Science of the Total Environment 466–467 (2014) 1112–1113
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Keywords:
Estrogen receptor
QSAR
Endocrine disruption
Polychlorinated biphenyls
We thank Dr. Rayne for the comments on our publication. It should
be stated at the outset that none of the comments Dr. Rayne makes
challenges the validity of the experimental data indicating that 25 of
the 28 hydroxylated PCB's tested bind to the calf uterine estrogen receptor with measurable affinity. The objectives as set out in our publication
were achieved.
Regarding the applicability of modeling the protonated form, the literature tells us that this is the form of 17β-estradiol that specifically
binds to the estrogen receptor with the phenolic proton forming a key
hydrogen bond (Tannenbaum et al., 1998). Thus, the molecular properties of the phenolic forms of the hydroxylated PCB's were modeled and
examined for relationships with binding affinity. It is true that the
anionic form may dominate in bulk solution for a subset of these
molecules. However, receptor binding occurs between one receptor
and one ligand and that reaction does not occur in bulk solution (it is
a desolvating process). The binding pocket of the receptor is hydrophobic. So, bulk properties that may dominate in solution do not necessarily
dominate at the level of the individual receptor and ligand binding reactions. Thus, it is appropriate to model the properties of the protonated
form for purposes of examining relationships of molecular properties
with receptor binding. The molecular modeling of the phenolic
structure reflected the ionizability of the oxygen atom as measured by
the OCHARGE parameter. So, that parameter does contain information
on the potential ionization state of the oxygen atom. The binding data
also incorporates the ionization state that was actually achieved in the
buffer system used. The ionization potential, which is related to pKa,
was incorporated in the QSBR and was discussed in the publication.
The bulk solution equilibrium influences the frequency of encounters between protonated ligand and receptor. However, the affinity
with which the protonated ligand binds to the receptor is not affected
by the bulk solution equilibrium since this is an entirely separate
chemical interaction (Eq. (1)).
Examination of this series of equations indicates that the equilibrium
concentration of OH–PCB is affected by both the pH in the dissociation
reaction and the concentration of estrogen receptor (ER) to which the
OH–PCB may become bound. Therefore, whereas the affinity (Keq)
with which binding occurs is separate from the dissociation reaction,
the measured IC50 is affected by the experimental conditions including
temperature, receptor concentration and pH of the reaction medium.
Two phenolic molecules with theoretically identical ligand binding affinity may have different IC50's if one is present at equilibrium at 1% versus 99% for the other. It may be anticipated that the IC50's in this
theoretical case would differ by two orders of magnitude. Experimentally, this effect could be demonstrated by altering either the pH and/or the
receptor concentration. In the experiments described in our publication,
pH and receptor concentration were held constant.
The model was optimized in a way to minimize susceptibility to bias.
The method for optimization is clearly described in the paper. It is not a
conventional method for optimization which is usually focused on a single measure of “goodness of fit”. We did not choose to apply an arbitrary
p value to the selection of parameters for the multiple regression model
because 1) any particular p value is arbitrary, 2) the parameter may be
contributing information to the overall model even if the p value is
not less than 0.05. The fact that the model predicts IC50 to within one
order of magnitude for all of the OH–PCB's that had measureable binding suggests some level of general applicability of the model to this set
of chemicals. Is the model perfect? Clearly the model is not perfect, as
reflected in residual plot provided by Dr. Rayne. No model is perfect.
Do the percent ionization predictions prepared by Dr. Rayne invalidate
the model? No. This question may be answered by calculating an adjustment to the predicted IC50's taking into account, as a worst case, the potential removal of the anionic species from the binding reaction. This
7
Ionization Adjusted Model
y = 0.7407x + 1.4862
R² = 0.5415
6.5
Predicted Log IC50 (nM)
Response to “Comment on Specific binding of hydroxylated
polychlorinated biphenyl metabolites and other substances
to bovine calf uterine estrogen receptor: Structure-binding
relationships [Kramer and Giesy, Sci Total Environ 1999;233:141–61]”
6
5.5
5
Original Model
y = 0.7081x + 1.3223
R² = 0.7077
4.5
4
3.5
3
2.5
2
2
Keq
½ER–OH–PCB ⇌ ER þ OH–PCB
Ka −
þ
OH–PCB þ H2 O ⇌ O–PCB þ H3 O
DOI of original article: http://dx.doi.org/10.1016/j.scitotenv.2013.03.017.
⁎ Corresponding author. Tel.: +1 3173373137; fax: +1 3173374880.
E-mail address: vjkramer@dow.com (V.J. Kramer).
0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.scitotenv.2013.07.116
ð1Þ
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
Observed Log IC50 (nM)
Original Predicted
Adjusted Predicted
Linear (Original Predicted)
Linear (Adjusted Predicted)
Fig. 1. Comparison of ionization-adjusted model to original model for bovine estrogen receptor binding affinity of hydroxylated PCB's.
V.J. Kramer / Science of the Total Environment 466–467 (2014) 1112–1113
1.5
effect on the model (Fig. 1), though there is an improvement in the
distribution of the residuals (Fig. 2).
Advanced receptor modeling, including docking models, may
improve predictions of binding of hydroxylated PCB's to the estrogen
receptor (Yang et al., 2010).
1
Original Model Residuals
y = 0.2919x-1.3223
R² = 0.2916
Residual
0.5
1113
0
Conflict of interest
Ionization Adjusted Residuals
y = 0.2593x-1.4862
R² = 0.1264
-0.5
-1
-1.5
The authors of this response declare that there is no conflict of interest in the preparation of this response.
References
-2
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
Observed Log IC50 (nM)
Original Residual
Adjusted Residual
Linear (Original Residual)
Linear (Adjusted Residual)
Fig. 2. Comparison of prediction residuals of ionization-adjusted model and original model
for bovine estrogen receptor binding affinity of hydroxylated PCB's.
adjustment then predicts what would the predicted IC50 have been if all
of the anionic species were unavailable for binding, hence shifting the
predicted IC50 to a higher concentration. This is an unrealistic, worst
case assumption because the dissociation reaction is a reversible reaction and will reach an equilibrium with the receptor binding reaction.
Therefore, these adjustments are maximum deviations from expectation. As shown by Dr. Rayne calculations, a significant shift in predicted
IC50 would be expected for 7 out of the 28 OH–PCB's examined (those
with predicted ionizations greater than 50% at pH = 7.4). None of the
OH–PCB's with low IC50's less than 104 nM (hence stronger binding
affinities) are significantly affected by these ionization calculations. Adjustment of the predicted IC50's for percent ionization has little overall
Rayne S. Comment on “Specific binding of hydroxylated polychlorinated biphenyl metabolites and other substances to bovine calf uterine estrogen receptor: structure-binding relationships [Kramer and Giesy, Sci Total Environ 1999;233:141–61]”. Sci Total Environ
2013;454–455:181–3.
Tannenbaum DM, Wang Y, Williams SP, Sigler PB. Crystallographic comparison of
the estrogen and progesterone receptor's ligand binding domains. Proc Natl Acad
Sci U S A 1998;95:5998–6003.
Yang W, Liu X, Liu H, Wu Y, Giesy JP, Yu H-X. Molecular docking and comparative molecular similarity indices analysis of estrogenicity of polybrominated diphenyl ethers
and their analogues. Environ Toxicol Chem 2010;29:660–8. http://dx.doi.org/10.
1002/etc.70.
Vincent J. Kramer
John P. Giesy
Dow AgroSciences LLC, Indianapolis, IN 46074, United States
Corresponding author. Tel.: +1 3173373137; fax: +1 3173374880.
E-mail address: vjkramer@dow.com.
Editor: Damia Barcelo
29 July 2013