Defining the Impact of Weakly Estrogenic Chemicals on the Action of Steroidal Estrogens

Nissanka Rajapakse, Delia Ong and Andreas Kortenkamp,1

Centre for Toxicology, School of Pharmacy, 29-39 Brunswick Square, London WC1N 1AX, United Kingdom

Received October 23, 2000; accepted January 2, 2001


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
We tested whether bisphenol A (BPA) or o,p'-DDT, when combined with 17ß-estradiol (E2), would contribute to the overall mixture effect using a yeast reporter gene assay, the yeast estrogen screen. Following comprehensive concentration-response analyses of the single agents, the pharmacologically well-founded models of concentration addition and independent action were used to predict entire concentration-response relationships for mixtures of the agents with a variety of fixed mixture ratios, assuming additivity. For molar mixture ratios proportional to the levels normally found in human tissues (i.e., below 1:5000, E2:BPA or o,p'-DDT), these predictions suggest that the effects of individual xenoestrogens are too weak to create an impact on the actions of steroidal hormones. However, at mixture ratios more in favor of the xenoestrogens, a significant contribution to the overall mixture effect was predicted. The predictions were tested experimentally. The observed combined effects of mixtures of E2 with either BPA or o,p'-DDT did not deviate from the additivity expectation. On combining E2 with either BPA or o,p'-DDT at approximately equieffective concentrations corresponding to molar mixture ratios between 1:20,000 and 1:100,000 (E2:BPA or o,p'-DDT), substantial modulations of the effects of E2 became discernible. The assumption that weak xenoestrogens are generally unable to create an impact upon the already strong effects of endogenous steroidal estrogens is not supported by our observations. Our studies indicate that the potential health implication of additive combination effects between xenoestrogens and steroidal estrogens deserve serious consideration.

Key Words: mixtures; xenoestrogens; 17ß-estradiol (E2); bisphenol A (BPA); o,p'-DDT; combination effects; concentration addition (CA); yeast estrogen screen (YES).


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
An impasse appears to have been reached in the field of xenoestrogens. Although increasing numbers of environmentally relevant compounds are being identified as estrogenic (Blair et al., 2000Go), the vast majority of these chemicals are considerably less potent than the steroidal estrogens and are present at low levels in human tissue. Their low potency in relation to 17ß-estradiol is often used to argue that xenoestrogens in combination with steroidal estrogens will not produce effects distinguishable from those of the steroid (Safe, 1995Go). Therefore, the mere demonstration of a compound's estrogenicity is proving insufficient to explain the induction of biological effects at the levels normally encountered in the environment.

To help resolve this dilemma, we set out to evaluate whether synthetic estrogenic chemicals, when combined with 17ß-estradiol, would contribute to the overall mixture effect. Our main interest was to define factors that may influence the ability of a weakly estrogenic compound to modulate the effects of the hormone 17ß-estradiol. We hypothesized that the potential impact of a weak xenoestrogen on 17ß-estradiol would depend predominantly on its concentration relative to the steroid hormone, i.e., the mixture ratio, and its relative potency. To test these ideas experimentally, bisphenol A and o,p'-DDT were selected for in-depth studies. Bisphenol A (BPA) is a monomer in polycarbonate plastics and a constituent of epoxy and polystyrene resins found extensively in food packaging and dental sealants. Human exposure to BPA is significant (Brotons et al., 1995Go; Olea et al., 1996Go), and its estrogenic activity is well established (Dodds and Lawson, 1936Go; Steinmetz et al., 1997Go). The ubiquitous organochlorine pesticide o,p'-DDT is also well documented as weakly estrogenic (Soto et al., 1995Go) and is present in human tissues (Hunter et al., 1997Go).

The fixed mixture ratio design described by Altenburger et al. (2000) and adopted in our earlier work on xenoestrogen mixtures (Payne et al., 2000Go) lends itself particularly well to achieve the goals of the present studies. Briefly, complete concentration-response relationships of mixtures of agents with fixed mixture ratios are predicted on the basis of concentration-response data of all individual mixture components. The predictions are made assuming additive combination effects and then tested experimentally. Agreement between prediction and observation is assessed statistically.

Evaluations of the combined effects of agents rely critically on the method used to estimate the expected effect of a mixture. Synergism and antagonism can be defined as deviations from expected effects, where synergistic mixtures show higher, and antagonistic mixtures lower, than expected effects. When expectations are met, the combined response can be called additive (Berenbaum, 1989Go). Thus, the problem becomes one of defining, on the basis of the potency of individual mixture components, what the expected effect of a mixture should be. A popular method of dealing with this challenge assumes that the combined effect of a mixture is the arithmetic sum of the individual effects of the mixture constituents (Arnold et al., 1997Go; Soto et al., 1995Go; Sumpter and Jobling, 1995Go). Although intuitively appealing, this approach is truly applicable only to agents with linear dose-response curves, and leads to unreliable predictions when used with agents that show sigmoidal curves, as is frequently seen with estrogenic agents (Kortenkamp and Altenburger, 1998Go).

In the present studies, we therefore employed concepts that can be applied to agents with nonlinear dose-response curves, namely the reference models of concentration addition and independent action. Both models have gained considerable acceptance (Greco et al., 1992Go).

The model of concentration addition (CA), introduced by Loewe and Muischnek (1926), assumes that the components of a mixture act in a similar way and have a common site of action. Thus, any effect can be obtained by replacing one substance totally or in part by the equieffective amount of any other. Simply stated, the contribution an agent makes to the overall observed effect of a mixture is proportional to its concentration within the mixture, even below effect thresholds.

Independent action (IA) was developed by Bliss (1939) to accommodate the observation that compounds may act on different subsystems within an organism, which may well involve different sites and modes of action. Individual mixture components are not assumed to contribute to the overall mixture effect if they are present at subthreshold doses.

Our chosen assay system, a recombinant yeast estrogen screen (YES), is an in vitro screen for agents that are capable of interacting with the {alpha}-human estrogen receptor (hER{alpha}). The DNA sequence of hER{alpha} was stably integrated into the main chromosome of yeast (Saccharomyces cerevisiae). Upon binding its ligand, the receptor-ligand complex interacts with the estrogen response element (ERE) that forms part of an hybrid promoter on a plasmid also containing the reporter gene Lac-Z. Expression of the gene leads to the enzyme ß-galactosidase being secreted into the medium, where it acts to convert the chromogenic substrate chlorophenol red-ß-galactopyranoside (CPRG) from the yellow parent compound to chlorophenol red (Routledge and Sumpter, 1996Go). The YES is ideal for use in the field of combination effects, as it is rapid and has been shown to be highly reproducible and sensitive (Beresford et al., 2000Go; Payne et al., 2000Go). As the assay monitors only events immediately following hER{alpha} activation, it is impossible to study the effects of converging signaling pathways, feedback loops, etc. The special features of the yeast cell wall will almost certainly complicate comparisons with uptake and transport phenomena typical for cell membranes in mammalian cells. In common with many in vitro systems, the assay is unable to model toxicokinetic interactions between test compounds that might occur at higher physiological levels.

In this paper we describe the results of investigations of factors that influence the impact of weak xenoestrogens on steroidal estrogens. We conclude that there is no experimental support for ideas that dismiss the possibility of modulations of the effects of 17ß-estradiol by weakly estrogenic chemicals.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Chemicals.
17ß-estradiol (E2; 98+% pure), o,p'-DDT [1-(2-chlorophenyl)-1-(4-chlorophenyl)-2,2,2-trichloroethane; 99+% pure], and bisphenol A (BPA, 4,4'-isopropylidene-diphenol; 97% pure) were purchased from Sigma Chemical Company Ltd. (Dorset, UK), Lancaster Synthesis (Morecambe, UK), and Acros Organics (Geel, Belgium), respectively. The agents were used as supplied and 1 mM stock solutions prepared in Baker HPLC-analyzed absolute ethanol (Mallinckrodt Baker, Deventer, Holland). Stock solutions of the mixtures were also made at 1 mM. Stocks and subsequent dilutions were kept in critically cleaned glass containers and stored at –20°C. All other chemicals used were research grade from Sigma Chemical Company Ltd. (Dorset, UK) unless otherwise stated.

The recombinant yeast estrogen screen.
A detailed description of the yeast estrogen screen can be found in Routledge and Sumpter (1996). Briefly, 50 ml of growth medium were inoculated with 125 µl of 10x concentrated yeast stock and grown overnight in an orbital shaker at 28°C until turbid (absorbance at 640 nm of 1.0). The assay medium consisted of 50 ml of growth medium, chlorophenol red-ß-galactopyranoside (10 mg/l, CPRG, Boehringer Mannheim, East Sussex, UK), and 2 ml of the overnight yeast culture.

Single agents and the mixture stock solutions were serially diluted in HPLC-analyzed ethanol. Aliquots of 10 µl of the dilutions were transferred to 96-well, optically flat bottom microtiter plates and allowed to evaporate to dryness. All plates included a row of ethanol controls (i.e., no test agent) and a row of assay medium without yeast cells (blanks). To each well, except the blanks, a volume of 200 µl of yeast-seeded assay medium was added. To minimize evaporation during the subsequent incubation time, the outer wells were not used for test agents, instead being filled with sterile water.

Plates were sealed with autoclave tape and shaken vigorously for 2 min on a microtiter plate shaker before incubating at 32°C in a humidified box for 72 h. During this period they were again shaken at 24 h and 71 h. Plates were then analyzed spectrophotometrically at 540 nm (color) and 620 nm (turbidity) using a Labsystem Multiskan Multisoft plate reader. Data shown in graphs are corrected for turbidity and constitutive Lac-Z expression seen in the ethanol-treated controls as follows:


(1)

Samples were run in duplicate and experiments were repeated at least twice. Nominal concentrations were used.

Dosimetry.
Scatter plots of corrected absorbance values ("effect") versus log concentration were constructed and analyzed using the best-fit approach (Scholze et al., 2000). The best fit from a number of nonlinear regression models was selected for final data analysis. In these studies we have used the asymmetric (or three-parameter) Hill function


(2)

where Min and Max are the minimal and maximal observed effects, respectively, c the concentration of test agent, EC50 the concentration of test agent yielding half maximal effects, and p the slope parameter; and the three-parameter Gompertz model


(3)

where b is the slope parameter. The 95% confidence intervals of the best estimate of mean effects were also calculated. Nonlinear curve-fitting was carried out using SigmaPlot (v. 5.0, SPSS Inc., U.S.).

Calculation of predicted mixture effects.
Once concentration-response curves for E2, o,p'-DDT, and BPA had been established, the responses of binary mixtures of xenoestrogen with E2 were predicted assuming additive combination effects. The expected effects for a range of mixture ratios were calculated using the models described below.

The model of concentration addition (CA) estimates concentrations of agents that yield a predetermined effect. Such estimates are derived from the relative prevalence of each agent in the mixture and from data on the concentrations of the mixture components that individually would produce this same predetermined effect. Thus, assuming that the combined effect of the mixture is additive, the following expression will hold for any effect level:


(4)

where ci denotes the concentration of the agent i in a mixture yielding an effect E, and ECi the concentration of i needed to produce effect E on its own. With synergistic mixture effects, the expression will give values < 1, with antagonistic mixture effects > 1 (Berenbaum, 1989Go). Equation (1) implicitly defines effect concentrations of a mixture of n agents. It can be used to calculate mixture concentrations that produce a predetermined effect provided the ECi of the individual mixture components and their prevalence in the mixture are known. Thus, the concentration ci of agent i in the mixture is related to the total mixture concentration by


(5)

where pi is the concentration of i relative to the total mixture concentration and ECmix the total concentration of the mixture required to produce effect E. Substitution of ci in Equation 4 gives


(6)

and rearranging yields


(7)

The parameters ECi were calculated from the concentration-response curves of single agents by using the inverse expression of the asymmetric Hill function or three-parameter Gompertz model as appropriate.

The model of independent action (IA) allows predicted effects of mixtures of known composition to be calculated using the expression


(8)

where E(ci) is the effect E produced by compound i at concentration c. Inherent in this expression is the fact that E(ci) cannot exceed 1, i.e., E(ci) is a fraction of a maximal possible effect, making independent action a probabilistic model.

Thus, when applying this model to our assay effects AE(ci), a maximal effect Emax has to be defined. The maximal absorbance value obtained in these investigations was to saturating concentrations of BPA, and this was chosen as a reference point. The effects of test agents are expressed relative to the maximal effect of BPA:


(9)

If the concentration-response relationships of all mixture constituents i are described by an appropriate regression model Fi (Hill or Gompertz), the assay effect AE(ci) can be estimated from the mean effect Fi (ci) predicted by the regression model. Thus,


(10)

substitution of E(ci) in Equation 8 yields


(11)

In order to ensure comparability of the independent action predictions with those of concentration addition, the fractional effects in Equation 11 were rescaled by multiplication with Emax. Thus,


(12)

Assessing mixture predictions.
To test the validity of the modeling, two E2-BPA and three E2-o,p'-DDT mixtures with differing mixture ratios were investigated experimentally. Master stocks (1 mM) were made and serially diluted to cover the range of concentrations modeled in the predictions.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Concentration-Response Analysis for the Single Agents
17ß-estradiol, BPA, and o,p'-DDT induced activation of the hER{alpha} in a concentration-dependent manner (Fig. 1Go). The experimental data were fitted to dosimetric functions (Gompertz for BPA and E2, Hill for o,p'-DDT). The findings for E2 and BPA are in agreement with previously published data (Routledge and Sumpter, 1996Go; Gaido et al., 1997Go), but in our hands the yeast responded to o,p'-DDT at lower concentrations than reported by Routledge and Sumpter (1996). The steroidal hormone was approximately 30,000 times more potent than BPA and about 17,000 times more potent than o,p'-DDT (median effect concentrations are 0.13 nM for 17ß-estradiol, 3.9 µM for BPA, and 2.2 µM for o,p'-DDT). However, the maximal effects produced were 1.59, 1.65, and 0.45 for 17ß-estradiol, BPA, and o,p'-DDT, respectively. Again, the low maximal effect for o,p'-DDT is in agreement with the literature.



View larger version (13K):
[in this window]
[in a new window]
 
FIG. 1. Concentration-response analysis for 17ß-estradiol (E2), bisphenol A (BPA), and o,p'-DDT in the recombinant yeast estrogen screen. Data (open symbols) were fitted to dosimetric functions: Gompertz for BPA and E2, Hill for o,p'-DDT (solid lines) with 95% confidence intervals for mean responses (shading). The dashed horizontal lines in this and in all other figures are the 95% confidence interval of mean responses in untreated control cultures.

 
Predicting the Effects of Binary Mixtures at Different Mixture Ratios
The parameters of the best regression lines of the single agents were used to calculate predictions for a number of mixtures with varying mixture ratios, assuming additive combination effects (Fig. 2Go). Predicted mixture responses were plotted against the sum of the concentrations of both mixture components (Fig. 2Go, curves on the right of each panel). For both the 17ß-estradiol–BPA and 17ß-estradiol–o,p'-DDT mixtures, the figures show shifts of the predicted mixture concentration-response curves toward lower concentrations, which became more pronounced with increases in the relative proportions of 17ß-estradiol in the mixtures. Thus, the models predict that the presence of 17ß-estradiol should increase the potency of the mixtures, to the point where the hormone completely swamps the effects of BPA or o,p'-DDT.



View larger version (33K):
[in this window]
[in a new window]
 
FIG. 2. Predictions as given by concentration addition (A and C) and independent action (B and D) for various molar mixture ratios of E2-BPA (A and B) and E2-o,p'-DDT (C and D) based on best fits to the single agent data (solid lines). Curves on the right of each panel are predictions against the sum of the concentrations of both mixture components. To highlight the contributions of the weak xenoestrogens, mixture effects have also been plotted against E2 concentration in the mixture (curves on left of each panel).

 
To highlight the contributions of the less potent xenoestrogens to the overall mixture effect, the predicted responses were also plotted against the 17ß-estradiol content of the mixtures (Fig. 2Go, curves on the left of each panel) and compared to the effects caused by the hormone on its own. With mixtures containing larger amounts of the hormone, the predicted differences between the effects of the mixture and those of 17ß-estradiol were small. For example, the responses calculated for the 1:5000 (17ß-estradiol:BPA, 17ß-estradiol:o,p'-DDT) mixtures (molar ratios) were almost indistinguishable from those of 17ß-estradiol, indicating that the impact of BPA or o,p'-DDT was negligible. However, the predicted combination effects increased markedly as the composition of the mixture changed in favor of the weaker xenoestrogens. At mixture ratios of 1:20,000 (17ß-estradiol:BPA, 17ß-estradiol:o,p'-DDT) and higher, the predicted combination effects were considerably larger than the response expected on the basis of the 17ß-estradiol content of the mixture. Here, the weak xenoestrogens are expected to contribute significantly to the overall mixture effect.

Although the two models yielded very similar curves for the 17ß-estradiol–BPA mixtures, the CA model consistently produced the more conservative predictions, independent of mixture ratio (Figs. 2A and 2BGo). On the basis of the model predictions, we chose to test experimentally the 1:20,000 and 1:50,000 (17ß-estradiol:BPA) mixtures.

The predictions for the mixtures of 17ß-estradiol and o,p'-DDT demonstrated clearly the main difference in the two models, namely, that the CA model is unable to predict effects higher than those of the mixture component with the lowest maximal effect, in this case the organochlorine (Fig. 2CGo). CA estimates concentrations of mixtures of agents that produce a predetermined effect, and these predictions are based on the effect concentration of each component that will individually produce the same effect. Thus, for mathematical reasons, CA predictions for 17ß-estradiol–o,p'-DDT mixtures cannot be made for effect levels above 0.45, i.e., the maximum response obtained in the YES with o,p'-DDT (Fig. 1BGo). The CA predictions all plateau at effect levels approaching 0.45. In contrast, the IA model predicted concentration-effect curves that began to plateau at an effect similar to that of 17ß-estradiol (Fig. 2CGo). There were peculiar inflection points near the median effect concentrations of the predicted curves for mixture ratios of 1:50,000 and 1:100,000 (17ß-estradiol:o,p'-DDT). From the predictions, three mixture ratios were chosen for experimental investigation, namely, the 1:20,000, 1:50,000, and 1:100,000 (17ß-estradiol:o,p'-DDT) mixtures.

Experimentally Observed Combination Effects: 17ß-Estradiol–Bisphenol A Mixtures
The predicted combination effects for the 1:20,000 and 1:50,000 (17ß-estradiol:BPA) mixtures were tested experimentally. Over the entire range of concentrations, the observations agreed excellently with the CA predictions (Fig. 3Go). There was a complete overlap of the CA predictions with the 95% confidence interval of the best-fit regression line of the observed responses. In both cases, the IA predictions were lower than the observations.



View larger version (19K):
[in this window]
[in a new window]
 
FIG. 3. Predicted and observed mixture effects of 1:20,000 (A) and 1:50,000 (B) E2:BPA mixtures. Best fits to single agent data are shown as solid lines with 95% confidence intervals of mean responses (shading). Predicted effects, computed using the models of concentration addition (long dashed lines) and independent action (short dashed lines) are plotted in terms of total mixture concentration (on the right) and estradiol component of the mixture (on the left of each panel). Mixture observations (solid symbols) are plotted as for predictions.

 
As expected, the 1:20,000 mixture data show clearly that the presence of BPA in the mixture had an impact upon the effect of 17ß-estradiol. When plotted against the 17ß-estradiol content of the mixture, the observed responses were considerably higher than those of the hormone alone and far exceeded the 95% confidence interval of the 17ß-estradiol regression line (Fig. 3AGo, left set of curves). This modulation of 17ß-estradiol by BPA could also be demonstrated for the 1:50,000 mixture (Fig. 3BGo).

Experimental Observed Mixture Effects: 17ß-Estradiol-o,p'-DDT Mixtures
The observed combination effects of the three tested 17ß-estradiol–o,p'-DDT mixtures agreed very well with the lower portions of the CA and IA predictions (Fig. 4Go). However, neither model was able to cope well with the marked changes in the maximal effects of the mixtures, which decreased as the proportion of o,p'-DDT rose in the mixtures. Independent of mixture ratio, the CA model predicted a leveling off at effect levels of approximately 0.45, the maximal effect of o,p'-DDT on its own. The observed responses actually plateaued at a level between the maxima of 17ß-estradiol and o,p'-DDT, the precise level of which depended on the mixture ratio (maximal effects of 0.95, 0.72, and 0.61 for the 1:20,000, 1:50,000, and 1:100,000 mixtures, respectively). Increasing amounts of 17ß-estradiol in the mixture also caused increases in the slopes of the observed concentration-response curves, a feature correctly predicted by CA. It was not possible to confirm experimentally the IA predictions of the maximal effects of the mixtures. Furthermore, the observed concentration-response curves did not show inflection points at median effect levels, as predicted by the IA model.



View larger version (20K):
[in this window]
[in a new window]
 
FIG. 4. Predicted and observed mixture effects of 1:20,000 (A), 1:50,000 (B), and 1:100,000 (C) 17ß-estradiol:o,p'-DDT mixtures. Best fits and 95% confidence intervals of mean responses to the single agents are shown (solid lines and shading, respectively). Observed mixture data (solid symbols) and predictions by concentration addition (long dashed lines) and independent action (short dashed lines) are also plotted in terms of the estradiol component of the mixture.

 
In summary, there was good agreement between prediction and observation in the low-effect range, but neither of the two concepts was able to accurately model the curvatures seen at high-effect levels. As was the case with BPA, o,p'-DDT contributed significantly to the overall mixture effect at mixture ratios of 1:20,000 (17ß-estradiol:o,p'-DDT) and higher.

The Influence of DMSO on 17ß-Estradiol–o,p'-DDT Mixture Effects
The preceding results led us to hypothesize that one reason for o,p'-DDT producing low maximal effects may have been its low solubility, coupled with a decreased ability to enter the yeast cells. To test this, we included DMSO in the assay medium, at a final concentration of 2%. DMSO has been shown previously to lead to marked increases in responses in the YES (Beresford et al., 2000Go). However, it was not established whether this was due to better solubilization of test chemicals, to changes in the properties of the yeast cell wall that enables xenoestrogens to enter and ß-galactosidase to leave more easily, or a combination of the two.

As demonstrated by the shift of the 17ß-estradiol curve to lower concentrations and the almost 40% increase in maximal response (Fig. 5AGo), the assay was much more sensitive in the presence of DMSO. The o,p'-DDT concentration-response curve showed a longer linear portion and also gave higher effects, with a slope similar to the one exhibited by the 17ß-estradiol curve. The effect elicited by the highest tested concentration was 1.1 (corrected absorbance units). Because of concerns about solubility, the tested concentrations were not extended into ranges that normally lead to a leveling-off of effects.



View larger version (21K):
[in this window]
[in a new window]
 
FIG. 5. Effect of DMSO in the assay medium on the single agent concentration-response curves (A) as well as on the predicted and observed mixture effects of 1:50,000 (B) and 1:100,000 (C) 17ß-estradiol:o,p'-DDT mixtures. Solid symbols are observations, with best fits and 95% confidence intervals being solid lines and shading, respectively. Predictions by concentration addition (long dashed lines) and independent action (short dashed lines) are also plotted in terms of the estradiol component of the mixture.

 
The increases in responses caused by o,p'-DDT individually led to changes in the mixture effect predictions. Most notable was the impact on the CA prediction, which modeled higher responses (Figs. 5B and 5CGo). As with the previous 17ß-estradiol–o,p'-DDT mixture responses, there was good agreement between the observations and the CA and IA predictions in the lower portion of the curves. The 1:50,000 mixture observations still exceeded the CA prediction at the point where the predicted responses began to plateau. The same was not seen in the 1:100,000 mixture, where agreement between CA and observation was good for the entire concentration-response curve.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Our data demonstrate clearly that the combined effect of 17ß-estradiol and BPA or o,p'-DDT is additive. There can be no doubt that weak xenoestrogens such as BPA or o,p'-DDT, when combined with 17ß-estradiol, are able to contribute to estrogenic mixture effects. We show that the impact of xenoestrogens on the actions of the steroid hormone depends on the mixture ratio and on its potency relative to 17ß-estradiol. When combined at approximately equieffective concentrations, substantial modulations of the effects of 17ß-estradiol by the xenoestrogen become discernible. This implies that the xenoestrogens will contribute to the combination effect at mixture ratios approximately equal to the ratios of the median effect concentrations. Considering that the molar median effect concentrations of BPA and o,p'-DDT are 30,000- and 17,000-fold higher in the YES assay than those of 17ß-estradiol, this should be the case with combinations showing mixture ratios of around 1:30,000 (17ß-estradiol:BPA) or 1:17,000 (17ß-estradiol:o,p'-DDT) and above. Our experimental results decisively confirm these expectations.

The magnitude of effect modulation by the weaker mixture component becomes apparent when the mixture responses are plotted in terms of hormone concentration and then compared to the effects of 17ß-estradiol on its own. A contribution of the weaker xenoestrogen to the overall combination effect reveals itself as a shift of the mixture concentration-response curves to the left of the 17ß-estradiol curve. The observed shifts with the 1:20,000 mixtures of 17ß-estradiol–BPA and 17ß-estradiol–o,p'-DDT are well outside the 95% confidence intervals of the best estimate regression line of the 17ß-estradiol concentration-response curves (Figs. 3, 4, 5B, and 5CGoGoGo). The effects predicted for 1:5000 mixtures, however, did not indicate that combination effects would be distinguishable from those of the hormone alone, within experimental error.

There was generally good agreement between the predictions of the CA and IA concepts and the experimental observations, although interesting deviations occurred. It is well established (Greco et al., 1992Go) that the CA and IA concepts often, but not always, produce differing predictions. There are no generally agreed criteria that would help decide a priori which of the two concepts should be applied to any given end point or assay system. For this reason, both concepts were used side by side in the present studies. However, in view of the intrinsic features of the YES, it was our expectation that the CA concept should yield the more valid mixture effect predictions. The assay measures the expression of a reporter gene upon activation of the hER{alpha} by ligands that bind to the same receptor domain, and thus seems ideally suited for the CA concept. The assay is blind to any other effects and precludes the monitoring of dissimilarly acting test agents.

The 17ß-estradiol–BPA mixture data confirmed our expectations. With the two tested mixtures, the best-fit regression lines for the observed effects were almost congruent with the curves predicted by CA. The IA concept slightly, but systematically, underestimated the observed combination effects for all effect levels (Fig. 3Go).

When considering the 17ß-estradiol–o,p'-DDT mixture data, a more complicated picture emerged. Again, the CA concept predicted slightly higher combination effects than the IA concept. For low-effect levels, the CA predictions overlapped with the 95% confidence interval of the best-fit mean mixture effects, whereas the IA concept yielded curves displaced to the right of the lower confidence intervals. Furthermore, there were inflection points in the predicted IA curves that did not occur in the experimentally observed concentration-response relationships (Fig. 4Go). However, for all mixtures, the CA concept failed completely to predict high-effect levels and was unable to model the observed maximal effects of the mixtures.

The possible reasons for this system failure deserve serious consideration. We hypothesized that processes related to the low water solubility of o,p'-DDT may have introduced problems with mixture effect predictions. Given that the aqueous solubility limit of o,p'-DDT is 0.35 µM at 35°C (Shiu et al., 1990Go), the actual concentrations of the compound in the culture medium may have been lower than the nominal concentrations used to plot the response curves shown. At some point, further increases in nominal concentrations will have led to disproportionately low rises in the actual concentrations, thus producing the apparent leveling off of the curves seen in Fig. 4Go.

A number of reasons lead us to suggest that data on the aqueous solubility of o,p'-DDT alone do not provide sufficient guidance to decide when that point will be reached.

The o,p'-DDT concentration-response relationship exhibited a linear portion. This signifies a range where increases in nominal concentrations are directly related to rises in the biologically effective concentrations of the agent in the culture medium. Crucially, this occurred at concentrations well above the aqueous solubility limit of o,p'-DDT (Fig. 1BGo).

Glucose metabolism by yeast cells will cause the production of ethanol during the 72-h period of incubation, which in turn is likely to aid the solubilization of o,p'-DDT. Transport processes may also work to shift equilibria and facilitate not only solubilizations of the solid material, but also increase the rate of secretion of ß-galactosidase into the medium. The events following estrogen receptor activation may be subject to further equilibria. Thus, the assay outcome, i.e., color development in the assay medium, is the net result of a large number of complex dynamic equilibria. The aqueous solubility of the test compound is only one, albeit important, factor in this interplay. It is for this reason that we believe that measurements of actual concentrations of sparingly soluble test compounds alone will not help address the problem. Nevertheless, the systematic deviations between mixture effect prediction and observation we encountered with the 17ß-estradiol–o,p'-DDT mixtures seem related to the poor water solubility of the organochlorine. Although beyond the scope of the present study, there is a need to explore systematically the relationship between water solubility of test compounds and assay outcomes in the YES.

Our experience with DMSO coincubations strongly supports these ideas. DMSO acts to increase solubility of the test compounds and may have an impact on the permeability of the yeast's cell wall, increasing both entry of test compound and secretion of ß-galactosidase. As a result, both the linear portion of the o,p'-DDT response curve and the maximal effect increased, and this had a positive influence on the predictability of mixture effects. Our experiences show that the predictability of combination effects with mixtures of sparingly soluble agents needs to be carefully explored.

Our results offer guidance in defining conditions where estrogenic chemicals may add substantially to estrogenic burdens, over and above the already strong effects of steroidal estrogens. From data about the levels of bioavailable 17ß-estradiol and o,p'-DDT in the serum of postmenopausal women (Deutsche Forschungsgemeinschaft, 1984Go; Zava et al., 1997Go), it can be estimated that the molar ratios of the two agents lie between 1:1000 and 1:6000 (17ß-estradiol:o,p'-DDT), far below the ratio of median effect concentrations observed in many bioassays. Our data suggest that the contribution of o,p'-DDT alone to combination effects with the hormone are negligible at physiological levels. Similar considerations apply to BPA, although data about human tissue levels are sparse.

However, human tissues contain many compounds with estrogenic activity. On the basis of our studies, it appears conceivable that a multitude of xenoestrogens, when present in sufficient number and/or concentration, might in principle act together to impact on the actions of steroidal estrogens. Whether such impacts will be physiologically relevant remains to be seen. Definitive answers to this question are currently hampered by our lack of knowledge about the full spectrum of estrogenic agents in human tissues. Furthermore, it is necessary to test experimentally whether xenoestrogens are able to act jointly when they are individually present at subthreshold levels. Finally, it is at present not possible to assess whether data about the type of joint action (additivity, synergism, etc.) at the molecular level of biological organization, amenable to study with in vitro assays, are useful for assessments of mixture effects at higher physiological levels.

In summary, our study does not provide support for the assumption that weak xenoestrogens are generally unable to create an impact upon the already strong effects of endogenous steroidal estrogens. In line with pharmacological principles, they will modulate significantly the actions of steroidal estrogens at mixture ratios similar to their relative potency.


    ACKNOWLEDGMENTS
 
We would like to thank Prof. John Sumpter and Nicky Beresford, Brunel University, Uxbridge, U.K., for providing the recombinant yeast cells and advice on running the assay. Nissanka Rajapakse is grateful for a School of Pharmacy Studentship.


    NOTES
 
1 To whom correspondence should be addressed. Fax: +44 207 753 5908. E-mail: andreas.kortenkamp{at}ams1.ulsop.ac.uk. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Altenburger, R., Backhaus, T., Boedeker, W., Faust, M., Scholze, M., and Grimme, L. H. (2000). Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri: Mixtures composed of similarly acting chemicals. Environ. Toxicol. Chem. 19, 2341–2347.[ISI]

Arnold, S. F., Vonier, P. M., Collins, B. M., Klotz, D. M., Guillette, L. J., Jr., and McLachlan, J. A. (1997). In vitro synergistic interaction of alligator and human estrogen receptors with combinations of environmental chemicals. Environ. Health Perspect. 105, 615–618.[ISI][Medline]

Berenbaum, M. C. (1989). What is synergy? Pharmacol. Rev. 41, 93–141.[ISI][Medline]

Beresford, N., Routledge, E. J., Harris, C. A., and Sumpter, J. P. (2000). Issues arising when interpreting results from an in vitro assay for estrogenic activity. Toxicol. Appl. Pharmacol. 162, 22–33.[ISI][Medline]

Blair, R. M., Fang, H., Branham, W. S., Hass, B. S., Dial, S. L., Moland, C. L., Tong, W. D., Shi, L. M., Perkins, R., and Sheehan, D. M. (2000). The estrogen receptor relative binding affinities of 188 natural and xenochemicals: Structural diversity of ligands. Toxicol. Sci. 54, 138–153.[Abstract/Free Full Text]

Bliss, C. I. (1939). The toxicity of poisons applied jointly. Ann. Appl. Biol. 26, 585–615.

Brotons, J. A., Olea-Serrano, M. F., Villalobos, M., Pedraza, V., Olea, N. (1995). Xenoestrogens released from lacquer coatings in food cans. Environ. Health Perspect. 103, 608–612.[ISI][Medline]

Deutsche Forschungsgemeinschaft. (1984). Rückstände und Verunreinigungen in Frauenmilch. Verlag Chemie: Weinheim.

Dodds, E. C., and Lawson, W. (1936). Synthetic estrogenic agents without the phenanthrene nucleus. Nature 137, 996.

Gaido, K. W., Leonard, L. S., Lovell, S., Gould, J. C., Babai, D., Portier, C. J., and McDonnell, D. P. (1997). Evaluation of chemicals with endocrine modulating activity in a yeast-based steroid hormone receptor gene transcription assay. Toxicol. Appl. Pharmacol. 143, 205–212.[ISI][Medline]

Greco, W., Unkelbach, H. D., Pöch, G., Sühnel, J., Kundi, M., and Bödeker, W. (1992). Consensus on concepts and terminology for combined-action assessment: The Saariselkä agreement. Arch. Complex Environ. Studies 4, 65–69.

Hunter, D. J., Hankinson, S. E., Laden, F., Colditz, G. A., Manson, J. E., Willet, W. C., Speizer, F. E., and Wolff, M. S. (1997). Plasma organochlorine levels and the risk of breast cancer. N. Engl. J. Med. 337, 1253–1258.[Abstract/Free Full Text]

Kortenkamp, A., and Altenburger, R. (1998). Synergisms with mixtures of xenoestrogens: A reevaluation using the method of isoboles. Sci. Total Environ. 221, 59–73.[ISI][Medline]

Loewe, S., and Muischnek, H. (1926). Über kombinationswirkungen. 1. Mitteilung: Hilfsmittel der Fragestellung. Naunyn-Schmiedebergs Arch. Exp. Pathol. Pharmakol. 114, 313–326.

Olea, N., Pulgar, R., Perez, P., Olea-Serrano, F., Rivas, A., Novillo-Fertrell, A., Pedraza, V., Soto, A. M., and Sonnenschein, C. (1996). Estrogenicity of resin-based composites and sealants used in dentistry. Environ. Health Perspect. 104, 298–305.[ISI][Medline]

Payne, J. P., Rajapakse, N. A., Wilkins, M., and Kortenkamp, A. (2000). Prediction and assessment of the effects of mixtures of four xenoestrogens. Environ. Health Perspect. 108, 983–987.[ISI][Medline]

Routledge, E. J., and Sumpter, J. P. (1996). Estrogenic activity of surfactants and some of their degradation products assessed using a recombinant yeast screen. Environ. Toxicol. Chem. 15, 241–248.[ISI]

Safe, S. H. (1995). Do environmental estrogens play a role in development of breast cancer in women and male reproductive problems? Human Ecol. Risk Assess. 1, 17–23.

Scholze, M., Boedeker, W., Faust, M., Backhaus, T., Altenburger, R., and Grimme, L. H. (2001). A general best fit method for concentration-response curves and the estimation of low effect concentrations. Environ. Toxicol. Chem. 20, 448–457.[ISI][Medline]

Shiu, W. Y., Ma, K. C., Mackay, D., Seiber, J. N., and Wauchope, R. D. (1990). Solubilities of pesticide chemicals in water. Part II: Data compilation. Rev. Environ. Contam. Toxicol. 116, 15–187.[ISI][Medline]

Soto, A. M., Sonnenschein, C., Chung, K. L., Fernandez, M. F., Olea, N., Serrano, F. (1995). The E-SCREEN assay as a tool to identify estrogens: An update on estrogenic environmental pollutants. Environ. Health Perspect. 103(S7), 113–122.[ISI][Medline]

Steinmetz, R., Brown, N. G., Allen, D. L., Bigsby, R. M., and Ben-Jonathan, N. (1997). The environmental estrogen bisphenol A stimulates prolactin release in vitro and in vivo. Endocrinology 138, 1780–1786.[Abstract/Free Full Text]

Sumpter, J. P., and Jobling, S. (1995). Vitellogenesis as a biomarker for estrogenic contamination of the aquatic environment. Environ. Health Perspect. 103, 173–178.[ISI][Medline]

Zava, D. T., Blen, M., and Duwe, G. (1997). Estrogenic activity of natural and synthetic estrogens in human breast cancer cells in culture. Environ. Health Perspect. 105(S3), 637–645.[ISI][Medline]