Environ. Sci. Technol. 2010, 44, 3992–3998
Binary Mixture Effects by PBDE Congeners (47, 153, 183, or 209) and PCB Congeners (126 or 153) in MCF-7 Cells: Biochemical Alterations Assessed by IR Spectroscopy and Multivariate Analysis ´ LIO TREVISAN,† V A L O N L L A B J A N I , †,‡ J U KEVIN C. JONES,† RICHARD F. SHORE,‡ A N D F R A N C I S L . M A R T I N * ,† Centre for Biophotonics, Lancaster Environment Centre, Lancaster University, Bailrigg, Lancaster LA1 4YQ, U.K., and NERC Centre for Ecology and Hydrology, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster LA1 4AP, U.K.
Received January 20, 2010. Revised manuscript received April 9, 2010. Accepted April 13, 2010.
Target organisms are continuously and variously exposed to contaminant mixtures in the environment. We noted that treatment with brominated diphenyl ether (BDE)47 or polychlorinated biphenyl (PCB)126 (toxic equivalency factor [TEF] ) 0.1) induces similar alterations in MCF-7 cells when these were determined usingattenuatedtotalreflectionFourier-transforminfrared(ATRFTIR)spectroscopywithmultivariateanalysis.Becausethismethod appears sensitive enough to signature low-dose effects, we examined how various test agents interact in binary mixtures to induce cell alterations. MCF-7 cells were exposed for 24 h to low concentrations (10-12 M) of polybrominated diphenyl ether (PBDE) congeners (47, 153, 183, or 209) with or without the coplanar PCB126 or nonplanar PCB153. Following treatment, ethanol-fixed cellular material was interrogated using ATRFTIR spectroscopy; derived IR spectra in the biochemical-cell fingerprint region (1800 cm-1-900 cm-1) were then subjected to principal component analysis-linear discriminant analysis. Assuming that if two test agents independently induce the same cell alteration that in combination they’ll give rise to an additive effect, we examined predicted versus observed differences in induced alterations by binary mixtures. Compared to corresponding control clusters, treatment with PBDE congener plus PCB126 appeared to cancel out their respective induced alterations. However, treatment with binary mixtures including PCB153 gave rise to an enhanced segregation. Our findings suggest that test agents which mediate their cellular effects via similar mechanisms might result in inhibition within a binary mixture whereas independently acting agents could exacerbate induced alterations in overall cell status.
* Corresponding author phone: +44 1524 510206; e-mail:
[email protected]. † Lancaster University. ‡ Lancaster Environment Centre. 3992
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010
Introduction Biological organisms are variously exposed to mixtures of contaminants via their environment. Contaminants include persistent organic pollutants (POPs) such as dioxins and furans, polychlorinated biphenyls (PCBs), and polybrominated diphenyl ethers (PBDEs); these accumulate in humans (1–3) and wildlife (4). The toxic equivalency factor (TEF) is employed to assess mixture effects for structurally similar agents, and compares compounds to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (5). TEF allows prediction of interactive toxic effects based on the assumption that dioxinlikeagentshavevariablebindingaffinitiestothearylhydrocarbon receptor (AhR) (6). It is ascertained by AhR binding affinity or induction of CYP1A1 (7). Coplanar agents have stronger binding affinity compared to nonplanar ones (8). PCBs are considered dioxin-like compounds and the TEF values of coplanar congeners have been calculated (9). PBDEs resemble PCBs in chemical structure and may induce dioxin-like effects (10). Some PBDE congeners possess weak binding affinity to AhR (11), so may be toxic at environmental levels (12, 13). Biomarkers include DNA adduct formation (14) or alterations ascertained by spectroscopy (15, 16). Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy has been employed to biochemically signature agent-induced cell alterations at subcytotoxic concentrations (16–18). Because biomolecules absorb in the infrared (IR), ATR-FTIR spectroscopy is used to fingerprint the biochemical state of interrogated material; alterations that might be induced by environmentally relevant chemical mixtures are manifest as an altered IR spectrum (19). Application of IR spectroscopy generates large complex data sets that are best analyzed by multivariate approaches such as principal component analysis (PCA) and/or linear discriminant analysis (LDA) (20). These data reduction methodologies allow one to identify distinctions that can then be related to discriminating wavenumbers (21) tentatively associated with protein conformational changes, DNA/RNA structural alterations, glycogen content, and protein phosphorylation. PCA and LDA are linear multivariate analysis techniques that generate new variables, called factors, which are linear combinations (weighted sums) of the original variables (wavenumbers). PCA can be employed to reduce the number of variables in a spectral data set (originally hundreds of absorption intensities at different wavenumbers) to just a few factors (e.g., 10-20 factors, that can capture more than 95% of the variance present in the whole data set). LDA is a transformation designed to maximize the between-class variance over the intraclass variance of its factors. Thus, in the LDA factor space, data classes tend to form clusters; nearness between clusters can be interpreted as class similarity, whereas cluster segregation may imply dissimilar classes. PCA can be applied before LDA (thus “PCA-LDA”) to avoid overfitting that can occur if LDA is directly applied to the spectral data set. LDA alone may be applied to larger data sets. In this study, we visualize multivariate analysis results in the form of scores plots and cluster vectors plots, based on a cluster vector approach (20). Scores plots are 1-, 2-, or 3-dimensional scatter plots where one, two or three factors are chosen as Cartesian coordinates (as they tend to be the most important). Cluster vectors plots are x-y plots where the x-axis is given in wavenumber units, and the y-axis corresponds to the cluster vectors coefficient values. High 10.1021/es100206f
2010 American Chemical Society
Published on Web 04/29/2010
cluster vector coefficients can reveal the most discriminating wavenumbers (i.e., responsible for segregation between classes). Our aim was to characterize in the human estrogen receptor-positive MCF-7 breast carcinoma cell line the alterations induced by binary mixtures containing environmental levels (10-12 M) of PBDE congeners (47, 153, 183, or 209) ( PCB congeners (126 or 153). These alterations were detected by ATR-FTIR spectroscopy and are analyzed computationally with PCA-LDA [see Supporting Information (SI) Figure S1]. Dose-related effects of PCB126, an AhR agonist, were compared to the nonplanar PCB153 (22) in order to determine magnitude of biochemical alteration versus concentration (10-12, 10-10, or 10-8 M). In addition, we compared the effects of PCB126 with those of PBDE congener, 2,2′,4,4′-tetrabromo diphenyl ether (BDE47), which may mediate dioxin-like actions on AhR (23). Finally, we investigated whether one could identify or estimate additive effects by summing PCA-LDA cluster vectors corresponding to individual treatments and comparing these to cluster vectors corresponding to experimentally observed alterations resulting from binary mixtures.
Experimental Section Cell Culture. MCF-7 cells were grown in Dulbecco’s modified essential medium supplemented with 10% heat-inactivated fetal calf serum (FCS), penicillin (100 U/ml) and streptomycin (100 µg/mL), as before (16). Test Agents. PCB congeners (purity >99%) were purchased from Wellington Laboratories Inc. (Guelph, Canada) and PBDE congeners (purity >99%) were purchased from Cambridge Isotope Laboratories Inc. (Andover, MA, US); PBDEs were predissolved in nonane at 50 µg/mL (50 ppm). PCB congeners 126 and 153 were purchased as powder and 2 µM stock solutions were made up in dimethylsulfoxide (DMSO). PBDE congeners (in nonane) were made up to 2 µM stock solutions with DMSO. From these stock solutions, serial dilutions were made. For all experiments, an appropriate negative control (NC) vehicle containing nonane (purity >95%; Promochem, Welwyn Garden City, Hertfordshire, UK) in DMSO was formulated. Vehicle solution did not exceed 1% (v/v), while ensuring treatment only differed from NC in test agent application. Cell Treatments and ATR-FTIR Spectroscopy. MCF-7 cells were disaggregated, resuspended in complete medium and seeded in T25 flasks; they were then given 24 h to attach prior to treatment with or without test agent for a further 24 h (24). Cells were treated with single agents or binary mixtures: 10-12 M of individual PBDE congeners (47, 153, 183, or 209) and/or 10-12 M, 10-10 M or 10-8 M of PCB congeners (126 or 153). Following treatment, cells were disaggregated again and the cell suspensions were immediately fixed with 70% ethanol (EtOH) and stored at 4°C until use; EtOH was employed to maintain the stability of the cellular material although it is acknowledged this may damage some component parts of biomolecular structures. Cellular material in 70% EtOH was then applied to 1 cm × 1 cm Low-E-reflective glass slides and allowed to air-dry prior to being desiccated. As the penetrating depth of the IR evanescent wave is limited, a sufficient amount of material was applied to the slide (not too thick) in order to generate a good signal-to-noise ratio (SNR). IR spectra were obtained using a Bruker Vector 22 FTIR spectrometer with Helios ATR attachment containing a diamond crystal (Bruker Optics Ltd., Coventry, UK). The ATR crystal was cleaned with sodium dodecyl sulfate (SDS; Sigma Chemical Co., Pool, Dorset, UK) prior to each new sample. In addition, a new background was collected prior to analysis of a new sample. Each experiment was independently repeated three times. For each
treatment slide 10 spectra were acquired, and thus 30 IR spectra were collected in total per treatment class. Spectral Processing and PCA-LDA. Resultant IR spectra were individually cut to the biochemical-cell fingerprint region (1800 cm-1-900 cm-1), resulting in 235 absorbance intensity values per IR spectrum (spectra were acquired at 3.84 cm-1 resolution). Spectra were then baseline-corrected (rubberband) and normalized to Amide I (1650 cm-1). These three preprocessing steps were carried out using OPUS software. Following on, PCA-LDA or LDA alone was carried out using in-house software developed in MATLAB r2008a (The MathWorks, Inc., US). Whenever PCA was applied, the first 10 PCA factors were retained. The first 10 PCA factors captured more than 99% of the variance of the original 235 variables (wavenumbers) in all the cases where PCA was employed. After LDA or PCA-LDA, the number of variables was reduced from 235 to 3-7, depending on the number of classes of the data set under analysis (i.e., 4-8 factors). A New PCA-LDA Cluster Vector Approach. Cluster vectors were originally developed (20) as tools to reveal biochemical alterations that are specific to each data class. Each class has a cluster vector; this was originally defined to be the vector that points from the origin of the PCA-LDA factor space to the center of its corresponding class. One drawback of this approach is that the “control” class itself has a cluster vector, making it difficult for one to identify chemically-induced alterations when compared to corresponding vehicle control. For the purposes of this study, a variant of the PCA-LDA cluster vectors approach was developed. The origin of the PCA-LDA factors space was shifted to the center of the “control” cluster (see SI Figure S2A). Thus, in the new version each cluster vector points from the center of the “control” cluster to the center of a cluster that corresponds to chemical treatment. Consequently, the cluster vector for the “control” class is itself the zero vector; that is, it represents no alteration, whereas the other cluster vectors express comparisons of data classes against their corresponding vehicle control. Statistical Significance Tests. To determine whether treatment groups were significantly different from corresponding controls, t-tests were carried out on the PCA-LDA scores. Thus, the t-tests were carried out taking two classes at a time, where one of the classes was always a control class. This was done by initially projecting the scores from the treatment class and its control class involved onto the corresponding cluster vector (which joins the centers of both classes), and then performing a t-test using the resulting scalar values as data (see SI Figure S2B). This test gives the probability that the current data points would be observed under the following null hypothesis: “both classes have the same mean”. A small P-value (usually below the most commonly used R ) 0.05) leads one to reject the corresponding null hypothesis; that is, the class means are significantly different (see SI Tables S1 and S2). In our case, if a significant difference between classes is obtained, it also indicates that one can proceed to the analysis of the corresponding cluster vector to find out distinguishing regions of the IR spectrum that separate the treated cells from corresponding controls. Models of Predictive Additive Effect. Based on the hypothesis that PBDE and PCB congeners induce additive effects, two predictive models were constructed toward comparison with observed alterations. The first model was obtained by adding together PCA-LDA cluster vectors from individual agent treatments (i.e., 10-12 M BDE47 or 10-12 M PCB126 alone) in order to derive a “pseudocluster vector” predictive of the alterations of both compounds acting together in MCF-7 cells. This predictive curve was then compared to the observed cluster vector of the BDE47 + PCB126 mixture effects. VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3993
FIGURE 1. Effects of 24-h treatment of MCF-7 cells with various doses (10-12, 10-10, or 10-8 M) of PCB126 or PCB153. PCA-LDA scores plots and resultant cluster vectors were derived from triplicate experiments (n ) 30 spectra per chemical treatment). Clusters for (A) PCB126 and (B) PCB153 are observed in scores plots; class distinguishing wavenumbers are identified in cluster vectors plots for (C) PCB126 and (D) PCB153. The second model is analogous to the first one, but here difference-between-means curves were used (instead of cluster vectors). Difference-between-means curves were calculated by subtracting the mean spectrum of a chemical treatment from the mean spectrum of the corresponding vehicle control. Therefore, following the same pattern as in the first model, difference-between-means curves of 10-12 M BDE47 and 10-12 M PCB126 were added together; this was then compared to the difference-between-means curve of the observed BDE47 + PCB126 mixture effects.
3. Results and Discussion Effects of Coplanar versus Nonplanar PCB Congener. Alterations in MCF-7 cells treated for 24 h with 10-12 M, 10-10 M or 10-8 M PCB126 or PCB153 were investigated (Figure 1; see SI Figure S3). In PCA-LDA scores plots the PCB126-treated cells (Figure 1A) generally segregate away from corresponding control more clearly compared to those of PCB153 (Figure 1B). The t-tests indicate that all treatments were significantly different from corresponding control at 5% level of significance (see SI Table S1). The degree of treatment-cluster segregation from control is not dose-related for PCB126; 10-8 M is associated with the highest effect followed by 10-12 M, with 10-10 M generating the cluster least segregated from control (Figure 1A). With the nonplanar PCB153 there appears to be a dose-related effect (Figure 1B), with the highest concentration (10-8 M) being the most segregated. PCB126 is a CYP isoform inducer unlike nonplanar PCBs (22), and this might be reflected in the cluster vectors plot (Figure 1C) that identifies marked protein changes following exposure to this test agent; these wavenumber-associated alterations are less profound following PCB153 exposure (Figure 1D). Exposure to 10-12 M PCB126 resulted in spectral alterations (compared to control) associated with 1508 cm-1 > 1570 cm-1 > 1616 cm-1 > 1404 cm-1 > 1099 cm-1 > 1254 cm-1 in 3994
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010
descending order of weight. Treatment with 10-8 M PCB126 induced the most pronounced spectral changes, and the main alterations in order of ranked importance were: 1545 cm-1 > 1481 cm-1 > 1666 cm-1 > 1616 cm-1 > 1750 cm-1 > 1350 cm-1 > 1165 cm-1 > 1030 cm-1 (Figure 1C). PCB153 (10-8 M) induced a similar, though less pronounced, profile of spectral alterations (Figure 1D). In the DNA/RNA region [e.g., 1225 cm-1 (asymmetric phosphate stretching vibrations; νasPO2-)], the highest effect was observed following treatment with 10-8 M PCB126. In the remaining spectral regions, 10-8 M PCB126 produced further biochemical alterations associated with secondary structure of proteins [Amide I (≈1650 cm-1) and Amide II (≈1550 cm-1)], lipid (≈1750 cm-1) or glycogen (≈1030 cm-1). Hypertrophy effects are mediated by this coplanar AhR agonist (23). The question of whether BDE47 induces a similar profile of cellular alteration to that of a PCB congener remains important. Although previous studies suggest that PBDE congeners are not AhR agonists (25, 26), recent work indicates that they do induce cellular effects (16). To this end, we compared the induced spectral alterations associated with 10-12 M BDE47 exposure to those observed following 24-h treatment with 10-12 M PCB126 (see SI Figure S4). Figure S4 shows that both agents induce a similar pattern of alteration. The main alterations are associated with the lipid region of the spectrum (≈1750 cm-1), Amide I (≈1650 cm-1), Amide II (≈1550 cm-1), carbohydrate (≈1150 cm-1), and symmetric phosphate stretching vibrations; (νsPO2-; ≈1080 cm-1). In fact, BDE47-induced spectral alterations were consistently greater than those associated with PCB126. There is evidence that BDE47 mediates dioxin-like effects (23), and these findings add to this, in contrast to other studies (25, 26). Single Agent versus Binary Mixture Effect. The possibility of detecting cellular alterations induced by test agents using ATR-FTIR spectroscopy with subsequent PCA-LDA led us to
FIGURE 2. PCA-LDA scores plots showing clusters representing IR spectra obtained from MCF-7 cells following 24-h treatment with 10-12 M PBDE congener (47, 153, 183, or 209) and/or 10-12 M PCB congener (126 or 153) individually or as mixtures. investigate the effects of treatments with binary mixtures of differences between “control” and any other given class are PBDE congeners (47, 153, 183, or 209) plus PCB126 or PCB153 highly significant). The profiles of spectral alterations (dis(Figure 2; see SI Figure S5). Binary-mixture effects in MCF-7 tinguishing wavenumbers compared with control), as indicells were compared to their individually induced alterations cated by the cluster vectors (Figure 3), were remarkably in the biochemical-cell fingerprint region (1800 cm-1-900 similar for both PBDE and PCB congeners. This raised the cm-1). Figure 2 is a PCA-LDA scores plot that shows the question as to whether binary mixtures of agents induce induced low-dose (10-12 M) alterations (i.e., treatment clusters additive effects (27) or mediate effects via AhR-independent segregating away from corresponding control) of PBDE mechanisms (28). congeners (47, 153, 183, or 209) ( PCB126 or PCB153. For Interestingly, PBDE congener plus PCB126 treatment all the test agent treatments, good segregation compared to resulted in a less-than-additive effect at shared biochemical control clusters for PBDE congeners (47, 153, 183, or 209) as targets (Figure 3); this was most noted in mixtures with higher well as PCB126 or PCB153 is observed in the PCA-LDA scores brominated PBDE congeners (Figure 2) as reflected by the plot (Figure 2), and this segregation is significant at 5% level coefficient values (vertical distance from the zero line) of the of significance (see SI Table S2; the table again emphasizes distinguishing wavenumbers in the cluster vectors plots VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3995
FIGURE 3. PCA-LDA cluster vectors plots where wavenumber-associated class-distinguishing alterations following 24-h treatment of MCF-7 cells with different combinations of 10-12 M PBDE congener (47, 153, 183 or 209) and/or 10-12 M PCB congener (126 or 153) can be identified. following binary mixture treatment compared to the summation of those observed with individual test agent treatment. The most distinguishing alterations (i.e., distinguishing wavenumbers compared to corresponding control on the cluster vectors plots) induced by BDE47 + PCB126 compared to single-agent treatments in order of descending importance were 1616 cm-1 > 1724 cm-1 > 1018 cm-1 > 1508 cm-1 > 1153 cm-1; a similar pattern was observed with BDE153 + PCB126 (Figure 3). The main biomarkers for the BDE183 + PCB126 binary mixture treatment ranked from highest to lowest included 1539 cm-1 > 1496 cm-1 > 1612 cm-1 > 1747 cm-1 > 1087 cm-1. A binary mixture of BDE209 + PCB126 also apparently reduced biomarker effects compared to individual test agent-induced alterations over the entire biochemical3996
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010
cell fingerprint region, except for 1400 cm-1 (CdO symmetric stretching vibrations of fatty acids and amino acid) where the mixture-induced effect is higher. The main spectral biomarkers altered by the BDE209 + PCB126 mixture included 1612 cm-1 > 1546 cm-1 > 1747 cm-1 > 1211 cm-1 and 964 cm-1. This finding suggests that the level of PBDE bromination might play an important role in modifying binary mixture effects with PCB congeners; higher brominated PBDEs appear to possess an increased ability to compete with PCB congeners for shared molecular targets (29). In contrast, a binary mixture of any of the PBDE congeners (47, 153, 183, or 209) with PCB153 induced a marked segregation of treatment cluster from the corresponding vehicle control along LD1 (Figure 2). The segregation of binary
FIGURE 4. Predictive analyses of the effects of binary mixtures (10-12 M PBDE congener (10-12 M PCB congener) in MCF-7 cells. (A) LDA scores plot representing IR spectra obtained from MCF-7 cells following 24-h treatment with PBDE congener ( PCB126. Confidence ellipsoids (95%) per cluster were drawn assuming normal distributions. Ellipsoids highlighted in light blue represent single test agent and those in red represent binary mixture; the vehicle control is in black. (B) LDA scores plot representing IR spectra obtained from MCF-7 cells following 24-h treatment with 10-12 M PBDE congener (10-12 M PCB153. Ellipsoids highlighted in light blue represent single test agent and those in red represent binary mixture; the vehicle control is in black. (C) Predictive model of additive effect using PCA-LDA cluster vectors. (D) Predictive model of additive effect using difference-between-means curves. mixture treatment away from the corresponding vehicle control cluster pointed to an additive effect of PBDE congener (47, 153, 183, or 209) with PCB153. Alterations induced by binary mixtures incorporating PCB153 generally appeared to be an approximate additive outcome of individual agents, irrespective of the level of PBDE congener bromination (Figure 3). For instance, BDE47 + PCB153 induced elevated weightings compared to either BDE47 or PCB153 individually, except at 1680 cm-1 and 1112 cm-1. Only BDE153 + PCB153 treatment appeared to result in an inhibition of spectral alterations compared to those induced by single agents; the main biomarkers where inhibition was noted included 1211 cm-1 > 1514 cm-1 > 1680 cm-1 > 1630 cm-1. In contrast, throughout the biochemicalcell region clear increases in biomarker alterations relative to those noted following single agent exposures were observed following exposure with BDE183 + PCB153: 1558 cm-1 > 1200 cm-1 > 1500 cm-1 > 1750 cm-1 > 1300 cm-1. Finally, BDE209 + PCB126 also induced increased spectral alterations compared to single test agent treatments throughout the spectral region: 1600 cm-1 > 1214 cm-1 > 1112 cm-1 > 1500 cm-1 > 1753 cm-1 (Figure 3). This raises the possibility that nonplanar PCB153 mediates its effects through mechanisms independent of dioxin-like compounds (28), such as the coplanar PCB126, and as such might not interfere or compete with PBDE congener-induced effects. Predicted versus Observed Additive Models of Binary Mixture Effect. In 3-dimensional LDA scores plots, the entire data set for 24-h treatment of MCF-7 cells with PBDE congener (47, 153, 183, or 209) ( PCB congener (126 or 153) relative to the control cluster was examined (Figure 4A and B). Consistent with previous observations, clusters for binary mixtures of PBDE congener + PCB126 were less distinct
compared to control than those for single-agent treatment (Figure 4A); in contrast, clusters for treatments with binary mixtures containing PCB153 segregated further away (Figure 4B). It appears that one component of the PBDE congener + PCB126 mixture attenuates the effect of the other, whereas approximately additive or independent mechanisms increase the level of spectral alterations associated with exposures to PBDE congener + PCB153. The predictive models using PCALDA cluster vectors (Figure 4C) or differences-betweenmeans spectra (Figure 4D) enable comparison of alterations predicted compared to those observed to determine if BDE47 + PCB126 induces an additive effect. In general, there was a markedly reduced level of observed alteration associated with the binary mixture compared to that predicted in the biochemical region (1800 cm-1-900 cm-1), except for 1700 cm-1, 1500 cm-1, and 1025 cm-1 (Figure 4C). A similar response was observed in the second model (Figure 4D) where 1700 cm-1 and 1500 cm-1 remained the only regions where an additive effect was noted. These independent methods each verify the respective findings of the other. The application of IR spectroscopy to monitor environmental agent-induced alterations in cellular systems is a novel approach (30). The potential to generate an integrated response profile to low-dose effects (16) that may be a nonlinear concentration-response relationship (20) is exciting. Given the multidimensional nature of IR data sets, one of the major challenges will be to extract features (i.e., altered biomarkers) in a computationally robust fashion (20). This methodological approach offers the possibility to investigate the mechanistic effects at sublethal levels of contaminants either singly or in mixtures. VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3997
Acknowledgments V.L. is a NERC-CEH algorithm student (NE/F008643/1).
Note Added after ASAP Publication This paper was published ASAP on April 29, 2010. The abstract was modified. The revised paper was reposted on May 13, 2010.
Supporting Information Available Rationale for algorithm development, further multivariate analyses and t-test results. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Sjodin, A.; Jones, R. S.; Focant, J. F.; Lapeza, C.; Wang, R. Y.; McGahee, E. E.; Zhang, Y. L.; Turner, W. E.; Slazyk, B.; Needham, L. L.; Patterson, D. G. Retrospective time-trend study of polybrominated diphenyl ether and polybrominated and polychlorinated biphenyl levels in human serum from the United States. Environ. Health Perspect. 2004, 112, 654–658. (2) Bi, X. H.; Thomas, G. O.; Jones, K. C.; Qu, W. Y.; Sheng, G. Y.; Martin, F. L.; Fu, J. M. Exposure of electronics dismantling workers to polybrominated diphenyl ethers, polychlorinated biphenyls, and organochlorine pesticides in South China. Environ. Sci. Technol. 2007, 41, 5647–5653. (3) Noren, K.; Meironyte, D. Certain organochlorine and organobromine contaminants in Swedish human milk in perspective of past 20-30 years. Chemosphere 2000, 40, 1111–1123. (4) Ecotoxicology of Wild Mammals. Ecological and Environmental Toxicology Series; Shore, R. F.; Rattner, B. A., Eds.; John Wiley & Sons: London, 2001. (5) Van den Berg, M.; Birnbaum, L.; Bosveld, A. T. C.; Brunstrom, B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa, R.; Kennedy, S. W.; Kubiak, T.; Larsen, J. C.; van Leeuwen, F. X. R.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.; Safe, S.; Schrenk, D.; Tillitt, D.; Tysklind, M.; Younes, M.; Waern, F.; Zacharewski, T. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ. Health Perspect. 1998, 106, 775–792. (6) IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Polychlorinated Dibenzo-Para-Dioxins and Polychlorinated Dibenzofurans, Lyon, France, 4-11 February 1997; IARC Press: Lyon, 1997; 69, 1-631. (7) National Toxicology Program. NTP toxicology and carcinogenesis studies of a binary mixture of 3,3′,4,4′,5-pentachlorobiphenyl (PCB 126) (CAS No. 57465-28-8) and 2,3′,4,4′,5-pentachlorobiphenyl (PCB 118) (CAS No. 31508-00-6) in female Harlan Sprague-Dawleyrats (Gavage studies). Natl. Toxicol. Program Tech. Rep. Ser. 2006, 531, 1-218. (8) Van den Berg, M.; Birnbaum, L. S.; Denison, M.; De Vito, M.; Farland, W.; Feeley, M.; Fiedler, H.; Hakansson, H.; Hanberg, A.; Haws, L.; Rose, M.; Safe, S.; Schrenk, D.; Tohyama, C.; Tritscher, A.; Tuomisto, J.; Tysklind, M.; Walker, N.; Peterson, R. E. The 2005 World Health Organization reevaluation of human and mammalian toxic equivalency factors for dioxins and dioxinlike compounds. Toxicol. Sci. 2006, 93, 223–241. (9) Gies, A.; Neumeier, G.; Rappolder, M.; Konietzka, R. Risk assessment of dioxins and dioxin--like PCBs in food - comments by the German Federal Environmental Agency. Chemosphere 2007, 67, S344–S349. (10) Luthe, G.; Jacobus, J. A.; Robertson, L. W. Receptor interactions by polybrominated diphenyl ethers versus polychlorinated biphenyls: A theoretical structure-activity assessment. Environ. Toxicol. Pharmacol. 2008, 25, 202–210. (11) Chen, G. S.; Konstantinov, A. D.; Chittim, B. G.; Joyce, E. M.; Bols, N. C.; Bunce, N. J. Synthesis of polybrominated diphenyl ethers and their capacity to induce CYP1A by the Ah receptor mediated pathway. Environ. Sci. Technol. 2001, 35, 3749–3756. (12) Kalantzi, O. I.; Hewitt, R.; Ford, K. J.; Alcock, R. E.; Thomas, G. O.; Morris, J. A.; Hewer, A.; Phillips, D. H.; Jones, K. C.; Martin, F. L. Inter-individual differences in the ability of human milkfat extracts to enhance the genotoxic potential of the procarcinogen benzo[a]pyrene in MCF-7 breast cells. Environ. Sci. Technol. 2004, 38, 3614–3622. (13) Safe, S. H. Polychlorinated biphenyls (PCBs): environmental impact, biochemical and toxic responses, and implications for risk assessment. Crit. Rev. Toxicol. 1994, 24, 87–149.
3998
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010
(14) Martin, F. L.; Piearce, T. G.; Hewer, A.; Phillips, D. H.; Semple, K. T. A biomarker model of sublethal genotoxicity (DNA single-strand breaks and adducts) using the sentinel organism Aporrectodea longa in spiked soil. Environ. Pollut. 2005, 138, 307–315. (15) Kelly, J. G.; Martin-Hirsch, P. L.; Martin, F. L. Discrimination of base differences in oligonucleotides using mid-infrared spectroscopy and multivariate analysis. Anal. Chem. 2009, 81, 5314–5319. (16) Llabjani, V.; Jones, K. C.; Thomas, G. O.; Walker, L. A.; Shore, R. F.; Martin, F. L. Polybrominated diphenyl ether-associated alterations in cell biochemistry as determined by attenuated total reflection Fourier-transform infrared spectroscopy: a comparison with DNA-reactive and/or endocrine-disrupting agents. Environ. Sci. Technol. 2009, 43, 3356–3364. (17) Barber, J. L.; Walsh, M. J.; Hewitt, R.; Jones, K. C.; Martin, F. L. Low-dose treatment with polybrominated diphenyl ethers (PBDEs) induce altered characteristics in MCF-7 cells. Mutagenesis 2006, 21, 351–360. (18) Ahmad, M. S.; Mirza, B.; Hussain, M.; Hanif, M.; Ali, S.; Walsh, M. J.; Martin, F. L. ATR-FTIR spectroscopy detects alterations induced by organotin(IV) carboxylates in MCF-7 cells at subcytotoxic/-genotoxic concentrations. PMC Biophys 2008, 1, 3. (19) Martin, F. L. Complex mixtures that may contain mutagenic and/or genotoxic components: a need to assess in vivo targetsite effect(s) associated with in vitro-positive(s). Chemosphere 2007, 69, 841–848. (20) Martin, F. L.; German, M. J.; Wit, E.; Fearn, T.; Ragavan, N.; Pollock, H. M. Identifying variables responsible for clustering in discriminant analysis of data from infrared microspectroscopy of a biological sample. J. Comput. Biol. 2007, 14, 1176–1184. (21) Bi, X.; Walsh, M. J.; Wei, X.; Sheng, G.; Fu, J.; Martin-Hirsch, P. L.; Thomas, G. O.; Jones, K. C.; Martin, F. L. Infrared spectral analysis of MCF-7 cells treated with serum-lipid extracts segregates predominantly brominated flame retardant-exposed subjects from those with mainly organochlorine exposures. Environ. Sci. Technol. 2007, 41, 5915–5922. (22) Matthews, J.; Wihle´n, B.; Heldring, N.; MacPherson, L.; Helguero, L.; Treuter, E.; Haldose´n, L. A.; Gustafsson, J. A. Co-planar 3,3′,4,4′,5-pentachlorinatedbiphenylandnon-co-planar2,2′,4,6,6′pentachlorinated biphenyl differentially induce recruitment of oestrogen receptor R to aryl hydrocarbon receptor target genes. Biochem. J. 2007, 406, 343–353. (23) Wahl, M.; Lahni, B.; Guenther, R.; Kuch, B.; Yang, L.; Straehle, U.; Strack, S.; Weiss, C. A technical mixture of 2,2′,4,4′-tetrabromo diphenyl ether (BDE47) and brominated furans triggers aryl hydrocarbon receptor (AhR) mediated gene expression and toxicity. Chemosphere 2008, 73, 209–215. (24) Jiao, H.; Allinson, S. L.; Walsh, M. J.; Hewitt, R.; Cole, K. J.; Phillips, D. H.; Martin, F. L. Growth kinetics in MCF-7 cells modulate benzo[a]pyrene-induced CYP1A1 up-regulation. Mutagenesis 2007, 22, 111–116. (25) Peters, A. K.; van Londen, K.; Bergman, A.; Bohonowych, J.; Denison, M. S.; van den Berg, M.; Sanderson, J. T. Effects of polybrominated diphenyl ethers on basal and TCDD-induced ethoxyresorufin activity and cytochrome P450-1A1 expression in MCF-7, HepG2, and H4IIE cells. Toxicol. Sci. 2004, 82, 488–496. (26) Peters, A. K.; Nijmeijer, S.; Gradin, K.; Backlund, M.; Bergman, A.; Poellinger, L.; Denison, M. S.; Van den Berg, M. Interactions of polybrominated diphenyl ethers with the aryl hydrocarbon receptor pathway. Toxicol. Sci. 2006, 92, 133–42. (27) Silva, E.; Rajapakse, N.; Kortenkamp, A. Something from “nothing”sEight weak estrogenic chemicals combined at concentrations below NOECs produce significant mixture effects. Environ. Sci. Technol. 2002, 36, 1751–6. (28) Kopec, A. K.; Burgoon, L. D.; Ibrahim-Aibo, D.; Mets, B. D.; Tashiro, C.; Potter, D.; Sharratt, B.; Harkema, J. R.; Zacharewski, T. R., PCB153-elicited hepatic responses in the immature, ovariectomized C57BL/6 mice: Comparative toxicogenomic effects of dioxin and non-dioxin-like ligands. Toxicol. Appl. Pharmacol. 2010, DOI 10.1016/j.taap.2009.12.003. (29) Meerts, I. A. T. M.; van Zanden, J. J.; Luijks, E. A. C.; van LeeuwenBol, I.; Marsh, G.; Jakobsson, E.; Bergman, Å.; Brouwer, A. Potent competitive interactions of some brominated flame retardants and related compounds with human transthyretin in vitro. Toxicol. Sci. 2000, 56, 95–104. (30) Malins, D. C.; Polissar, N. L.; Gunselman, S. J. Infrared spectral models demonstrate that exposure to environmental chemicals leads to new forms of DNA. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 3611–3615.
ES100206F
