed as those of the CoMFA model. But, the additional hydrophobic, hydrogen bond donor, and hydrogen bond acceptor contours (because the contributions of donor and acceptor field were very low, the donor and acceptor contours are not presented in this paper) of CoMSIA hint that the CoMSIA model should be more accurate than the CoMFA model. The reason is perhaps that the activity data used in the present study was in vitro. The results, therefore, had contributions not only from steric and electrostatic fields but also from hydrophobicity, which accounts for the transport phenomenon. Although R2 (0.982) of the CoMSIA model was lower than that for the CoMFA model (R2 ) 0.995), CoMSIA provided a slightly better QSAR model than CoMFA with this alignment as indicated by the Q2 values (0.68 as compared to a Q2 value of 0.58 for CoMFA). In these compounds, BDE 209 should be watched because it is the primary component in a commonly used flame retardant, which accounts for approximately 80% of the world market demand for PBDEs. Some studies show that the BDE 209 molecule is so large and hydrophobic that it is not readily bioavailable (38-40). But, in other studies, it is indicated that eggs from the wild peregrine populations had higher BDE 209 concentrations than the captive population feeding on chicken, which is perhaps an indication that this congener is bioavailable (41). Chen et al. (12) failed to determine the RBA of the commercial decaBDE because of its very low solubility. The calculated log Kow of BDE209 is 11.39 by using the of HyperChem Release 7.0 software (Hepercube, 2002). QSAR models in our studies showed that it has a relative lower pI (2.126 from CoMFA and 1.904 from CoMSIA). Although the bioavailability of BDE209 is still in debate and its implications need to be clarified, there is a trend that the strength of
EROD induction by HACs parallels the strength of AhR binding (42). Large quantities of BDE 209 have been manufactured and are being used (43). Despite the low volatility of BDE 209, its capability of long-range transport is evidenced by its presence in the remote regions on Earth (7, 44). There is also other evidence that photolytic debromination of BDE 209 is a possible pathway for the formation of more bioavailable, lower brominated BDEs (45). In conclusion, all the contour analyses suggest that much 3-D space information such as energy differences and the distribution of steric, electrostatic, and hydrophobic fields should be taken into account more when we examine AhR binding affinities and properties of PBDE congeners than those of PCBs. As can be seen in Figure 2, the conformation of the lowest energy for PBDEs is not coplanar. This indicates that the relative AhR of the PBDE congeners is not critically related to the planarity of the molecules. If these molecules bind to the intercellular AhR by their planar conformation, more energy would be needed because the large size of the bromine atoms expands the Ah receptor’s binding site. Future research on this is warranted.
Acknowledgments This work was jointly supported by the National Basic Research Program of China (2003CB415001) and the National Natural Science Foundation of China (20329701). The authors thank Profs. Man-cang Liu (Department of Chemistry, Lanzhou University) and An Li (School of Public Health, University of Illinois at Chicago) for helpful discussions. The authors also thank Dr. Guo-sheng Chen (Environmental Health Center, Ottawa, Canada) for providing the toxicity data of PBDEs and for his valuable suggestions.
Literature Cited (1) WHO; Environmental health criteria 162: Brominated diphenyl ethers; World Health Organization: Geneva, 1994; pp 31-34. (2) Renner, R. Increasing levels of flame retardants found in North American environment. Environ. Sci. Technol. 2000, 34, 452A453A. (3) Hites, R. A. Polybrominated Diphenyl Ethers in the Environment and in People: A Meta-Analysis of Concentrations. Environ. Sci. Technol. 2004, 38, 945-956. (4) Sellstro¨m, U.; Kierkegaard, A.; de Wit, C.; Jansson, B. Polybrominated diphenyl ethers and hexabromocyclododecane in sediment and fish from a Swedish river. Environ. Toxicol. Chem. 1998, 17, 1065-1072. (5) Wolkers, H.; van Bavel, B.; Derocher, A. E.; Wiig, Ø.; Kovacs, K. M.; Lydersen, C.; Lindstro¨m, G. Congener-specific accumulation and food chain transfer of polybrominated diphenyl ethers in two arctic food chains. Environ. Sci. Technol. 2004, 38, 16671674. (6) Hale, R. C.; La Guardia, M. J.; Harvey, E. P.; Mainor, T. M.; Duff, W. H.; Gaylor, M. O. Polybrominated diphenyl ether flame retardants in Virginia freshwater fish (USA). Environ. Sci. Technol. 2002, 35, 4585-4591. (7) Song, W. L.; Ford, J. C.; Li, A.; Mills, W. J.; Buckler, D. R.; Rock, K. J. Polybrominated diphenyl ethers in the sediments of the Great Lakes. 1. Lake Superior. Environ. Sci. Technol. 2004, 38, 3286-3293. (8) Booij, K.; Zegers, B. N.; Boon, J. P. Level of some polybrominated diphenyl ether (PBDEs) flame retardants along the Dutch coast as derived from their accumulation in SPMDs and blue mussels (Mytilus edulis). Chemosphere 2002, 46, 683-688. (9) Vives, I.; Grimalt, J. O.; Lacorte, S.; Guillamo´n, M.; Barcelo´, D. Polybromodiphenyl ether flame retardants in fish from lakes in European high mountains and Greenland. Environ. Sci. Technol. 2004, 38, 2338-2344. (10) Klasson Wehler, E.; Hovander, L.; Bergman, Å. New organohalogens in human plasma identification and quantification. Organohalogen Compd. 1997, 33, 420-425. (11) Nore´n, K.; Meironyte, D. Certain organochlorine and organobromine contaminants in Swedish human milk in perspective of past 20-30 years. Chemosphere 2000, 40, 1111-1123. (12) Chen, G. S.; Konstantinov, A. D.; Chittim, B. G.; Joyce, E. M.; Bols, N. C.; Bunce, N. J. Synthesis of polybrominated diphenyl VOL. 39, NO. 13, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4965
(13)
(14)
(15)
(16) (17) (18)
(19)
(20) (21) (22)
(23)
(24) (25) (26)
(27)
(28)
(29) (30)
ethers and their capacity to induce CYP1A by the Ah receptor mediated pathway. Environ. Sci. Technol. 2001, 35, 3749-3756. Eriksson, P.; Viberg, H.; Jakobsson, E.; Orn, U.; Fredriksson, A. A brominated flame retardant, 2,2′,4,4′,5-pentabromodiphenyl ether: uptake, retention, and induction of neurobehavioral alterations in mice during a critical phase of neonatal brain development. Toxicol. Sci. 2002, 67, 98-103. Hallgren, S.; Darnerud, P. O. Polybrominated diphenyl ethers (PBDEs), polychlorinated biphenyls (PCBs), and chlorinated paraffins (CPs) in rat-testing interactions and mechanisms for thyroid hormone effects. Toxicology 2002, 177, 227-243. Arulmozhiraja, S.; Morita, M. Structure-Activity Relationships for the Toxicity of Polychlorinated dibenzofurans: approach through density functional theory-based descriptors. Chem. Res. Toxicol. 2004, 17, 348-356. Bradley, M.; Waller, C. L. Polarizability fields for use in threedimensional quantitative structure-activity relationship (3-DQSAR). J. Chem. Inf. Comput. Sci. 2001, 41, 1301-1307. Lukacova, V.; Balaz, S. Multimode ligand binding in receptor site modeling: implementation in CoMFA. J. Chem. Inf. Comput. Sci. 2003, 43, 2093-2105. Puri, S.; Chickos, J. S.; Welsh, W. J. Three-dimensional quantitative structure-property relationship (3-D-QSPR) models for prediction of thermodynamic properties of polychlorinated biphenyls (PCBs): enthalpy of sublimation. J. Chem. Inf. Comput. Sci. 2002, 42, 109-116. Tuppurainen, K.; Ruuskanen, J. Electronic eigenvalue (EEVA): a new QSAR/QSPR descriptor for electronic substituent effects based on molecular orbital energies. A QSAR approach to the Ah receptor binding affinity of polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs), and dibenzofurans (PCDFs). Chemosphere 2000, 41, 843-848. SYBYL, version 6.9; Tripos Associates: St. Louis, MO, 1999; http:// www.tripos.com/. Agarwal, A.; Taylor, E. W. 3-D-QSAR for intrinsic activity of 5-HT1A receptor ligands by the method of comparative model field analysis. J. Comput. Chem. 1993, 14, 237-245. Baurin, N.; Vangrevelinghe, E.; Morin-Allory, L.; Me´rour, J. Y.; Renard, P.; Payard, M.; Guillaumet, G.; Marot, C. 3-D-QSAR CoMFA study on imidazolinergic I2 ligands: a significant model through a combined exploration of structural diversity and methodology. J. Med. Chem. 2000, 43, 1109-1122. Wold, S.; Ruhe, A.; Wold, H.; Dunn, W. J. The covariance problem in linear regression. The partial least squares (PLS) approach to generalized inverses. SIAM J. Sci. Stat. Comp. 1984, 5 (3), 735-743. Murray, J. S.; Politzer, P. Electrostatic potentials of some dibenzop-dioxins in relation to their biological activities. Theor. Chim. Acta. 1987, 72, 507-517. Politzer, P. Computational approaches to the identification of suspect toxic molecules. Toxicol. Lett. 1988, 43, 257-276. Murray, J. S.; Evans, P.; Politzer, P. A comparative analysis of the electrostatic potentials of some structural analogues of 2,3,7,8-tetrachlorodibenzo-p-dioxin and of related aromatic systems. Int. J. Quantum. Chem. 1990, 37, 271-289. McKinney, J. D.; Fawkes, J.; Jordan, S.; Chae, K.; Oatley, S.; Coleman, R. E.; Briner, W. 2,3,7,8-Tetrachlorodinenzo-p-dioxin (TCDD) as a potent and persistent thyroxine agonist: a mechanistic model for toxicity bases on molecular reactivity. Environ. Health Perspect. 1985, 61, 41-53. McKinney, J. D. Multifunctional receptor model for dioxin and related compound toxic action: possible thyroid hormoneresponsive effector-linked site. Environ. Health Perspect. 1989, 82, 323-336. Okey, A. B.; Riddick, D. S.; Harper, P. A. The Ah receptor: mediator of the toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and related compounds. Toxicol. Lett. 1994, 70, 1-22. Mhin, B. J.; Lee, J. E.; Choi, W. Understanding the congenerspecific toxicity in polychlorinated dibenzo-p-dioxins: chlo-
4966
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 13, 2005
(31)
(32)
(33)
(34)
(35)
(36)
(37)
(38)
(39)
(40)
(41)
(42) (43) (44)
(45)
rination pattern and molecular quadrupole moment. J. Am. Chem. Soc. 2002, 124, 144-148. Bonati, L.; Fraschini, E.; Lasagni, M.; Pitea, D. Toward a mechanistic understanding of PCDD biological activity based on molecular electrostatic potential modeling. J. Mol. Struct. (THEOCHEM) 1994, 303, 43-54. Bonati, L.; Fraschini, E.; Lasagni, M.; Modoni, E. P.; Pitea, D. A hypothesis on the mechanism of PCDD biological activity based on molecular electrostatic potential modeling. Part 2. J. Mol. Struct. (THEOCHEM) 1995, 340, 83-95. Fraschini, E.; Bonati, L.; Pitea, D. Molecular polarizability as a tool for understanding the binding properties of polychlorinated dibenzo-p-dioxins: definition of a reliable computational procedure. J. Phys. Chem. 1996, 100, 10564-10569. Arulmozhiraja, S.; Morita, M. Structure-activity relationships for the toxicity of polychlorinated dibenzofurans: approach through density functional theory-based descriptors. Chem. Res. Toxicol. 2004, 17, 348-356. Golas, C. L.; Prokipcak, R. D.; Okey, A. B.; Manchester, D. K.; Safe, S.; Fujita, T. Competitive binding of 7-substituted-2,3dichlorodibenzo-p-dioxins with human placental Ah-receptorQSAR analysis. Anal. Biochem. Pharmacol. 1990, 40, 737-741. Kodavanti, P. R.; Ward, T. R.; Mckinney, J. D.; Waller, C. L.; Tilson, H. A. Increased [3H] phorbol ester binding in rat cerebellar granule cells and inhibition of Ca2+-sequestration in rat cerebellum by polychlorinated diphenyl ether congeners and analogues: structure-activity relationships. Toxicol. Appl. Pharmacol. 1996, 138, 251-261. Kliebe, G.; Abraham, U.; Mietzner, T. Molecular similarity indices in a comparative analysis (CoMSIA) of drug molecules to correlate and predict their biological activity. J. Med. Chem. 1994, 37, 4130-4146. Stapleton, H. M.; Baker, J. E. Comparing polybrominated diphenyl ether and polychlorinated biphenyl bioaccumulation in a Lake Michigan food web. Arch. Environ. Contam. Toxicol. 2003, 45, 227-234. Stapleton, H. M.; Alaee, M.; Letcher, R. J.; Baker, J. E. Debromination of the flame retardant decabromodiphenyl ether by Juvenile Carp (Cyprinus carpio) following dietary exposure. Environ. Sci. Technol. 2004, 38, 112-119. Tomy, G. T.; Palace, V. P.; Halldorson, T.; Braekevelt, E.; Danell, R.; Wautier, K.; Evans, B.; Brinkworth, L.; Fisk, A. T. Bioaccumulation, biotransformation, and biochemical effects of brominated diphenyl ethers in Juvenile Lake Trout (Salvelinus namaycush). Environ. Sci. Technol. 2004, 38, 1496-1504. Lindberg, P.; Sellstro¨m, U.; Ha¨ggberg, L.; de Wit, C. A. Higher brominated diphenyl ethers and hexabromocyclododecane found in eggs of Peregrine Falcons (Falco peregrinus) breeding in Sweden. Environ. Sci. Technol. 2004, 38, 93-96. Whitlock, J. P. Induction of cytochrome P4501A1. Annu. Rev. Pharmacol. Toxicol. 1999, 39, 103-125. Betts, K. S. Deca in the atmosphere. Environ. Sci. Technol. 2004, 38, 81A-82A. Muir, D.; Teixeira, C.; Chigak, M.; Yang, F.; D’Sa, I.; Cannon, C.; Pacepavicius, G.; Alaee, M. Current deposition and historical profiles of decabromodiphenyl ether in sediment cores. Organohalogen Compd. 2003, 61, 77-80. So¨derstro¨m, G.; Sellsto¨m, U.; De Wet, C. A; Tysklind, M. Photolytic debromination of decabromodiphenyl ether (BDE 209). Environ. Sci. Technol. 2004, 38, 127-132.
Received for review January 4, 2005. Revised manuscript received April 28, 2005. Accepted April 29, 2005. ES050017N
