(Received for publication, July 15, 1996, and in revised form, October 15, 1996)
From the Department of Biology and Biochemistry, Brunel University, Uxbridge, Middlesex UB8 3PH, United Kingdom
The ability of certain man-made chemicals to
mimic the effects of natural steroid hormones and their potential to
disrupt the delicate balance of the endocrine system in animals are of increasing concern. The growing list of reported hormone-mimics includes the alkylphenolic (AP) compounds, a small number of which have
been reported to be weakly estrogenic. In their most basic form, APs
are composed of an alkyl group, which can vary in size, branching, and
position, joined to a phenolic ring. The aim of this project was to
identify the important structural features responsible for the
estrogenic activity of AP chemicals. This was achieved by incubating
APs with different structural features in a medium containing a
previously described estrogen-inducible strain of yeast
(Saccharomyces cerevisiae) expressing the human estrogen
receptor and comparing their activity spectrophotometrically by the
resulting color change of the medium. The results were compared to the
effects of the main natural estrogen 17-estradiol. The data indicate
that both the position (para > meta > ortho) and branching (tertiary > secondary = normal) of the alkyl group affect estrogenicity. Optimal estrogenic
activity requires a single tertiary branched alkyl group composed of
between 6 and 8 carbons located at the para position on an
otherwise unhindered phenol ring. The results are discussed in relation
to the purity and composition of the chemicals tested.
In the early 1930s, scientists began synthesizing compounds based
on the phenanthrene nucleus of steroidal estrogens in an attempt to
produce substances with similar properties and considerable clinical
value. Estrogenic activity was assessed by subcutaneous administration
of chemicals (dissolved in sesame oil) to ovariectomized rats that were
then observed for the onset of estrus (1). It soon became apparent that
the phenanthrene condensed ring structure was not required for
estrogenic activity, with the discovery that diphenyl and diphenyl
methane derivatives (2) and stilbene derivatives (3-6) containing two
hydroxyl groups in the para positions (e.g.
4,4-dihydroxy diphenyl) were also active. The substituted derivative
of stilbene (4,4
-dihydroxystilbene) was renamed stilbestrol and was
used as the parent compound in the production of another series of
estrogenic compounds, including 4,4
-dihydroxydiethylstilbene (later
known as diethylstilbestrol), which is still one of the most potent
man-made estrogens. During this period, the first references to active
compounds containing only one benzene ring, such as
para-hydroxy propenyl benzene (anol) (7) and some of the
alkylphenolic compounds, are found (8). It seems that this information
was long forgotten, until it recently resurfaced within the context of
endocrine disruption.
Some members of a surprisingly large group of natural and man-made
chemicals are estrogenic (9-17). The unforeseen biological activity
and exposure to man-made xenoestrogens have been implicated in the
increased incidence of certain cancers and also with various disorders
of the male reproductive system, including reduced testicular size and
sperm production in humans and wildlife (18, 19), through both direct
(receptor-mediated) and indirect actions on the endocrine system. For
example, dioxins produce physiological responses in vivo
similar to those seen by administration of potent synthetic estrogens
and antiestrogens, without interacting directly with the estrogen
receptor (17). In receptor binding studies and transfected
chicken embryo fibroblast (CEF) cells, alkylphenols have been shown to
interact with the estrogen receptor directly (20) and act in an
identical way to 17-estradiol in stimulating receptor transcription
(21).
One pathway via which exposure of the developing male fetus to xenoestrogens could result in reduced testicular size and sperm production in adult life has been proposed. Ultimate testicular size and fertility of adult animals is determined shortly after testicular differentiation by the final number of Sertoli cells present. Exposure of the fetus to estrogens at this critical time point could reduce the number of Sertoli cells. Recent studies with rats (22) and rainbow trout (23) have demonstrated an inhibition of testicular growth after exposure to alkylphenolic chemicals. These findings support the contention that exposure to low levels of weakly estrogenic environmental chemicals, including alkyphenols, could affect gonadal development in the fetus. In view of this, it is important to isolate the structural features responsible for the activity of xenoestrogens, so that chemicals with similar features can be identified.
In this study we have screened alkylphenolic compounds of known structure for estrogenicity using a previously described recombinant yeast screen (24). Shifts in potency derived from specific structural changes are used to identify the important structural features responsible for their activity.
Details of the estrogen-inducible
expression system in yeast and preparation of the medium components
have been described previously (24). In brief, the DNA sequence of the
human estrogen receptor was integrated into the yeast genome, which
also contained expression plasmids carrying estrogen-responsive
sequences controlling the expression of the reporter gene Lac-Z
(encoding the enzyme -galactosidase). Thus, in the presence of
estrogens,
-galactosidase is synthesized and secreted into the
medium, where it breaks down the chromogenic substrate chlorophenol
red-
-D-galactopyranoside (CPRG),1 which is initially yellow, into a
red product that can be measured by absorbance.
As described previously in Routledge and
Sumpter (24), stock solutions of alkylphenolic chemicals and the
estradiol standard were serially diluted in ethanol, and 10-µl
aliquots of each concentration were transferred in duplicate to
optically flat 96-well microtiter plates and allowed to evaporate to
dryness. Aliquots (200 µl) of the assay medium containing recombinant
yeast and the chromogenic substrate CPRG were then dispensed to each
sample well. Each plate contained at least one row of blanks (assay
medium only). Possible spurious results resulting from conversion of
the CPRG by the chemicals alone was addressed by incubating identical
plates in assay medium without yeast. The plates were sealed with
autoclave tape and shaken vigorously for 2 min on a titer plate shaker
before incubation at 32 °C. After an 84-h incubation, the color of
the medium was read at an absorbance of 540 nm (using a Titertek
Multiskan MCC/340 plate reader). At this stage, control wells (blanks)
appeared light orange in color, due to background expression of
-galactosidase, and turbid, due to growth of the yeast. Positive
wells were indicated by a deep red color with turbid yeast growth.
Clear yellow wells indicated lysis (a toxic response), and the color
may vary. All data were corrected for turbidity using a second reading
at 620 nm. The plates were left at room temperature for an additional 7 days and observed for possible enzyme-inhibiting action, indicated by
responses below background.
The specificity of the screen was
determined using steroids purchased from Sigma. Stock
solutions (2 × 107 M) were made in
ethanol, and all serial dilutions were carried out in ethanol to
achieve final concentrations of 1 × 10
8 to 5 × 10
12 M in the microtiter wells. The
ability of the yeast screen to detect known environmental estrogens has
been described previously in Routledge and Sumpter (24).
To demonstrate that the chemicals tested were acting through the
estrogen receptor, the ability of tamoxifen, an estrogen antagonist
known to act via the estrogen receptor, to inhibit the activity of the
alkylphenolic chemicals was investigated. A 40 mM stock
solution of tamoxifen (Aldrich) was serially diluted in ethanol, and
10-µl aliquots of each concentration were transferred in duplicate to
optically flat 96-well microtiter plates and allowed to evaporate to
dryness. Aliquots (200 µl) of the assay medium containing recombinant
yeast, the chromogenic substrate CPRG, and 17-estradiol (or
alkylphenolic chemical) were then dispensed to each sample well. The
concentration of 17
-estrogen added to the medium was sufficient to
produce an obvious, but not maximal, response. Controls were assay
medium alone (largest attainable response) and assay medium without the
addition of estradiol or alkylphenols (normal background expression),
which was also incubated concurrently with tamoxifen. The plates were
sealed with autoclave tape and shaken vigorously for 2 min on a titer
plate shaker before incubation at 32 °C. After an 84-h incubation,
the color of the medium was read at an absorbance of 540 nm. At this
stage, medium without 17
-estradiol or alkylphenolic chemicals
appeared light orange in color, due to background expression of
-galactosidase, and turbid, due to growth of the yeast. Wells
containing only 17
-estradiol or alkylphenolic chemicals appeared red
in color, due to increased synthesis of
-galactosidase. The
antiestrogenic activity of tamoxifen was observed as a
dose-dependent inhibition of the color change of the medium
from yellow to red.
Alkylphenolic chemicals with particular
structural features were obtained from commercial and industrial
sources and assessed for estrogenic activity. Chemicals tested were
chosen on the basis of purity and knowledge of their composition. The
activities of alkylphenolic chemicals were compared to the effects of
the main natural estrogen 17-estradiol (purchased from
Sigma). A 200 nM stock of 17
-estradiol
was prepared in ethanol, after which serial dilutions were carried out
to produce final concentrations of 10 nM to 5 pM in the wells.
The alkylphenolic compounds 4-methylphenol (99% pure), 4-ethylphenol (99% pure), 4-propylphenol (99% pure), 3-tert-butylphenol (99% pure), 4-tert-butylphenol (99% pure), and 4-tert-amylphenol (99% pure) were purchased from Aldrich. 2-phenylphenol (99% pure), 3-phenylphenol (99% pure), and 4-phenylphenol (99% pure) were purchased from MTM (Lancashire, United Kingdom). 2-tert-butylphenol (99% pure), 2-sec-butylphenol (99% pure), 4-sec-butylphenol (98% pure), 4-n-pentylphenol (98% pure), 4-tert-hexylphenol (99% pure), 4-tert-heptylphenols (80% pure, 15% 2-tert-heptylphenols), 4-n-heptylphenol (98% pure), 2-tert-octylphenol (99% pure), 4-tert-octylphenol (98% pure), 2-nonylphenol (95% pure), 4-nonylphenol (95% pure), 2-sec-decylphenol (98% pure), 4-sec-decylphenol (90% pure), 2,4 di-nonylphenol (75% pure, 25% 4-nonylphenol), 2-methyl 4-nonylphenol (>95% pure), and 4-methyl 2-nonylphenol (95% mixed isomers) were supplied by Schenectady International, Inc. (Schenectady, NY). 4-nonylphenol diethoxylate (4NP + 2EO; 95% pure), propoxylated 4-nonylphenol (4NP + 1.5PO; 95% pure), 4-sec-dodecylphenol (95% pure), 4-sec-dodecylphenol dipropoxylate (4-DDP + 2PO; 95% pure) were supplied by Witco (Houston, TX). Stock solutions of the chemicals were made up in ethanol, after which serial dilutions were carried out to achieve final concentrations of between 5 mM and 60 nM in the wells.
Relative PotenciesThe estrogenic activity of alkylphenolic
chemicals was assessed by dividing the concentration of the chemical
producing a half-maximal response by the concentration of
17-estradiol required to produce the same response. The data
presented in this paper were the product of a single assay to ensure a
fair representation of the potencies of the chemicals relative to one
another.
The specificity of
the screen was assessed by the ability of a range of steroids and
steroid metabolites, diluted from 1 × 108
M, to stimulate synthesis of
-galactosidase in the
yeast. The data presented in Fig. 1 show that the
strongest response was seen in the wells containing 17
-estradiol.
The next most potent steroids were estrone, the man-made estrogen
diethylstilbestrol, 17
-estradiol, and estriol, which had potencies
approximately 2, 4, 40, and 300 times less than that of
17
-estradiol, respectively. All the other steroids failed to induce
-galactosidase activity. Sensitivity and reproducibility were
assessed by measuring the response of the screen to triplicate
dilutions of 17
-estradiol, over a concentration of 3072 ng/liter to
1.5 ng/liter, compared to triplicates with assay medium only. Variance
did not exceed 0.5% of each mean value at each point across the
dilution range, and a detectable increase in
-galactosidase
production was consistently produced with 2 ng of
17
-estradiol/liter.
Assay Response to Known Environmental Estrogens
The response
of the screen to genistein (phytoestrogen), bisphenol-A (epoxy resin
precursor), and o, p-DDT (pesticide) tested over
a concentration range of 100 mg/liter to 50 µg/liter has been
reported previously in Routledge and Sumpter (24). All produced a
dose-dependent elevation in
-galactosidase production. Thus, the assay is capable of detecting known xenoestrogens.
The ability of the antiestrogen tamoxifen to inhibit
the effect of 17-estradiol and alkylphenolic chemicals was used to
illustrate the requirement of these compounds to interact with the
estrogen receptor to stimulate the synthesis of
-galactosidase.
Yeast cells were incubated in the presence and absence of
17
-estradiol, 4-sec-butylphenol,
4-n-heptylphenol, and 4-tert-octylphenol at concentrations required to produce a distinct, but not maximal, synthesis of
-galactosidase and concurrently with varying
concentrations of tamoxifen. Fig. 2 illustrates that in
the absence of 17
-estradiol or alkylphenolic compounds, the final
background absorbance of the medium (A540)
remained at approximately 0.9. In the presence of 17
-estradiol or
alkylphenolic chemicals, the increased synthesis of
-galactosidase
elevated the absorbance of the medium to approximately 2.0 during the
same incubation period. Wells in which the chemicals were incubated
together with tamoxifen all showed a dose-dependent inhibition of expression of
-galactosidase, indicating that all the
compounds are acting via the estrogen receptor. Tamoxifen on its own
behaved as a weak partial agonist,2
producing a maximal stimulatory response (approximately half that
produced by 17
-estradiol) at a concentration 4 orders of magnitude
greater than that of 17
-estradiol.
Estrogenicity of Alkylphenolic Chemicals
After having previously assessed the response of the screen to a variety of natural and synthetic estrogens, we proceeded to test alkylphenolic chemicals with different structural features for estrogenic activity. All the compounds were tested over a concentration range of 5 mM down to 60 nM. There was no indication that any of the chemicals reacted with the CPRG or affected the performance of the enzyme. Different structural features of alkylphenolic chemicals were assessed for their ability to confer estrogenic activity.
Size of the Alkyl GroupFig. 3 shows the
effect of increasing alkyl group size of 4-tert substituted
alkylphenolic chemicals on estrogenic activity. The small alkyl group
size of 4-methylphenol and 4-ethylphenol (1 and 2 carbons,
respectively) means these exist only as normal branching forms.
4-propylphenol (3 carbons) exists only as normal and secondary branched
forms. 4-methylphenol and 4-ethylphenol did not stimulate
-galactosidase production in the yeast; however, 4-propylphenol (3 carbons in the alkyl group) produced a weak response (20,000,000 times
less potent than estradiol). The response increased with each
additional carbon, up to a maximum of 8 carbon atoms, but all were much
less potent than 17
-estradiol: 4-tert-butylphenol (4 carbons; 1,500,000-fold less potent), 4-tert-amylphenol (5 carbons; 100,000-fold less potent), 4-tert-hexylphenol (6 carbons; 6,000-fold less potent), 4-tert-heptylphenol
(7 carbons; 3,000-fold less potent), and 4-tert-octylphenol
(8 carbons; 1,000-fold less potent). Activity seems to decrease when
the carbon number exceeds 8 because 4-nonylphenol (9 carbons;
30,000-fold less potent) was 30 times less potent than
4-tert-octylphenol.
Degree of Branching of the Alkyl Group
Fig. 4
compares the effect of 4-secondary and 4-tertiary alkyl group branching
structures on estrogenic activity. Chemicals with both types of
branching structure stimulated -galactosidase production in the
yeast and therefore are estrogenic in vitro. 4-tert-butylphenol was approximately 1,500,000 times less
potent than 17
-estradiol and 2.5 times more potent than its
secondary branched equivalent (Fig. 4A).
4-tert-octylphenol was 1,000 times less potent than
17
-estradiol and 60 times more potent than its secondary branched
equivalent (Fig. 4B). This indicates that tertiary branching
structures are more active in vitro and that potency increases with increasing alkyl group size irrespective of the type of
branching. 4-tert-octylphenol was 1,500 times more potent than 4-tert-butylphenol, whereas
4-sec-octylphenol was only 65 times more potent than
4-sec-butylphenol, indicating that secondary branched
structures have a more limited potency range in vitro compared to 4-tert equivalents.
Fig. 5 compares 4-normal (unbranched) alkylphenolic
chemicals to their 4-tert branched equivalents. Chemicals
with both types of branching were active, but in this case the trend
was not so well defined. Of the normal branched structures available,
the most potent was 4-n-amylphenol, which was approximately
30,000 times less potent than estradiol and about 3 times more potent than its 4-tert equivalent (Fig. 5A). The
addition of 2 carbons to the alkyl group reversed this pattern because
4-tert-heptylphenol was approximately 3,000 times less
potent than estradiol and 25 times more potent than its normal branched
equivalent (Fig. 5B).
Fig. 6 summarizes the relative potency of all the
4-branched APs tested compared to 17-estradiol. In general,
4-tertiary branching structures containing 6-8 carbons are
approximately 30 times more potent than 4-normal and 4-secondary
equivalents, which are similar in potency in vitro.
Estrogenicity increases with alkyl group size up to 8 carbons
(tertiary), or between five and six carbons (normal and secondary),
after which it appears to decrease (tertiary), or remain fairly
constant up to 10 carbons (normal and secondary). This plateau in
activity observed by the normal and secondary branching forms may not
continue indefinitely, as 4-sec-dodecylphenol, containing 12 carbons in the alkyl group, was 100-fold less active than the 10 carbon
equivalent.
Position of the Alkyl Group on the Phenol Ring
Fig.
7 shows the effect of positional changes of alkylation
on estrogenic potency. Fig. 7A shows that 2 (ortho)- and 3 (meta)-tert-butylphenol did not stimulate -galactosidase production in the screen and therefore possess no detectable estrogenic activity.
4-tert-butylphenol produced a dose-dependent
response from 5 × 10
5 M, making it
approximately 1,500,000 times less potent than 17
-estradiol. 4-tert-octylphenol was 1,000 times less potent than
17
-estradiol and 10,000 times more potent than the 2-tert
equivalent (Fig. 7B). 4-sec-decylphenol was
100,000 times less potent than estradiol (Fig. 7C), whereas
no observable activity was detected with 2-sec-decylphenol (>100,000K times less potent than 17
-estradiol). This indicates that the alkyl group must be at the para or 4th position on
the phenol ring relative to the hydroxyl group for optimal estrogenic activity. The slight response seen with 2-tert-octylphenol
may be attributed to the higher potency that this size alkyl group gives to the molecule compared to the other alkylphenolic chemicals tested. It is possible that 2- and 3-tert-butylphenol and
2-sec-decylphenol are also estrogenic, but that they must be
present in the assay in many grams/liter to produce a response; that
is, if they are active, they are extremely weak
estrogens.
Effect of Methylation
Fig. 8 illustrates the
effect of adding of a small nonestrogenic methyl group (see Fig. 6)
onto the phenolic ring of an existing AP. Addition of a methyl group to
the second (ortho) position on the phenolic ring of
4-nonylphenol did not affect the activity because 2-methyl
4-nonylphenol and 4-nonylphenol were both approximately 30,000 times
less potent than 17-estradiol. If the alkyl groups change position
so that the methyl group is in the 4th position and the nonyl group is
in the 2nd position, there is a 33-fold reduction in activity.
2-nonylphenol is about 3 times less potent than 4-nonylphenol (data not
shown), so the addition of the methyl group in the para
position was responsible for an additional 11-fold reduction in
activity not attributable to the positional change of the nonyl
(9-carbon alkyl) group alone. Moreover, 3-methyl 2-nonylphenol (in
which the methyl group is at the meta or 3rd position) and
2-nonylphenol are equipotent (data not shown). This indicates that the
para or 4th position (opposite the hydroxyl group on the
phenolic ring) is an important feature for estrogenicity because
activity is reduced only when this position, rather than another
position, is methylated.
Effect of Di-Substitution
Fig. 9 illustrates
the effect of di-substitution on estrogenic activity. However,
interpretation of these results is complicated by the fact that that
the 2,4-di-nonylphenol was not pure; it contained about 25% "free"
nonylphenol. The response produced by the 2,4-di-nonylphenol may be
entirely attributable to the nonylphenol in the sample. It is possible,
therefore, that 2,4-di-nonylphenol has no intrinsic activity.
2,4-di-butylphenol (99% pure) and 2,6-di-butylphenol (99% pure) both
failed to stimulate -galactosidase production in the assay (data not
shown). It therefore seems likely that di-substitution (or addition of
another alkyl group larger than a methyl group) greatly reduces or
completely abolishes estrogenic activity.
Effect of Altering the Size and Branching of the Ethoxylate Side Chain
Fig. 10 shows the effect of ethoxylate
chain length and branching on estrogenic potency of 4-substituted
alkylphenolic compounds. The short chain ethoxylates are particularly
relevant because they are some of the main breakdown products of
nonionic surfactants. Fig. 10A shows that the addition of 2 ethoxylates to 4-nonylphenol results in a 165-fold reduction in
activity, indicating that a free ring hydroxyl group is required for
optimal activity. Moreover, addition of a branched propylene oxide
group results in at least a 35,000-fold reduction in activity because
no -galactosidase production was observed. Fig. 10B
indicates that the same is true for secondary branched alkylphenolic
chemicals because the addition of two propylene oxide groups to
4-sec-dodecylphenol results in a complete loss of activity
over the concentration range tested.
Estrogenic Activity of Different Isomers of Phenylphenol
Many
of the features required for estrogenic activity of alkylphenolic
chemicals may apply to other groups of chemicals, such as those
composed of two benzene rings. Fig. 11 again
illustrates the importance of positional changes on estrogenicity in
three phenylphenol derivatives. Analogous to the situation observed with alkylphenolic chemicals, the most active derivative
(4-phenylphenol) is 20,000 times less potent than 17-estradiol and
bears the phenyl group opposite to the hydroxyl group. Each successive
movement of the phenyl group around the phenolic ring results in an
approximately 7-fold reduction in activity.
A large number of test systems now exist to identify estrogenic activity. These include traditional bioassays relying on relatively complex biological responses, such as stimulation of water imbibition, uterine growth and vaginal cornification in ovariectomized laboratory rodents (16, 25), and the expression of the egg yolk precursor vitellogenin in male oviparous vertebrates (26, 27), in vitro assays including hepatocyte cell cultures (11), and proliferation of MCF-7 breast cancer cell lines (28) and some crystallographic and computer modeling studies (29, 30). More recently, the advancement of molecular biology has allowed the production of recombinant systems. In these systems, the DNA sequence of the estrogen receptor is stably integrated and expressed in host cells that also contain reporter genes that monitor the activity of the receptor by the production of easily quantifiable end points, such as the expression of an enzyme or the production of light by luciferase (31-33).
In vitro studies using a range of different cell lines and
transfection assays have shown that certain xenoestrogens can produce biological responses similar to 17-estradiol itself, albeit only when present at a 1,000- to 100 million-fold greater concentration. The
actual discovery of many weak xenoestrogens in vitro relies on the simple fact that these systems combine extreme sensitivity with
the ability to operate over an extremely large concentration range
(34).
The information available on the estrogenic activity of
alkylphenolic chemicals in vitro and in vivo
is surprisingly consistent, despite species differences in affinity of
receptors for 17-estradiol (35) and variations in uptake,
bioavailability, and metabolism in different systems. In accordance
with our findings (Fig. 3), early experiments by Dodds and Lawson (8)
indicated that a single 100-mg dose of 4-tert-amylphenol
(4-tert-pentylphenol) was more potent than
4-tert-butylphenol. Contrary to our results, however, none
of the 4-normal (unbranched) alkylphenolic compounds (except the propyl
derivatives) were active at the same dose, possibly due to metabolic
conversion to nonestrogenic derivatives in vivo. The method
of administration (the vehicle and route of administration) may also
affect the response to certain xenoestrogens (8).
Data derived from fish (hepatocyte culture), avian (CEF cells), and
mammalian (MCF-7) systems (21, 28) are also concordant with our
findings. All agree that 4-tert-octylphenol is approximately 10-fold more potent than 4-nonylphenol and between 103 and
104 times less potent than 17-estradiol. This suggests
that species differences in the amino acid sequence of the estrogen
receptor are likely to occur in regions that are not important for AP
binding. The discoveries that 4-ethylphenol was inactive, but
4-propylphenol produced a weak proliferative response in MCF-7 cell
culture, and that 4-sec and 4-tert-butylphenol
were equipotent and over 10 times less potent than 4-nonylphenol (28)
also correspond to our findings (see Figs. 3 and 4A).
The basic molecular mechanisms of hormone-dependent transcription activation by the human estrogen receptor in mammalian cells (including DNA binding, hormone binding, nuclear localization, and transcriptional enhancement of a gene bearing estrogen-responsive sequences) have been shown to operate in yeast (36). Yeast-based systems have, therefore, been widely used to establish the mechanisms of ligand binding, as well as dissecting receptor function and transcriptional activation using site-directed mutagenesis (37-39). In vitro tests, however, represent only a part of the metabolic system present in entire animals and may not illustrate the effect of bioconcentration, both of which may affect estrogenic potency. Thus, the potency of a chemical in vitro may not be relevant to an in vivo situation. Chemicals identified as being estrogenic in vitro should therefore be tested in in vivo studies retrospectively. The advantages and limitations of yeast-based systems are discussed in Routledge and Sumpter (24).
The prerequisite of alkylphenolic compounds to interact with the
estrogen receptor to stimulate the production of -galactosidase in
this system was tested by incubating the yeast cells in the absence or
presence of alkylphenols together with the antiestrogen tamoxifen.
Tamoxifen has been shown to be a competitive inhibitor of estradiol
binding to estrogen receptors (40) and is used in the treatment of
hormone-dependent breast cancer. Although tamoxifen is able
to inhibit the action of estradiol, it itself exhibits slight
estrogenic activity and acts as a partial agonist (41, 42), which is in
agreement with our findings. Fig. 2 illustrates the ability of
tamoxifen to inhibit the response of the screen to both 17
-estradiol
and alkylphenolic chemicals in a dose-dependent manner.
This substantiates the requirement for both the estrogen receptor and
an active ligand to stimulate the production of
-galactosidase.
Certain structural motifs are now realized to be important for
estrogenicity. Xenoestrogens are generally fat-soluble mono or
diphenolic chemicals in which hydrophobicity is achieved by either
chlorination (for example o, p-DDT,
o, p
-DDE, and some polychlorinated biphenyls are
all weakly estrogenic) or by the addition of a bulky hydrocarbon side
chain to the meta (3rd carbon) or para (4th
carbon) position of an otherwise unhindered ring (such as
alkylphenols). The present study indicates that the size and degree of
branching of the alkyl group, as well as its position relative to the
hydroxyl group on the phenol ring, are also important features for
estrogenic activity of APs and illustrates the effectiveness of this
yeast-based system in identifying structure-activity relationships of
xenoestrogens. However, it should always be borne in mind that the
chemical structures illustrated within the figures represent only a
two-dimensional image. It is likely that certain aspects of the
structures such as charge distribution and "twist" may produce
novel structural features that could only be revealed in
three-dimensional images and that may also be important in determining
the estrogenic potency of this class of chemicals.
Estrogenicity requires that the alkyl group is composed of 3 or more
carbons. The addition of each carbon increases the hydrophobicity of
the molecule, which may explain, in part, why short chain alky groups
(containing 3 or 4 carbons) are active. A stepwise increase in activity
then occurs with the addition of each carbon (Fig. 3), but divergent
activities occur between different branching groups (Figs. 4, 5, 6). Only
a limited number of 4-secondary and 4-normal branching structures were
available to test, but the results indicate that they are likely to be
similar in potency. This is not surprising, given the structural
similarity of their alkyl groups. The 4-tert branching
structures, however, were generally much more potent, indicating that
the branching on the -carbon (the carbon attached to the phenol
ring) may be an important feature for estrogenicity. This observation
is further supported by the fact that the 30-fold reduction in activity
seen between 4-tert-octylphenol and 4-nonylphenol may be due
in part to the fact that branching also occurs in places outside the
-carbon in the case of 4-nonylphenol (43).
Activity was greatest with 4-tert-octylphenol (8 carbons),
in which 1,000 molecules produced a response similar to that obtained by 1 molecule of 17-estradiol, but 4-nonylphenol (9 carbons) was 30 times less active (Fig. 3). This drop in activity only occurs in
4-tert-branching structures and therefore may be due to the
isomeric heterogeneity of 4-nonylphenol compared with that of some
other 4-alkylphenolic compounds. Recent high-resolution gas
chromatographic analyses of para-nonylphenol have identified 22 para-isomers (43). Isomeric purity of alkylphenolic
chemicals is largely dependent on the nature of the starting olefin. It is possible that due to the augmentation of possible branching forms
(44), 4-nonylphenol contains isomeric branching structures that are
both greater in activity and relatively inactive compared to
4-tert-octylphenol, resulting in an overall reduction in
observed effect.
The position of alkylation on the phenolic ring (irrespective of branching form) also affects estrogenicity, possibly because this affects the alignment of the alkylphenol in the receptor. The potent ortho-para directing influence of the hydroxyl group on the phenol ring does not favor the formation of meta-isomers. However, results from those meta-substituted alkylphenols that were available, along with the data obtained from the phenylphenols (Fig. 11), indicate that estrogenicity increases as the alkyl group is moved from ortho to meta to para, respectively (Fig. 7). The importance of the para position was further illustrated by substitution of an existing alkylphenol with a small inactive methyl group (Fig. 3). Exchanging places of a methyl and a nonyl (9-carbon) group situated on the 2nd (ortho) and 4th (para) position on the ring results in a reduction in activity (Fig. 8) that is not entirely accounted for by the movement of the active nonyl group away from the para position alone. It seems, therefore, that the size and type of branching, as well as the position of the alkyl group on the phenolic ring, are key features in the estrogenicity of AP compounds.
Di-substitution with an alkyl group of 4 carbons or greater greatly reduces or completely abolishes activity, which emphasizes the importance of an unhindered ring structure (Fig. 9). Because no decrease in activity was observed between 4-nonylphenol and 2-methyl-4-nonylphenol, it seems that the alkyl groups must be similar in size for this effect to be noticeable.
Substitution of the hydroxyl group also affects estrogenicity of APs. The compounds tested in Fig. 10 are more relevant to the biodegradation of alkylphenolic surfactants, such as nonylphenol polyethoxylates (NPnEO, where n = 6-40). The parent surfactant itself is not active, but during sewage treatment, APnEO are biodegraded via shortening of the ethoxylate chain to short-chain ethoxylates and carboxylic acid derivatives and finally APs (45), which are all estrogenic (24). Fig. 10 indicates that short-chain ethoxylates are less active than the unsubstituted equivalent, but substitution of the hydroxyl group with a similar-sized propylene oxide derivative results in a much greater reduction in activity. As biodegradation of the primary branched ethoxylated alkylphenols results in the production of estrogenic metabolites, but secondary hydroxyl groups cannot be oxidized to carboxylates by bacteriological activity,3 the biodegradation of propoxylated alkylphenols is therefore less likely to result in the formation of estrogenic intermediates.
Ethoxylated alkylphenolic compounds (APEs) comprise one of the largest volume surfactants in production. Approximately 450 million pounds of APE surfactants were sold in the United States in 1988 alone (46). APEs are used in both household and institutional cleaning products (47). The four largest industrial uses of APEs are in plastics (12) and elastomers, textiles (cleaning, spinning, weaving, finishing), agricultural chemicals (wetters and emulsifiers), and paper (pulping). In the household market the main reported uses of APEs are in laundry detergents and hard-surface cleaners. In addition, nonylphenol polyethoxylates are widely used as a spermicide in contraceptive creams and jellies and in prophylactics. APEs have also been found in hair-care products, including shampoos, hair colorings, and hair styling aids, and certain APs are used in the manufacture of flavors and fragrances, all of which are possible routes for human exposure.
The potential impact of AP compounds and other xenoestrogens on humans
and wildlife is unclear but may be influenced by factors such as the
level and route of exposure, metabolism, bioconcentration, lifestyle,
and stage of development. Unlike natural steroidal estrogens, many
man-made xenoestrogens are lipophilic and may bioaccumulate; in fish,
an average bioconcentration factor for nonylphenol would seem to be
approximately 300 (48). Moreover, circulating xenoestrogenic chemicals
may not be recognized and sequestered by plasma steroid-binding
proteins such as sex hormone binding globulin and -feto protein
(49), which are critical in preventing early estrogen exposure of the
fetus, and thus unmodulated action of APs on target cells and tissues
may occur. All these questions must be addressed if the potential risk
from exposure to xenoestrogens is to be determined with any
accuracy.
We are extremely grateful to research scientists at GlaxoWellcome, particularly Ian Purvis and Doreen Burt, who developed the yeast screen we have used. We thank Robert Yunick, Scott Smith, and Charles Green for both supplying chemicals and commenting on earlier drafts of the paper and Philip Lightowlers for suggesting screening the phenylphenols.