ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
INTEGRATED SUMMARY REPORT
for
Validation of an
Estrogen Receptor Binding Assay
using
Rat Uterine Cytosol as Source of Receptor
as a Potential Screen in the
Endocrine Disruptor Screening Program Tier 1 Battery
March 2009
U.S. Environmental Protection Agency
Office of Science Coordination and Policy
and
Office of Research and Development
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Table of Contents
I. Introduction 1
A. Purpose of the Endocrine Disrupter Screening Program 1
B. Tiered approach to screening 1
C. The Tier 1 battery of assays 2
D. Validation 4
E. Purpose of this report 7
II. Purpose and brief description of the assay 8
A. Purpose of the assay 8
B. Overview of the assay 8
C. Review of literature 9
D. Assay components other than test chemical 10
1. Solvent 10
2. Reference estrogen - 17(3-Estradiol 10
3. Marker/tracer- Radiolabeled 17(3-estradiol 11
4. Positive control - Norethynodrel 11
5. Negative control - Octyltriethoxysilane 11
6. Rat uterine cytosol 11
E. Saturation binding assay 12
F. Competitive binding assay 15
1. One-site binding model 17
2. Statistical analysis 19
III. Assay standardization and optimization 20
A. Buffer composition and receptor concentration 20
B. Separation technique (HAP vs. DCC) 22
C. Post-incubation temperature 23
D. Assay Volume 25
E. Cytosol Source 26
F. Cytosol shelf life 26
G. Concentration of radiolabeled estradiol 27
H. Maximum solvent concentration 28
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
IV. Interlaboratory validation 29
A. First interlaboratory study 29
1. Selection of laboratories 29
2. Study design 29
3. Results 30
a. Saturation binding 30
b. Competitive binding - standards 33
c. Competitive binding -test chemicals 38
B. Second interlaboratory study 45
1. Selection of laboratories 45
2. Selection of test chemicals 49
3. Preparation of rat uterine cytosol 53
4. Results 53
a. Saturation binding 54
b. Competitive binding 55
5. Analysis 55
a. Model fitting 56
b. Evaluation of runs for acceptability 56
c. Comparison across laboratories 57
d. Development of performance standards 61
6. Discussion 63
V. Additional considerations 67
VI. Summary 68
A. Strengths 68
B. Weaknesses 69
C. Conclusion 70
VII. References 72
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
List of Appendices
Appendix 1. Protocol (as revised following the second interlaboratory validation study)
Appendix 2. ICCVAM Background Review Document on the Estrogen Receptor
Binding Assay. Executive Summary and Conclusions.
Appendix 3. Eldridge CJ. 2007. Final report: Development of a standardized
approach for evaluating environmental chemicals with low solubility in the
estrogen receptor (ER) binding assay.
Appendix 4. Report on statistical methods for evaluating variability in and setting up
performance criteria for receptor binding assays
Appendix 5. Overall report on second interlaboratory validation study
Appendix 6. Final report from second interlaboratory validation study: Lab X
Appendix 7. Final report from second interlaboratory validation study: Lab Y
Appendix 8. Final report from second interlaboratory validation study: Lab Z
Appendix 9. Detailed statistical report
Appendix 10. Curve fits after normalization: Lab X
Appendix 11. Curve fits after normalization: Lab Y
Appendix 12. Curve fits after normalization: Lab Z
Appendix 13. Graphs of acceptable runs for reference standard (estradiol), weak
positive (norethynodrel), and test chemicals, by laboratory
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
List of Tables
Table 1. Tier-1 screening assays recommended by the EDSTAC
Table 2. Alternative assays recommended by the EDSTAC for the Tier-1 Screening
Battery
Table 3. Performance criteria for competitive binding, reference and weak positive
controls.
Table 4. Saturation assays comparing buffer composition and receptor concentration
Table 5. Competitive assays comparing buffers
Table 6. Comparison of separation systems (DCC vs. HAP) in competitive assays
Table 7. Comparison of total assay volume, competitive binding assay
Table 8. Age of animal source of receptor
Table 9. Intra-laboratory variability of the saturation assay with centrally supplied
cytosol preparation (series a and b)
Table 10. Intra-laboratory variability of the saturation assay using individual laboratory
cytosol preparation (series c and d)
Table 11. Intra-laboratory variability of the competitive binding assay with centrally
supplied cytosol
Table 12. Competitive assay results for the individual-laboratory prepared cytosol,
standard and weak positive chemicals
Table 13. Test chemical intra-laboratory variability using centrally-supplied cytosol
(series b)
Table 14. Test chemical inter-laboratory variability (series b and d)
Table 15. Test chemical intra-laboratory variability using individual-laboratory-prepared
cytosol (series d)
Table 16. Performance criteria for second interlaboratory study
Table 17. Qualification runs: Saturation binding assays
Table 18. Qualification runs: Competitive binding assays
Table 19. Qualification runs: Summary of estradiol competitive binding data
Table 20. Qualification runs: Summary of norethynodrel competitive binding data
Table 21. Chemicals selected for the second interlaboratory validation study
IV
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Table 22. Rat uterine cytosol preparations
Table 23. Saturation binding results, Lab X
Table 24. Saturation binding results, Lab Y
Table 25. Saturation binding results, Lab Z
Table 26. Correspondence between experimentally determined binding category and
expected affinity
Table 27. Comparison of classifications across labs, ranked by expected affinity
Table 28. Slope, top, bottom, RBA: Tolerance interval bounds to contain at least 80% of
population of test runs with 95% confidence. Outliers deleted.
Table 29. Ln(residual standard deviation), residual standard deviation: Tolerance
interval bounds to contain at least 80% of population of test runs with 95%
confidence. Outliers deleted.
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
I. Introduction
A. Purpose of the Endocrine Disruptor Screening Program
Section 408(p) of the Federal Food Drug and Cosmetic Act (FFDCA) requires the
U.S. Environmental Protection Agency (EPA) to
develop a screening program, using appropriate validated test systems
and other scientifically relevant information, to determine whether
certain substances may have an effect in humans that is similar to an
effect produced by a naturally occurring estrogen, or other such
endocrine effect as the Administrator may designate [21 U. S. C.
346a(p)].
Subsequent to passage of the Act, the EPA formed the Endocrine Disruptor
Screening and Testing Advisory Committee (EDSTAC), a committee of scientists and
stakeholders that was charged by the EPA to provide recommendations on how to
implement its Endocrine Disruptor Screening Program (EDSP). The EDSP is described
in detail at http://www.epa.qov/scipoly/oscpendo/.
Upon recommendations from the EDSTAC (1998), the EPA expanded the EDSP
using the Administrator's discretionary authority to include the androgen and thyroid
hormonal systems as well as wildlife.
B. Tiered approach to screening
The EPA accepted the EDSTAC's recommendations for a two-tier screening
program in a Federal Register Notice in 1998 (USEPA (1998)). The purpose of Tier 1 is
to identify the potential of chemicals to interact with the estrogen, androgen, or thyroid
(EAT) hormonal systems. A negative result in Tier 1 would be sufficient to put a
chemical aside as having low to no potential to cause endocrine disruption, whereas a
positive result would require further testing in Tier 2. The purpose of Tier 2 is to confirm
the interaction, identify and characterize any adverse effects, and to provide information
that will be useful in risk assessment based, in part, on dose-response relationships.
Tier 2 is expected to comprise multigeneration tests in various taxa (i.e., mammals,
birds, fish, amphibians, and invertebrates).
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
C. The Tier 1 battery of assays
The EDSTAC concluded that a Tier-1 battery should be comprised of a suite of
complementary screening assays having the following characteristics:
• Maximum sensitivity to minimize false negatives while permitting an as yet
undetermined, but acceptable, level of false positives.
• Range of organisms representing known or anticipated differences in metabolic
activity and include assays from representative vertebrate classes to reduce the
likelihood that important pathways for metabolic activation or detoxification of
parent substances or mixtures are not overlooked.
• Capacity to detect all known modes of action (MOAs) for the endocrine endpoints
of concern. All chemicals known to affect the action of EAT hormones should be
detected.
• Range of taxonomic groups among the test organisms. There are known
differences in endogenous ligands, receptors, and response elements among
taxa that may affect the endocrine activity of chemical substances or mixtures.
• Diversity among the endpoints and within and among assays to reach
conclusions based on "weight-of-evidence" considerations. Decisions based on
the screening battery results will require weighing the data from several assays.
• Inexpensive, quick, and easy to perform.
To detect chemicals that may affect the EAT hormonal systems through any one
of the known MOAs — interruption of hormone production or metabolism, binding of the
hormone with its receptor, interference with hormone transport, etc. — the EDSTAC
recommended the in vitro and in vivo assays shown in Table 1 for inclusion in the Tier-1
screening battery.
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 1. Tier-1 screening assays recommended by the EDSTAC
Assays
Estrogen receptor (ER) binding
ortranscriptional activation
Androgen receptor (AR) binding
ortranscriptional activation
In vitro steroidogenesis
Uterotropic (rat)
Hershberger (rat)
Pubertal female (rat)
Frog metamorphosis
Fish screen
Reasons for consideration
A sensitive in vitro test to detect chemicals that may affect the endocrine
system by binding to the ER.
A sensitive in vitro test to detect chemicals that may affect the endocrine
system by binding to the AR.
A sensitive in vitro test to detect chemicals that interfere with the synthesis of
the sex steroid hormones.
An in vivo assay to detect estrogenic chemicals. It offers the advantage over
the binding assay of incorporating absorption, distribution, metabolism, and
excretion (ADME)
An in vivo assay to detect androgenic and anti-androgenic chemicals. It offers
the advantage over the binding assay of incorporating ADME and
differentiating between AR agonists and antagonists.
An assay to detect chemicals that act on estrogen or through the
hypothalamus-pituitary-gonadal (HPG) axis that controls the estrogen and
androgen hormone systems. It also detects chemicals that interfere with the
thyroid system.
A sensitive assay for detection of chemicals that interfere with the thyroid
hormone system.
Fish are the furthest removed from mammalians among vertebrates both from
the standpoint of evolution — their receptors and metabolism are different from
mammals — and exposure/habitat, since they would be subject to exposure
through the gills, whole body, and diet. Thus, the fish assay would augment
information found in the mammalian assays and would be more relevant than
the mammalian assays in triggering concerns for fish and perhaps other non-
mammalian taxa.
In addition, the EDSTAC recognized there were other combinations of screening
assays that may be suitable and therefore recommended that the EPA validate the
alternative screening assays shown in Table 2.
Table 2. Alternative assays recommended by the EDSTAC for the Tier-1 Screening Battery
Assays
In vitro placental aromatase
Pubertal male (rat)
Adult male (rat)
Reasons for consideration
The aromatase assay detects chemicals that inhibit aromatase and would be
needed if either of the two following assays using males were substituted for
the female pubertal assays. The male is not believed to be as sensitive to
alterations in aromatase as the female and would not therefore be sufficient to
detect interference with aromatase in the screening battery.
The assay detects chemicals that act on androgen or through the HPG axis
that controls the estrogen and androgen hormone systems. It ialso detects
chemicals that interfere with the thyroid system. This assay could in part
substitute for the female pubertal assay.
The assay is also designed to detect chemicals that act on androgen or
through the HPG axis that controls the estrogen and androgen hormone
systems. It is also designed to detect chemicals that interfere with the thyroid
system..
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
D. Validation
As noted, Section 408(p) of the FFDCA requires the EPA to use validated test
systems and other scientifically relevant information. Validation has been defined as
"the process by which the reliability and relevance of a test method is evaluated for a
particular use" (OECD (1996); NIEHS (1997)).
Reliability is defined as the reproducibility of results from an assay within and
between laboratories.
Relevance describes whether a test is meaningful and useful for a particular
purpose (OECD (1996)). ForTier-1 EDSP assays, relevance can be defined as
the ability of an assay to detect chemicals with the potential to interact with the
estrogen, androgen, and/or thyroid hormonal systems.
The EDSTAC considered the ER binding assay to have "gained sufficient
general acceptance within the field of endocrine toxicology to be considered de
facto validated (reliable and relevant)" (EDSTAC 1998, Appendix R).
Nevertheless, it continued, "variations in protocols for [this screen] can produce
disparate results. Therefore, standardization of the protocol ... should be
accomplished by EPA before [this assay is] implemented as [a] screening
[requirement] for endocrine activity or disruption." (Ibid.) As a result, EPA began
efforts to standardize the assay. These standardization efforts are described in this
document, along with two additional validation studies.
Federal agencies are also instructed by the Interagency Coordinating Committee
for the Validation of Alternative Methods (ICCVAM) Authorization Act of 2000 to ensure
that new and revised test methods are valid prior to their use.
In general, the EPA has followed a five-stage validation process outlined by the
ICCVAM (NIEHS (1997)) for validation EDSP assays. The stages of the process
outlined by the ICCVAM are as follows:
First Stage - Test Development, an applied research function which culminates in
an initial protocol. As part of this phase, the EPA prepares a Detailed Review Paper
(DRP) to explain the purpose of the assay, the context in which it will be used, and the
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
scientific basis upon which the assay's protocol, endpoints, and relevance rest. The
DRP reviews the scientific literature for candidate protocols and evaluates them with
respect to a number of considerations, such as whether the candidate protocols meet
the assay's intended purpose, the costs and other practical considerations. The DRP
also identifies the developmental status and questions related to each protocol;
provides the information needed to answer the questions; and, when possible,
recommends an initial protocol for the initiation of the second stage of validation.
Second Stage - Standardization and Optimization, in which the protocol is
refined, optimized, standardized and initially assessed for transferability and
performance. Several different types of studies are conducted during this second phase
depending upon the state of development of the method and the nature of the questions
that the protocol raises. The initial assessment of transferability is generally a trial in a
second laboratory to determine that another laboratory besides the lead laboratory can
follow the protocol and execute the study.
Third Stage - Inter-laboratory Validation studies are conducted in independent
laboratories with the optimized protocol. The results of these studies are used to
determine inter-laboratory variability and to set or cross-check performance criteria.
Fourth Stage - Peer Review, an independent scientific review by qualified
experts. The EPA has developed extensive guidance on the conduct of peer reviews
because the Agency believes that peer review is an important step in ensuring the
quality of science that underlies its regulatory decisions (USEPA (2007)).
Fifth Stage - Regulatory Acceptance, adoption for regulatory use by an agency.
Criteria for the validation of alternative test methods (in vitro methods designed to
replace animal tests in whole or in part) have generally been agreed upon in the United
States by the ICCVAM, in Europe by the European Centre for the Validation of
Alternative Methods (ECVAM), and internationally by the Organisation for Economic Co-
Operation and Development (OECD). These criteria, as stated by ICCVAM (NIEHS
(1997)), are as follows:
1. The scientific and regulatory rationale for the test method, including a clear
statement of its proposed use, should be available.
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
2. The relationship of the endpoints determined by the test method to the in vivo
biologic effect and toxicity of interest must be addressed.
3. A formal detailed protocol must be provided and must be available in the
public domain. It should be sufficiently detailed to enable the user to adhere
to it and should include data analysis and decision criteria.
4. Within-test, intra-laboratory and inter-laboratory variability and how these
parameters vary with time should have been evaluated.
5. The test method's performance must have been demonstrated using a series
of reference chemicals preferably coded to exclude bias.
6. Sufficient data should be provided to permit a comparison of the performance
of a proposed substitute test to that of the test it is designed to replace.
7. The limitations of the test method must be described (e.g., metabolic
capability).
8. The data should be obtained in accordance with Good Laboratory Practices
(GLPs).
9. All data supporting the assessment of the validity of the test methods
including the full data set collected during the validation studies must be
publicly available and, preferably, published in an independent, peer-reviewed
publication.
The EPA has adopted these validation criteria for the EDSP as described
elsewhere (USEPA (2007)). Although attempts have been made to thoroughly comply
with all validation criteria, the various in vitro and in vivo screening assays are not
replacement assays (Validation Criterion No. 6). Many of them are novel assays;
consequently, large data bases do not exist as a reference to establish their predictive
capacity (e.g., determination of false positive and false negative rates). It is expected
that a review of results from the testing of the first group of 50 to 100 chemicals, which
was recommended by the Scientific Advisory Panel (SAP) (USEPA (1999)), will allow a
more complete assessment of the performance of the Tier-1 screening battery.
For technical guidance in developing and validating the various Tier-1 screens
and Tier-2 tests, the EPA chartered two federal advisory committees: the Endocrine
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Disrupter Methods Validation Subcommittee, or EDMVS (from 2001 to 2003), and the
Endocrine Disrupter Methods Validation Advisory Committee, or EDMVAC (from 2004
to 2006). These committees, composed of scientists from government, academia,
industry, and various interest groups, were charged to provide expert advice to the EPA
on protocol development and validation. The EPA also cooperates with member
countries of the OECD to develop and validate assays of mutual interest to screen and
test for endocrine effects.
Even though assays are being developed and validated individually and peer
reviewed on an individual basis (i.e., their strengths and limitations are being evaluated
as stand-alone assays), the Tier-1 assays will be used in a battery of complementary
screens. An individual assay may serve to strengthen the weight of evidence in a
determination (e.g., positive results in an ER binding assay in conjunction with positive
results in the uterotropic and pubertal female assays would provide a consistent signal
for estrogenicity) or to provide coverage of MOAs not addressed by other assays in the
battery. Information supporting the validation of individual assays was used at the
Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) SAP for peer review of the
EPA's recommendations for a Tier-1 battery (USEPA 2008). The Tier-1 battery peer
review focused, in part, on the extent of coverage and overlap that the suite of assays
will have with one another in detecting endocrine-related effects associated with the
EAT hormonal systems.
E. Purpose of this report
The purpose of this Integrated Summary Report is to provide an historical
summary of the standardization and validation of a protocol for the estrogen receptor
(ER) binding assay using rat uterine cytosol (RUC) as source of receptors. The
reasoning and judgments leading to the various studies, and conclusions concerning
the strengths and weaknesses of the assay in its current form, are presented.
7
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
II. Purpose and brief description of the assay
A. Purpose of the assay
This assay will be used to determine the ability of a compound to interact with the
ERs isolated from rat uteri. It will be used in conjunction with other in vitro and in vivo
assays in the EDSP to determine whether there is a potential for the chemical to interact
with the estrogen hormonal system. Because it is intended for screening for potential
interaction of any sort, the assay is not being standardized for distinguishing one-site
competitive binding from other kinds of interaction. In addition, the assay is not being
standardized for use in Quantitative Structure-Activity Relationship models.
As explained in section I.B, information from Tier 1 screens such as the ER
binding assay will be used as indications of the need for further testing in Tier 2. It is
only after Tier 2 results are available that risk assessment will be undertaken.
Generating dose-response information for use in risk assessment is not one of the
purposes of this assay.
B. Overview of the assay
This assay evaluates the inhibition of radiolabeled estradiol binding to rat ER by
a test chemical. Rat uterine cytosol (RUC) is the source of the estrogen receptor. The
assay consists of two sets of experiments: a saturation binding assay to characterize
receptor activity, followed by competitive binding assays that measure the competition
of test compounds and control chemicals (the native ligand, 17|3-estradiol, as the
reference ("standard") chemical; a weak positive control, norethynodrel; and a negative
control, octyltriethoxysilane) against radiolabeled 17|3-estradiol for the receptor.
The purpose of the saturation binding assay is to characterize the specificity and
activity of the cytosol preparation and ensure that the ER activity is sufficient for the
competitive assay. The saturation assay measures the affinity of the receptor for its
natural ligand (17(3-estradiol, radiolabeled), quantified by the dissociation constant (Kd,
nM); and the concentration of active receptor sites, quantified by the maximum specific
binding number (Bmax, fmoles of estradiol/100 ug of protein). (See Section E for further
discussion of these quantities.) The saturation binding assay tests eight increasing
concentrations of labeled estradiol across two orders of magnitude, each in the
8
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
presence of a 100-fold higher concentration of unlabeled estradiol. Both total and
nonspecific binding are measured and specific [3H]-estradiol binding is calculated by
subtracting non-specific from total. Kd and Bmax are calculated through nonlinear
regression to a one-site binding model. (The Scatchard plot is not used for quantification
although it is recommended as a visual aid in evaluating the performance of the assay
since the linearity of the plot is a useful indicator of a well-performed run.)
The competitive binding assay measures the affinity of an unlabeled chemical
(i.e., reference "standard", weak positive control, or test chemical) in competition with
high affinity radioligand (tritiated 17|3-estradiol) for the estrogen receptor. It is quantified
by the concentration of competitor which inhibits 50% of the binding of the radioligand
(IC50) and frequently by relative binding affinity (RBA, % relative to estradiol). The
competitive assay measures the binding of [3H]-estradiol at a fixed concentration in the
presence of a wide range (eight orders of magnitude) of test chemical concentrations.
The data are then fit, where possible, to the Hill Equation (Hill 1910), which describes
the displacement of the radioligand by a one-site competitive binder. The extent of
displacement of the radiolabeled estradiol is used to characterize the test chemical as
interacting, not interacting, or generating an equivocal response. (See Section F.1 for
further discussion.)
C. Review of literature
The EPA asked the Interagency Coordinating Committee on the Validation of
Alternative Methods (ICCVAM) to prepare a review of the literature on ER binding and
ER transcriptional activation (as well as their androgen receptor counterparts). ICCVAM
identified 72 publications with an appropriate level of detail, containing data on 638
substances from 14 different ER binding assays over four species (rat, mouse, rabbit,
and human). Relatively few chemicals were tested more than once in the same assay
or in multiple assays, and no formal validation studies to assess the reliability or
performance of ER binding assays were found. The review found no published
guidelines for conducting in vitro ER binding studies.
The Background Review Document (BRD) that documents the review provided
procedural standards for in vitro ER binding assays and proposed that the rat uterine
cytosol (RUC) assay, which has been the most widely used method for identifying
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
substances with ER binding activity, be used as the standard against which new assays
for ER binding activity be evaluated. The BRD further recommended that "[b]ased on a
consideration of such factors as relative performance, elimination of animal use, the use
of the ER from the species of interest, and the use of alternatives to radioactive
substanced, the [human recombinant ER alpha (hrERa), human recombinant ER alpha
- fluorescence polarization, and human recombinant ER beta] assays should have the
highest priority for validation as screening assays for human health-related issues, while
the GST-rtERdef assay might be preferred when screening for substances that pose a
hazard to wildlife." EPA is currently chairing an OECD-sponsored validation effort for
the hrERa binding assay. If the hrERa assay is validated, EPA may consider using it to
replace the ER-RUC assay in the Tier 1 battery.
ICCVAM's BRD became available in 2002 (ICCVAM 2002). It serves as the
Detailed Review Paper (DRP) referred to in the description of the validation process
above (section I.D). The Executive Summary and the Conclusions of the BRD are
attached as Appendix 2.
D. Assay components other than test chemical
The following is a general description of the major components of the assay other
than the test chemical. Brief descriptions of conduct of the assay are given in Sections
E and F. The full protocol is attached as Appendix 1.
1. Solvent
The best solvent for the test chemical is chosen from among dimethylsulfoxide
(DMSO), ethanol, or water. The solvent used for a test chemical must also be used for
the reference chemical (inert 17(3-estradiol) and the control chemicals (norethynodrel
and octyltriethoxysilane) unless the solvent is water. If the test chemical is run in water,
the controls are run in ethanol since they are not soluble in water. The total volume of
ethanol allowed is no more than 3% of the total assay volume; DMSO, if used, may not
exceed 10%. (See Section III.H for further discussion of solvent maxima.)
2. Reference estrogen - 17(3-Estradiol
17(3-Estradiol is the native ligand that binds with high affinity to the estrogen
receptor of rat uterus. In the saturation assay, the final unlabeled estradiol
10
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
concentrations in the assay tubes are 3, 6, 8, 10, 30, 60, 100, and 300 nM. Unlabeled
estradiol concentration is 100x the tritiated estradiol concentration to bind all the high-
affinity ER binding sites so that the tritiated estradiol competes only at the low-affinity
sites, thus providing a measure of non-specific binding. For the competitive assay, the
seven concentrations for estradiol include serial dilutions at each log unit (with half-log
spacing around the log(IC50)) from 10~7 to 10~11 M. This establishes a standard
reference curve whose characteristics can be compared to performance criteria to
ensure that the binding assay is working correctly.
3. Marker/tracer-Radiolabeled 17(3-estradiol
Tritiated estradiol is used as the marker/tracer in the assay. Before preparing the
dilutions of the [3H]-17(3-estradiol, the specific activity is adjusted for decay over time
since certification by the manufacturer. In the saturation assay, the final [3H]-estradiol
concentrations in the assay tubes are 0.03, 0.06, 0.08, 0.1, 0.3, 0.6, 1, and 3 nM. For
competitive assays, the final [3H]-estradiol concentration is a constant 1 nM.
4. Positive control - Norethynodrel
Norethynodrel is used as the weak positive control in the competitive binding
assay. It was chosen because it was one of the weakest positive chemicals that reliably
produced a full binding curve through 10~4 M in preliminary studies. Norethynodrel final
concentrations are 8 dilutions that cover log units between 10~4 to 10~85 M, with half-log
units around the IC50.
5. Negative control - Octyltriethoxysilane
Octyltriethoxysilane is used as the negative control in the competitive binding
assay. It was chosen because it reliably showed no competition with the radiolabeled
estradiol, and little variability across the entire range of concentrations tested, in multiple
laboratories. The eight concentrations cover each log unit from 10~3 to 10~10 M inclusive.
6. Rat uterine cytosol
Uteri are collected from Sprague-Dawley rats (85 to 100 days of age)
ovariectomized seven to ten days prior to being humanely killed. Weighed and trimmed
uterine tissues are placed in ice-cold buffer prepared with Tris(hydroxymethyl)amino-
methane, Ethylenediaminetetraacetic acid, Dithiothreitol, and Glycerol (TEDG) with
11
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
phenylmethylsulfonyl fluoride (PMSF). The final extraction volume has a ratio of 0.1 g
of tissue per 1.0 ml buffer. The tissues are homogenized and the cytosol pooled,
aliquoted, and stored at -80 °C. The protein content for each batch of cytosol is
determined using a method compatible with buffers that contain DTT. Typical protein
values are 1 to 4 mg/ml. Care is taken to thaw only amounts needed and to discard
rather than refreeze unused portions.
E. Saturation binding assay
Estrogen receptor saturation binding experiments measure total and nonspecific
binding of increasing concentrations of [3H]-estradiol, at equilibrium. Three replicate
data points are collected at each concentration. Total binding is calculated by
converting the disintegrations per minute (dpm) from samples containing [3H]-estradiol
(no radioinert estradiol). Nonspecific binding is calculated by converting the dpm from
tubes containing [3H]-estradiol + 100-fold molar excess of radioinert estradiol, assuming
that the excess of radioinert estradiol will occupy all of the available estrogen receptor
binding sites. Specific binding is calculated as the difference between the nonspecific
binding and total binding at each of the tested concentrations. The saturation assay
conditions are:
Source of receptor
Concentration of radioligand (as serial dilutions)
Concentration of inert ligand (100 x [radioligand])
Concentration of receptor
Temperature
Incubation time
Assay buffer
Tris
EDTA
Dithiothreitol
Glycerol
Phenylmethylsulfonyl fluoride
Rat uterine cytosol
0.03- 3 nM
3 - 300 nM
50 ug protein/tube*
4°C
16-20 hours
10 mM (pH7.4)
1.5mM
1 mM
10%
1 mM
*Protein concentration may need to be adjusted to minimize ligand depletion while maintaining
an adequate radioactivity signal.
Reversible binding between 17(3-estradiol and the estrogen receptor can be
described using a single site binding model:
Receptor + Ligand
Receptor* Ligand
12
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
where kon and k0ff represent the rates of the binding and dissociation events respectively.
The dissociation constant (Kd) is defined as:
•on
At equilibrium, the ratio of free to bound estradiol is constant and the fraction bound can
be calculated:
[Ligand]
Fraction of receptor bound
[Ligand]+KC
Thus the fraction of receptor bound starts at 0 when ligand concentration is zero and
has an upper asymptote of 1 as ligand concentration grows to greatly exceed the
receptor concentration.
Ligand binding is conceptually represented as specific binding and nonspecific
binding. Specific binding is binding to the ligand binding domain of the receptor while
nonspecific binding is binding to sites other than the ligand binding domain, including
the reaction tube walls or other components of the reaction mixture. The greater the
concentration of free radioligand, the higher the nonspecific binding will be. Nonspecific
binding is adjusted for by subtracting the disintegrations per minute (dpms) of the
nonspecific binding tubes corresponding to each total ligand concentration, from the
total binding dpms for that concentration.
The degree of nonspecific binding is determined by including a parallel set of
tubes that contain the same concentrations of radioligand as the total-binding tubes,
plus a sufficiently large concentration of unlabeled substance that will bind with all the
receptor binding sites and leave no receptors remaining to bind with the radioligand.
Any bound dpms must then necessarily correspond to nonspecific binding of the
radioligand.
In the current assay, three parallel tubes are run for each total concentration of
radioligand and the averages of the radioactive decay (dpm) among the three total
binding tubes and among the three total added tubes are determined. The total free
radioligand is determined by subtracting the total bound dpm from the average total
13
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
added dpm. The specific bound decay is determined by subtracting the average
nonspecific binding decay from each of the total bound decay determinations.
The specific bound ligand concentration is related to the total free ligand
concentration by the nonlinear regression relation model:
y - ^max* | £
X + Kd
which is based on the law of mass action (i.e., the law that reaction rate is proportional
to reactant concentration for reactions with a single mechanistic step). In this relation X
is the free radioligand concentration and Y is the concentration of radioligand specific-
bound to the receptor. Bmax is the maximum concentration bound as the concentration
of free radioligand goes to oo. Kd is the equilibrium dissociation constant discussed
above; it corresponds to the radioligand concentration at which half the receptor binding
locations are filled, e represents the random variation about the model and is often
modeled as independently distributed with mean 0 and constant variance a2.
The model relies on the assumption that there is such an excess of free
radioligand available that its concentration does not change appreciably when some of it
binds to receptor. There appears to be no general agreement in the literature about the
extent of "ligand depletion" that can be tolerated, but a value of 10 to 20% is often
considered high. Reducing the amount of cytosolic protein that is used in the assay
reduces the amount of receptors in the mix and thus reduces ligand depletion (all else
being equal) but it also reduces the number of bound dpms available for measurement
and can therefore add variability to the measurement. EPA has chosen to aim at a level
of protein that binds 25% to 35% of the total radiolabeled estradiol that is added to the
tube, and to rely on the method developed by Swillens (1995) to compensate for ligand
depletion when calculating Kd and Bmax. Carter et al. (2007) provide evidence that the
Swillens correction "appears to be the most appropriate method for estimating ligand
affinity in situations of ligand depletion."
In general, when evaluating data from ER saturation assays, the following points
should be considered:
14
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
• As increasing concentrations of [3H]-estradiol were used, does the specific
binding curve reach a plateau? Maximum specific binding must be reached,
indicating saturation of ER with ligand.
• Does the data produce a linear Scatchard plot (a plot of bound/free ligand as a
function of specific binding)? Non-linear plots generally indicate a problem with
the assay such as ligand depletion or incorrect assessment of non-specific
binding.
• Is the Kd within an acceptable range? Literature values for Kd using rat uterine
cytosol preparations have varied from 0.05 to 0.5 nM. The variation in Kd may be
a reflection of different laboratories using radiolabeled estradiol with a wide range
of specific activity ([3H]-17(3-estradiol versus [125l]-17(3-estradiol). In addition,
lower Kd may be observed when assay conditions minimize ligand depletion, and
slightly different Kd values exist for ERa and ER(3. Rat uterine cytosol prepared
using this protocol will typically yield a Kd of 0.03 to 1.5 nM.
• Are runs consistent? That is, are the standard errors for the Kd or Bmax
excessive?.
• Is nonspecific binding excessive? In general, the value for nonspecific binding
should be less than 50% of the total binding at the highest concentration.
F. Competitive binding assay
The competitive binding assay measures the binding of a single concentration of
[3H]-estradiol in the presence of increasing concentrations of a test substance. If the
test substance interacts with the receptor, it inhibits the binding of increasing amounts of
radiolabeled estradiol. EPA requires three concurrent replicates per run at each
concentration, and three non-concurrent runs, to characterize the potential of a test
substance to interact with the estrogen receptor.
Control samples are included for each assay run. These include:
• Graded concentrations of unlabeled 17(3-estradiol. The behavior of unlabeled
estradiol in competing with labeled estradiol is well known and provides a
standard by which to measure performance of the assay. The highest
concentration of this series (100 nM) is 100 times the concentration of the
radiolabeled estradiol (1 nM) and serves as the measure of non-specific binding.
• Graded concentrations of a positive control (norethynodrel). The behavior of the
weak binder norethynodrel in competing with labeled estradiol is well known and
provides assurance that similarly weak binders can be detected.
• Graded concentrations of a negative control (octyltriethoxysilane). This chemical
does not interact with the estrogen receptor and provides assurance that a well-
performed run does not falsely classify negatives as positives.
15
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
• Solvent control. Binding of the radiolabeled estradiol in the absence of any
competitor is the baseline condition (100% binding) to which displacement by
competitor can be compared.
The competitive binding assay conditions are:
Source of receptor
Concentration of radioligand
Concentration of receptor
Concentration of test substance (as serial dilutions)
Temperature
Incubation time
Assay buffer
Tris
EDTA
Glycerol
DTT
Phenylmethylsulfonyl fluoride
Rat uterine cytosol
1.0 nM
50 ug protein/tube*
100pM-1 mM
4°C
16-20 hours
10 mM (pH 7.4)
1.5mM
10%
1 mM
1 mM
* Receptor concentration may need to be adjusted for each batch of cytosol.
** Range and spacing of test substance concentrations may need to be adjusted
depending on solubility and strength of interaction, if any.
Receptor concentration is adjusted to keep ligand depletion below 15% (although
it is recommended that the protein concentration not be reduced below 35 ug per assay
tube since this can result in the loss of centrifuge pellets during the separation of bound
estradiol from free estradiol). This adjustment is required for each batch of cytosol
prepared.
The test substance is initially tested at concentrations from 1 mM to 100 pM (i.e.,
10~3 to 10"10M inclusive), in ten-fold (i.e., log) increments. Ethanol, DMSO, or water
may be used as solvent. If the highest concentration cannot be prepared in any of
these solvents (e.g., because there is precipitate in the stock solution or forms upon
addition to 4° C assay buffer, and adding more solvent would cause the final solvent
concentration in the tube to be greater than the acceptable limit of 3% ethanol or 10%
DMSO), that concentration may be omitted. Evidence must be provided in the report
showing measures taken at each highest-concentration-attempted to obtain full
solubility, such as gentle heating or using a different solvent. If the test substance is
such a strong binder that a full curve is not obtained in the default range of
16
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
concentrations tested, additional dilutions are required so that the curve is adequately
characterized.
Tubes are loaded (on ice, to prevent degradation of the receptor) as follows:
Volume
(nL)
390
10
100
500
Constituent
Master mixture (TEDG + PMSF assay buffer + [JH]-
17(3-estradiol)
Unlabeled 17(3-estradiol, weak positive control,
negative control, or test substance
Uterine cytosol (at concentration determined to be
appropriate for that batch of cytosol)
Total volume in each assay tube
The tubes are vortexed and incubated at 4° C for 16 to 20 hours. Hydroxyapatite
slurry (60% in cold TEDG+PMSF buffer) is added to each assay tube and the mixture
vortexed at 5 minute intervals for 15 minutes (kept cold between vortexes), then
centrifuged at 4° C for 10 minutes at 1000 x g. After centrifugation, the supernatant is
decanted and the pellet containing the bound [3H]-17(3-estradiol is re-suspended in cold
buffer. The wash is repeated twice more in the same manner. The final pellet is
suspended in ethanol and allowed to come to room temperature, centrifuged at 1000 x
g for 10 minutes, and a measured aliquot of the supernatant is counted in a scintillation
counter for determination of dpms/vial.
1. One-site binding model
Although the nature of the interaction, if any, between a test substance and the
estrogen receptor is not usually known beforehand, data from this screening program
will be fit to a one-site competitive binding model for the sake of standardization. One-
site competitive inhibition of the native ligand estradiol is a mechanism by which many
Pharmaceuticals interact with the estrogen receptor in vivo, and a test substance that
displays a good fit to the one-site competitive model will likely be of interest. A
substance that interacts with the receptor but is not a true one-site competitive inhibitor
may not fit the model well but is still likely to demonstrate behavior that will allow
classification as interacting with the receptor when this model is used. Such
17
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
compounds would also likely be of interest for further exploration of the compound's
estrogenic properties.
If the radioligand and the inhibitor both bind reversibly to the same single binding
site on the receptor, then specific binding at equilibrium follows a four parameter relation
between percent bound (Y) and logarithm of inhibitor concentration (X). The
concentration response relation is described by a sigmoid curve (a variation of the
commonly used Hill equation):
1 + 10
The parameters in the equation represent the following quantities:
• B is the bottom plateau, i.e., the least expected percent bound.
• T is the top plateau, i.e., the greatest expected percent bound.
• (3 is the "Hill slope," i.e., the steepness with which the curve declines. Since
the curve declines with increasing X, (3 is necessarily negative.
• logioOCso) is the logarithm of the concentration at which the expected value of
Y = 50%. LogioOCso) always corresponds to the same percentile of the
concentration response and so can be directly compared between the test
compound and the standard.
• s is the random variation about the concentration response relation, with
mean 0 and variance a function of the expected value of Y (often modeled as
a constant, a2).
For an ideal response by a one-site competitive binder,
B = 0,
T=100,
and p = -1.
The competitive binding assay is functioning correctly if all of the criteria in Table
3 have been met. The criteria apply to each individual run. If a run does not meet all of
the performance criteria, the run must be repeated. Results for test chemicals in
18
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
disqualified runs are not used in classifying the ER interaction potential of those
chemicals.
Table 3. Performance criteria for competitive binding, reference and weak positive controls.
Parameter
Loge(Syx)
(i.e., Loge(Residual
Std.Dev)
Bottom plateau level
Top plateau level
Hill slope
Unit
-
% binding
% binding
log10(M)-n
Estradiol
Lower
limit
NA
-4
94
-1.1
Upper
limit
2.35
1
111
-0.7
Norethynodrel
Lower
limit
NA
-5
90
-1.1
Upper
limit
2.60
1
10
0.7
Octyltriethoxysilane
Lower
limit
NA
NA
NA
NA
Upper
limit
2.60
NA
NA
NA
These performance criteria reflect the fact that estradiol and norethynodrel are
one-site competitive binders for the estrogen receptor and thus should display behavior
consistent with one-site competitive binding in each run. Specifically, the curve fitted to
the data points should descend from 90 - 10% over approximately an 81-fold increase
in concentration (i.e., this portion of the curve will cover approximately 2 log units). A
binding curve for either of these two standards that drops dramatically (e.g., from 90 -
0%) over one order of magnitude should be questioned, as should one that is U-shaped
(i.e., percent bound is decreasing with increasing concentration of competitor but then
begins to increase again). In both cases, something has happened to the dynamics of
the binding assay and the reaction is no longer following the law of mass action. The
values shown for the performance criteria are based on data generated using this ER-
RUC assay and judged to be acceptable runs by EPA.
2. Statistical analysis
For each test run the one-site competitive binding model is fit to the data by
nonlinear regression analysis. The model fits result in parameter estimates and
associated standard errors as well as estimates of residual variability.
Nonlinear regression analysis can be carried out using PRISM 5 software
(Motulsky 2003, 2007) or general purpose statistical systems such as SAS (2003).
Prism, however, does not have a model for estimating log(ICso), and it must be entered
by the user. EPA is supplying a Prism template that includes the manually-entered
formula. It is important not to use log(ECso), which is the value Prism supplies, as if it
were log(ICso). The ECso (effective concentration, 50%) is the concentration at which
19
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
50% of the effect of the chemical is seen - that is, the concentration halfway between
the top plateau and bottom plateau. Log(EC50) values are generally not comparable
across chemicals.
III. Assay standardization and optimization
In 1998 the EDSTAC already considered the ER binding assay to be validated
for use in the Endocrine Disrupter Screening Program but recommended
standardization of the assay before implementation as a screening tool. (See Section
I.D above.) Therefore, EPA standardized and optimized several experimental
conditions including buffer composition, extraction method and radiolabel concentration.
Some of the parameters were standardized and optimized in response to
recommendations made in the ICCVAM Background Review Document.
A. Buffer composition and receptor concentration
Experiments were performed to determine the utility of adding 1 mM
phenylmethylsulfonyl fluoride (PMSF) and/or 10 mM sodium molybdate. PMSF is a
protease inhibitor and serves to protect the receptor from degradation by native
enzymes present in the cytosol preparation. Sodium molybdate is thought to have
protein stabilizing activity to prevent denaturing of the receptor at high (physiological)
temperatures. The effect at the lower temperature used in this protocol was unclear. Its
use had been suggested in the ICCVAM BRD but without explanation.
Also as part of this standardization and optimization, the effect of receptor
concentration was evaluated. The original design called for 100 mg of protein per assay
tube. Protein concentrations from 25 to 100 mg/tube were evaluated. A series of
saturation runs (Table 4) were performed.
Table 4. Saturation assays comparing buffer composition and receptor concentration
Buffer
Run
Protein
(ug/tube)
Kd
(nM)
Dmax
(fmole/100 ug)
TEDG Only
344 J
100
0.050
17.71
345 L
100
0.055
21.41
350-J
50
0.048
14.51
355-J
50
0.058
15.06
351-L
25
0.061
11.20
20
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Buffer
Run
Protein
(ug/tube)
Kd
(nM)
(fmole/100ug)
TEDG + sodium molybdate + PMSF
346-J
100
0.089
38.85
347-L
100
0.098
40.33
348-J
50
0.041
32.33
352-J
40
0.031
25.53
353-L
40
0.037
27.33
349-L
25
0.032
18.05
Based on the linearity of the Scatchard plots and ligand depletion (data not
shown), the optimal conditions were 50 |jg protein/tube.
The receptor concentrations of cytosol preparations are likely to vary with batch
and the 50 |jg value will not necessarily be optimal for all batches. These data
established a reasonable starting point around which laboratories should determine
their own optimum for each cytosol preparation.
Additionally, a series of competitive assays were run using the estradiol and
norethynodrel standards to compare the buffer systems. The ICsoS of the standard
curves, and the IC50s and RBAs of norethynodrel are presented in Table 5. The RBAs
obtained in the different buffers were observed to be in the same range. There did not
appear to be any greater variation in ICso values between buffer systems than that
observed within buffer systems. The dpm values of the 100% binding tubes in these
assays ranged from 1 to 4 percent of those of the hot tubes, indicating that ligand
depletion was not significant.
Table 5. Competitive assays comparing buffers
Buffer
Run
Estradiol IC50 (nM)
Norethynodrel ICso(nM)
RBA (percent)
Buffer
Run
Estradiol IC5o (nM)
Norethynodrel IC5o(nM)
RBA (percent)
TEDG Only
364 J
0.94
4362
0.022
367 J
1.47
5630
0.026
TEDG + sodium molybdate + PMSF
361 L
1.44
8650
0.017
362 J
1.02
5627
0.018
366 J
0.67
5371
0.012
21
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
The results of these experiments demonstrated that the combination of
molybdate and PMSF improved the assay performance. After a review of literature
regarding the effect of molybdate on the estrogen receptor (e.g., Mauck et al. (1982);
Murayama and Fukai (1985); Pettersson et al. (1985)), it was felt that since this protocol
was performed at low temperature the addition of molybdate to prevent thermal
degradation of the receptor was not necessary and could perhaps lead to unpredictable
effects. For this reason, and to simplify the number of steps needed in the protocol,
only the protease inhibitor PMSF was included in the final protocol.
B. Separation technique (HAP vs. DCC)
Separation of free and bound radioactivity was compared using dextran-coated
charcoal (DCC) and hydroxyapatite (HAP) using the TEDG buffer with PMSF. A series
of competitive assays were run using DCC and HAP for estradiol, norethynodrel and
bisphenol A (Figure 1). ICsoS are presented in Table 6. Ligand depletion was less than
10% in all assays. The RBAs obtained in the different separation systems were
observed to be in the same range. There did not appear to be any greater variation in
ICso values between the two separation systems than that observed within separation
systems. Given indications in the literature of potential problems such as dilution-
induced, dramatic, and variable loss of apparent estrogen receptor content with use of
DCC with cytosol-derived receptor (Pettersson et al. (1985)), EPA chose to use HAP as
the separation medium in the final protocol.
22
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
125-
£ 75=)
•o
I a
CO
iii
I 25-
0-
-12
Buffer= TEGD plus PMSF
Extraction: HAP (open symbols, solid line);
(solid symbols, broken line)
f\
*
\ :
\ f
\.
-10 -8 -6 -4
Competitor Concentration (log^ M)
8/05/03 (400-J)
8/11/03 (401-J)
8/14/03 (404-L)
8/28/03 (414-L)
-2
Figure 1. Competitive assay comparing separation media for estradiol (left), norethynodrel
(center) and bisphenol A (right).
Table 6. Comparison of separation systems (DCC vs. HAP) in competitive assays
Standard Curve Positive Control Bisphenol A
8/05/03 (400-J)
8/11 703 (401 -J)
8/1 4/03 (404-L)
8/28/03 (414-L)
log10ICEQ
-8.949
-9.046
-8.964
-9.051
C. Post-incubation temperature
In initial assays using DCC as the separation medium, it was observed that there
was drift in the values obtained for the 100% tubes placed at the beginning vs. the end
of a run. According to the DCC protocol, assay tubes are refrigerated (4°C) or held on
ice during the performance of the assay. The addition of the DCC terminates the
reaction. Following the DCC addition step the tubes are mixed well, incubated for 15
minutes at 4° C and centrifuged at 4000 RPM for 15 minutes at 4° C. 300 ul of
23
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
supernatant is transferred from each assay tube to counting vials. It is at this step,
when the tubes were not held on ice, that a drift in counts was observed. In an assay
with many tubes, there would be an opportunity for the samples to increase in
temperature.
To determine if temperature was the cause, the following experiment was
conducted. Two separate assays composed of a series of 100% tubes were prepared
in triplicate. On the first day, l[3H]-estradiol (10ul), 100 ul of cytosol in buffer (100 ug of
protein), 10 ul of absolute ethanol and 380 ul of buffer were added to each tube, mixed
and incubated for 18-20 hours at 4° C on a rotating mixer. The total volume was 500 ul.
On the second day, 300 ul of DCC suspension was added to tubes that were then
mixed and incubated for 15 minutes at 4° C and centrifuged. 300 ul of each
supernatant was removed from tubes at >5, 20, 40, 60, 140 and 160 minutes after
centrifugation. One set of tubes remained on ice and the other set of tubes were held at
room temperature. A decrease in supernatant counts (Figure 2) was observed over
time for the tubes held at room temperature when compared to tubes held on ice. At 1
hour the counts had dropped ~5% and by 2 hours it was at ~10 % when compared to
the "< 5 minutes" time point.
3000 -,
2800 -
2600 -
2000
Room Temp
Linear (Room Temp)
lice
Linear (Ice;
20
60
100
120
Time (P.iinutes)
Figure 2. Influence of post-incubation temperature on bound-dpm counts when DCC is used
Although this information was not used in the interlaboratory validation studies,
which were conducted using HAP rather than DCC, in retrospect it may be appropriate
to investigate whether post-incubation temperature may have been the cause of the drift
in solvent control tubes seen in the second interlaboratory validation study.
24
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
D. Assay Volume
A set of experiments compared the assay volume at 300 and 500 uL using
cytosol at 0.2, 0.1, and 0.05 mg protein per assay tube in saturation assays. The
results in both volumes were similar: the Bmax decreased about 50% with decreased
protein concentration, and the Kd also decreased with each reduction in amount of
protein used, in both volumes. The reduction of the Bmax is normal but the reduction in
the Kd probably reflects a better estimate of the Kd as the assay approached linearity.
At 300 uL the Hill coefficients improve from approximately 2.0 to 1.5 with decreasing
protein concentration as would be expected if the receptor concentration was initially too
high, but at 500 uL the Hill coefficient improved even further, to 1.2 at the lowest
concentration of protein. Saturation assay results using a different batch of protein at
0.065 mg per tube at assay volumes of 300 and 500 uL showed equivalent Bmax (0.02
and 0.022 pmole) and KdS (0.097 and 0.095 nM) at the different volumes, but
improvement of the Hill coefficient from 1.4 at 300 uL to 1.08 at 500 uL.
The competitive assay results across the two volumes are shown in Table 7.
The test chemical was 17(3-estradiol. ICso values for this chemical using rat uterine
cytosol as source of ER range from 1 nM to 8 nM in the BRD summary of the literature
(ICCVAM 2002) but protein concentrations and assay volumes were not reported in that
summary.
Table 7. Comparison of total assay volume, competitive binding assay
Total
Volume
(ML)
500
500
500
300
300
300
Cytosol
Protein
(mg/ml)
4.905
4.905
4.905
4.905
4.905
4.905
Protein
(mg/tube)
0.200
0.150
0.100
0.200
0.150
0.100
IC50
(M)
0.88E-9
0.87E-9
0.89E-9
0.97E-9
0.93E-9
0.93E-9
95% Confidence Interval IC50
(M)
8.3470e-010 to 9.2870e-010
7.69906-010 to 9.78706-010
7.95206-010 to 1.00506-009
9. 19406-010 to 1.01 306-009
8.56306-010 to 1.01206-009
8.32906-010 to 1.04206-009
Given the clearly superior results in the saturation binding assay and no
significant difference in results in the competitive binding assay, an assay volume of 500
uL was chosen as the most appropriate assay volume.
25
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
E. Cytosol Source
A comparison of cytosols from retired breeders or 80-90 day old female virgin
rats was conducted. A set of saturation binding assays was run with assay tube
volumes of 500 ul per tube and 50 ug protein per assay tube. In comparing values
obtained for Kd, Bmax or Hill coefficients, no significant differences were seen for these
cytosols (Table 8). EPA chose to specify use of younger animals based on other
considerations such as more-consistent general health but recognizes that retired
breeders may be acceptable sources of receptor.
Table 8. Age of animal source of receptor
Cytosol Source
Retired Breeder
Retired Breeder
Retired Breeder
Retired Breeder
Retired Breeder
80-90 d virgin
80-90 d virgin
80-90 d virgin
Protein
Con.
(mg/ml)
3.990
4.295
4.295
4.295
4.075
2.927
2.927
2.927
Protein
Con.
(mg/tube)
0.050
0.050
0.065
0.050
0.050
0.050
0.065
0.050
pmole/m
g protein
0.24
0.42
0.46
0.40
0.44
0.40
0.40
0.34
Dmax
(pmole)
0.012
0.021
0.030
0.020
0.022
0.020
0.026
0.017
Kd
(nM)
0.06
0.07
0.08
0.05
0.06
0.06
0.07
0.05
F. Cytosol shelf life
Originally the assay included an arbitrary 30-day storage life for cytosol. In an
effort to determine whether this could be extended, a series of competitive ER receptor
assays were conducted with the same cytosol for a period of time ranging from 17 days
to 127 days. The assays included standard curve, positive controls and a weak binder
(bisphenol A). In general, the activity of the cytosol was very stable and no apparent
loss of activity was observed over the measured time period (17-127 days). The
cytosol was stored at -80° C for this period and it was not thawed and refrozen. Based
on these data, a recommended storage limit of 90 days at -80°C was adopted.
26
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
150-
100-
o
CO
n
50-
0-
_ * 6/26/03 (379-J)
» 6/26/03 (380-L)
» 7/3/03 (382-J)
• 7/3/03 (3S3-L)
• 10/14/03 (432J)
-12
-10 -8 -6
Competitor Concentration (log-in M)
-4
Figure 3. Competitive assay comparing cytosol storage time for estradiol (left), norethynodrel
(center) and bisphenol A (right).
G. Concentration of radiolabeled estradiol
The radiolabeled estradiol concentration was tested at 0.5 and 1.0 nM to
determine if sensitivity of the competitive binding assay is improved at lower [3H]-
estradiol concentrations. The TEDG + PMSF buffer and the HAP separation system
were used for these studies. The maximum bound dpms in the 100% tubes ranged
from 6000 to 6500 for 0.5 nM [3H]-estradiol and 6700 to 7500 for 1.0 nM [3H]-estradiol.
The bound counts using 0.5 nM represent a ~ 12-15 % decrease from data obtained
using 1.0 nM. The observed RBAs for both the positive control and bisphenol A were
comparable across the two concentrations of radiolabeled estradiol. In both assay
conditions (0.5 and 1.0 nM [3H]-estradiol) ligand depletion did not exceed 10%. Given
that the RBAs were similar under the two conditions, the [3H]-estradiol was kept at 1.0
nM in order to maximize the number of counts available (and thus reduce variability).
27
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Standard Curve, Positive Control & Bisphenol A
Buffer= TEGD plus PMSF
Hot: 1.0 nM (open symbols, solid line);
0.5 nM (solid symbols, broken line
Protein: 100 ug per tube
100-
5?
c
O 50'
CO
0-
- 9/15/03 (420-J)
9/29/03 (426-LF)
10/02/03 (429-J)
° 10/09/03 (431-LF)
-12
-10 -8 -6
Competitor Concentration
-4
-2
M)
Standard Curve Positive Control Bisphenol A
9/1 5/03 (420-J)
9/29/03 (426-LF)
10/02/03(429-0)
10/09/03 (431-LF)
log13IC50
-9.255
-9.217
-8.950
-8.922
Figure 4. Competitive assay comparing radioligand concentration for estradiol (left),
norethynodrel (center) and bisphenol A (right).
H. Maximum solvent concentration
The effect of increasing concentrations of absolute ethanol and of DMSO on the
binding of 17(3-estradiol in the competitive binding assay was examined (Eldridge 2007,
attached as Appendix 3). The IC95 was approximately 5% ethanol and 20% DMSO. To
be conservative, EPA chose 3% ethanol and 10% DMSO to use as the maximum
solvent concentrations.
28
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
IV.Interlaboratory validation
Following assay optimization, the protocol was tested for transferability to
laboratories and for reliability across laboratories. Due to wider than expected
intralaboratory variability in the first interlaboratory study - which resulted in dropping
two of five laboratories as well as higher coefficients of variation than expected for
RBAs - the protocol was modified and a second interlaboratory validation study was
undertaken.
A. First interlaboratory study
1. Selection of laboratories
Four independent laboratories were recruited that had experience in reliable
performance of in vitro receptor binding assays using biological materials that they
obtained from appropriate biological tissues, although the experience did not have to be
specifically with estrogen receptor binding assays. The fifth laboratory was the
laboratory that had produced the optimization data described above.
2. Study design
Four steps were taken in this study to evaluate the intra- and inter-laboratory
variability of results among five independent laboratories:
a. Saturation and competitive assays using centrally supplied cytosol (reference
and weak positive chemical only)
b. Competitive assays using centrally supplied cytosol (9 test chemicals)
c. Saturation and competitive assays using cytosol prepared by individual
laboratory (reference and weak positive chemical only)
d. Saturation and competitive assays using cytosol prepared by individual
laboratory (5 test chemicals)
[3H]-estradiol and dilutions of 17(3-estradiol and norethynodrel were centrally
supplied. Cytosol was either centrally supplied (series "a" and "b") or prepared by the
individual laboratories (series "c" and "d"), to examine the laboratories' proficiency in
preparing cytosol and running the complete assay. The centrally supplied cytosol was
prepared by Lab E.
The results of this study are shown for historical purposes only. Competitive
binding data from the second interlaboratory validation study (Section IV.B.4) were
analyzed differently from the data from this first validation study, so the two validation
29
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
studies are not directly comparable. Specifically, in the first study, the tops and bottoms
of competitive binding curves were constrained to 100% and 0% respectively, while for
the second study the tops and bottoms were not constrained when fitting the data.
3. Results
a. Saturation binding
Saturation data for this study were fit to the one-site binding model as described
in Section II.E. The fit provided estimates of the dissociation constant (log(Kd)) and
maximum number of receptors (Bmax). The intra- and inter-laboratory means and
coefficients of variation were evaluated to ensure that each laboratory was using the rat
uterine cytosol preparations correctly and could reliably measure the Kd and Bmax. The
goodness of fit to the one site model was also calculated.
The laboratories first ran the saturation assay with cytosol provided by the lead
laboratory (series "a" and "b"). The protein concentration was 2.52 and 3.10 mg/mL for
series a and b cytosol preparations, respectively. The intra-laboratory variability is
reported in Table 9.
The mean goodness-of-fit to the one-site binding equation was 88% for the 27
runs. The range of Bmax (fmole/100 ng) was 12.30 to 52.5 with a mean value of 36.5 for
series a, and 20.9 to 56.7 with a mean of 40.1 for series b.. The range of log(Kd) values
was -10.37 to -9.36 with a mean value of -9.99. The inter-laboratory variability was 11 %
forKd.
The purpose of the series a and b saturation assays was to compare each lab's
ability to accurately and reproducibly conduct the saturation assay on a standard cytosol
preparation. Lab B was not able to successfully produce saturation assays consistent
with the other laboratories and did not participate beyond the saturation assays of series
a. The other laboratories were relatively consistent in these saturation assays.
For the series c and d saturation assays, the goal was to establish the variability
of results among the remaining independent laboratories when cytosol was prepared by
each laboratory. As noted, Lab B did not participate due to difficulties in the series a
saturation assay. Lab A also did not participate due to difficulties in the series a and b
competitive assays (discussed below). Lab E did not participate in series c studies as
its cytosol had been characterized previously in series a and b.
30
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
The protein concentration of the cytosol preparations is shown in
31
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 10. The protein concentration varied by a factor of approximately 2 between the
laboratories.
The mean goodness-of-fit to the one-site binding equation was 96% for the 15
runs. The range of Bmax (fmole/100 ng) was 23.89 to 55.6 with a mean value of 43.2 for
series c, and 28.3 to 58.0 with a mean of 39.0 for series d. The range of log(Kd) values
was -10.15 to -9.48 with a mean value of -9.85. The inter-laboratory variability was 11 %
forKd.
Table 9. Intra-laboratory variability of the saturation assay with centrally supplied cytosol
preparation (series a and b)
Statistic
log Kd (M)
Run
1a
2a
3a
1b
2b
3b
Lab A
-10.18
-10.14
-10.20
-10.18
-10.02
-9.97
LabB
-9.74
-10.37
-10.07
-
-
-
LabC
-10.14
-10.15
-9.36
-10.08
-10.10
-9.86
LabD
-10.06
-9.96
-10.21
-10.20
-10.18
-10.22
LabE
-9.90
-9.77
-9.85
-9.65
-9.63
-9.57
Mean log Kd (M)
CVKd
-10.11
17%
-10.06
11%
-9.95
8%
-10.14
17%
-9.73
13%
Mean Bmax
(fmole/100 |jg)
CV Bmax
a
a
35.1
13%
16.2
22%
46.0
17%
49.1
4%
35.9
4%
Mean Bmax
(fmole/100 |jg)
CV Bmax
b
b
25.1
23%
-
-
54.2
5%
48.5
12%
32.6
11%
Average
Goodness of Fit
83%
77%
92%
83%
98%
32
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 10. Intra-laboratory variability of the saturation assay using individual laboratory cytosol
preparation (series c and d)
Statistic
Protein Cone.
(mg/mL)
Assay
c
d
LabC
1.60
1.79
LabD
2.83
3.52
LabE
3.55
log Kd (M)
1c
2c
3c
1d
2d
3d
-9.97
-10.03
-9.99
-9.82
-9.83
-9.83
-9.77
-9.96
-9.93
-9.83
-10.00
-10.15
-
-
-
-9.48
-9.51
-9.65
Mean log Kd (M)
CVKd
-9.91
16%
-9.94
15%
-9.55
13%
Mean Bmax
(fmole/100 MS)
CV Bmax
c
c
54.5
3%
32.0
37%
-
-
Mean Bmax
(fmole/100 |jg)
CV Bmax
d
d
56.2
3%
30.7
6%
30.0
10%
Average
Goodness of Fit
96%
95%
99%
b. Competitive binding - standards
The competitive assays in the first validation study followed the same structure
as the saturation assays described in the previous section. The laboratories first ran the
competitive assay with cytosol provided by the lead laboratory for the reference
standards (series a and b). Each of the five participating laboratories conducted three
independent competitive binding runs using a standard and a weak positive control, with
three replicates at each concentration. The data for Laboratory B is not included in the
analysis due to that lab's inability to complete sufficient acceptable runs in the time
allotted in series a.
The estimated log(ICso) for the standard (estradiol) and weak positive
(norethynodrel) are shown in Table 11 for both series a and b. For series a, the range
of log(ICso) values for the standard was -9.12 to -8.80 with a median value of -8.90. The
range of log(ICso) values for the weak positive control was -6.78 to -6.39 with a median
value of -6.59. The resulting RBAs ranged from 0.29% to 0.66% with a median value of
33
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
0.46%. The intra-laboratory CVs for RBA ranged from 15% to 36% with a median of
20%.
The inter-laboratory variability of the three competitive binding measurements
was 0.4%, 1.6%, and 22.7% for the standard and weak positive log(ICso) values and
RBA, respectively. The variability in these measurements was fairly small as can be
inferred from the small variability in the fitted one-site competitive curves (Figure 5).
For series b, the range of log(ICso) values for the standard (estradiol) was -9.25
to -8.79 with a median value of -8.96. The range of log(ICso) values for the weak
positive control was -9.49 to -6.25 with a median value of -6.45. Lab A, Run 2, with the
weak positive had the poorest fit to the one-site competitive curve and produced the
smallest log(ICso) value (Figure 6). The percentage bound for this run dropped to less
than 20% at a log M concentration of -9 and continued to drop for two more
concentrations. The percentage bound then increased at log M concentration of -6 and
decrease in a pattern consistent with the two other runs. The weak positive results for
this run were removed from the statistical analysis. Lab C, Run 2, weak positive results
produced the second smallest log(ICso) value. The standard and weak positive results
for Lab D and Lab E were similar to each other. The resulting RBAs for the weak
positive ranged from 0.21% to 0.57% with a median value of 0.27%.
Table 11. Intra-laboratory variability of the competitive binding assay with centrally supplied
cytosol
Series "a"
Iog(ic50)
Standard
log(IC50)
Weak Positive
RBA
Mean
CV
Mean
CV
Mean
CV
Lab A
-8.94
2%
-6.57
2%
0.38%
9%
LabC
-8.98
1%
-6.47
4%
0.33%
38%
LabD
-9.17
1%
-6.57
1%
0.25%
14%
LabE
-8.89
1%
-6.29
1%
0.25%
11%
Series "b"
log(IC50)
Standard
log(IC50)
Weak Positive
RBA
Mean
CV
Mean
CV
Mean
CV
Lab A
-8.99
2%
-6.67
2%
0.51%
36%
LabC
-8.93
1%
-6.53
1%
0.40%
15%
LabD
-8.92
1%
-6.45
1%
0.35%
25%
LabE
-8.91
1%
-6.67
0%
0.58%
16%
34
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Standard
Curve
125-
Weak
Positive
-10 -9 -8 -f -6 -6
Competitor Concentration (M)
4-A-5/13/04 «
5-A-5/24/04 *
6-A-5/26/04 °
5305-C-4/28/04 °
5306-C-5/4/05 *
5307-C-5/6/04 "
1-D-4/19/04 •
2-D-4/19/04 *
3-0^/29/04 "
442-E-4/6/04 *
443-E-4/7/04 *
444-E-4/8/04 °
Figure 5. Inter-laboratory variability of standard and weak positive and associated 95%
confidence bands (light gray) for Labs A, C, D and E (series a).
Lab A
Com petit orConcentrati on (M)
_ IS'
i'
£!
± 7'
LabD
i--1;-' := : =
D-1T-1.O&O5
D-19-1,'10,05 "
-11 -10 -9 -3 -7 -6 -5 -i
Competitor Concentration (Mf
-11 .-to -9 -5 -7 -6 -6 -t
Com petit or Concentration (M)
Figure 6. Intra-laboratory variability of the standard and weak positive chemicals in the
competitive binding assay (series b).
35
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
As can be seen from Figure 6 the data from Lab A on estradiol and norethynodrel
display somewhat more intralaboratory variability than the data from other laboratories,
but it was the variability of the data on the test chemicals using centrally-supplied
cytosol, discussed in section IV.A.S.c below, that led Lab A to be dropped from the
remainder of the study.
For series c and d, Labs C and D conducted the competitive binding assay with
the individual-laboratory-prepared cytosol. One run from Lab D was removed from
statistical analysis because it did not meet certain protocol criteria. No other data were
removed from statistical analysis.
The results for series c and d standard and weak positive are shown in Table 12
and Figure 7 and Figure 8. For Lab C, the mean log(ICso) values were similar for both
series for the standard and weak positive, with correspondingly similar RBA values. For
Lab D, the mean log(ICso) values differed between series but the RBAs were not
extraordinarily dissimilar from the RBAs produced by the other laboratories.
Table 12. Competitive assay results for the individual-laboratory prepared cytosol, standard and
weak positive chemicals
Iog(ic50)
Standard
log(IC50)
Weak Positive
RBA
Mean
CV
Mean
CV
Mean
CV
Series "c"
LabC
-9.08
0.50%
-6.62
0.10%
0.35%
9.50%
LabD
-9.02
0.80%
-7.06
1 .90%
1.16%
38%
Series "d"
LabC
-8.95
0.30%
-6.81
3.40%
0.79%
44%
LabD
-9.68
1.10%
-6.77
1 .50%
0.12%
10%
LabE
-8.89
0.30%
-6.44
0.40%
0.36%
4.20%
36
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
125n
100-
3
o
m
CM
LJJ
-10 -9 -8 -7 -6 -5
Competitor Concentration (M)
-4
Standard
Curve
C-1-5/10/05
C-2-5/16/05
C-3-5/18/05
D-5-5/23/05
D-6-5/24/05
D-7-5/31/05
-3
Weak
Positive
C-1-5/10/05
C-2-5/16/05
C-3-5/18/05
D-5-5/23/05
• D-6-5/24/05
D-7-5/31/05
Figure 7. Competitive binding assay for individual-laboratory-prepared cytosol for the standard
and weak positive chemicals in Labs C and D (series c)
37
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
standard curve
* E-495-8/31/05
• E -496-9/1 /OS
• E-497-9/7/05
* D5-8/24/05
• D6-8/29/05
• D7-8/31/05
• D8-9/6/05
* D9-9/7/05
* C4-9/22/05
• C5-9/26/05
• C6-9/28/05
weak positive
*
0
7
0
D
0
A
7
E-495-8/31/05
E-496-9/1/05
E-497-9/7/05
D5-8/24/05
D6-8/29/05
D7-8/31/05
D8-9/6/05
D9-9/7/05
C4-9/22/05
C5-9/26/05
C6-9/28/05
-3
Competitor Concentration (M)
Figure 8. Competitive binding assay for individual-laboratory-prepared cytosol for the standard
and weak positive chemicals in Labs C, D and E (Series d)
c. Competitive binding - test chemicals
In addition to the standard chemicals, the laboratories conducted competitive
assays with centrally supplied cytosol (series b - 9 test chemicals) and laboratory
prepared cytosol (series d - 5 chemicals).
The goodness-of-fit to the one-site competitive equation for the nine test
chemicals reflected the characteristics of the chemical being tested. Thus, the lack of
convergence did not cause any data to be removed from the analysis. Most of the R2
values were greater than 80%. However, the progesterone results fit the curve poorly
or did not converge (as would be expected from a substance that does not interact with
the estrogen receptor). The mean log(IC5o), R2, and RBA and intra-laboratory CVs for
the standard, weak positive, and test chemicals are presented in Table 13. The intra-
laboratory CVs for the weak positive RBA ranged from 10% to 38% with a median of
38
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
12%. The large CV of 38% was directly related to Lab C, Run 2. There was no obvious
relationship between the test chemical run and the resulting intra-laboratory CV for the
RBA (Figure 9). Each laboratory produced the smallest intra-laboratory CV for a
different test chemical.
As can be seen in Table 13, Lab A had higher intralaboratory variability as
measured by coefficients of variation, for both log(ICso)s and RBAs for most chemicals.
While CVs were higher than expected for other labs for many of the chemicals, the
difficulties that Lab A experienced were considered sufficient to disqualify it from further
participation in the study.
The inter-laboratory CV for the standard and weak positive log(ICso) and RBA in
series b was 1.4% 2.0% and 20%, respectively (Table 14). Estrone had the largest
inter-laboratory mean RBA (7.59%) of the test chemicals. The inter-laboratory CVs for
the test chemical RBAs were similar and averaged 60%.
39
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 13. Test chemical intra-laboratory variability using centrally-supplied cytosol (series b)
Mean
Lab A
LabC
Lab D
Lab E
CV
Lab A
LabC
LabD
LabE
Standard
log(ICso)
-8.94
-8.98
-9.17
-8.89
log(ICso)
2%
1%
1%
1%
R2
0.97
0.99
0.99
0.99
Weak Positive
log(ICso)
-6.57
-6.47
-6.57
-6.29
log(ICso)
2%
4%
1%
1%
R2
0.96
0.98
0.99
0.99
RBA
0.38%
0.33%
0.25%
0.25%
RBA
9.49%
38.1%
14.0%
10.6%
Bisphenol B
log(IC50)
-6.07
-6.16
-6.72
-6.09
log(IC50)
5%
1%
1%
1%
R2
0.94
0.98
0.97
0.99
RBA
0.18%
0.14%
0.41 %
0.16%
RBA
91 .0%
49.4%
27.0%
3.41 %
4-Cumylphenol
log(ICso)
-4.85
-5.16
-5.44
-4.64
log(ICso)
2%
5%
3%
1%
R2
0.90
0.97
0.97
0.95
RBA
0.01%
0.01%
0.02%
0.01%
RBA
45.3%
34.2%
44.4%
12.4%
Mean
Lab A
LabC
Lab D
Lab E
CV
Lab A
LabC
LabD
LabE
Estrone
log(ICso)
-7.70
-7.53
-8.30
-7.85
log(ICso)
0%
1%
1%
1%
R2
0.93
0.99
0.98
1.00
RBA
5.38%
4.17%
1 1 .6%
9.16%
RBA
38.9%
25.4%
25.4%
12.5%
Coumestrol
log(ICso)
-7.50
-7.68
-8.22
-7.34
log(ICso)
4%
1%
1%
2%
R2
0.92
0.99
0.98
1.00
RBA
4.0%
5.8%
1 1 .9%
2.9%
RBA
64%
19%
34%
30%
Progesterone
log(IC50)
R2
No Convergence
No Convergence
No Convergence
-2.52
log(IC50)
0.02
No Convergence
No Convergence
No Convergence
12%
RBA
0.00%
RBA
58.3%
Daidzein
log(ICso)
-4.96
-5.07
-5.64
-4.93
log(ICso)
4%
1%
3%
2%
R2
0.80
0.84
0.90
0.78
RBA
0.01%
0.01%
0.03%
0.01%
RBA
57%
19%
32%
12%
Mean
Lab A
LabC
Lab D
LabE
CV
Lab A
LabC
LabD
Lab E
Tamoxifen citrate
log(ICso)
-5.71
-6.52
-7.32
-6.50
log(ICso)
1%
1%
1%
2%
R2
0.92
0.90
0.99
0.97
RBA
0.05%
0.40%
1 .36%
0.43%
RBA
17%
7%
23%
44%
4-t-Octylphenol
log(ICso)
-5.07
-5.29
-5.72
-5.19
log(ICso)
3%
1%
1%
0%
R2
0.87
0.98
0.97
0.97
RBA
0.01 %
0.02%
0.03%
0.02%
RBA
47%
8%
14%
14%
Bisphenol A
log(IC50)
-4.87
-5.06
-5.57
-4.84
log(IC50)
3%
1%
1%
0%
R2
0.95
0.98
0.97
0.97
RBA
0.01 %
0.01 %
0.03%
0.01 %
RBA
41%
31%
13%
12%
= Poor fit to the competitive binding curve
40
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 14. Test chemical inter-laboratory variability (series b and d)
Test Chemical
Standard
Weak Positive
Bisphenol B
4-Cumylphenol
Estrone
Coumestrol
Progesterone
Daidzein
Tamoxifen citrate
4-t-Octylphenol
Bisphenol A
4-t-Butylphenol
Statistic
log(IC50)
log(IC50)
RBA
log(IC50)
RBA
log(IC50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Iog(ic50)
RBA
Centrally supplied
Cytosol (series b)
Mean
-8.99
-6.47
0.30%
-6.26
0.22%
-5.02
0.01%
-7.85
7.59%
-7.69
6.15%
-2.52
0.00%
-5.15
0.02%
-6.51
0.56%
-5.32
0.02%
-5.09
0.01%
CV
1 .40%
2.00%
20%
5.00%
57%
7.00%
56%
4.20%
45.20%
5.00%
65%
NA
NA
6.40%
52%
10.10%
100%
5.30%
37%
6.60%
66%
Individual-lab-
prepared Cytosol
(series d)
Mean
-9.17
-6.68
0.42%
CV
4.78%
3.03%
79%
-8.02
8.92%
6.93%
48%
No Convergence
No Convergence
-7.54
6.69%
10.90%
149%
-5.42
0.02%
-3.83
0.00%
12.80%
69%
17.60%
84%
= Poor fit or lack of convergence
41
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
QOA
7% -
6% -
5% -
<
ffl 4%
o:
3%
2%
1 %
«<*
x
i — i
—
a
Oc
/
^
^
/
n ,— ,
<* ..d- ef VV- -o" oP
''// "///
V V
120% n
100%
80%
o
0%
•A®
<&
0°
Figure 9. Inter-laboratory mean (top) and CV (bottom) for the weak positive and nine test chemical
RBAs in order of greatest to least binding (series b)
For series d, the goodness-of-fit to the one-site competitive equation for the five
test chemicals reflected the characteristics of the chemical being tested. Thus, the lack
of convergence did not cause any data to be removed from the analysis. Most of the R2
values were greater than 70%. However, the progesterone results fit the curve poorly
or did not converge. The mean log(ICso), R2, and RBA and intra-laboratory CVs for the
standard, weak positive, and test chemicals are presented in Table 15. The intra-
laboratory CVs for the weak positive RBA ranged from 4% to 44% with a median of
9.6%. The large CV of 44% was directly related to Lab C, Run 6. There was no
42
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
obvious relationship between the test chemical and the intra-laboratory CV for the RBA
(Figure 10). Lab E had intra-laboratory CVs consistently less than 11%, and Lab C
averaged about 45%.
The inter-laboratory CV for the standard and weak positive log(ICso) and RBA in
series d was 4.78% 3.03% and 79%, respectively (Table 14). The greatest difference in
RBA for the weak positive was between Lab C and Lab D with Lab E results between
the two. Estrone had the largest inter-laboratory mean RBA (8.92%) of the test
chemicals. The inter-laboratory CV for the test chemical RBAs was the least for estrone
(48%) and the greatest for tamoxifen citrate (149%).
Table 15. Test chemical intra-laboratory variability using individual-laboratory-prepared cytosol
(series d)
Mean
LabC
LabD
Lab E
CV
LabC
Lab D
Lab E
Standard
log(IC50)
-8.95
-9.68
-8.89
log(ICso)
0.27%
1 .09%
0.25%
R2
98%
93%
99%
Weak Positive
log(IC50)
-6.81
-6.77
-6.44
log(ICso)
3.41 %
1 .45%
0.38%
R2
92%
99%
100%
RBA
0.79%
0.12%
0.36%
RBA
43.6%
9.62%
4.15%
4-t-Butylphenol
log(IC50)
-4.01
-4.39
-3.08
log(ICso)
5.76%
3.16%
0.86%
R2
97%
91%
84%
RBA
0.00%
0.00%
0.00%
RBA
43.6%
8.37%
7.67%
Tamoxifen citrate
log(IC50)
-6.61
-7.88
-8.15
log(ICso)
3.09%
0.69%
0.50%
R2
97%
92%
100%
RBA
0%
1%
18%
RBA
54.3%
3.01 %
10.6%
Mean
LabC
LabD
LabE
CV
LabC
LabD
LabE
Bisphenol A
log(ICso)
-5.23
-6.19
-4.84
log(ICso)
2.37%
1 .59%
0.76%
R2
0.82
0.85
0.99
RBA
0.02%
0.04%
0.01 %
RBA
27.0%
44.1%
8.15%
Estrone
log(ICso)
-7.56
-8.64
-7.85
log(ICso)
3.1%
4.5%
0.3%
R2
0.89
0.94
1.00
RBA
4.49%
13.1%
9.22%
RBA
56.2%
84.7%
7.2%
Progesterone
log(ICso)
R2
RBA
No Convergence
No Convergence
No Convergence
log(ICso)
RBA
No Convergence
No Convergence
No Convergence
= Poor fit or lack of convergence to the competitive binding curve
43
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
90%
80%
Figure 10. Intra-laboratory CV for five test chemicals and the weak positive RBAs (series d)
1 0 00% -|
rf 6 00% -
5 00% -
re
S 4 00% -
3 00% -
2 00% -
1 00% -
i — i
Estrone Tamoxifen Weak Bisphenol A 4-t- Progesterone
citrate Positive Butylphenol
Estrone Tamoxifen Weak Bisphenol A 4-t- Progesterone
citrate Positive Butylphenol
Figure 11. Inter-laboratory mean (top) and CV (bottom) for the weak positive and five test
chemical RBAs in order of greatest to least binding (series d)
44
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
The intralaboratory variability was higher than expected in most laboratories.
Laboratory E produced the most consistent and acceptable results, but this lab had
worked closely with EPA on previous ER binding assay studies (most notably the
optimization studies) and thus could have benefited from "coaching" that was not written
in the protocol. It was expected that most laboratories should be able to produce the
low variability exemplified by Laboratory E if the protocol were written in sufficient detail.
Thus the protocol was significantly expanded with details and examples, and a second
interlaboratory study undertaken.
B. Second interlaboratory study
For this study, significantly more detail was provided in the protocol. For
example, specific dilution "recipes" were provided and analysis templates were
improved. Attention was drawn to particular steps that need particular care, such as
use of multi-tube decanting racks during separation of bound from free tracer in order to
minimize the time that receptor is exposed to room temperature. Also, the procedure
for making dilutions of test chemical was changed. Dilutions in the first study were
made in small quantities of solvent first, and a prescribed amount of each dilution was
added to the assay tube to obtain the final concentrations. Since all the dilutions were
made separately this might have accounted for a significant amount of variability in the
results in the first interlaboratory study. Therefore in the second study one stock
solution was made in solvent and this solution diluted sequentially with buffer.
The study was conducted using fewer laboratories (three rather than five), but
more chemicals (23 rather than 9). Each laboratory prepared its own cytosol; there was
no centrally-prepared cytosol in this study.
An attempt was made to establish a base level of competence in performing the
assay before allowing full participation in the study. Laboratories were to qualify by
meeting performance criteria that had been established from acceptable runs generated
in the first interlaboratory study.
1. Selection of laboratories
Laboratories were required to be competent in laboratory methods relevant to in
vitro receptor binding assays. They were also required to be mutually independent, and
45
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
unconnected with the first interlaboratory validation study so that transferability of the
new protocol could be assessed without influence from prior experience with the assay.
Also in order to evaluate the transferability of the protocol, EPA requested that
"coaching" of laboratories be minimized and that all questions about the protocol be
referred to the EPA rather than discussed between the labs. EPA refrained from visiting
labs to observe technique and offer suggestions for improvement for most of the study,
but towards the end of the study such a trip was made to two labs to observe technique
during the critical stages of separation of bound from free radioligand, in order to try to
pinpoint the cause of the variability that had been seen in the data. No obvious cause
was found, and no significant changes were recommended.
As noted above, laboratories were to demonstrate proficiency in performing this
assay by meeting performance criteria during a qualification step, before generating
data in the main study. The performance criteria are shown in Table 16. They were set
separately for estradiol and norethynodrel, based on the performance of Labs C, D, and
E of the first interlaboratory validation study. They were meant to include 80% of the
values from these labs, with 95% confidence. Details of the method used are provided
in Appendix 4. Table 17 through Table 20 describe the results of this qualification step.
Table 16. Performance criteria for second interlaboratory study
Parameter
^ I-' within-run
(within-run variation)
Bottom plateau level
Top plateau level
(Hill) Slope
Unit
% binding
% binding
% binding
log10(M)-1
Estradiol
Lower
limit
NA
-5.0
90.0
-1.1
Upper
limit
5.0
1.0
110.0
-0.7
Norethynodrel
Lower
limit
NA
-5.0
90.0
-1.1
Upper
limit
5.7
1.0
110.0
-0.7
R1881
Lower
limit
NA
NA
NA
NA
Upper
limit
10
NA
NA
NA
46
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 17. Qualification runs: Saturation binding assays
Number of runs
Bmax(fmole/100ug protein)
Protein/assay tube (ug)
LabX
4
0.1 137 [nM]
0.461 [nM]
(CV2.89%)
(CV 3.6%),
89.97
(CV 8.3%)
50,46
LabY
5
0.13 ± 0.006 nM
(mean of 5 runs)
25.85 ± 0.56
(mean)
10 or 25
LabZ
5
0.6588 nM*
[CV 29.5%]
(last assay)
"Lab reported "6.69
E+08", unitless. EPA
retrieved number above
from data file.
4918dpm**
[CV21.5%]
(last assay)
**Lab reported "5.487
E+15", unitless. EPA
retrieved number above
from data file.
37, 37, 40, 50
Table 18. Qualification runs: Competitive binding assays
Number of runs
Ki
Log(IC50) for 17p-estradiol
(M)
Log(IC50) for norethynodrel
(M)
RBA
(norethynodrel compared
with estradiol)
Protein/assay tube (ug)
LabX
7
0.2865
0.3289
0.3565
0.4037
-9.042
-8.982
-8.947
-8.893
-6.083
-6.042
-5.971
-5.859
1.099E-03
1.148E-03
1.06 E-03
9.247 E-04
50, 35, 35, 46, 46, 46,
46
LabY
5
-9.04 ± 0.04
-6.01 ±0.05
9.20E-04 ± 6.63E-05
25
LabZ
4
-9.7
-8.9
-9.09
-8.92
-8.90
-6.11
-5.98
-5.89
1.718E-10
1.216E-09
6.339 E-04
1.271 E-09
First assay 40 ug,
next 3 at 50 ug
47
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 19. Qualification runs: Summary of estradiol competitive binding data
LabX
Mean
SE
LabY
Mean
SE
LabZ
Mean
SE
Bottom
-1.1
-0.7
-1.7
-1.4
-1.2
0.5
Bottom
0.1
0.3
0.3
1.1
-0.1
0.3
0.2
Bottom
-1.6
-0.3
0.0
-0.6
0.5
Top
103
100
104
103
103
1
Top
88
101
103
119
112
105
5
Top
110
110
117
112
2
Log(IC50)
-8.89
-8.95
-8.98
-9.04
-8.97
Log(IC50)
-9.16
-9.05
-8.94
-9.01
-9.05
-9.04
Log(IC50)
-8.92
-9.09
-8.90
-8.97
IC50 (nM)
1.29
1.12
1.05
0.91
1.09
0.08
IC50 (nM)
0.69
0.89
1.15
0.98
0.89
0.92
0.07
IC50 (nM)
1.20
0.81
1.26
1.09
0.14
Hill Coef
-0.99
-0.97
-0.91
-0.89
-0.94
0.02
Hill Coef
-1.10
-1.40
-0.98
-1.10
-0.90
-1.10
0.08
Hill Coef
-0.92
-1.07
-0.99
-0.99
0.04
48
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 20. Qualification runs: Summary of norethynodrel competitive binding data
LabX
Mean
SE
LabY
Mean
SE
LabZ
Mean
SE
Bottom
2.8
2.8
-2.3
-0.2
0.8
1.3
Bottom
-5.9
-5.2
-3.9
-1.2
-6.4
-4.5
0.9
Bottom
3.5
13
6.5
7.7
2.8
Top
102
99
100
94
99
2
Top
100
111
116
113
104
109
3
Top
110
110
105
108
2
Log(IC50)
-5.86
-5.97
-6.04
-6.08
-5.99
Log(IC50)
-6.16
-6.11
-5.88
-5.98
-5.93
-6.01
Log(IC50)
-5.98
-5.89
-5.88
-5.92
0.03
IC50 (nM)
1380
1072
912
832
1049
121
IC50 (nM)
687
785
1309
1057
1180
1004
117
IC50 (nM)
1047
1288
1318
1218
86
Hill Coef
-0.84
-1.01
-0.91
-1.04
-0.95
0.05
Hill Coef
-1.08
-0.74
-0.66
-1.10
-1.06
-0.93
0.09
Hill Coef
-0.91
-0.92
-1.2
-1.01
0.10
The highlighted cells show values which did not meet the performance criteria.
Lab X met the performance criteria; the other two labs did not. Note that it was primarily
the top plateaus that were exceeded.
While these results were disappointing, the study continued with these
laboratories. Given the experience in finding these laboratories, finding other
laboratories would have delayed the study - and the Screening Program - significantly,
and the deviations were judged to be marginally acceptable in this context.
2. Selection of test chemicals
Chemicals were selected to cover a wide range of binding strengths, indicated by
their Relative Binding Affinities (RBAs) for the estrogen receptor as reported in the
literature. Strengths were assigned as follows:
49
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
• Very strong: RBA > 100 (where RBA of 17(3-estradiol=100)
• Strong: RBA between 100 and 1
• Moderate: RBA between 1 and 0.1
• Weak: RBA < 0.1
• Negative: no RBA achieved
These assignments are only approximate as they are based on the median value
of reported RBAs, some of which span two or more orders of magnitude and others of
which are based on a single value. See Table 21. Values from human recombinant ER
binding assays reported in the literature are also reported in Table 21 for comparison.
RBAs from hrERa studies, where available, also vary by an order of magnitude.
The chemicals selected for this study had been agreed upon by an international
group of experts that is managing the validation of the human recombinant estrogen
receptor (hrER) binding assay under the auspices of the Organisation for Economic
Cooperation and Development The EPA used this chemical list so that results could be
compared between the ER-RUC study and the parallael hrERa study. A draft list of
proposed chemicals was submitted by the OECD group to an independent Chemical
Advisory Board (CAB) of three international experts. The CAB generally concurred with
the strategy of getting a wide range of strengths and chemical structures but
recommended removing estrone, a strong binder, from the proposed list and adding
enterolactone and benz(a)anthracene (weak binders), and atrazine (non-binder but
estrogen-active). The CAB noted that this places more emphasis on weak binders,
increases structural diversity, and provides the possibility of differentiating receptor-
based from non-receptor-based modes of estrogenic action. The CAB's
recommendations were adopted, and the final list of chemicals is as recommended by
the CAB with the exception of the negative control chemical. The ER-RUC study
planned to use R1881 as the negative control, while the hrER study had decided upon
dibutylphthalate (DBP) as its negative control. DBP was regarded as inappropriate for
use in the ER-RUC study because of prior indications that it did not give a clear
negative signal in the RUC assay (Zacharewski et al. 1998).
The test chemicals were coded before shipment to the laboratories so as not to
reveal their identity, although molecular weight was disclosed so that molar
concentrations could be prepared. 17(3-Estradiol (the reference standard),
50
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
norethynodrel (the weak positive control), and R1881 (the negative control) were
included among the blinded test chemicals.
51
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 21. Chemicals selected for the second interlaboratory validation study
Code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Chemical
17b-estradiol
17a-ethynylestradiol
DES
meso-hexestrol
genistein
norethynodrel
butyl paraben
4-nonylphenol
o,p'-DDT
corticosterone
equol
zearalenone
tamoxifen
5a-dihydrotestosterone
Bisphenol A
4-heptylphenol
kepone (chlordecone)
benz(a)anthracene
enterolactone
progesterone
octyltriethoxysilane
atrazine
R1881
Binding
affinity
Strong
Very strong
Very strong
Very strong
Moderate
Moderate
Weak
Weak
Weak
Negative
Moderate
Strong
Strong
Weak
Weak
Weak
Weak
Weak
Weak(t)
Negative
Negative
Negative
Negative
RUC* Historical RBAs (ICCVAM)
median RBA (range of values)
100 (reference estrogen)
148(100-867)
124(0.003-5000)
234 (58 - 302)
0.56 (0.45 - 0.67)
0.22 (0.2 - 0.23)
0.002 (0.0009 - 0.002)
p-nonylphenol 0.033 (0.0025 - 0.5)
0.013(0.001 -0.09)
negative (one study)
0.15 (tested once)
44.07 (tested once)
3.1 (0.13-6)
0.0135(0.001 -0.26)
0.056(0.008-0.1793)
no data
0.03(0.0035-0.2)
no data
no data
0.0003 (1 positive/8 tests)
no data
1/2 studies negative
no data
hrERa historical RBAs
median RBA
(range of values)
100 (reference chemical)
no data
236 (66.7 - 468)
no data
2.36 (0.7 - 5)
0.7 (tested once)
no data
0.026 (0.05, 0.001)
0.055(0.01 -0.1)
negative (one study)
0.33 (tested once)
8.5(7-10)
4 (2.94 - 7)
0.05 (tested once)
0.01 (0.003-0.05)
no data
0.06 (tested once)
no data
no data
no data
no data
negative (one study)
no data
Description
physiological estrogen
synthetic estrogen
synthetic estrogen
phenol (bisphenol)
flavonoid, phytoestrogen
steroid, nonphenolic
paraben
alkylphenol, intermediate cmpd
organochlorine
steroid
phytoestrogenic metabolite
resorcylic acid lactone, mycotoxin
antiestrogen
steroid, nonphenolic
phenol
alkylphenol
organochlorine
aromatic hydrocarbon
phytoestrogen, lignan
steroid
silane
triazine (herbicide)
classic androgen receptor binder
* RUC = Rat Uterine Cytosol (values are from ICCVAM's "Evaluation of In Vitro Test Methods for Detecting Potential Estrogen Receptor and Androgen Receptor
Binding and Transcriptional Activation Assays", NIH document No. 03-4503, May 2003).
t Enterolactone added per recommendation by independent Chemical Selection Board, which characterized it as a weak binder.
52
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
3. Preparation of rat uterine cytosol
The rat uterine cytosol (RUC) was prepared by the laboratories according to the
protocol. The following table (Table 22) shows similarities and differences between the
cytosol preparations. The wide range of protein concentrations is noteworthy.
Table 22. Rat uterine cytosol preparations
Source of rats
Age of rats (days) at
ovariectomy
Days after
ovariectomy that
uteri were removed
Strain of rats
Rats ovariectomized
at source
Number of cytosol
preparations
Date ovariectomized
Uteri removed at
source and frozen
Dates uteri removed
Dates cytosol made
Protein
concentration of
cytosol batches
(mg/mL)
LabX
Charles River
94/94
8/7
Sprague-Dawley
yes
2
1/14/08,3/4/08
no
1/22/08,3/11/08
1/23/08,3/19/08
2.2, 2.3
LabY
Harlan
88-93
8
Sprague-Dawley
yes
3
2/19/08
yes
2/27/08
2/29/08, 4/20/08,
6/12/08
8.09,6.67,2.5
LabZ
laconic
84-90
8-10
Sprague-Dawley
yes
5
2/8/08,2/13/08,
2/22/08, 5/28/08
no
2/18/08,2/21/08,
2/22/08, 3/3/08,
6/6/08
2/18/08,2/21/08,
2/22/08, 3/3/08,
6/6/08
1.16,3.16,2.69,
2.75, 2.52
4. Results
The contractor's summary report of the results from the three labs is attached as
Appendix 5, and the individual laboratories' reports are attached as Appendix 6 through
Appendix 8.1
Note, however, that the analyses done by the individual laboratories differ from the uniform analysis on
which this ISR is based. Thus the summaries and graphs in the laboratories' reports are different from
what is presented here. See Section IV.B.5 below.
53
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
a. Saturation binding
Saturation binding assays were performed on each batch of cytosol used in this
study. The results are shown in Table 23 (Lab X), Table 24 (Lab Y), and Table 25 (Lab
Z):
Table 23. Saturation binding results, Lab X
Assay
ID
Sat1
Sat 2
Sat 3
Average
Sat 4
Sat 5
Sat6
Average
(4,5,6)
Average
(4.6)
Cytosol
Prep
1/23/08
1/23/08
1/23/08
3/19/08
3/19/08
3/19/08
Plateau
reached
?
Y
Y
Y
Y
Y
Y
Linear
Scatchard
?
Y
Y
Y
Y
Y
Y
Kd
(nM)
0.1508
0.1466
0.1813
0.1596
0.6457
1.1810
0.3691
0.7319
0.5074
RSEof
Kd
(%)
10.9%
12.5%
18.9%
6.8%
28.8%
7.0%
Bmax
(fmole/
100 MS)
71.03
58.02
59.84
52.09
84.96
57.16
RSEof
Bmax
(%)
3.2%
3.9%
5.6%
2.7%
13.1%
2.4%
NSB
acceptable?3
Y
Y
Y
Y
Y
Y
NSB is acceptable if it is <50% of total binding.
Table 24. Saturation binding results, Lab Y
Cytosol Prep
3
3
4
5
ug Protein/tube
25
50
50
50
Kd
(nM)
0.042
0.061
0.138
0.235
Bmax
(fmol ER/100 ug protein)
10.99
17.43
15.17
36.46
Table 25. Saturation binding results, Lab Z
Cytosol Batch ID
RUC2-18-08
RUC2-21-08
RUC2-22-08
RUC3-3-08
RUC6-6-08
Protein
Concentration
(mg/ml)
1.16
3.16
2.69
2.75
2.52
Est. Kd
(nM)
1.4
1.1
1.5
1.0
0.8
Est. Bmax
(fmole/100
g protein)
146.0
85.2
107.9
41.2
113.2
Scatchard plots are provided in the final reports (Appendix 6 through Appendix 8).
54
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
b. Competitive binding
Competitive binding experiments were performed by each of the three labs for
the estradiol (strong) standard, the norethynodrel (weak) standard, and 23 test
chemicals. EPA dropped R1881 from testing as a standard chemical (negative control)
when it became clear that it was a weak binder. (This chemical was tested at higher
concentrations in this study than previously tested by EPA, and this revealed its weak
binding affinity for the ER.)
Each test run consisted of the estradiol and norethynodrel standards and one or
more test chemicals. EPA required that all chemicals be run in ethanol as solvent for
this study (with the exception of DMSO for benzanthracene) in order to facilitate
comparisons across laboratories, but this requirement was not followed by the labs. In
addition, apparently labs did not always run the standards in the same solvent as was
used for the test chemical - a requirement of the protocol. Since the labs did not
always report the solvent used for the standard even when the solvent for the test
chemical was reported, EPA was unable to examine solvent effects.
Each laboratory analyzed its own data, evaluated the acceptability of individual
runs, and classified the test chemicals as binders, non-binders, or equivocal according
to the data interpretation guidelines in the protocol.
The labs did not adhere to the performance criteria for the standards as EPA had
required. In particular, the tops of the standard estradiol curves often were significantly
higher than allowed (e.g., 130% of binding in the solvent control tubes and higher). This
may have been due to an unexplained drift downward in radioactivity counts between
solvent control tubes at the beginning of the run and tubes at the end of the run.
5. Analysis
To ensure comparability of results across laboratories, EPA had the data
reanalyzed uniformly, normalizing each run to the mean value of binding at the lowest
concentration of estradiol standard for each run. (See the discussion in Section IV.B.6.)
It is the result of this uniform analysis that is presented in this Integrated Summary
Report.
The analysis for the second interlaboratory study was substantially different from
the analysis for the first. In the analysis of the second study, tops and bottoms were not
55
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
constrained to 100% and 0%, respectively. Not constraining the tops and bottoms
allows performance criteria to be used for these quantities, and provides a more realistic
estimate of the Hill slope. Outliers were identified and excluded, using the automatic
outlier elimination procedure that had not been available for the first study. For these
reasons, the analyses of the first and second interlaboratory studies are not
quantitatively comparable.
a. Model fitting
The renormalized percent binding values were calculated for each run and
transferred to PRISM Version 5 for model fitting. The four-parameter one-site
competitive binding model was fitted to the data using PRISM'S algorithm for nonlinear
least squares with automatic outlier detection. The algorithm is discussed in detail in
Motulsky and Brown (2006). Details of the algorithm, free parameters in the algorithm,
and PRISM default values are discussed in the detailed statistical methods report
(Appendix 9).
Separate model fits were carried out for each test run of each of the standard
chemicals and the 23 test chemicals. These fits are shown as Appendix 10 (Lab X),
Appendix 11 (Lab Y), and Appendix 12 (Lab Z).
b. Evaluation of runs for acceptability
Since the original performance criteria were not followed, EPA reviewed the
individual renormalized fits for the standard chemicals (estradiol and norethynodrel) and
judged each run as "acceptable" or "unacceptable", as described in section IV.B.S.d
EPA also reviewed the test chemical runs for acceptability. It would have been
better to accept test chemical runs if and only if both of the standard chemicals for that
run were judged to be acceptable, but this would have resulted in too few acceptable
runs to evaluate.
Test chemicals were evaluated as if they were true unknowns; that is, it was not
assumed that these were one-site competitive binders whose slopes, tops, and bottoms
were expected to conform to expected behavior. Instead, runs were discarded only if
there was extreme scatter across data points; or if a run was markedly different from
two runs, judged acceptable for that chemical, that were similar to each other. If unclear
whether to keep or discard a run, the run was kept. Tables showing the EPA
56
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
evaluations for test chemicals compared to the evaluations of the standards for that run
as well as to the labs' own designations of acceptability are included as Tables 14 (Lab
X), 15 (Lab Z), and 16 (Lab Y) of the Detailed Statistical Report (Appendix 9).
c. Comparison across laboratories
For each test laboratory the individual test runs of the test chemicals were
designated as "binder", "equivocal", "non-binder", "non-testable"2. The criteria are given
below.
• Binder - A binding curve can be fitted with "slope approximately -1". The lowest
point on the response curve within the range of the data is less than 50%.
The criterion "slope approximately -1" was determined as follows: The set
of estradiol standard fits for the runs that were considered by EPA to be
acceptable runs were compiled across the test laboratories. There were 134
such runs. The range of slopes within these runs was defined to be the range of
slopes that are "approximately -1". This range was -1.78 to -0.67.
• Non-binder - The run is testable (see discussion below concerning "non-
testable") and
a) a binding curve can be fitted (with any slope) and the lowest point on the
fitted response curve within the range of the data is above 75%.
or
b) a binding curve cannot be fitted and the lowest average of replicates at
any concentration is above 75%.
• Equivocal - Any testable run that is neither a binder nor a non-binder is
equivocal. The run might or might not have a fitted model. In general, this
category covers those compounds which appear to be interacting with the
receptor at high concentrations but are so weak that they displace only 25 to
50% of the radioactive estradiol. It also covers those compounds that appear to
have precipitous slopes since such slopes may be an artifact of the fitting
algorithm rather than a reflection of the behavior of the chemical. Such curves
deserve a closer look in the weight-of-evidence determination of interaction with
the receptor.
2 The terminology has been changed in the protocol to "interacting" and "not interacting" instead of
"binder" and "non-binder" to reflect that this assay does not fully characterize the interaction.
Nevertheless, the old terms are retained here.
57
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
• Non-Testable - There are no data points at or above concentration 10"6 M and
one of the two following conditions hold:
a) A binding curve can be fitted but the binding curve is not lower than 50%
by concentration 10~6.
or
b) A binding curve cannot be fitted and the lowest average of replicates at
any concentration is above 50%.
The "non-testable" designation is meant to cover chemicals that could not be put
into solution at a high enough concentration to determine whether or not they are
binders. It does not cover strong binders, which typically reach 50% inhibition of the
radioligand by 10~6 M.
The classifications for each run are shown in Table D (i.e., Appendix D) of the
Detailed Statistical Report (Appendix 9).
Classifications for chemicals were obtained by assigning the following values to
each run and averaging:
• Testable runs:
binder = 2
equivocal = 1
non-binder = 0
• Non-testable runs: missing value, not used in averaging
The average score determined the overall binding category for that laboratory
and test chemical as follows:
Binder: Averages 1.5
Equivocal: 0.5 < Average < 1.5
Non-binder: Average < 0.5
The binding categorizations were combined across labs by a majority voting rule.
If two or more labs reported the same binding categorization it was taken as the overall
binding categorization. If each lab reported a different binding categorization (e.g. Test
chemicals 8, 14) the overall binding designation was reported as "Inconclusive".
The "Expected Affinity" of each test chemical is, as explained in Section IV.B.2,
based on Relative Binding Affinities for the estrogen receptor reported in the literature.
Table 27 summarizes the classification of each chemical by lab, and compares
results across labs. It also compares the majority result to the expected result.
58
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Because there are only three categories in the assay (binder, non-binder, equivocal) but
five categories in the expected affinity designation (very strong, strong, moderate, weak,
negative), a map between the two was required. The convention adopted is displayed
in Table 26.
Table 26. Correspondence between experimentally determined binding category and expected
affinity
Binding Category
Binder
Equivocal
Non-binder
Expected Binding Affinities
Considered Equivalent to the
Binding Category
Very Strong, Strong, Moderate, Weak
Moderate, Weak, Negative
Negative
59
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 27. Comparison of classifications across labs, ranked by expected affinity
Chemical
Code
2
3
1
4
13
12
11
5
6
16
14
18
15
7
19
17
8
9
22
10
21
20
23
Chemical
(all were run blind)
17-Ethynylestradiol
DES
17beta-Estradiol (blinded)
Meso-Hexestrol
Tamoxifen
Zearalenone
Equol
Genistein
Norethynodrel (blinded)
4-n-heptylphenol
Salpha-Dihydrotestosterone
Benz(a)anthracene
Bisphenol A
Butyl paraben
Enterolactone
Kepone (Chlordecone)
Nonylphenol (mixture)
o,p'-DDT
Atrazine
Corticosterone
Octyltriethoxysilane
Progesterone
R1881
Expected
Affinity
Very Strong
Very Strong
Strong
Very Strong
Strong
Strong
Moderate
Moderate
Moderate
Weak
Weak
Weak
Weak
Weak
Weak
Weak
Weak
Weak
Negative
Negative
Negative
Negative
Negative5
Majority of
Labs
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Inconclusive
Inconclusive
Binder
Binder
Binder
Binder
Binder
Binder
Equivocal
Binder
Non-Binder
Equivocal
Binder
LabX
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Equivocal
Non-Binder
Binder
Binder
Binder
Binder
Binder
Equivocal
Equivocal
Binder
Non-Binder
Equivocal
Binder
LabY
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Non-Binder
Binder
Binder
Binder
Binder
Binder
Binder
Binder
Non-Binder
Binder
Non-Binder
Non-Binder
Binder
LabZ
Binder
Binder
Equivocal3
Binder
Binder
Binder
Binder
Binder
Binder
Equivocal
Binder
Equivocal
Binder
Binder
Equivocal
Binder
4
Binder
Equivocal
Binder
Non-Binder
Equivocal
Binder
Legend
Black (standard) font
Reef (italics) font
Green cell (dark shading)
Yellow cell (light shading)
Agreement with the "Majority of Labs"
Disagreement with the "Majority of Labs"
Agreement with the "Expected Affinity"
Disagreement with the "Expected Affinity"
Lab Z did not test more-dilute solutions than the default and thus did not obtain a full curve. The data that it
obtained showed full displacement of the radioligand at higher concentrations. The protocol has been adjusted to
emphasize that a full curve must be obtained where there are clear indications of binding.
Lab Z had no acceptable test runs for test chemical 8.
5 During the validation study, higher concentrations of R1881 were tested than previously used and the chemical was
demonstrated to be a binder.
60
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
cf. Development of performance standards
The results of each individual run for estradiol and norethynodrel were reviewed
by EPA and were designated as "acceptable" or "unacceptable". These judgment calls
were based on review of fitted curves and the numerical values for slopes, tops, and
bottoms as provided by the Prism software, based on data normalized to the binding
value of the run's lowest estradiol concentration. Graphs superimposing all of the runs
judged "acceptable" for each lab, estradiol separately from norethynodrel, are attached
as Appendix 13. EPA's designations were used in the determination of performance
criteria for the standards. The performance criteria were defined as the tolerance
bounds that include 80% of the acceptable runs with 95% confidence, for each of the
binding curve parameters (top, bottom, slope, residual standard deviation; and in the
case of the weak positive, the RBA), across all of the laboratories. Graphs showing
where each of the runs (both acceptable and unacceptable) falls in relation to the
tolerance bounds are included as Figures 1 through 9 of the Detailed Statistical Report
(Appendix 9).
The tolerance bounds reflect the variance components among the runs. The
laboratories were assumed to be a random sample from the population of acceptable
laboratories that might carry out the ER assay in the EDSP. The laboratories were
treated as random effects. The variance components incorporated in the tolerance
bounds are:
• Within run variation (reflected in the standard errors of the fitted binding curve
parameters)
• Run to run variation within labs
• Lab to lab variation
The variance components were estimated by mixed models analysis of variance.
The analysis of variance models and fitting methods are discussed in Appendix 9.
The tolerance bounds are normal theory tolerance bounds, constructed to
include 80% of the population of parameter values with 95% confidence. The tolerance
bounds for top, bottom, Hill slope, and log™ (Relative binding affinity) are two sided
bounds. The tolerance bounds for residual variation are one sided upper bounds.
The tolerance bounds are summarized in Table 28 and Table 29.
61
-------
ER-RUC ISR, v3.16c.doc
printed 3/11/2009 6:49 PM
Table 28. Slope, top, bottom, RBA: Tolerance interval bounds to contain at least 80% of population of test runs with 95% confidence.
Outliers deleted.
Chemical
Estradiol
Norethynodrel
Parameter
Hill Slope
Top
Bottom
Hill Slope
Top
Bottom
logeRBA
Average
Estimate
-1.0006
102.43
-1.4549
-0.9874
99.8434
1.4100
-2.9256
Std Error
Estimate
0.02319
0.5857
0.1611
0.03832
1.7742
1.6191
0.01889
Lower Tolerance Limit
-1.3434
94.3303
-3.1906
-1 .4048
74.0351
-19.1099
-3.22953
Upper Tolerance Limit
-0.658
110.533
0.281
-0.570
125.652
21.930
-2.62167
Table 29. Ln(residual standard deviation), residual standard deviation: Tolerance interval bounds to contain at least 80% of population
of test runs with 95% confidence. Outliers deleted.
Chemical
Estradiol
Norethynodrel
Mean
Log(Syx)
1.2999
1.6175
Std Error
Log(Syx)
0.1964
0.1927
Upper
Tolerance Limit
Log(Syx)
2.34866
2.59228
Upper
Tolerance
Limit
Syx
10.4715
13.3601
62
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
The tolerance bounds for estradiol are remarkably similar to the performance
criteria established after the first interlaboratory study (Table 16) but the values for
norethynodrel are significantly wider. Although the reason for this is not known, the
differences might be attributable to solubility problems with norethynodrel. The
laboratories in the second study reported consistent difficulty in preparing the highest
concentration of norethynodrel and were often forced to drop that concentration from
the run. If the highest concentration that ultimately was used was not fully soluble, the
serial dilutions prepared from that mixture would not have the expected concentrations,
thus shifting the plotted curve. Also, the lack of a value at the highest concentration (10"
4 M) could have caused the curve to be fit differently than it would have been if that data
point had been available. A further difference between the first study and the second
was that in the second study dilutions were made serially in assay buffer whereas in the
first study they were done serially in ethanol so that all of the concentrations had the
same amount of solvent. Labs in the first interlaboratory validation study did not report
difficulties in preparing the 10~4 M concentration, and a review of the data from that
study (not shown) indicates only a few isolated runs where solubility might have been a
problem. Pending an ongoing investigation into the solubility of norethynodrel in this
assay, EPA is retaining the more restrictive performance criteria established from the
first study.
6. Discussion
Results generally matched expectations (Table 27). All of the very strong,
strong, and moderate binders were correctly determined to interact with the estrogen
receptor, and seven of nine weak binders were also correctly identified as interacting.
Most chemicals that were expected not to be binders were "equivocal" in their
responses, and in the case of at least one if not both of the chemicals that were
expected to be negative but consistently showed interaction (R1881 and corticosterone)
it is likely that the expectation of non-binding was incorrect. One chemical
(octyltriethoxysilane) produced consistently negative results across all three laboratories
indicating that the assay can correctly identify compounds that do not interact with the
estrogen receptor.
63
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
There are, however, several results in Table 27 that deserve comment. The
finding that estradiol was "equivocal" rather than clearly interactive in one laboratory
when tested blindly appeared to be a serious deficiency of the assay but is explained by
the fact that the laboratory did not adjust the test concentration range to a more dilute
range that would have allowed characterization of the full binding curve as required by
the protocol. Binding at the concentrations tested showed clear interaction with the
receptor. Although the data interpretation procedures could be adjusted to
accommodate cases like this where there is clear interaction even though the top is not
fully characterized, the Agency finds it more appropriate to require that the full curve be
generated than that partial datasets be accepted. This will increase confidence that the
chemical has been adequately characterized, by minimizing the effect of variability over
a reduced set of datapoints.
The highest concentration at which a chemical is tested in this protocol (1 mM)
may help explain why several chemicals that were expected to be negative produced
equivocal responses or even showed evidence of interaction with the estrogen receptor.
As noted in the Background Review Document, "[h]istorically, the highest dose tested ...
has ranged generally from 1 to 100 uM, with some tests conducted at doses as high as
1 mM." Thus, some chemicals that were reported in the literature to be negative may
have shown evidence of interaction had they been tested at this high concentration.
The EPA included the 1 mM concentration based on an analysis in the BRD.
The BRD explains that the ability of the assay to identify weakly positive chemicals rises
with the highest concentration tested, and notes that"... if testing for ER binding
substances requires the ability to detect substances with an ICso that is at least six
orders of magnitude lower than that of 17(3-estradiol, then the limit dose...should be
above 4 mM (e.g., 10 mM) to allow for the detection of an ICso in the concentration
range of interest. However, if five orders of magnitude are sufficient for RBA values,
then the limit dose would have to be above 400 uM (e.g., 1 mM). Decreasing the limit
dose to 100 uM would limit the sensivity of the assay to RBA values that cover
approximately four orders of magnitude." The EPA believes that five orders of
magnitude is appropriate for this Tier 1 screening assay and thus is requiring 1 mM
concentration (where achievable in solvent). The EPA recognizes that the 1 mM
64
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
concentration is of questionable relevance to in vivo systems, and that other interactions
besides one-site competitive binding may be occurring at such levels. Such interactions
would be "false positives" for one-site competitive binding even though they are indeed
"interactions" with the estrogen receptor. The EPA finds the emphasis on "interaction"
rather than one-site competitive binding perse to be appropriate for a Tier-1 screening
assay, and preferable to attempting to identify the specific mechanism of interaction
using this assay. Since the data on octyltriethoxysilane produced unequivocal negative
responses in all three laboratories, it is known that the use of 1 mM as the highest
concentration will not result in positive results for all chemicals. The assay is specific to
interaction with the estrogen receptor.
The protocol calls for a negative control in each run (along with reference
chemical and weak positive control) but except for a few runs in the beginning, no
negative controls were included in this study. This is because R1881, the chemical
chosen to be the negative control, turned out to be a weak binder during the first few
runs in this study. As discussed in the previous paragraph, testing at higher
concentrations than previously tested may show that a chemical previously thought not
to interact is a weak binder. Evidence that R1881 interacts with the estrogenic system
(i.e., is not just anti-androgenic) includes a positive uterotropic study (Ojasoo and
Raynaud 1978). This suggests strongly that R1881 indeed interacts with the estrogen
receptor rather than that the ER-RUC assay falsely characterized a non-interactor as an
interactor. EPA has replaced R1881 with octyltriethoxysilane, a compound which tested
consistently negative in all laboratories, as negative control.
In the analyses produced by the individual laboratories in this study, the top
plateaus for the standard chemicals (estradiol and norethynodrel) often exceeded the
performance criteria by several tens of percentage points. The reason for the high
plateaus was not determined but may be related to the solvent control tubes. Such
tubes placed at the end of the run (of several test chemicals run simultaneously) often
yielded lower dpms than similar tubes placed at the beginning of the run. The average
of all solvent control tubes was therefore lower than it would have been had only the
first solvent control tubes been included. The lower average could have contributed to
the appearance of higher-than-solvent-control values for the estradiol and norethynodrel
65
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
standards, which were always placed at the front of the runs. To compensate for this
phenomenon, EPA requested that the data be analyzed using the binding at the lowest
concentration of estradiol, rather than the binding in the solvent control tubes, as the
value for 100% binding of the radiolabeled estradiol (i.e., zero displacement by
unlabelled estradiol). (The lowest concentration of estradiol (10~11 M) had previously
been determined to be sufficient to establish the top plateau for estradiol.) The results
of this analysis are what was presented in this Integrated Summary Report.
Use of the binding values at the lowest concentration of estradiol to establish
100% binding of the radioligand may have contributed to the variability of results both
within laboratories and across laboratories in this study inasmuch as there were only a
maximum of three replicates per run to establish the 100% binding value as opposed to
the six solvent control tubes. If, as happened in a few cases, one or more of the
replicates was unusable, the 100% binding value was likely to have a larger standard
error of the mean even though the standard deviations may not have been unusually
high.
As explained in the previous section, laboratories had significant difficulty getting
the weak positive control chemical, norethynodrel, to stay in solution at the highest
concentration (1 mM), and this data point is frequently missing from a run. Loss of this
data point can significantly affect the estimate of the bottom plateau and can cause a
run to miss the performance criterion for the bottom even though the remaining data
points are consistent with a good run. Solubility problems with norethynodrel had not
been encountered before in any of the preliminary studies. EPA is currently performing
solubility studies to determine whether the solubility of norethynodrel is likely to pose a
significant problem in the future. It is also investigating other weakly positive estrogen
receptor binders as possible substitutes. Alternatively, it may specify a lower
concentration to use as the maximum for this weak positive control, and establish
modified performance criteria using that concentration.
Although the influence of the number of chemicals per run on intralaboratory
variability of the results was not studied, EPA notes that most laboratories ran 3 or 4
chemicals per run. Each test chemical requires 24 tubes (8 concentrations x 3
replicates) in addition to the 87 tubes for controls, so a run of 3 chemicals consists of
66
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
159 tubes. (In this study, such a run consisted of 132 tubes since 27 R1881 tubes were
omitted.) Processing this large number of tubes may have increased variability due to
such factors as increased duration of exposure to room temperature (and subsequent
denaturation of the receptor), and diminished ability to monitor partial pellet loss after
centrifugation. This potential source of variability is expected to be less of a factor for
laboratories in the EDSP if only one chemical is being tested at a time.
Finally, it should be remembered that while intralaboratory variability was
disappointingly high for at least one laboratory in this study, such variability is not
expected to be as much of a problem for laboratories that demonstrate the ability to
meet the required performance criteria. The limited time available to run this large study
on 23 chemicals apparently did not allow development of the proficiency necessary to
obtain precise runs in all laboratories. The fact that results were almost all in accord
with expectations when screening for interaction despite the variability in quantitative
values shows that the assay is robust for this use.
V. Additional considerations
An ER transcriptional activation assay has recently been validated by the OECD
for use in screening chemicals. There is overlap between receptor binding assays and
transcriptional activation assays inasmuch as binding is the first step in transcriptional
activation. Both assays are relatively simple and inexpensive in vitro assays, and the
EPA is likely to require use of both, at least during the initial stages of the EDSP. The
transcriptional activation assay is specific to the a isoform of the estrogen receptor while
rat uterus contains both the a and (3 isoforms. Thus the ER-RUC binding assay may
respond to substances that are specific to the (3 isoform while the transcriptional
activation assay cannot respond to such substances.
Several changes have been made to the protocol since the second
interlaboratory validation study. Small revisions were necessary to clarify wording that
laboratories had found confusing, such as the upper limit for ligand depletion. The three
significant changes are: 1) substitution of octyltriethoxysilane for R1881 as the negative
control (discussed above); 2) introduction of an optional solubility testing step before
running the assay; and 3) adjusting the test chemical dilution scheme so that dilutions
are made in solvent rather than buffer. This last change was made in order to keep the
67
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
concentration of solvent constant (2%) across all test chemical concentrations. This
change is not expected to affect the results of the assay given the limit on solvent
concentration included in the protocol and the results of the solvent concentration study
(Eldridge 2007) discussed above. The dilution scheme is now similar to the scheme
used in the first interlaboratory study.
EPA tested additional chemicals at the weak end of the expected affinity
spectrum in a later portion of this study. Those results are not being released at this
time in order not to compromise the identity of the substances in the on-going, parallel
hrER validation study. However, the pressure of processing 20 additional substances in
addition to the 23 chemical reported here, within a defined time period, may have led to
greater variability in results than will typically be seen when attention can be focused on
one chemical in the Endocrine Disrupter Screening Program.
VI.Summary
The EPA agrees with the 1998 conclusion of the EDSTAC that the estrogen
receptor binding assay using rat uterine cytosol is a validated assay for simple
screening for interaction with the estrogen receptor in the context of a battery of assays.
As recommended by the EDSTAC and ICCVAM, EPA has optimized and standardized
the most important parameters of this assay and has shown that the resulting protocol is
transferable to other laboratories. The variability of results may not support use for
quantitative structure-activity relationship model development at this time, but if further
work were to be undertaken to validate for such a use in the future, it may be that only
the performance criteria rather than the protocol itself need to be adjusted.
A. Strengths
As an in vitro assay, the ER-RUC assay provides direct contact between
chemical and the estrogen receptor without modification through absorption,
distribution, metabolism, and excretion (ADME) considerations. The assay therefore
has the potential of being more sensitive than the in vivo assays in the EDSP Tier 1
Battery, which usually involve ADME.
68
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
The assay provides consistent responses at the simple screening level, across
laboratories, and these responses are in line with expectations for those chemicals
tested whose estrogen receptor binding behavior is well-established.
The criteria used for classifying a chemical as interactive or not err on the side of
classifying a chemical as interactive, which is appropriate for a screening assay. Since
additional assays will be used in Tier 1 screening before a final determination of the
potential for interaction is assigned, and additional Tier 2 assays will be performed to
confirm the interaction and provide dose-response information before risk is assessed,
the bias towards false positives is appropriate.
Despite the bias towards false positives, chemicals that truly do not interact in
any way with the estrogen receptor (whether by one-site competitive binding or any
other mechanism) consistently test negative in this assay.
The assay is short and inexpensive compared to the in vivo assays in the Tier 1
Battery.
Rat uterus contains both a and (3 isoforms of the estrogen receptor (Kuiper et al.
1997) and unlike current assays using recombinant receptor, which are specific to the
alpha isoform, may therefore respond to substances which are specific to either isoform.
8. Weaknesses
The assay is sensitive to many details of preparation and technique and can
show wide variability if not performed exactly as stated in the protocol. It is, for
example, subject to problems if the receptor concentration in the cytosol is too low or
too high, or the tubes are not kept cold at all times during preparation, incubation, and
separation of bound from free tracer. However, the data suggest that a lab that meets
the performance criteria for the standard and weak positive is likely to generate data
that is much less variable than laboratories that do not meet the performance criteria.
Thus the data from the EDSP, which requires adherence to the performance criteria,
are likely to be of less variability than the data obtained in the second interlaboratory
study.
Because of the sensitivity of this assay to technique, it is not consistently
possible to characterize the probable mechanism of action of an "interactor" as one-site
competitive binding even when the substance is known to be a one-site competitive
69
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
inhibitor. Nevertheless, since the purpose for which this assay will be used in the EDSP
is only to identify interactors, not the mechanism by which they interact or to develop
quantitative descriptors of the interaction such as the log(ICso), this weakness is not
acute.
Insolubility of test chemicals can be a significant problem when trying to identify
weak interactors. When a high concentration of test chemical in solvent cannot be
obtained, it may not be possible to test adequately the ability of the chemical to interact
with the estrogen receptor. The choice of three possible solvents (DMSO, ethanol, or
water) should help mitigate this potential weakness.
The lack of metabolic activity can cause chemicals which require metabolic
activation to test negative. Results from in vivo assays in the Tier 1 Battery may not be
consistent with results from this assay. It will be important not to tally simple "positives"
and "negatives" in the Tier 1 Battery of assays but to evaluate the entire dataset when
judging the weight of the evidence for interaction with the estrogen system.
The analysis of datasets might be more complicated than necessary. When the
analysis was developed, the expectation was that standardization of the assay would
allow precise and replicable quantitative analysis of log(ICso)s and Relative Binding
Affinities. The expectation was also that precise, standardized methods of analysis
would contribute to reproducibility and therefore use in other applications such as
structure-activity relationship models. However, in the face of the variability
encountered, EPA is assessing whether such analysis could be replaced with a simpler
analysis and still meet the needs of the Screening Program.
Finally, the assay requires the use of animals. Although it is an in vitro assay,
the receptor is obtained from uteri. EPA is cooperating with an OECD effort to validate
a binding assay that uses human recombinant estrogen receptor rather than receptor
from animals.
C. Conclusion
EPA believes that the standardized estrogen receptor binding assay using rat
uterine cytosol as source of receptor has proven to be transferable, sensitive to
chemicals known to interact with the estrogen receptor, specific to chemicals which
interact, and reproducible in contract laboratories in terms of classifying chemicals as
70
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
interacting with the estrogen receptor or not. The assay is appropriate for use in its
standardized form in a screening program to identify interaction with the endocrine
system even though it may not be appropriate for other uses such as development of
quantitative structure-activity relationship models.
71
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
VII. References
Carter, C.M.S., Leighton-Davies, J.R., Charlton, S.J. (2007). "Miniaturized receptor
binding assays: complications arising from ligand depletion." J Biomol Screen 12(2):
255-66.
EDSTAC (August 1998). Endocrine Disrupter Screening and Testing Advisory
Committee (EDSTAC) final report. U.S. Environmental Protection Agency, Washington,
DC. Available at: http://www.epa.gov/scipoly/oscpendo/pubs/edspoverview/finalrpt.htm
Eldridge, J.C. (2007). Final report: Development of a standardized approach for
evaluating environmental chemicals with low solubility in the estrogen receptor (ER)
binding assay. USEPA Contract number EP-D-05-076. (Attached to this ISR)
FFDCA. (1996). Federal Food, Drug, and Cosmetic Act, as amended by the Food
Quality Protection Act of 1996 (Pub. L. 104-170), 21 U.S.C. 346a(p). Section 408(p).
Hartung, J. and Makambi, K.H. (2001). Simple non-iterative t-distribution based tests
for meta-analysis South African Statistical Journal 35: 1-17.
, A.V. (1910). The possible effects of the aggregation of the molecules of
haemoglobin on its dissociation curves. J. Physiol. (Lond.) (1910) 40, iv-vii.
ICCVAM. ICCVAM Authorization Act of 2000. Interagency Coordinating Committee on
the Validation of Alternative Methods. 42 U.S.C. 285. Public Law 106-545, December
19, 2000. http://iccvam.niehs.nih.gov/docs/about_docs/PL106545.pdf
ICCVAM (2002). Background Review Document - Current Status of Test Methods for
Detecting Endocrine Disrupters: In Vitro Estrogen Receptor Binding Assays (NIH 03-
4504). U.S. Department of Health and Human Services, National Toxicology Program,
Interagency Coordinating Committee on the Validation of Alternative Methods.
http://iccvam.niehs.nih.gov/docs/endo docs/final 1002/erbndbrd/ERBd034504.pdf
Kuiper, G., Carlsson, B., Grandien, K. etal. (1997). Comparison of the ligand binding
specificity and transcript tissue distribution of estrogen receptors a and (3.
Endocrinology 138(3): 863-870.
Mauck, L.A., Day, R.N, Notides A.C. (1982). Molybdate interaction with the estrogen
receptor: effects on estradiol binding and receptor activation. Biochemistry 21 (8): 1788-
1793.
Motulsky, H. (2003). GraphPad PRISM User's Guide Version 4.0, GraphPad Software,
Inc. http://www.graphpad.com/prism/Prism.htm
72
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Motulsky, H.J. and Brown, R.E. (2006). Detecting outliers when fitting data with
nonlinear regression a new method based on robust nonlinear regression and the false
discovery rate. BMC Bioinformatics, Vol 7, pp 123-142.
Motulsky, H. (2007). GraphPad PRISM Version 5 software, San Diego, CA.
http://www.graphpad.com/prism/Prism.htm
Murayama A. and Fukai F. (1985). Effect of molybdate on the molecular organization
of the estrogen receptor system of porcine uterus. J. Biochem. 98(4): 1083-1090.
NIEHS (October 2002). Background Review Document: Current Status of Test
Methods for Detecting Endocrine Disrupters: In Vitro Estrogen Receptor Binding
Assays, National Toxicology Program Interagency Center for the Evaluation of
Alternative Toxicological Methods, National Institute of Environmental Health Sciences
NIH Publication No. 03-4054. Available at:
http://iccvam.niehs.nih.gov/docs/endo docs/final 1002/erbndbrd/ERBd034504.pdf
NIEHS (March 1, 1997). Validation and regulatory acceptance of toxicological test
methods: a report of the ad hoc Interagency Coordinating Committee on the Validation
of Alternative Methods (NIH 97-3981). Available at
http://iccvam.niehs.nih.gov/docs/about docs/validate, pdf
OECD (August 1996). Final report of the OECD workshop on harmonization of
validation and acceptance criteria for alternative toxicological test methods.
Organisation for Economic Co-operation and Development. Stockholm, Sweden,
January 22-24, 1996.
Ojasoo T., and Raynaud, J.-P. 1978. Unique steroid congeners for receptor studies.
Cancer Research 38:4186-4198.
Pettersson, K.S., Vanharanta, R.M., Soderholm, J.R. (1985). Pitfalls in the dextran-
coated charcoal assay of estrogen receptors in breast cancer tissue. J. Steroid
Biochem. 22(1 ):39-45.
SAS, Version 9.1.3 (2003). SAS Institute, Inc., Gary, NC.
Swillens, S. (1995). Interpretation of binding curves obtained with high receptor
concentrations: practical aid for computer analysis. Molecular Pharmacology 47(6):
1197-1203.
USEPA. (December 28, 1998). Endocrine Disrupter Screening Program; Proposed
statement of Policy, Federal Register Notice, vol. 63, issue 248, pp. 71542-71568.
Available at: http://www.epa.gov/scipoly/oscpendo/pubs/122898frnotice.pdf
73
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
USEPA (July 1999). Review of the Endocrine Disrupter Screening Program by a Joint
Subcommittee of the Science Advisory Board and Scientific Advisory Panel. EPA-SAB-
EC-99-013.
http://yosemite.epa.gov/sab/sabproduct.nsf/C8ABD410E357DBCF85257193004C42C4/
$File/ec13.pdf
USEPA (July 2007). Validation of screening and testing assays proposed for the EDSP.
Version 5.4. Available at:
http://www.epa.gov/scipoly/oscpendo/pubs/edsp validation paper v%205.4.pdf
USEPA (June 2008). Meeting minutes of the FIFRA Scientific Advisory Panel
Meeting held on March 25-26, 2008 to review and consider the Endocrine
Disrupter Screening Program (EDSP) Proposed Tier 1 Screening Battery.
http://www.epa.gov/scipolv/sap/meetings/2008/march/minutes2008-03-25.pdf
Zacharewski, T.R., Meek, M.D., demons, J. H., Wu, Z. F., Fielden, M. R., Matthews, J.
B. (1998). Examination of the in vitro and in vivo estrogenic activities of eight
commercial phthalate esters. Toxicological Sciences 46(2): 282-293.
74
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 1. Protocol (as revised following the second interlaboratory
validation study)
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 2. ICCVAM Background Review Document on the Estrogen
Receptor Binding Assay. Executive Summary and
Conclusions.
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 3. Eldridge CJ. 2007. Final report: Development of a
standardized approach for evaluating environmental
chemicals with low solubility in the estrogen receptor (ER)
binding assay.
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 4. Report on statistical methods for evaluating variability in and
setting up performance criteria for receptor binding assays
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 5. Overall report on second interlaboratory validation study
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 6. Final report from second interlaboratory validation study:
LabX
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 7. Final report from second interlaboratory validation study:
LabY
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 8. Final report from second interlaboratory validation study:
LabZ
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 9. Detailed statistical report
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 10. Curve fits after normalization: Lab X
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 11. Curve fits after normalization: Lab Y
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 12. Curve fits after normalization: Lab Z
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
Appendix 13. Graphs of acceptable runs for reference standard (estradiol),
weak positive (norethynodrel), and test chemicals, by
laboratory
-------
ER-RUC ISR, v3.16c.doc printed 3/11/2009 6:49 PM
------- |