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
We tested whether bisphenol A (BPA) or o,p*-DDT, when com-
bined with 17b-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 pre-
dict 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 com-
bined 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 concen-
trations 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 po-
tential health implication of additive combination effects between
xenoestrogens and steroidal estrogens deserve serious consider-
ation.
Key Words: mixtures; xenoestrogens; 17b-estradiol (E2); bis-
phenol A (BPA); o,p*-DDT; combination effects; concentration
addition (CA); yeast estrogen screen (YES).
An impasse appears to have been reached in the field of
xenoestrogens. Although increasing numbers of environmen-
tally relevant compounds are being identified as estrogenic
(Blair et al., 2000), 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 17b-estradiol is often used to argue that xenoestro-
gens in combination with steroidal estrogens will not produce
effects distinguishable from those of the steroid (Safe, 1995).
Therefore, the mere demonstration of a compound’s estroge-
nicity is proving insufficient to explain the induction of bio-
logical effects at the levels normally encountered in the envi-
ronment.
To help resolve this dilemma, we set out to evaluate whether
synthetic estrogenic chemicals, when combined with 17b-es-
tradiol, 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 17b-estradiol. We hypothesized that the
potential impact of a weak xenoestrogen on 17b-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,p9-DDT were selected for in-depth studies. Bisphenol A
(BPA) is a monomer in polycarbonate plastics and a constitu-
ent of epoxy and polystyrene resins found extensively in food
packaging and dental sealants. Human exposure to BPA is
significant (Brotons et al., 1995; Olea et al., 1996), and its
estrogenic activity is well established (Dodds and Lawson,
1936; Steinmetz et al., 1997). The ubiquitous organochlorine
pesticide o,p9-DDT is also well documented as weakly estro-
genic (Soto et al., 1995) and is present in human tissues
(Hunter et al., 1997).
The fixed mixture ratio design described by Altenburger et
al. (2000) and adopted in our earlier work on xenoestrogen
mixtures (Payne et al., 2000) 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 pre-
dictions 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 mix-
ture. 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
1
To whom correspondence should be addressed. Fax: 144 207 753 5908.
E-mail: andreas.kortenkamp@ams1.ulsop.ac.uk.
TOXICOLOGICAL SCIENCES 60, 296–304 (2001)
Copyright 2001 by the Society of Toxicology
296

additive (Berenbaum, 1989). 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.,
1997; Soto et al., 1995; Sumpter and Jobling, 1995). 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, 1998).
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., 1992).
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 mix-
ture 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 a-human estrogen receptor (hERa). The
DNA sequence of hERa was stably integrated into the main
chromosome of yeast (Saccharomyces cerevisiae). Upon bind-
ing 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 b-galactosidase
being secreted into the medium, where it acts to convert the
chromogenic substrate chlorophenol red-b-galactopyranoside
(CPRG) from the yellow parent compound to chlorophenol red
(Routledge and Sumpter, 1996). 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., 2000;
Payne et al., 2000). As the assay monitors only events imme-
diately following hERa 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 17b-estradiol by weakly estrogenic chemicals.
MATERIALS AND METHODS
Chemicals. 17b-estradiol (E2; 981% pure), o,p9-DDT [1-(2-chlorophe-
nyl)-1-(4-chlorophenyl)-2,2,2-trichloroethane; 991% pure], and bisphenol A
(BPA, 4,49-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 mix-
tures 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 mlof103 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-b-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 ml 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 ml 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:
Corrected absorbance 5 test
540 nm
2 test
620 nm
1 control
620 nm
2 control
540 nm
(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
Effect 5 Min 1
~Max 2 Min!
F
1 1
S
c
EC
50
D
2p
G
(2)
where Min and Max are the minimal and maximal observed effects,
respectively, c the concentration of test agent, EC
50
the concentration of test
297IMPACT OF XENOESTROGENS ON ESTRADIOL