n. \r .1?

if

5 1.1593
v

(to!!!

I: IL.
£52....

t,
. .. n o.
52:52.

 

'1":

.L!
a. x. . ;
13.33;: v.

I}.
{:31,‘ I. A 3.

.613.

1
hr... . an...
.1 z): (3...; AI
.12. .. .. :1 52:13:
9:15.25: L I).

5;... w.

I I
... 5.3 :39!
.. «1...: on
53.9 ..R
it nu...
2.2. . x..fl&o.....wn3.§.. V
55:35.5 «ma . am. airway:
€337.23. .rlhxa. .1. .Pv 2X
.. . r .i v .ltfil32...
aw. .i .3.
3:932, 13
, hiaaawifi.‘ fining”
3%.; Jung}... 5......
. .i‘. (1.1.39. .1
Hun-3m»... In. 4.59.1...
! a ) is, ’—

but.

>I‘N...
:.

99.... .
i
.IOJP. I.
s..-
33...:
., 5:

J

1 I

it.”
1.

3......
: .f #51-

l. :l
5 l2) ....av9.31
.3 \V. i

THESIS

(1'7"?

 

LIBRARY

Michigan State '
University

 

 

This is to certify that the

dissertation entitled

APPLICBXTIQV OF MECHANISVI-SPECIFIC IN VITRO
BIOASSAYS TO ADDRESS QUESTIONS IN
ENVIRCNMENI‘AL 'IOXImIDGY

presented by

Daniel Lawrence Villeneuve

has been accepted towards fulfillment
of the requirements for

Ph. D. degree in _ZQQ]_Qg;L__

(MA F‘fuw

Major prd‘ssor

Date April 24, 2000

MS U i: an Afflrmariw Action/Equal Opportunity Institution 0-12771

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

12 07 08
JAN 2 2 2007

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRCfDateDuepes-p. 15

APPLICATION OF MECHANISM-SPECIFIC IN VI T R0 BIOASSAYS TO ADDRESS
QUESTIONS IN ENVIRONMENTAL TOXICOLOGY

By

Daniel Lawrence Villeneuve

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Zoology

2000

ABSTRACT

APPLICATION OF MECHANISM-SPECIFIC IN VITRO BIOASSAYS TO ADDRESS
QUESTIONS IN ENVIRONMENTAL TOXICOLOGY

By

Daniel Lawrence Villeneuve

This dissertation presents four studies centered around the application of
mechanism-specific in vitro bioassays in the field of environmental toxicology. The first
study used in vitro bioassays to screen and rank 11 chemicals based on their affinity for
the estrogen receptor (ER) and ability to elicit an estrogenic response in vitro. Six of the
eleven chemicals tested were able to displace tritiated 17B-estradiol from the ER. Their
rank order of affinity was 17B-estradiol (E2) > coumestrol > l7B-ethynyl estradiol (EE2)
> nonylphenol (NP) > octylphenol (OP) > bisphenol A (EPA). The rank order of potency
for inducing reporter gene expression in the MCF-7-luc in vitro bioassay was E2 > EE2 >
NP > OP > coumestrol > atrazine > BPA. ER binding and MCF—7-luc gene expression
results were compared to generate hypotheses regarding potential mechanisms of action
for the compounds analyzed. The second study focused on the development and
characterization of a recombinant rainbow trout cell line (RLT 2.0)-based assay for
assessing dioxin-like potency. Methods were adapted for a 96-well microplate format,
and assay specific relative potencies (REPS) were derived for a number of halogenated
aromatic hydrocarbons. Overall, the rank order of RLT 2.0-derived REPS was
comparable to those generated based on other fish and mammalian bioassays, both in
vitro and in vivo. A sensitivity analysis was also conducted to estimate the uncertainty

associated with 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQ) calculated using

the REPS generated. Result indicated that variability in the RLT 2.0-derived REPS could
yield up to lO-fold uncertainty in RLT 2.0-based TEQ estimates. The third study focused
on the current debate surrounding the derivation, use, and misuse of in vitro bioassay-
based REP estimates. A systematic approach for deriving REP estimates from non-ideal
in vitro bioassay results was presented and demonstrated using example data sets. The
final study was designed to examine the relevance of current in vitro models for
predicting estrogenic potencies in fish. Sexually mature male common carp (Cyprinus
carpio) were exposed to 4-nonylphenol. Potential indirect mechanisms of action
involving the modulation of plasma steroid concentrations were examined. The study
detected no treatment related increases in concentrations of plasma E2, testosterone, or
vitellogenin, despite measurable levels of NP in the plasma and tissues of exposed carp.
Unexpectedly high variability among fish, and plasma E2 and VTG concentrations below
method detection limits limited the ability to resolve effects, however. The lack of a
detectable estrogenic effect in vivo hindered the ability to calibrate the known in vitro
potency of NP to its potency for producing estrogenic effects in sexually mature male
carp. The lack of estrogenic response also raised questions regarding the utility of
estimating plasma or tissue concentrations of 17B-estradiol equivalents as a means of
predicting the potential for estrogenic effects in vivo. Overall, the studies described in
this dissertation exemplify research which is being done to deve10p and establish the
utility of mechanism-specific in vitro bioassays as tools for environmental toxicology

research and risk characterization.

Dedicated to my parents, Kathy and Larry Villeneuve, for a lifetime of unconditional

support and encouragement.

iv

ACKNOWLEDGEMENTS

My sincere thanks to Dr. John Giesy for providing me with incredible
opportunities to develop both professionally and personally over the past five years. I
will always appreciate Dr. Giesy’s efforts to make those opportunities available. I also
thank Dr. Giesy for his tolerance of dyed beards, loud music, and other elements of
personal expression.

Thanks to my graduate committee members, Dr. Larry Fischer, Dr. Robert Roth,
and Dr. Juli Wade, for their patience, advice, and assistance.

I acknowledge Dr. Ron Crunkilton for providing me the undergraduate research
opportunities that led me down this path. Dr. Crunkilton’s mentorship was invaluable, as
was the advice and support of Dr. Kent Hall and Dr. C. Edward Gasque.

I owe a debt of gratitude to all the graduate students, post-docs, student workers,
and visiting scholars at MSU’S Aquatic Toxicology Laboratory, l995-2000. They were
all great colleagues and fi'iends who made my time at MSU both productive and
enjoyable.

I am forever indebted to my parents, Kathy and Larry, who provided a lifetime of
unconditional support and encouragement. This would not have been possible without
them. Thanks also to my sister Renee, Aunt Jane, Aunt Barb, and all my friends from
back home for staying in touch over the years.

Finally, I thank my family of fiiends, Rose, Norm, Celeste, Stephanie, Anne-
Marie, Clare, for helping me see more of what life has to offer and helping me stay sane,

but not too sane, over these last few years.

TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES
GENERAL INTRODUCTION
Chapter 1.

INTERACTIONS BETWEEN ENVIRONMENTAL XENOBIOTICS AND
ESTROGEN RECEPTOR-MEDIATED RESPONSES

INTRODUCTION
Environmental Estrogens
Description of the ER
Mechanism of Action For Estrogen Receptor Agonists/Antagonists
Rapid, Non-genomic Effects of Estrogens
Interactions and Cross-talk from other Signaling Pathways
SCREENING and MONITORING
Predictive Quantitative Structure Activity Relationships
In Vitro Assays
Receptor Binding Assays
Relationships Between In Vitro Receptor Binding and In
Vivo Assays
Cell proliferation and diflerentiation
Expression Assays
MCF- 7-luc Bioassay Method
Applications of the MCF- 7-luc Bioassay (a Model
Expression Assay)
Other Expression Assays
Environmental Monitoring
In Vivo Biomarkers
ACKNOWLEDGMENTS
REFERENCES

Chapter 2.

RAINBOW TROUT CELL BIOASSAY-DERIVED RELATIVE POTENCIES
FOR HALOGENATED AROMATIC HYDROCARBONS: COMPARISON
AND SENSITIVITY ANALYSIS

Abstract

INTRODUCTION

MATERIALS AND METHODS
RLT 2.0 cell culture

vi

ix

xi

14
15
19
19
20
21
24
26
28

32
33
34
35

38
43
49
50
51

64

65
65
69
69

Chemicals and Reagents
Exposure of RLT 2.0 cells
Luciferase assay
Luciferase assay: flash method
Luciferase assay: glow method
Calculation of RPS
Sensitivity Analysis
RESULTS
RLT 2. 0-specific RPS
DISCUSSION
Comparison of methods
Comparison of relative potencies
Sensitivity analysis
Conclusions
ACKNOWLEDGEMENT
REFERENCES

Chapter 3.

DERIVATION AND APPLICATION OF RELATIVE POTENCY
ESTIMATES BASED ON IN VITRO BIOASSAY RESULTS

Abstract
INTRODUCTION
Relative potency estimation
METHODS
Multiple Point Estimates
Relative Potency Bands
Interpretive framework
Example data sets
Data analysis
RESULTS and DISCUSSION
Applying the Framework
Conclusions
Acknowledgment
REFERENCES

Chapter 4
EFFECTS OF WATERBORNE EXPOSURE TO 4-NONYLPHENOL ON
PLASMA SEX STEROID AND VITELLOGENIN CONCENTRATIONS IN
SEXUALLY MATURE MALE CARP (CYPRINUS CARPIO)

Abstract

1. Introduction
2. Materials and Methods

vii

69
70
71
71
73
74
77
77
78
80
80
84
89
96
97
97

102

103
104
105
111
111
113
114
118
121
122
122
127
129
129

134

135
136
140

2.1 Exposure
2.2 Sample Collection
2.3 Estradiol Radioimmunoassay (RIA)
2.4 Testosterone Enzyme Immunoassays (EIA)
2.5 Vitellogenin ELISA
2.6 Histological Examination
2. 7 Tissue and plasma NP concentrations
3. Results
3.] Exposure
3.2 Plasma Estradiol
3.3 Plasma Testosterone
3.4 Plasma Vitellogenin
3.5 Histopathology
3.6 Tissue and Plasma NP Concentrations
4. Discussion
4.1 Nonylphenol exposure
4.2 MorphologicaI/Histological Endpoints
4.3 Estrogenic Potency
4.4 Eflect of NP Exposure on Plasma Steroid Concentrations
4.5 Study Limitations
4.6 Conclusions
5. Acknowledgements
6. References

SUMMARY AND CONCLUSIONS

APPENDIX A: RELATED PUBLICATIONS

APPENDIX B: RAW DATA FROM WATERBORN E EXPOSURE OF
SEXUALLY MATURE MALE COMMON CARP (C YPRINUS CARPIO) TO
4-NONYLPHENOL

APPENDIX C: VITELLOGENIN ELISA PLASMA INTERFERENCES
EXPERIMENT

viii

140
141
143
144
145
146
147
148
148
150
152
155
158
158
160
160
162
163
166
168
169
170
171

179

184

187

204

LIST OF TABLES

Chapter 1
Table 1. Compounds implicated as xenoestrogens or antiestrogens

Table 2. Comparison of factors affecting bioavailability of estrogenic
compounds.

Table 3. Relative binding affinity (RBA) for calf estrogen receptor and
luciferase expression potency and efficacy in MCF-7-1uc cells

Chapter 2

Table 1. RLT 2.0 bioassay-derived relative potencies (RPs) for halogenated
aromatic hydrocarbons presented for two different bioassay methods. RPs
based on both EC-SOs and EC-ZOs are presented.

Table 2. RLT 2.0 bioassay-derived potency of halogenated aromatic
hydrocarbons. ECSO and ECZO estimates (pM in well) generated using flash
and glow methods are presented. Estimates are the mean of n replicates :t one
standard deviation (SD).

Table 3. Comparison of RLT 2.0 bioassay derived relative potencies (RPS) to
rainbow trout-specific RPS based on other in vitro and in vivo endpoints and
World Health Organization (WHO) toxic equivalency factors (TEFs) for fish.
A correlation matrix is presented. Correlations were calculated using only
those congeners for which point estimates of RP were available for both
columns being compared.

Table 4. Comparison of RLT 2.0 bioassay derived relative potencies (RPs) to
RPs based on in vitro bioassay with H4IIE-rat liver cells (recombinant [Inc]
and wildtype [wt]) and international toxic equivalency factors (TEFs).

Table 5. Sensitivity analysis: percent contribution to total variance in
tetrachlorodibenzo-p-dioxin (TCDD) equivalent (TEQs), calculated for three
separate samples, caused by congener specific uncertainties in RLT 2.0
relative potency (RP) estimates. Determined by Monte Carlo Simulation (n =
2000 independent trials). RLT 2.0-RPS for all congeners listed in Table 1

ix

10

13

18

79

81

86

88

92

were allowed to vary independently over empirically defined, congener
specific, frequency distributions.

Chapter 3

Table 1. Relative potency estimates for example data sets (Fig. 2).

Chapter 4

Table 1. Average water quality conditions during exposure.

Table 2. Measured concentrations of 4-nonylphenol (NP) in treatment tanks.

Table 3. Nonylphenol (NP) concentrations in tissue and pooled plasma
samples.

Appendices

Table 8.1. Results of weekly water quality monitoring over the course of the
exposure of sexually mature male carp to 4-nonylphenol (NP)

Table B.2. Blood volume collected, length, mass, gonad mass,
hepatopancreas (hp) mass, gonado-somatic index (GSI), hepato-somatic index
(H81), and duration of exposure (time d) for individual sexually mature male
carp exposed to 4-nonylphenol at the nominal concentrations indicated.

Table B.3. Concentrations of 17B-estradiol (E2) in plasma from individual
sexually mature male carp exposed to 4-nonylphenol (NP). Mean, standard
deviation (SD), and coefficient of variation (CV) across three replicate
determinations is presented. Method detection limit was 175 pg E2/ml.
Values less than 175 pg E2/ml may not be accurate.

Table B.4. Concentrations of testosterone (T) in plasma from individual
sexually mature male carp exposed to 4-nonylphenol (NP). Mean, standard
deviation (SD), and coefficient of variation (CV) across three replicate
determinations is presented. a = 1:30 dilution of plasma extract; b = 1:90
dilution of plasma extract

Table 3.5. Concentrations of Vitellogenin (VTG) in plasma from individual
sexually mature male carp exposed to 4-nonylphenol (NP). Mean across three
replicate determinations is presented. Value reported is for the dilution which
yielded a %-bound closest to 50%. The method detection limit (MDL) was
1.0 pg VTG/ml. Values less than 1.0 ug VTG/ml may not be accurate.

125

149

149

159

190

191

195

199

203

LIST OF FIGURES

Chapter 1

Fig. 1. Model depicting mechanisms for activation of the estrogen receptor
(ER) in an MCF-7 cell line stably transfected with an ER-controlled luciferase
reporter gene construct (MCF-7—luc). Refer to the text for a full description. In
brief, pathway 1 depicts an ER agonist entering a cell and interacting with
intracellular ER, which then binds to estrogen responsive elements (EREs) in
the promoter region of ER-responsive genes. Pathway 2 depicts ligand-
independent activation of the ER through a protein phosphorylation pathway.

Fig. 2. Competitive receptor binding assay. a) Competitor (compound being
tested) competes with tritiated 17-B-estradiol (3HE2) for binding to calf uterine
estrogen receptor (ER) in a test tube. b) Hydroxyapatite (HAP) is added to
complex with proteins (and bound ligands). Proteins with their bound ligands
are trapped on filter paper while the unbound ligand passes through. c) Filter
paper transferred to a scintillation vial and counted. Successful competitors
cause reduction in counts relative to a control using 3HE; only.

Fig. 3. Estrogen-responsive element-mediated induction of luciferase activity
in MCF-7-luc cells. Cells were treated with 17-B-estradiol (E2), nonylphenol,
dieldrin, endosulfan I, dieldrin plus endosulfan I (E+D 1:1 mixture), or
solvent only for 3 (1. Relative luciferase activity is expressed as a percentage
of control, with each point representing the mean of at least three replicates
(standard deviations are represented by error bars).

Fig. 4. Dose-dependent inhibition of estrogen-responsive element-mediated
induction of luciferase activity by 2,3,7,8-tetrachlorodibenzo-p-dioxin
(TCDD) in MCF-7-1uc cells. Cells were treated with E2 either alone or in
combination with TCDD for 3 (1. Relative luciferase activity is expressed as
relative light units, with each point representing the mean of at least three
replicates (standard deviations are represented by error bars).

Fig. 5. A comparison of several in vitro bioassays for estrogenic activities.
Adapted from Zacharewski (6).

Fig. 6. Generalized toxicant identification and evaluation scheme for
environmental samples using in vitro bioassay-directed fractionation and

xi

16

25

36

37

42

47

instrumental analysis.

Chapter 2

Fig 1. Examples of replicate dose-response curves generated using the RLT 76
2.0 bioassay (glow method). (a.) Replicate 2,3,7,8-tetrachlorodibenzo-p-
dioxin (TCDD) standard curves (n = 4). (b.) Replicate dose-responses for
l,2,3,7,8-pentachlorodibenzofuran (PeCDF) (n = 3). Worst-case encountered
for application of probit analysis for estimating EC-ZOs and EC-SOS. (c.)
Replicate dose-responses for 1,2,3,4,7,8-hexachlorodibenzo-p-dioxin
(HxCDD) (n = 3). (d.) Example of replicate dose-responses obtained for
mono- and di-ortho PCB congeners (n=3). No responses were significantly
different from background. Note, negative values for percentage of the mean
maximum response observed for TCDD standard curves generated on the
same day (%-TCDD-max) are due to subtraction of mean solvent control
response prior to calculation of %-TCDD-max.

Fig. 2. Frequency distributions for alewife (top), walleye (middle), and carp 91
(bottom) tetrachlorodibenzo-p-dioxin (TCDD) equivalents (TEQS) generated

by Monte Carlo Simulation (n = 2000 independent trials). RLT 2.0-relative

potencies (RPS) for all compounds listed in Table 1 were allowed to vary

independently over empirically defined, congener specific, frequency

distributions.

Fig. 3. Comparison of tetrachlorodibenzo-p-dioxin (TCDD) equivalents 95
(TEQ) estimates based on instrumental analyses of Saginaw Bay carp,

walleye, and alewife samples [22], calculated using RLT 2.0 (RLT), H4IIE-

wild type (H4IIE), or in vivo rainbow trout early life stage mortality (ELSM)

relative potencies or international toxic equivalency factors (INT) (Tables

3,4). TEQ estimates shown here are based on concentrations of PCBs 77 and

126, and the dioxin and furan congeners listed in Table 1.

Chapter 3

Fig. l. A. Illustration of relative potency estimation for dose response 108
relationships which conform to the assumptions of indirect bioassay (i.e.

equal efficacy and parallel slopes). Relative potency is constant over the

effective response range. B. Illustration of relative potency estimation for

dose-response relationships which do not conform to the assumption of

parallel slopes. Relative potency varies over the effective response range.

xii

Fig. 2. Systematic framework to guide derivation and application of relative
potency (RP) estimates from in vitro bioassay data using multiple point
estimates (MPE). RP band = the range of relative potency values derived
over a standard range of responses (20-80%—std.-max.). %-std.-max. =
sample response expressed as a percentage of the maximum response
magnitude achieved for the standard compound.

Fig. 3. Example data sets. (a.) data set 1 — H4IIE-luc responses to 2,3,7,8-
tetrachlorodibenzo-p—dioxin (TCDD), 1,2,3,7,8-pentachlorodibenzofiiran
(PCDF), l,2,3,4,7,8-hexachlorodibenzofi1ran (HxCDF), and 2,3,7,8-
tetrachlorodibenzofuran (TCDF). (b.) data set 11 - H4IIE-luc responses to
extracts of sediment from Masan Bay, Korea and corresponding TCDD
standard curves. (c.) data set 111 - RLT 2.0 responses to extracts of sediment
collected from a superfund site contaminated with Aroclor 1268 and
corresponding TCDD standard curves. Response magnitudes presented as a
percentage of the mean maximum response observed for the corresponding
TCDD standard (%-TCDD-max.)

Fig. 4. Relative potency (RP) bands for three example data sets. RP-20, RP-
50, and RP-8O refer to RPS calculated as a ratio of potency estimates
(Equation 1) where the defined level of response (Yi) was 20-, 50-, and 80%-
TCDD-max. respectively. RP-max refers to the RP calculated at Y, = the
maximum magnitude of response observed for the sample expressed as %-
TCDD-max. Bars indicate regions of the RP band within the range of

empirical data. A line extending beyond the bar indicates the region of the RP

band which is based on extrapolation beyond the range of the empirical data.

Chapter 4

Fig. 1. Accuracy and parallelism tests for l7B-estradiol radioimmunoassay.
Accuracy test was determined over the range of 20 to 80%-bound. The slope
of the accuracy test plot was not significantly different from 1.0 (p<0.749).

The parallelism test was conducted using five serial dilutions of three separate
plasma extracts (P-1, P-2, P-3). A dilution factor of 1.0 corresponds to 200 pl

of plasma extract or 800 pg of 17B-estradiol standard per sample tube.

Fig. 2. Plasma estradiol (E2) concentrations, measured by
radioimmunoassay, for sexually mature male carp exposed to 4-nonylphenol
(NP). X-axis = nominal exposure concentrations in ug NP/L. SC = solvent
control exposure. C= control exposure. Circles represent the values
measured for individual fish. Columns represent the mean concentration for
each treatment group. Error bars represent 1 standard deviation above and

xiii

115

119

124

151

153

below the mean. The method detection limit (175 pg E2/ml plasma) is
represented by a horizontal line.

Fig. 3. Accuracy and parallelism tests for testosterone enzyme immunoassay. 154
Accuracy was determined over the range of 20-80%-bound. The slope of the

accuracy test plot was not significantly different from 1.0 (p<0.969). The

parallelism test was conducted using 8 serial dilutions of a pooled plasma

sample (n=3). A dilution factor of 1.0 corresponds to 50 pl of plasma extract

or 250 pg/ml of testosterone standard per test well.

Fig. 4. Plasma testosterone (T) concentrations, measured by enzyme 156
immunoassay, for sexually mature male carp exposed to 4-nonylphenol (NP).

X-axis = nominal exposure concentrations in pg NP/L. SC = solvent control

exposure. C= control exposure. Circles represent the values measured for

individual fish. Columns represent the mean concentration for each treatment

group. Error bars represent 1 standard deviation above and below the mean.

Fig. 5. Plasma Vitellogenin (VTG) concentrations, measured by ELISA, for 157
sexually mature male carp exposed to 4-nonylphenol (NP). X-axis = nominal

exposure concentrations in pg/L. SC = solvent control exposure. C= control

exposure. Circles represent the values measured for individual fish. Columns

represent the mean concentration for each treatment group. Error bars

represent 1 standard deviation above and below the mean. The method

detection limit (1.0 pg/ml plasma) is represented by a horizontal line.

Appendices

Fig. B.1. Temperature profiles over the course of the exposure of sexually 188
mature male carp to 4-nony1phenol (NP)

Fig. C.l. VTG ELISA results for plasma samples (C9 and H8) diluted in 206
buffer containing 268 ng VTG/ml. Mean of three replicate determinations i

one standard deviation is presented. Plots A and C depict responses expressed

as %-bound for dilutions of samples C9 and H8, respectively. Plots B and D

depict responses expressed as final calculated concentration of VTG for

dilutions of samples C9 and H8, respectively. Dilution refers to the ratio of

whole plasma to the total volume of dilute plasma sample (ex. 0.08 = 1 pl

plasma/ 12.5 p1 dilute sample).

xiv

INTRODUCTION

There are currently over 100,000 chemicals in commerce (Trapp and Matthies
1998). Over 1000 of these have annual productions exceeding 1000 tons (Trapp and
Matthies 1998). Hundreds of new chemicals enter the market each year (Zeeman et al.
1995, Trapp and Matthies 1998). Of these chemicals, a portion are released to the
environment either directly (i.e. pesticides) or indirectly as byproducts of their use,
accidents, etc. (i.e. automobile emissions, industrial effluents, oil spills, etc.). Once in the
environment additional compounds may be formed as environmental processes such as
photolysis, hydrolysis, oxidation, reduction, and biotransformation, transform parent
compounds. Some compounds released into or formed in the environment have the
potential to cause adverse effects in exposed organisms. In some cases, obvious adverse
effects, such as acute toxicity, may occur. In other cases, effects may be subtle, but just
as detrimental to the long term survival of exposed populations of organisms and the
function of the ecosystems they are part of (Carson 1962). Eggshell thinning in birds
exposed to DDT was a classic example (Carson 1962, Bowerman et al. 1995). In order to
help avoid environmental tragedies, there is clearly a need to develop rapid, cost
effective, and reliable methods to screen chemicals for their potential to elicit adverse
biological effects, whether obvious or subtle. Furthermore, there is need to monitor and
characterize the potential hazards posed by chemicals in the forms and complex mixtures
in which they are found in the environment.

In vitro bioassays are ideally suited to this role. Field and laboratory studies with

whole animals are critical for linking exposure to biologically relevant effects. They are,

however, impractical for routine, high throughput, screening of individual compounds or
environmental samples. Procedurally, in vitro bioassays are typically performed more
quickly and at significantly less cost than in vivo studies. Use of simplified biological
systems circumvents much of the inter-individual, seasonal, and temporal variability
which can confound interpretation of in vivo responses. Additionally, in vitro bioassays
avoid many of the complex socio-political and ethical issues associated with whole
animal studies (Stokes and Marafante 1998). Thus, although they are not a replacement
for in vivo studies, in vitro bioassays are more amenable for routine use than in vivo
assays.

In vitro bioassays also overcome some of the key limitations of instrumental
analysis. While instrumental analyses are essential for the identification and quantitation
of compounds in a sample, they provide no information regarding the capacity and
potency of those compounds to elicit a biological effect. In vitro bioassays measure
mechanistically-based biological responses. As a result, they can provide information
regarding the biological relevance of a sample. In vitro bioassays provide an integrated
measure of the combined potency of all compounds in a sample. This is a practical
advantage, since instrumental analysis of complex mixtures can be very expensive,
difficult, and time consuming. Additionally, from a functional stand-point, it means that
in vitro bioassays can account for both compounds for which there are no analytical
methods available, and potential additive and non-additive interactions between
compounds. When the biological response being measured is highly sensitive, in vitro
bioassays can also account for compounds which can exert a biological effect at

concentrations below analytical detection limits. As a result, in vitro bioassays can serve

as simple, rapid, and sensitive tools for detecting the presence and mutual interactions of
chemicals which function through a Specific mode of action. In vitro bioassays applied in
an iterative fashion along with successive chemical fractionation and instrumental
analysis can be used to identify putative causative agents in complex mixtures. Thus, in
vitro bioassays provide information which can complement instrumental analyses.

In recent years, a number of factors have spurred the development of in vitro
bioassays as tools for addressing questions in environmental toxicology. Developments
in molecular and cellular biology have led to significant advances in the understanding of
molecular and cellular mechanisms of toxicity (Stokes and Marafante 1998). These have
been paralleled by technological advances in tissue culture, genetic engineering, and
automated testing equipment (Stokes and Marafante 1998). The development of in vitro
bioassays has also been stimulated by public concern about the use of animals for testing
(Stokes and Marafante 1998). In 1993, the US. National Institute of Environmental
Health Sciences was directed, by law, to develop alternative methods that can reduce or
eliminate the use of animals in acute and chronic safety testing (U .8. Code 1993). Thus,
there appears to be an important and increasing role for in vitro bioassays to play in the
field of environmental toxicology.

This dissertation presents research which centered around the application of
mechanism-specific in vitro bioassays to address questions in environmental toxicology.
In particular, it focuses on in vitro bioassays developed for the assessment of dioxin-like
and estrogenic effects.

Chapter 1 provides an overview of the potential uses of in vitro bioassays in

environmental toxicology, using the issue of environmental estrogens as an example. A

mechanistic basis for the deve10pment of in vitro bioassays to characterize estrogenic
effects is presented. The advantages and disadvantages of a variety of in vitro testing
methodologies are discussed. Experimental results presented in chapter 1 provide
examples of the application of in vitro bioassays to screen chemicals for their potential to
elicit a mechanism-Specific biological response, rank them based on their potency to
produce that response, and generate hypotheses regarding their potential mechanisms of
action. Furthermore, estrogenic potencies reported in chapter 1 served as the basis for
mass balance analyses conducted in later studies aimed at the identification and
characterization of xenoestrogens in various environmental samples (Appendix A).

Chapter 2 presents methods for assessing dioxin-like potency using recombinant

rainbow trout hepatoma cells (RLT 2.0). It addresses several important issues in in vitro
bioassay development and application. Assay specific relative potencies, needed for mass
balance analyses, were derived and reported for a number of important dioxin, furan, and
polychlorinated biphenyl (PCB) congeners. The issue of predictive relevance was
addressed both in terms of species specificity and calibration of in vitro and in vivo
potencies by comparing RLT 2.0-derived relative potencies to other fish and mammalian
bioassay-derived values. Finally, the relative potencies derived were used to calculate
2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQ) for representative environmental
samples and a sensitivity analysis was performed to evaluate the range of uncertainty in
the TEQ value resulting from uncertainties in the congener specific relative potency
estimates. Such analyses are important for mass balance applications in which a
researcher must determine what magnitude of difference between bioassay derived toxic

equivalents and instrumentally based equivalents is significant.

Chapter 3 presents a standardized procedure for deriving and interpreting relative
potency estimates based on in vitro bioassay results. Relative potency estimates are
widely used in environmental toxicology for ranking and comparing the potencies of
chemicals and environmental samples. Currently, however, there is no consensus method
for the derivation of relative potency estimates. Furthermore, such estimates are based on
a number of assumptions which are ofien both violated and ignored. This chapter
discusses the assumptions involved, the uncertainties that are generated when
assumptions are violated, and proposes methods for addressing these issues when
deriving estimates and presenting results.

Chapter 4 revisits the issue of environmental estrogens. Although the research
presented is primarily in vivo in nature, it is closely tied to the application of in vitro
bioassay methods in environmental toxicology. In vitro bioassays utilize vastly
simplified biological systems. In order to be useful, in vitro bioassay results must be
rigorously calibrated with in vivo endpoints of toxicological relevance and be shown to
be predictive. The extent to which an in vitro assay can accurately model or predict the
in vivo potency of a chemical or environmental sample is directly dependent on the
relevance of the mechanism of action modeled for mediating the in vivo effects. For
example, an in vitro assay which examines the ability of a compound or sample to elicit
estrogen response element (ERE)-mediated gene expression would not detect compounds
which exert estrogenic responses in vivo by acting as an anti-androgen, or by altering
endogenous estradiol levels through effects on the hypothalamus or pituitary. The study
presented in chapter 4 was designed to investigate whether 4-nonylphenol was able to

elicit estrogenic effects in fish through an indirect mechanism which nearly all current in

vitro assays for estrogenic activity would be unable to model. Similar studies, aimed at
establishing the predictive relevance of in vitro bioassay responses are critical for
enhancing and increasing the utility of in vitro assays in environmental research,
monitoring, and risk assessment.

Studies presented in this dissertation, and referenced in Appendix A provide
examples of the variety of applications for in vitro bioassays in the field of environmental
toxicology. Additionally, they demonstrate the type of research and development
required to fully develop their potential as research, monitoring, and risk assessment
tools. Although it does not provide all the answers or address all the issues, the research
presented in this dissertation should help facilitate and enhance the applicability exisiting
in vitro bioassay methods and the interpretation of their results for addressing questions

in environmental toxicology.

References

Blaauboer BJ, Balls M, Barratt M, Casati S, Coeke S, Mohamed MK, Moore J, Rall D,
Smith KR, Tennant R, Schwetz BA, Stokes WS, Younes M. 1998. 13th meeting of the
scientific group on methodologies for the safety evaluation of chemicals (SGOMSEC):

alternative testing methodologies and conceptual issues. Environ Health Perspect (Suppl

2):4l3-418.

Bowerman WW, Giesy JP, Best DA, Kramer VJ. 1995. A review of factors affecting
productivity of bald eagles in the Great-Lakes region —- implications for recovery.

Environ. Health Perspect. 103(suppl. 4):51-59.

Carson R. 1962. Silent Spring. Houghton Mifflin, Boston, MA, USA.

Stokes WS, Marafante E. 1998. Introduction and summary of the 13th meeting of the
scientific group on methodologies for safety evaluation of chemicals (SGOMSEC):

alternative testing methodologies. Environ Health Perspect (Suppl 2):405-412.

Trapp S, Matthies M. 1998. Chemodynamics and Environmental Modeling An

Introduction. Springer, New York, NY, USA.

United States Code. NIH/National Institutes of Health Revitalization Act. Public Law

103-43. 42 USC. Washington: US. Government Printing Office, 1993.

Zeeman M, Auer CM, Clements RG, Nabholtz JV, Boething RS. 1995. US. EPA
regulatory perspectives on the use of QSAR for new and existing chemicals SAR QSAR.

Environ. Res. 3: 179-201 .

Chapter 1

INTERACTIONS BETWEEN ENVIRONMENTAL XENOBIOTICS AND ESTROGEN

RECEPTOR-MEDIATED RESPONSES

Daniel L. Villeneuve‘, Alan L. Blankenship, John P. Giesy.

Dept. of Zoology, Pesticide Research Center, and Institute for Environmental

Toxicology, Michigan State University.

Published In: T oxicant-Receptor Interactions. Denison MS, and Helferich W.G., eds.

Taylor and Francis, Philadelphia, PA, USA, pp. 69-99.

INTRODUCTION

Environmental Estrogens

Endocrine "disruption" by environmental contaminants has become a cause for
concern among scientists, environmental advocates and politicians alike (1-3). A number
of compounds released into the environment by human activities can modulate endogenous
hormone activities and have been termed endocrine-disrupting compounds (EDCs) (3-5). It
has been hypothesized that such compounds may elicit a variety of adverse effects in both
humans and wildlife including promotion of horrnone-dependent cancers, reproductive
tract disorders, and a reduction in reproductive fitness (6).

The neuroendocrine system is a primary mechanism by which organisms maintain
homeostasis (7). Thus, generalized adaptations to stress, which allow organisms to resist
perturbations from normal homeostatic ranges, typically involve a variety of endocrine and
physiological responses (8). As a result, any stressor could be loosely defined as an
“endocrine disruptor”. Furthermore, there are a number of receptor-mediated hormonal
responses to chemicals. These include xenobiotic effects on thyroid hormone receptor (9),
epidermal grth factor (EGF) receptor (10), aryl hydrocarbon receptor (AhR), as well as
the estrogen and androgen receptor (ER and AR) mediated mechanisms. In this chapter, we
will restrict our discussion to direct-acting estrogenic and antiestrogenic compounds. These
will be defined as those compounds that bind competitively to the estrogen receptor (ER)
and cause or inhibit estrogen-like responses in vitro or in vivo. Although our discussion
focuses on estrogen agonists and antagonists, compounds that can cause tissue-level

responses without ER binding are also covered.

Table 1. Compounds implicated as xenoestrogens or antiestrogens

 

 

Compounds Reference
Organochlorine Pesticides
DDT/DDE 32,33
toxaphene 34
dieldrin 34
chlorodecone 35
Polychlorinated biphenyls (PCBS) 36-38
Polycyclic aromatic hydrocarbons (PAHs) 39
Polychlorinated dibenzo dioxins (PCDDS) 40,41
Plasticizers (e.g. Bisphenol A) 42
Phthalates 43,44
Surfactants (e. g.alkylphenol ethoxylates and alkylphenols) 44,45
Synthetic steroids (e.g. DES; ethynyl estradiol) 47
Phytoestrogens (e.g. genistein, coumestrol, etc.) 11-14

 

There are a wide variety of compounds in the environment that have been shown to
bind to the ER and function as either agonists or antagonists. These include both natural
products (1 1-14) and synthetic compounds (3,4). The synthetic compounds include both
chlorinated and non-chlorinated compounds (3,4). Some act as xenoestrogens which either
mimic or antagonize the effects of endogenous estrogen (Table 1). Others act as androgens
in the case of tributyltin (TBT) or as anti-androgens in the case of the fungicide vinclozolin
(15) and 2,2-bis(p-chlorophenyl)-1,1 dichloroethane (p, p’-DDE) (16). Some compounds
such as 2,3,7,8-tetmchlorodibenzo-p-dioxin (TCDD) can modulate a number of hormone
firnctions, by acting as both estrogens and anti-androgens (17), as well as affecting thyroid
hormone function (18,19) epidermal growth factor (EGF) (10), insulin/insulin-like growth
factor-I (20), transforming growth factor-or (21) or other Signal transduction pathways
including protein kinases (22-26). Thus, it can be seen that a wide variety of types of

xenobiotics can exert endocrine modulating effects through many different mechanisms

10

including altered steroid receptor function, estrogen-androgen ratios, and changes in
concentrations of hormones in specific tissues.

Of all the endocrine modulating compounds, those that are direct-acting estrogen
receptor (ER) agonists (xenoestrogens or estrogen mimics), direct acting ER antagonists
(antiestrogens), or androgen antagonists have received the greatest attention (27,28). This
is due to their importance in embryonic development (29). It also reflects the fact that some
xenobiotics that have been released into the environment seem to act through mechanisms
that affect the expression of sex steroid hormones (30,31). Examples of some compounds
that have received attention as potential xenoestrogens or antiestrogens are presented (Table
1). The “estrogen hypothesis” stemmed from the observation that some of the effects
observed in oviparous wildlife exposed to persistent and bioaccumulative chemicals were
similar to those that could be caused by injecting estrogen into eggs. This hypothesis was
supported by the fact that naturally occurring exoestrogens, such as phytoestrogens, could
cause reproductive dysfunction in animals (47,48). Further support came from the
observation that some xenobiotics which can bind the ER were weak estrogen agonists or
antagonists in in vitro expression assays (6,27).

Estrogen agonists are compounds that mimic the effects of estrogen. The classic
definition of an estrogen is a compound that produces changes in an estrogen-responsive
tissue, such as cornification of vaginal epithelium (49). Other physiologic endpoints have
been used to define estrogenicity including increased uterine weight, uterine glycogen
content, protein induction (50) and cellular proliferation (51). It has been recognized that
numerous synthetic as well as naturally-occurring compounds fit the classic definition of an

estrogen (52). A refined definition of an estrogen, recognizing the role of nuclear hormone

ll

receptors in gene expression, is a compound that binds to the estrogen receptor, induces
dirnerization of the receptors that specifically bind to and activate transcription of genes
under the regulatory control of trans-acting estrogen responsive elements (53).

Estrogen antagonists block the action of estrogens by interfering with the normal
functioning of the estrogen receptor. Anti-estrogenic compounds act by several
mechanisms, not necessarily related to estrogen receptor binding and activation. Classical
estrogen antagonists such as ICI 164,3 84 and tamoxifen bind competitively to the estrogen
receptor, displacing the natural ligand E2, and blocking or reducing the effectiveness of the
ligand-bound receptor to enhance gene expression (54-56). Aryl hydrocarbon receptor
agonists, such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) and non-ortho-
chlorinated PCBS, cause down regulation of estrogen receptor and may also interfere with
DNA binding (6). Aromatase inhibitors, such as aminoglutethimide, block conversion of
testosterone to E2 and are used in the treatment of metastatic estrogen-dependent breast
cancer (57). Inducers of Phase I and Phase II metabolic detoxification enzymes, such as
2,3,7,8-TCDD and non-ortho-chlorinated PCBS, reduce the estrogen-dependent expression
of proteins by enhancing the enzymatic conversion of latent E2, thus eliciting an anti-
estrogenic response (37).

Another important factor in determining a compound’s ability to modulate ER
function is bioavailability. Three of the most important characteristics for determining
bioavailability of ER ligands are lipid solubility, biological half-life, and amount of protein
binding. Comparisons of these properties for natural estrogens, synthetic estrogens,
phytoestrogens, and o,p ’-DDT are presented in Table 2. Some researchers have suggested

that xenoestrogens have greater bioavailability than E2, due mostly to differences in protein

12

 

 

 

I ewegs £53 I a: .3:

 

32 .3688 fig Ease: 3 En 325.5
I I 9.: 3:
mm 9 339:8
£53 325
3:22 .x. R a: 3: a S v 52250
323
I babe 32 5638 I
:3; am 55 bags so 3: 6mm:
seem 28-8 9 2 5% 2o 3: F_ a 358953505
a. as .x. S _ 3.3 tongue an
.x. NM: a. :8 so? .823
s. a: .\. as .322 ESEE
so 3: 5:53
an .3: 3:
.3 N3 .x. on: $92 so 3: 82 an 3: a we a; 8633-:
a as: me:
2:85 Essa 82% 5.320 32$ £53: $528 98888
“50:2 .050 :5 5532 08.80: 223m 3335 was

 

memefim 528m :0 “SEE.“

 

63:38.58 3:63.39? basemgeofi Manuel? E885? zoatamfiob .N 03:.

13

binding and biological half-lives (66,67). It is important to note, however, that this
hypothesis has yet to be tested in vivo.

It has been suggested that relative to E2, o,p '-DDT, a weak estrogen agonist, may be
more active in vivo since E2 binds to steroid hormone binding globulins (SHBGS). Because
it does not bind to SHBGS o,p’-DDT is thought to be free in plasma and thus more
bioavailable. Except in the third trimester of pregnancy, however, less than 40% of E2 is
bound to SHBGs (Table 2). Most E2 is bound to albumin and other serum proteins, with
only about 2% being available as free E2 (Table 2). While exact values for o,p’-DDT
distribution in blood are not available, there is a high capacity for binding to serum proteins
(48). Furthermore, DDT is more lipophilic than E2. The octanol-water partitioning
coefficient (Kow) for DDT is over 300,000 times that of E2 (58,61). Thus, DDT is likely to
be bound as tightly as E2, and may, in fact, be less available. Assuming differences on the
order of two orders of magnitude are significant for risk assessment, efforts should be made

to determine the binding characteristics of xenobiotics relative to E2.

Description of the ER

The estrogen receptor (ER) is an intracellular receptor belonging to the steroid
hormone receptor superfamily (68,69). It functions as a ligand-activated transcription
factor mediating the effects of estrogens, which regulate the growth, differentiation, and
functioning of diverse target tissues. To date, two isoforms, ER-a and ER-B, have been
described (69,70). ER-a has been well-characterized and is a highly conserved protein of

approximately 66 kDa (69). ER-B has only recently been described and has a calculated

14

molecular weight of 54.2 kDa (70). The ligand binding Specificities are similar between the
two isoforms, however, there are differences in the distribution and relative binding
affinities which could contribute to selective actions of ER agonists and antagonists in

different tissues (70).

Mechanism of Action For Estrogen Receptor Agonists/Antagonists

A simplified mechanism of action for activation of the ER is shown in figure 1.
First, an estrogen receptor ligand must enter the cell and bind to the ER. The non-ligand
bound form of the ER is predominantly localized to the nucleus (71) and is part of a
complex with associated proteins (72). Upon ligand binding, the associated proteins
dissociate, and then ligand-receptor complexes associate with additional nuclear factors
(73) and bind to estrogen responsive elements (ERES) as homodimers. Binding of the
transformed complex to the ERE results in production of mRNA for a number of estrogen-
responsive proteins, such as PS-2, and cathepsin D (6).

In addition to endogenous responses, exogenous reporter systems, such as firefly
luciferase under control of ERES, can be stably transfected into cells such that upon
exposure to estrogens, luciferase expression is induced (74). Luciferase or other such
reporter systems provide a rapid and convenient means for receptor-mediated dose-
response assessment as well as allowing for mechanistic investigations into transcriptional
activation.

There is considerable evidence for hyperphosphorylation of the ER after ligand
binding, supporting the hypothesis that phosphorylation is an important regulatory

mechanism for ER function. Phosphorylation of key serines (75-77) and at least one

15

 

 

 

Estrogen _______
orxegrestrogen/ 1 V" """" \
4 + In
VVV {

 

 

\ >% 7?? Non-genomic ® ® I Nuclear Factors

effects / v
'1 /
.’
O
I
.' r V l.’ 'C.' o.‘
. .0. . or 0

 

 

 

Protein Phosphorylation of ER?‘.\ DNA Binding I."
Peptide Growth Ligand-Independent Activation

Factor
- ““~ 1
\E l/ ---------------

/ Luciferase <
/”
K‘Eflmgemc Effects” “—— ER-Responsive Genes /

 

 

 

 

 

 

 

 

 

 

Fig. 1. Model depicting mechanisms for activation of the estrogen receptor (ER) in an
MCF-7 cell line stably transfected with an ER-controlled luciferase reporter gene construct
(MCF-7-luc). Refer to the text for a fiJll description. In brief, pathway 1 depicts an ER
agonist entering a cell and interacting with intracellular ER, which then binds to estrogen
responsive elements (ERES) in the promoter region of ER-responsive genes. Pathway 2
depicts ligand-independent activation of the ER through a protein phosphorylation
pathway.

16

tyrosine residue of the ER, Tyr 527 (78), play an important role in regulating the
transcriptional activity of the ER. Phosphorylation increases the negative charge and
acidity of a region of a protein, thereby modifying interactions with other proteins and
DNA. Hypo- and hyper-phosphorylation at the same time in different regions of the ER
could potentially explain differential transcriptional regulation of certain genes, in addition
to tissue- and cell-specific regulatory mechanisms (79). In addition, differential
phosphorylation of other nuclear factors with which the ER interacts to mediate
transcription potentially play a significant role (80).

There are also ligand-independent pathways for modulating the transcriptional
activity of the ER through activation of protein kinases (Fig. 1, pathway #2). For example,
upon treatment with growth factors, such as epidermal growth factor, insulin-like grth
factor-1, and platelet-derived growth factor, or agents that increase cAMP levels, the ER
can be phosphorylated, bind to ERES, and activate transcription in the absence of ligand
(81-83). The fact that the pure antiestrogen ICI 164,384 blocks the effects of these agents
supports an ER-mediated mechanism (83). Likewise, protein kinase inhibitors block the
effects of these agents and E2 (69). 1C1 164,384 also stimulates phosphorylation of the ER
without a Similar increase in transcriptional activation. This indicates that overall
phosphorylation does not necessarily result in increased transcriptional activity (69)
although this could be due to impaired receptor dimerization by ICI 164,384 (55). Given
the fact that some xenobiotics can modulate protein phosphorylation (26,84,85), this ligand-
independent mechanism for modulating ER function has a potentially important impact on
screening for chemicals with estrogenic activity which will be discussed later, especially if

the emphasis is placed on receptor binding alone as a first tier screen. For example, a

17

TABLE 3. Relative binding affinity (RBA) for calf estrogen receptor and luciferase
expression potency and eflicacy in MCF- 7-luc cells

 

 

 

MCF-7-luc
Compound RBA ICso Potency Efficacy (fold induction
(nM) (EDSO nM) relative to solvent controls)
17B-Estradiol 6.9 0.00625 4.5
Coumestrol 24 23 15 2.8
17B-Ethynyl-estradiol 37.3 0.0555 4.5
Nonyl-phenol 2300 8.56 1 .9
Octyl-phenol 12300 16 2.3
Bisphenol A 24300 3640 4.3
Indole-3-Carbinol 2.3x10lo 21 100 2.3
Atrazine >2.3x10‘° 2470 2.8
o,p’-DDE >2.3x1010 NE _
p,p’-DDE >2.3x10lo NE _
_2,3,7,8-TCDD >2.3x10lo NE _

 

ICso-concentration of competitor to reduce total binding control counts by 50%;
EDso-dose to elicit 50% of maximal effect. NE, potency estimate was not possible based on
the dose-response obtained.

18

compound like 2,3,7,8-tetrachlorodibenzo-p-dioxin, which is a potent antiestrogen (86),
would be missed in an initial screen because it does not bind to the ER (87). Similarly, the
triazine herbicide atrazine, which does not bind to the ER, can cause estrogen-like
responses in in vitro expression assays (Table 3). Neither of these in vivo or in vitro effects

would be predicted from receptor binding assays.

Rapid, Non-genomic Effects of Estrogens

In addition to the mechanisms described above, estrogens produce rapid (within
seconds to minutes), non-transcriptional responses which are similar to those evoked by
growth factors. Recent evidence suggests the existence of a membrane ER (88) which may
play a role in these rapid, non-genomic effects of estrogens, which include prolactin release
in GH3/B6 rat pituitary tumor cells (88), intracellular calcium release in chicken granuloma
cells (89), stimulation of protein tyrosine phosphorylation in MCF-7 mammary carcinoma
cells (90) activation of the p21m/MAP kinase pathway in MCF-7 cells (91) and stimulation

of adenylate cyclase and CAMP-regulated gene transcription (75).

Interactions and Cross-talk from other Signaling Pathways

“Cross-talk” between the estrogen receptor and other signaling pathways provide
important mechanisms for modulating biological responses. Interactions and “cross-talk”
with the ER have been described for the progesterone receptor (69), aryl hydrocarbon
(Ah) receptor (92), epidermal grth factor receptor (81), insulin-like grth factor I

(69), and as discussed above, pathways involving agents affecting protein kinase

19

activities, particularly those affecting protein kinase C (PKC), CAMP-dependent protein
kinase (PKA), and tyrosine kinases (81,83).

As interest in monitoring the environment for environmental estrogens increases,
it is likely that more than one mechanism will need to be assessed. Some environmental
mixtures contain compounds that modulate the responsiveness of multiple receptor-
mediated pathways, such as the aryl hydrocarbon (Ah) receptor, estrogen receptor,
androgen receptor, epidermal grth factor receptor, etc. Thus, the complexities of
environmental mixtures require innovative methods and approaches to assess the

potential for adverse effects.

SCREENING AND MONITORING

Concern over xenoestrogens has created a need to both screen and monitor for
compounds which can modulate endocrine effects. This need is underscored by recent
legislation mandating that chemicals and formulations be screened a priori for potential
estrogenic activity before they are manufactured or used in certain processes (Safe Drinking
Water Act Amendments of 1995 - Bill Number 8.1316; Food Quality Protection Act of
1996 - Bill Number PL. 104-170). Monitoring a posteriori is needed in order to assess
concentrations of estrogenic compounds in both abiotic matrices like soil, sediments, and
water, and biotic matrices such as human and wildlife tissues an food stuffs (93,94).

Both biological and instrumental methods can be applied to monitor or screen for
estrogenic compounds. Instrumental methods are useful for measuring the uptake,
disposition, and concentrations of specific compounds (monitoring). They can also be
applied to help discern metabolic pathways. They are generally not useful, however, for

discerning biological efficacy (screening). A variety of bioassays, both in vitro and in vivo
20

can be used to screen individual compounds, formulations, or environmental samples for
potential estrogenic/antiestrogenic activity. Models may also be used to predict the
potential estrogenicity based on the physico/chemical properties of the compound of
interest.

A number of techniques for screening and monitoring for the effects of
xenoestrogens are discussed below, with most attention being focused on bioassay methods.
The advantages and disadvantages of each technique are discussed. Although this
discussion is focused on assays used to determine the estrogenicity or anti-estrogenicity of a
given compound or sample, the same type of techniques can be applied to other receptor-

mediated processes, as long as the mechanism of action is known.

Predictive Quantitative Structure Activity Relationships

The ideal screening tool is one that is rapid, inexpensive, objective, applicable for a
wide range of compounds acting through a given mechanism of action, and capable of
accurately predicting a compound’s potential to elicit (adverse) effects in vivo. One of the
most attractive screening tools is computer modeling, based on quantitative structure
activity relationships (QSAR). Once a suitable computer model is constructed, compounds
can potentially be screened for possible activity in less time than it takes to conduct even
the most simple in vitro bioassays, and at a fraction of the cost. For this reason,
development of an accurate QSAR model is certain to be a goal of any large-scale
screening program. Some of the disadvantages of such models is the difficulty in collecting

necessary input parameters that are representative of all potential ER ligands, and the

21

reliance on receptor binding assays in many of these models rather than looking at ER
fimction (to be discussed in more detail in following sections).

Structure activity relationships for estrogenic compounds have been studied
extensively (50,51,95,96). Correlations have been made between certain structural features
and both affinity for the estrogen receptor, and expression of estrogen modulated genes
(50,95,96). Evidence of stereochemical recognition and stereospecific modulation of gene
expression has been reported (50). Though estrogenic substances vary widely in structure,
some common characteristics of most estrogens include 1) a sterically unhindered phenol
group, and; 2) a hydrophobic substituent of greater than three carbons bonded para to the
phenolic hydroxyl (56,97). The aromatic A-ring with it’s free hydroxyl group has been
cited as a key for affmity of endogenous steroid hormones to the estrogen receptor (95,96).
Structure activity relationships identified have been based on both simple structural
features, and more advanced computer based models such as the electrostatic models used
by VanderKuur et al. (96); comparative molecular field analysis (CoMF A) used by Waller
et al. (98,99); and computer graphic and energy based models for fit into DNA used by
Hendry et al. (100,101).

QSAR-based computer models for predicting estrogenic and/or androgenic
potential have been developed. Two of the most promising models are discussed here.
CoMFA/3D-QSAR based models for predicting ER or AR binding affinity have been
proposed (98,99). These models consider the overall steric and electrostatic properties of
the compound of interest. Empirically derived ER binding affinity for seven classes of
potentially estrogenic compounds, both natural and synthetic were used for the construction

and validation of the model for ER binding affinity (98). Average errors of less than 2 log

22

units for predicting empirically derived ER or AR affinity based on the CoMFA/3D—QSAR
model have been reported (98,99). Such a range of error may or may not be within the
range of certainty required by a tier I screening tool.

Another QSAR approach has been proposed by Hendry et al. (100,101). This
approach is based on the hypothesis that ligand fit into DNA (or DNA complementarity) is
an important property affecting hormone-like activity in vivo (100,101). The model
developed uses both 3D computer graphics and energy (electrostatic and van der Waals)
calculations to estimate fit into DNA (100,101). Degree of fit into DNA has been found to
correlate with in vivo responses to estrogens such as uterotropic activity (101) and in vitro
responses like proliferation of MCF-7 human breast cancer cells (100). Hendry et al. found
that, without exception, compounds which fit into DNA better than E2 were more active in
vivo than E2, while those that fit more poorly exhibited less uterotropic activity in vivo
(101).

Both of the models discussed above appear to hold promise as potential tools for
screening compounds for potential estrogenic or endocrine modulating activity. QSAR
approaches could be used in a tiered screening system for direct-acting estrogenic
compounds acting as agonists or antagonists. QSAR models based on receptor binding
affinity alone, will have little or no utility in identifying estrogenic compounds acting
through non-ER mediated mechanisms. Furthermore, additional validation will be
necessary before QSAR techniques can be applied with certainty. Both models need to be
validated with a larger set of compounds. Additional in vitro and in vivo studies are needed
to establish relationships between predicted receptor or DNA fit and relevant in vivo

endpoints in multiple species. Finally, if they are to be applied, such models Should be used

23

as 3 tier I tool to narrow the range of compounds to be examined more thoroughly using in
vitro and in vivo bioassays. Regardless of their utility as screening tools, QSAR approaches

are useful in understanding of mechanisms and scaling doses in experiments.

In Vitro Assays

There are a number of types of in vitro assays available for measuring the estrogenic
activity of single compounds or complex mixtures (102). The range of responses includes
everything from simple receptor binding to expression of endogenous or exogenous genes to
cell proliferation and differentiation (52). In vitro systems are attractive as screening tools
because they are rapid, inexpensive, and fairly reproducible. For these reasons precise
estimates of the relative potency of a great many samples or compounds can be obtained in a
rather short period of time. One of the greatest utilities of in vitro assays is for studying
mechanisms of action of compounds or assessing the potential for synergisms among direct-
acting estrogen agonists. Using simple in vitro cell systems it is often possible to determine
the mechanism of action of a compound. Knowledge of a compounds mechanism of action
can reveal ways to monitor using in vivo biomarkers.

In vitro assays are limited by the fact that they do not completely represent the in
vivo situation. Pharmacokinetics, biotransformation, and binding to carrier proteins may not
be accurately represented by in vitro systems. In addition, some xenoestrogens may be
activated or deactivated by enzymatic conversion during metabolism, conjugation, and/or
excretion. While there are methods to adjust for and minimize some of these limitations,

they must be considered when applying the results of in vitro screening tests.

24

HAP added b Unbound

Cg. %\

 

 

 

     

Competitor
W ER
. 3

Fig. 2. Competitive receptor binding assay. a) Competitor (compound being tested)
competes with tritiated 17-B-estradiol (3HE2) for binding to calf uterine estrogen receptor
(ER) in a test tube. b) Hydroxyapatite (HAP) is added to complex with proteins (and bound
ligands). Proteins with their bound ligands are trapped on filter paper while the unbound
ligand passes through. c) Filter paper transferred to a scintillation vial and counted.
Successful competitors cause reduction in counts relative to a control using 3HE2 only.

25

Receptor Binding Assays

For direct-acting estrogen agonists or antagonists to exert an effect, it is necessary for
them to bind to the estrogen receptor (103). The affinity with which such compounds bind
to the estrogen receptor may be related to the potency of the compound relative to
endogenous estrogen. For this reason, receptor binding assays have been used as a method
to screen for potential estrogenic compounds (16,44). It is important to note, however, that
while binding of the ligand to the ER is necessary, it alone may not be sufficient for eliciting
estrogenic responses in tissues.

Receptor binding assays are conducted as competitive binding assays where a
compound with unknown binding affinity is allowed to compete for estrogen receptor
binding sites with a labeled standard of known affinity (Fig. 2). Operationally, a known
amount of the compound of interest is added at varying concentrations to a fixed amount of
receptor and competitor. The compound of interest is allowed to compete with radiolabelled
competitor for a fixed number of binding sites. (Fig 2 - a). The receptor-ligand complexes
are separated from suspension by filtration (Fig. 2 - b) or centrifiigation and the amount of
activity determined via liquid scintillation (Fig. 2 - c) (104). Compounds with affinity for
the estrogen receptor will “displace” or compete for more of the available binding sites,
thereby decreasing the radioactivity of the receptor-ligand complex fiaction accordingly.
Scatchard (105) and/or Woolf (106) analyses can be used to determine the concentration of
compound required to displace 50% of the endogenous compound (hormone) or it’s
synthetic analogue. Relative binding affinities can be derived for compounds of interest.
These values, generally reported as the effective concentration needed to reduce the binding

of the labeled competitor by 50% (IC50), are measures of the relative potency of the

26

compound. For instance, if the concentration of a compound required to compete for 50% of
the binding sites is 1000 times (on a molar basis) greater than the endogenous or synthetic
reference compound, it would have a relative potency 1000 fold less than the reference
compound. In theory, this information can be used to predict concentrations of competing
compounds in organisms that would be required to cause a given level of effect. Several
competitors have been used in receptor binding assays. These include endogenous ligands
such as, E2 (14,63), in the case of the ER. Alternatively, synthetic competitors such as
ethinylestradiol (EE2) or diethystilbestrol (DES) can be used. The use of synthetic
competitors is useful because they are oflen more stable and are not as likely to bind non-
specifically to endogenous non-receptor proteins.

Receptor binding assays are attractive because they are Simple and inexpensive to
conduct. The only materials necessary are a preparation of the purified receptor of interest
and an agonist of known binding affinity. In the case of the estrogen receptor (ER) the
receptor is similar among species (107). Thus, results obtained with a preparation from one
vertebrate species can generally be extrapolated to other species. However, this may not be
true for all receptor-mediated responses. ER preparations from estrogen-responsive tissues
of rodents, calf, humans and other vertebrates such as fish have been used in binding assays
(44,47,108). In addition, receptor does not necessarily need to be harvested fiom animals or
their tissues. Receptor can also be harvested from in vitro cell cultures. Transformed yeast
cells (109,110) and transfected cell lines (14) have been used to produce relatively large
quantities of uniform receptor.

Although they are attractive and simple, receptor binding assays are limited in the

amount of information they can provide. The affinity of binding to a receptor is only one

27

step in a complex series of events that occur during endocrine modulation (Fig. 1). In
addition, there needs to be transformation and translocation of the receptor followed by
binding to accessory ligands or proteins and subsequent transcription (Fig. 1). Receptor
binding assays cannot account for such steps in the expression process. In addition, a
compound may be present in sufficiently great concentrations to cause effects even though it
has a relatively weak affinity for the receptor. In vitro receptor binding assays do not
consider pharmacokinetic processes that are important determinants of exposure in vivo.
Enzymatic modification, differential turnover rates, and binding of endogenous and
exogenous ligands are often different in the in vitro and in vivo situation. Finally, from
binding affinity one can not infer whether a compound will be an agonist or antagonist. One

can only identify the potential for modulation of the receptor-mediated process.

Relationships Between In Vitro Receptor Binding and In V ivo Assays

For simple in vitro bioassays such as receptor binding to be useful for screening, it is
necessary for the responses to be correlated to physiologically-relevant responses in vivo.
Risk assessment based on receptor binding affinity would be meaningless without such
correlation. Thus, the use of receptor binding affinity as a predictive tool is predicated on the
assumption that endocrine modulation proceeds through a receptor-mediated process (111).
There may, however, be alternative signal transduction pathways. Here, the relationship
between receptor binding affinity and in vitro gene expression will be discussed.

In an attempt to elucidate the potential for receptor binding assays to predict effects
in vivo, affinity for binding to calf ueterine ER was compared to responses in a simple in

vitro gene expression assay (discussed in more detail later in this chapter). The expression

28

assay used MCF-7-luc cells which are MCF—7 human breast tumor cells (ATCC # HTB-22)
stably transfected with a DNA construct that includes an exogenous reporter gene, luciferase,
under control of estrogen responsive elements (ERES) and a mouse mammary tumor virus
(mmtv) promoter (74). The comparison was made for a range of potentially estrogenic
compounds including endogenous estrogen, synthetic estrogen, xenoestrogens and a natural
product (Table 3). The hypothesis that receptor binding affinity can be used to predict ERE
mediated gene expression in vitro was tested, to gain insight regarding the potential
predictive power of receptor binding assays. Results indicated that relative binding affinity
(RBA) for calf uterine ER was correlated with the potency for expression in the MCF-7-luc
bioassay (r2=0.711; Table 3). RBA was not, however, very predictive of eflicacy, or
magnitude of expression, (r2= -0.356). Furthermore, there was no relationship between
potency and efficacy among the compounds studied using the expression assay (r2= -0. 165).
Since the response of an organism is a complex interaction between both potency and
efficacy, these results suggest that screening of compounds or mixtures with a simple
receptor binding assay will not be very predictive of other in vitro or in vivo responses.
Results from this in vitro study correspond with reports in the literature that suggest a
lack of correlation between RBA and ligand activity in vivo (112,113). It has been stated
that “interaction with the estrogen receptor may not necessarily be an absolute, or sole
requirement for the expression of estrogenic activity” (114). It has also been reported that
there is little relationship between receptor binding of E2 analogues and either the character
or extent of response (95). Similarly, a poor relationship between receptor binding and
biological activity of non-steroidal DES analogues has been found. Researchers have

concluded that ligand structure may be important in processes other than binding to and

29

activating the ER (50). Alkaline degradation products of some steroids are potent estrogens,
even though they do not bind with great affinity to the ER (115). One of the most active
estrogen analogues developed has a receptor binding affinity of less than 1% of that of E2
(116). The triazine herbicide atrazine was found to have no measurable affinity for the ER,
yet caused a significant response in the MCF-7-luc assay (Table 3). Ligand independent
activation of the ER by growth factor signaling and protein kinase activation can occur (69).
Some potent antiestrogens bind the ER poorly or not at all (117,118). In some cases there is
a need for metabolic activation of compounds, thus receptor binding affinity of the parent
compound may not relate to in vivo activity (1 12,1 19).

In vitro and in vivo estrogenicity of o,p ’-DDT, o,p '—DDE and chlordecone, based on
vitellogenesis in fish, has been compared (33). While both in vitro (ER binding affinity) and
in vivo (plasma Vitellogenin) assays indicated that these compounds were weakly estrogenic,
the potency of the substances, based on the two assays was not correlated. ER binding
studies indicated that o,p '-DDT and o, p ’-DDE were less potent competitors for moxestrol
(synthetic estrogen) than was chlordecone. However, in vivo studies, based on plasma
Vitellogenin normalized for tissue residue levels, indicated that o,p ’-DDT and o,p ’- DDE
were more potent than chlordecone.

Together, the results of our studies and reports in the literature suggest that the
assumption upon which application of receptor binding assays is based may not be valid.
The information available suggests that ER binding may be a poor predictor of more
complex in vitro and in vivo responses. One reason for the discrepancy between receptor
binding and hormonal activity in vitro could be the multiple steps involved in expression

beyond receptor binding. These include both transcriptional and post-transcriptional events.

30

Lack of concordance between receptor binding assays and in vivo responses may also be due
to nongenomic mechanisms or influences through other pathways, which were discussed
earlier in this chapter. Slight differences in receptors or receptor expression among different
species and tissues (79,120,121) must also be considered a potential source of the
discrepancy. While the structure of the ER is well conserved among species, there are
species, tissue-specific, and even temporal differences in the affinity to estrogen agonists
(69,121).

One alternative working hypothesis that could explain the lack of correlation
between binding affinity and response is the “ligand insertion hypothesis” (122). This
hypothesis states that upon binding to both the ligand and the DNA (at the ERE), the
receptor shifts conformation in a way that facilitates insertion and release of the ligand
into the DNA where it acts as an additional transcription factor (122). Thus, the
characteristic structure of the compound and it’s fit into DNA would influence
transcriptional activity in a manner not necessarily related to it’s receptor binding affinity
(122). The results reported here do not provide a rigorous test of this hypothesis, but they
are consistent with it. Binding affinity could determine the concentration of ligand
needed to produce a response (potency) while the structure of the ligand itself would
determine its effectiveness as a transcription factor (which would have bearing on the
magnitude of response generated - efficacy). This hypothesis is only now being developed
and tested with computer modeling and simple expression assays. If it is validated it would
further support the contention that measures of receptor binding would be poor predictors of
the effects of endocrine disrupters. In light of the information reviewed here, the use of

empirically determined receptor binding affinity is unlikely to be very accurate for screening

31

and risk assessment purposes. Furthermore, QSAR relationships based solely on ER binding

may not be sufficiently accurate to be used in a tiered screening approach.

Cell proliferation and differentiation

One of the in vitro assays most frequently used to determine the relative potency for
EDCs acting through the ER is cell proliferation of estrogen responsive tissues (123). The
MCF-7 and ZR-75 cell lines are human breast cancer cells that have been used extensively
for screening of estrogenic effects (34,44,124). The primary cell line used to assess
estrogenic effects is the MCF-7 line, which was originally derived from a hormone-
dependent metastatic breast cancer (125). These cells were demonstrated to contain ER
(126) and express a number of responses that are under control of the ER. MCF-7 cells
have been used in the E—screen to screen for estrogenicity of a number of compounds
(34,51,127,128). This assay is based on estrogen (E2)-dependent proliferation of MCF—7
cells.

In the E-screen assay, MCF-7 cells are cultured in media stripped with dextran
coated charcoal so that it is E2-deficient. A test compound is interpreted as an estrogen
agonist if it Significantly increases cell proliferation relative to a control upon its addition to
the stripped media. In the original E-screen assay the response was determined by counting
cell nuclei in trypsinized cells (129). Other responses of cell proliferation such as metabolic
reduction of dirnethylthiazolyldiphenyltriazolium bromide (MMT), reduction of thiamine
blue, and incorporation of [3H]thyrnidine have also been used.

The E-screen assay has been extensively used (15,34,42,124), but it has some

limitations. Since it measures cell proliferation, the E—screen assay is, technically, a

32

mitogenicity assay. Thus, a positive response can not be strictly attributed to estrogen
agonists. Variation in media preparation and relative concentrations of E2 and other growth
factors can greatly affect the responses obtained by different laboratories. Also, because it
depends on a proliferation response, ER antagonists are not easily detected using this assay.
Furthermore, secondary types of endocrine modulators that could result in responses that are
similar to those involving E2 at the ER (such as anti-androgens) would not be measured by
this assay. For these reasons, if used alone, the E—screen assay could result in a significant
number of false negative determinations of EDCS. A positive response in the E-screen assay
should be confirmed by in vivo studies before a conclusion about environmental activity of a

compound is made.

Expression Assays

Because it has been demonstrated that, in some situations, binding of a ligand to the
ER is necessary but not sufficient for endocrine modulation while in other situations
modulation can occur without significant ER binding, more complex in vitro assays have
been developed. Knowledge of ER binding characteristics is only part of the information
necessary to interpret the potential endocrine modulatory effects of compounds. A
compound that binds with high affinity may be either an agonist or an antagonist.
Furthermore, proliferation is a potentially nonspecific response. For this reason a series of in
vitro receptor-dependent transcriptional expression assays have been developed. Those
designed to measure potency of estrogenic, antiestrogenic, androgenic, and anti-androgenic
compounds are the best developed. A number of these types of systems have been

developed for ER-dependent responses.

33

MCF- 7-luc Bioassay Method

Expression assays examine induction or suppression of proteins encoded by genes
whose transcription is thought to be modulated through an ER mediated mechanism.
Increases or decreases in the activity of the protein of interest upon exposure to a single
compound or complex mixture, such as an environmental extract, suggest the presence of
one or more ligands with the potential to modulate a broad range of genomically controlled
estrogenic responses. The MCF-7-luc bioassay method (see discussion of receptor binding,
this chapter) is detailed here as a prototypic expression assay (Fig. 1). MCF-7-1uc cells are
seeded into 96 well microplates. Compounds or environmental samples are then delivered to
the plates. Active compounds enter the cell where they affect the transcription of ERE
regulated genes, presumably through mechanisms discussed earlier in this chapter. Estrogen
agonists cause upregulation of transcription of the luciferase reporter gene. Since there is
sufficient E2 in the medium to allow constitutive luciferase expression, the assay can also be
used to detect antagonists (103). Luciferase mRNA is produced and translated at the
ribosomes into the luciferase enzyme. The activity of this enzyme is then measured by
adding the exogenous substrate luciferin which interacts with luciferase in a light producing
reaction. The magnitude of light production, measured by a luminometer, serves as a gauge
of reporter gene expression and thus a gauge of the estrogenic potential of the sample.
Potency of the sample can be reported in terms of the concentration of test substance needed
to yield a significant amount of light (LOEC) or concentration needed to yield 50% of
maximal light production (EC-50). The maximum amount of light produced provides a

measure of the compound’s efficacy. This means that relative potencies of both agonists and

34

antagonists can be determined singly or as a net potency for a complex mixture of both
agonists and antagonists. Though there are other expression assays (6), the MCF-7-luc
method illustrates many of the features common to expression assays (i. e. examines changes
in expression of a gene whose activity is modulated by a known mechanism of action upon
exposure to a ligand). Most of the differences among expression assays are the result of

differences in the particular cells and/or reporter gene/protein, used in the assay system.

Applications of the MCF- 7-luc Bioassay (a Model Expression Assay)

The MCF-7-luc bioassay has been successfully utilized to screen for both
estrogen agonists and antagonists, assess interactions within mixtures, conduct
mechanistic studies, and perform environmental monitoring of a wide variety of biotic
and abiotic matrices. Recently, there has been controversy over the potential for
mixtures of weakly estrogenic pesticides to act in a synergistic (130) or additive manner
(131). When tested in the MCF-7-luc, dieldrin, endosulfan I, and a 1:1 mixture of
endosulfan I and dieldrin (E+D) all displayed similar dose-response curves (Fig. 3).
Potencies for individual compounds and the binary mixture were not significantly
different and were approximately a million-fold less potent than E2. Thus, synergism was
not observed between endosulfan I and dieldrin in MCF-7-luc cells. In addition, it is

important to point out that the magnitude of response (efficacy) elicited by these

35

    
   
   

--E§
300 2_ -D-Nonylphenol fl--- , ,_ ,
+Dieldrin

-°- Endosulfan 1
-0-IE+D 11

 

250 ‘I.

150 ‘-

100 ..

 

Relative luciferase activity (% of control)

 

0 "—ffi-fimf—Wmé—mf—wwmd—WI—WM—fl-mfl—a—fiml—rW—W
10'” 10"2 10'“ 10‘“) 10'9 10“ 10‘7 10° 10" 10" 10‘3

Concentration (M)

Fig. 3. Estrogen-responsive element-mediated induction of luciferase activity in MCF-7-
luc cells. Cells were treated with 17-[3-estradiol (E2), nonylphenol, dieldrin, endosulfan I,
dieldrin plus endosulfan I (E+D 1:1 mixture), or solvent only for 3 (1. Relative luciferase
activity is expressed as a percentage of control, with each point representing the mean of
at least three replicates (standard deviations are represented by error bars).

36

0.07

  
     

 

 

I 0 E2 alone
0'06 ‘I— - plus 30 pg TCDD
A 005 . plus 10 pg TCDD
D —‘I-
Q . plus 3 pg TCDD
;’ 0.04
'E
D
E 0.03
.99
._l
é’ 0.02
E
D
9‘ 0.01
0 l l l l l
I I I II IIIII I IjIIITII I I TIIIIIT I I I IIIIII I I IIIIIII
0.01 0.1 l 10 100 1000

E 2 concentration (pM)

Fig. 4. Dose-dependent inhibition of estrogen-responsive element-mediated induction of
luciferase activity by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in MCF-7-luc cells.
Cells were treated with E2 either alone or in combination with TCDD for 3 (1. Relative
luciferase activity is expressed as relative light units, with each point representing the
mean of at least three replicates (standard deviations are represented by error bars).

37

chlorinated pesticides is considerably less than that caused by E2. Another
environmental estrogen, nonylphenol, was found to possess an efficacy approximately
90% of E2 but with a relative potency approximately 10,000-fold less than E2. Therefore,
when comparing estrogenic activities of potential xenoestrogens, it is important to
consider both potency and efficacy.

Since estrogen antagonist potency can be determined using the MCF-7-luc bioassay,
the dose-dependence for TCDD’s antiestrogenicity was evaluated for the full dose-
response curve of E2 (Fig. 4). In general, parallel dose-response curves were obtained
and there was a dose-dependent increase in the EC50 for E2 (i.e., a decrease in potency)

with increasing concentrations of TCDD.

Other Expression Assays

Cell lines used for expression assays include chicken fibroblasts, both primary and
transformed rainbow trout hepatocytes (6,44,132,l33), MCF-7 (42,131), HeLA (6,14), yeast
(6,130,131), and others. There is considerable evidence that in vitro expression of ER
modulated genes is influenced by cell and promoter context (69,121,134,135). The type of
cell used in the expression assay system largely determines the number of confounding
factors that must be considered when interpreting results and determining the relevance of
the assay for predicting effects in vivo. This is due to differences in the intracellular
“machinery” found in different types of cells. Some cells, such as yeast, contain none of the
receptors, DNA response elements, or steroid responsive genes involved in the gene
expression assay. These cells simply act as housing for exogenous expression systems (6).

Others contain parts of the machinery but lack one or more components, such as receptors or

38

regulatory sequences of DNA. Finally, some cells contain all the machinery necessary for
expression.

Yeast, such as Sacchromyces cerevisiae, represent the stripped-down type of cell.
While they contain no steroid receptors of their own, steroid receptors expressed in yeast
have been shown to function normally (136,137). Thus, they must be transformed with
plasmids coding for estrogen receptor and a reporter gene to be used for expression assays.
Yeast based expression assays have been used by a number of researchers (6,130,136,138-
140). They have a number of advantages. First, they tend to be relatively simple and rapid
assays Since yeast iS more easily cultured than more complex cells. Because they have no
endogenous hormone responsive machinery, they are specific for compounds that act
directly through estrogen receptor mediated control of gene expression. Responses are not
modulated by other regulatory factors. This also means that yeast cells are isolated from
many of the confounding factors that affect more complex cells (140). The tradeoff,
however, is the inability of yeast-based expression systems to account for many of the
factors at the cellular level that may affect a compounds activity in vivo. Furthermore, yeast
cells have some morphological (i.e. cell walls) and physiological properties not found in
most animal cells (6). These factors may affect the accumulation and response dynamics of
the xenobiotics. Indeed, significant differences in response between yeast cell based assays
and mammalian cell assays have been reported (6).

Another level of complexity are expression assays which use complex cells such as
mammalian hepatoma cells which do not express steroid receptors of their own. Such cells
transfected with plasmids coding for estrogen receptor and an appropriate reporter gene

retain the specificity of yeast based-assays but incorporate some of the more complex

39

phannacokinetics, metabolic potential, etc. of more complicated cells. Such cells are
considered particularly useful for studying receptor function (140).

Cells containing all the necessary machinery such as MCF-7 cells or rainbow trout
hepatocytes are the most realistic in terms of cellular level factors that can influence
expression. This, however, increases the complexity of interpretation by expanding the
number of possible avenues via which compounds in a sample may influence the endpoint
measured. The choice of whether to use an exogenous or endogenous reporter gene/protein
as an endpoint provides another level of complexity/realism within cells containing a
complete set of endogenous estrogen modulated gene expression machinery.

The different reporter proteins used as endpoints for expression assays are nearly as
numerous as the cell types used. Common reporters include endogenous proteins such as
p82 (42,141) Vitellogenin (44,132,133) and sex hormone binding globulins (6) and
exogenous reporters like luciferase (74), and B-galactosidase (130,140), and
chloramphenicol acetyltransferase (CAT) (14). Like cell type, the type of reporter protein
used as an endpoint can affect the number of confounding factors to consider. Endogenous
proteins may have a direct affinity for the ligands being studied, may be linked to metabolic
processes or feedback mechanisms not directly controlled by the estrogen expression
mechanism, may be affected by other steroid hormones, etc. Exogenous proteins like
luciferase are less likely to be affected by such factors.

The type of reporter protein or endpoint used also has a major impact on the
sensitivity and responsiveness of the assay, since it affects the type of instruments or
analytical methods that can be used to detect it. Due to the availabilty of sensitive detectors

for light and the high quantum efficiency of the luciferase reaction, the light producing

40

endpoint for a luciferase based expression assay can be very sensitively detected using a
luminometer (142). The sensitivity of MCF-7 assays using a luminescent endpoint are in the
low pM ranges (6). Colorimetric endpoints, such as the B-galactosidase endpoint used in
many yeast based assays tend to be nearly 100 fold less sensitive (140).

Expression assays have several advantages over other potential in vitro screening
assays. First, they tend to be more sensitive than receptor binding assays (Fig. 5) (6), though
their degree of sensitivity varies considerably with both cell type and reporter protein used.
Unlike receptor binding or cell proliferation assays, expression assays can be used to detect
both estrogen agonists and antagonists (14,74,143). Because protein expression requires the
transcription of a gene to mRN A followed by translation at the ribosomes, expression assays
incorporate transcriptional and post-transcriptional processes that may be important
determinants of ligand-influenced response. This means expression assays can detect
ligands that influence any part of the gene expression pathway, not just those that bind to the
estrogen receptor. This makes such assays more robust in light of uncertainties that remain
regarding mechanisms of action. Expression assays should also be more specific than
proliferation assays, since many factors that can affect proliferation (i.e. temp, pH, serum
quality, other mitogens, etc.) should have little or no bearing on expression of estrogen
modulated genes.

The primary disadvantages of expression assays are those common to all in vitro
bioassays. Like all in vitro systems, expression assays do not incorporate the entire range of
metabolic and phannacokinetic factors that can affect the disposition of target compounds in

a whole organism. The inability to account for such complexities makes the in vivo

41

1 0000

2 1000
5'
m0.
3-
.2
'5'
3 100
E
8
=
O
0
US
m 10

1

o—MCF-7 - Competitive ligand binding
o—Yeast - hER and URA3 reporter gene

 

’9—Rainbow Trout Hepatocvtes - Vtg expression

<—-Yeast - hER and LacZ reporter gene

F—MCF-7 - Chimeric receptors

<—MCF-7 - other ERE-regulated reporter genes

*— lshikawa Cells - Alkaline phosphatase protein
X MCF-7 - Cell proliferation

MCF-7 ERE-Luc - luciferase activity

 

Fig. 5. A comparison of several in vitro bioassays for estrogenic activities. Adapted from

Zacharewski (6).

42

relevance of such tests unclear. More work is needed to examine and develop correlations
between expression assay results and effects in vivo if indeed they exist. Relative to other in
vitro bioassays, potential disadvantages include the need to develop and characterize
recombinant cell lines, in some cases longer assay duration (6), and greater system
complexity. Expression assays vary considerably in sensitivity, specificity, and
responsiveness (in terms of fold induction), due to the wide variety of constructs used (6).

This can make it difficult to compare results from different assay systems and laboratories.

Environmental Monitoring

In addition to the need to screen chemicals being produced and used, there is a need
to monitor levels of estrogenic compounds in the environment. Monitoring is an essential
part of the risk assessment process. It is also necessary to evaluate the need for and
effectiveness of remediation and/or regulation efforts. Tools such as instrumental/analytical
chemistry analyses and in vivo biomarkers, though not particularly suitable for screening
applications, are useful for monitoring. Many of the in vitro methods discussed previously
can be applied to monitor the environment for estrogenic compounds as well.

Instrumental analysis can be used to monitor the environment for known estrogen
agonists or antagonists. Extracts can be prepared from water, sediments, soil, and biological
matrices. The extracts can then be analyzed using gas chromatography (GC), high pressure
liquid chromatography (HPLC), GC/mass spectrometry (GC/MS), etc. Such analysis can be
very sensitive for common estrogen agonists, particularly due to the ability to concentrate

samples during the extraction and clean-up process. The key advantage to this type of

43

analysis is the ability to precisely quantify concentrations of compounds of interest in the
environmental matrix being studied.

Instrumental analyses alone, however, are rather limited as monitoring tools.
Instrumental analyses, though quantitative, provide no information regarding the biological
activity of the sample. Only compounds which have previously been identified as estrogen
agonists or antagonists can be monitored in this fashion. This problem is further complicated
by the fact that complex interactions may occur between various agonists, antagonists, and
other compounds found in a complex environmental mixture. Analytical methods provide
no mechanism by which to evaluate the overall activity of a mixture, unless strict additivity
of effects is assumed. Non-target compounds in a complex environmental mixture may also
interfere with proper resolution and quantification of target compounds. Thus, although
instrumental analyses are powerful and sensitive tools, they cannot, alone provide the
information needed for comprehensive monitoring and risk assessment.

Some of the in vitro bioassays used for screening are also useful for monitoring.
Extracts from a variety of biotic and abiotic matrices can be tested. Unfortunately, as for
instrumental analysis, the extraction and clean up process is often the most tedious and time
consuming step in monitoring using in vitro bioassays. In some cases, however, less
extensive sample preparation is necessary Since the need to resolve individual compounds is
not an issue for bioassays. The primary advantage to using in vitro bioassays for monitoring
is that they should respond to any and all active compounds in the sample, not just a limited
set of known or target compounds. Because of this, they can integrate the activity of the

entire mixture, accounting for all potential interactions, whether they be additive or not.

One distinct advantage of monitoring with in vitro bioassays is the potential to derive
and compare relative potencies. The concentration or volume of environmental extract
needed to elicit a given level of effect can be compared to that of a standard, such as E2, and
expressed relative to it (i. e. E2 equivalents = ECSO sample / EC50 E2). This type of approach
has been used extensively for dioxin-like compounds (144-147). It provides a simple
method for comparing estrogenic potency from sample to sample, even when sample
composition cannot be determined. Comparison of relative potencies of different fractions
from the same sample can also be used to describe complex interactions that may be
occurring in mixtures.

Unfortunately, however, it is not always easy or straight forward to calculate
comparable potencies for environmental samples based on bioassay results. For comparison
of potencies to be accurate and usefirl, the same response must be achieved with each
sample. That is, the concentration associated with a particular magnitude of response needs
to be estimated, and that magnitude must be the same for all samples being compared. This
generally requires obtaining a complete dose-response for all samples being compared, or at
least assuming that given a sufficient dose of sample, all will reach the same maximum level
of response. In practice, however, these conditions are often difficult to meet and the
assumptions do not usually hold. The inability to control the concentration of active
compounds in the extract beyond the limits of sample concentration and/or dilution often
precludes the ability to generate a complete dose-response curve such that maximal activity
is achieved. Variation in the efficacy of different estrogen agonists within their range of
solubility generally precludes the ability to assume that all samples will reach the same

maxima (Fig. 3). This generally means that in order to estimate or compare potencies, the

45

concentration needed to elicit a minimum statistically significant response (LOEC) or
threshold is about the only comparable point to use for all samples. This is not ideal,
however, since there is minimal statistical confidence at this point.

This discussion illustrates the advantages and deficiencies of both instrumental and
in vitro bioassay methods for monitoring estrogenic compounds in the environment. From
the discussion, it Should also be apparent, that both types of tools complement one another.
Analytical and in vitro tools can be used effectively together to provide information needed
for monitoring, as well as risk assessment, via bioassay-directed fractionation and
identification. Such complementary application of instrumental and bioassay techniques is
often referred to as toxicant identification and evaluation (TIE) (Fig. 6).

In the first step of a TIE scheme, samples from an abiotic or biotic matrix of
interest are collected and chemicals are selectively extracted and fractionated,
chromatographically, to separate the complex mixture from the matrix and into groups of
compounds with nominal characteristics. In monitoring for the presence of estrogen
agonists in water, samples of water could be passed through a solid phase sampling
device, such as Sep-pacs or EmporeTM disks. Alternatively, passive samplers such as
semi-permeable membrane devices (SPMDS), can be placed in the environment to collect
a time integrated sample of the materials of interest (148). Once a sample has been
separated from its matrix it can be analyzed in an in vitro assay. This is generally a
functional assay such as the MCF-7-luc assay used to test for estrogenicity. A battery of
assays based on different functional end points could be used (52). If there were a

positive response in the assay, the sample could be further fractionated based on

46

 

Environmental
Sample

 

 

 

 

M etabolic Activationl

i..e S9 Mix «FractionationJ ‘—

9 Sig nificant Response I

in in vitro Bioassay

 

 

 

 

 

 

 

 

Potential Complex
Interactions

O

 

 

 

l Chemical Synthesis/

Purification I“
' Sum Equals Total I

Significant Response » [Sum Activity I ’_ I Acuvny ?
in in vitro Bioassay
, [Test In Viva ]‘ -

Active Components
via Instrumental Analysis

 

Identification of l

*E

 

 

 

 

 

 

 

Fig. 6. Generalized toxicant identification and evaluation scheme for environmental samples
using in vitro bioassay-directed fractionation and instrumental analysis.

47

molecular size, polarity or a combination of the two. Each fraction would then be
subjected to bioassay. In this way the types of compounds contributing to a positive
response could be narrowed to members of operationally-defined functional classes.
Instrumental analyses would then be applied to the fraction(s) which elicited activity in
vitro. Typically, active fractions are analyzed by several methods, such as HPLC with
both fluorescence and UV-visible detection, gas chromatography with flame ionization
detection (FID), electron capture detection (ECD), and/or mass-selective detection
(MSD). The particular methods used depend on the properties of the fraction of interest.
For instance, if the activity were in the non polar fraction, it could be analyzed by gas
chromatography directly. However, if a more polar fraction contained the activity it
might be necessary to use HPLC-based analysis or derivatization techniques. Additional
separation of active fractions can be achieved using high pressure liquid chromatography
(HPLC). Through additional iterations of fractionation, bioassay, and instrumental
analysis, it should be possible to identify the specific bioactive agents present in the
original extract. Once a tentative identification is achieved, standards would be used to
quantify the mass of material in the sample. If authentic standards were not available
commercially they can be synthesized. Pure compounds could then be analyzed by in
vitro bioassay to determine if they are, indeed, active. If so, relative potency factors
(RPFS) are derived and total equivalents determined for the sample by multiplying the
RPFs by the molar concentrations of the compound in the sample. Predicted and
measured activity in the assay are compared in a mass balance of potency. If they are

equal, it indicates that all of the activity has been accounted for by the assay and that

48

there are no interactions among compounds. Alternatively there could be more activity
or less activity in the measured or predicted values. More activity in the assay than
predicted could indicate that there were additional unidentified compounds or
superadditive interactions in the mixture. Less activity in the bioassay might indicate an
infraadditive interaction among compounds. Use of selective isobolar additions and
selective removal of compounds can confirm the presence or absence of interactions or
unidentified compounds. A similar technique has been proposed for use to separate the

relative contributions of endogenous and exogenous hormones in plasma (124).

In Vivo Biomarkers

In vivo biomarkers are also viable monitoring tools. Reports of in vivo Vitellogenin
induction (149), altered secondary sex characteristics (150,151,152) lowered sperm counts
(44), etc. are responsible for much of the attention that endocrine disrupting compounds have
drawn recently. These and other in vivo responses can be monitored as indicators of
potential exposure to estrogenic and other endocrine disrupting compounds. Such indicators
have the advantage of incorporating all the complex pharmacokinetic and metabolic factors
that can affect uptake and disposition of compounds in a whole organism, giving them more
biological significance. In some cases, however, it is not known whether the changes
observed are adverse or not. This limits the risk assessment utility of some biomarkers.
Because in vivo endpoints are influenced by so many additional variables, both physiological
and environmental, it is much harder to establish clear cause effect relationships between

responses and specific compounds in the enviromnent. Thus, in vivo biomarkers are most

49

usefirl for monitoring and risk assessment as part of a comprehensive, tiered approach that
incorporates both instrumental analyses and rapid in vitro screening assays.

For each receptor-mediated mechanism, an agonist/antagonist screen can potentially
be developed. This can serve as the first tier in an integrated monitoring approach. In vivo
models can provide a second tier for evaluating the toxicological significance of active
samples. In conjunction with chemical analysis, this tiered approach could be used to
monitor for and identify specific bioactive compounds in the environment and evaluate their

potential toxicological relevance.

ACKNOWLEDGMENTS

Portions of the research presented herein and preparation of the chapter were
supported by the Chlorine Chemistry Council of the Chemical Manufacturers Association,
cooperative agreement NO. CR 822983-01-0 between Michigan State University and the
US. EPA - Office of Water Quality, the NIEHS - Superfund Basic Research Program (NIH-
ES-04911), Michigan State University Distinguished Fellowship and an NIEHS postdoctoral
fellowship for financial support. We thank Dr. M.D. Pons for providing MCF-7-luc cells.
We acknowledge Rebecca Zmyslo and Emily Nitsch and the members of MSU’s Aquatic
Toxicology Laboratory - Estrogen Workgroup for their assistance and comments on the

manuscript. Finally, we acknowledge the editors for their patience.

50

REFERENCES

l.

10.

ll.

12.

13.

Wolff MS, Toniolo PG, Lee EW, Rivera M, Dubin N. Blood levels of
organochlorine residues and risk of breast cancer. J Natl Cancer Inst 1993;85:648-
652.

 

Dibb S. Swimming in a sea of estrogens, chemical hormone disrupters. T_he
Ecologist 1995;25:27-31.

McLachlan JA, Arnold SF. Environmental estrogens. Amer Sci 1996;84:452-461.

Colbom T, Vom Saal FS, Soto AM. Developmental effects of endocrine-

disrupting chemicals in wildlife and humans. Environ Health Perspect
1993;101:378-384.

 

Adlercreutz H. Phytoestrogens: Epidemiology and a possible role in cancer
protection. Environ Health Perspect 1995;103 (suppl. 7):102-112.

Zacharewski T. In vitro bioassays for assessing estrogenic substances. Environ
Sc. Techno] 1997;31:613-623.

Mazeaud MM, Mazeaud F. Adrenergic responses to stress in fish. In: Pickering
AD, ed. Stress in Fish. New York:Academic Press, 1982;49-75.

 

Selye H. Stress in Health and DiSease. I History and General Outline of the
Stress Concept. BostonzButterworths, 1976; 3-34

Ness DK, Schantz SL, Moshtaghian J, Hansen LG. Effects of perinatal exposure
to specific PCB congeners on thyroid hormone concentrations and thyroid
histology in the rat. Toxicol Lett 1993;68:311-323.

 

Madhukar BV, Brewster DW, Matsumura F. Effects of in vivo administration of
2,3,7,8-TCDD on receptor binding of epidermal growth factor in the hepatic
plasma membrane of rat, guinea pig, mouse, and hamster. Proc Natl Acad Sci
1984;81:7407-741 l.

 

Makela S, Santti R, Salo L, McLachlan JA. Phytoestrogens are partial estrogen
agonists in the adult male mouse. Environ Health Perspect 1995;103(supp1
7): 123-127.

Mellanen P, Petanen T, Lehtimaki J, et al. Wood-derived estrogens: studies in
vitro with breast cancer cell lines and in vivo in trout. Toxicol Appl Pharmacol
1996;136:381-388.

Barrett I. Phytoestrogens: Friends or foes? Environ Health Perspec 1996;104:478-
482.

51

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Miksicek RI. Commonly occurring plant flavonoids have estrogenic activity.
Molec Pharmacol 1993;44:37-43.

Kelce WR, Monosson E, Gamcsik MP, Laws SC, Gray LE. Environmental
hormone disruptors: Evidence that vinclozolin developmental toxicity is mediated
by antiandrogenic metabolites. Toxicol Appl Pharmacol 1994;126:276-285.

Kelce WR, Stone CR, Laws SC, Gray LE, Kemppainen JA, Wilson EM.
Persistent DDT metabolite p, p ’-DDE is a potent androgen receptor agonist.
Nature 1995;375:581-585.

Astroff B, Romkes M, Safe S. Mechanism of action of 2,3,7,8-TCDD and 6-
methyl-l,3,8-trichlorodibenzofuran (MCDF) as antiestrogens in the female rat.
Chemosphere 1989; 19:785-788.

Rozman K, Rozman T, Greim H. Effect of thyroidectomy and thyroxine on
2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) induced toxicity. Toxicol Appl
Pharmacol 1981; 72:372-376.

Janz DM, Bellward GD. In ovo 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure in
three avian species. Effects on thyroid hormones and grth during the perinatal
period. Toxicol Appl Pharmacol 1996;139:281-291.

Liu H, Biegel L, Narasimhan TR, Rowlands C, Safe S. Inhibition of insulin-like
growth factor-I responses in MCF-7 cells by 2,3,7,8-tetrachlorodibenzo-p-dioxin
and related compounds. Mol Cell Endocrinol 1992;87:19-28.

Choi E, Toscano D, Ryan J, Riedel N, Toscano W. Dioxin induces transforming
growth factor-0t in human keratinocytes. J Biol Chem 1991;266:9591-9597.

Carrier F, Owens RA, Nebert DW, Puga A. Dioxin-dependent activation of
murine CypIa-I gene transcription requires protein kinase C-dependent
phosphorylation. Molec Cell Biol 1992;12:1856-1863.

Reiners JJ, Cantu AR, Scholler A. Phorbol ester-mediated supression of
cytochrome P450 CYPlal induction in murine Skin: involvement of protein
kinase C. Biochem Biophys Res Communicat 1992;186:970-976.

Aoki Y, Matsumoto M, Suzuki K. Expression of glutathione S-transferase P-forrn
in primary cultured rat liver parencymal cells by coplanar biphenyl congeners is
supported by protein kinase inhibitors and dexamethasone. Fed European
Biochem Soc 1993;333:114-118.

DeVito M], Ma X, Babish JG, Menache M, Birnbaum LS. Dose-response
relationships in mice following subchronic exposure to 2,3,7,8-

‘ tetrachlorodibenzo-p-dioxin: CYP1A1, CYP1A2, estrogen receptor and protein

tyrosine phosphorylation. Toxicol Appl Pharmacol 1994;124:82-90.

52

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

Blankenship A, Matsumura F. 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)
causes an ah receptor-dependent and ARNT-independent increase in membrane
levels and activity of p60Src. Environ Toxicol Pharmacol 1997;(in press).

Jobling S, Sumpter JP. Detergent compounds in sewage effluents are weakly
oestrogenic to fish: an in vitro study using rainbow trout (Onchorhyncus mykiss)
hepatocytes. Aquat Toxicol 1993; 27:361-372.

 

Hoare SA, Jobling S, Parker MG, Sumpter JP, White R. Environmental persistent
alkylphenolic compounds are estrogenic. Endocrinology 1994; 135:175-182.

Guillette LJ, Crain DA, Rooney AA, Pickford DB. Organization versus
activation: the role of endocrine disrupting contaminants (EDCS) during

embryonic development in wildlife. Environ Health Perspect 1995;103 (suppl.
7): 157-164.

Colborn T. Environmental estrogens: Health implications for humans and
wildlife. Environ Health Perspect 1995;103 (suppl. 7):l35-136.

F eldman D, Krishnan A. Estrogens in unexpected places: Possible implications
for researchers and consumers. Environ Health Perspect 1995;103 (suppl. 7):129-
133.

Duby RT, Travis HF, Terrill CE. Uterotropic activity of DDT in rats and mink
and its influence on reproduction in the rat. Toxicol Appl Pharmacol
1971;18:348-355.

Donohoe RM, Curtis LR. Estrogenic activity of chlordecone, o,p ’-DDT and o,p ’-
DDE in juvenile rainbow trout: induction of vitellogenesis and interaction with
hepatic estrogen binding sites. Aquatic Toxicol 1996;36:31-52.

 

Soto AM, Chung KL, Sonnenschein C. The pesticides endosulfan, toxaphene and
dieldrin have estrogenic effects in human estrogen-sensitive cells. Environ Health
Perspect 1994; 102:3 80-3 83.

Eroschenko VP. Estrogenic activity of the insecticide Chlordecone in the
reproductive tract of birds and mammals. J Toxicol Environ Health 1981;8z731-
742.

Korach KS, Sarver P, Chae K, McLachlan JA, McKinney JD. Estrogen receptor-
binding activity of polychlorinated hysroxybiphenyls: conformationally restricted
structural probes. Molecular Pharmac_ol 1988;33:120-126.

Spink DC, Johnson JA, Connor SP, Aldous KM, Gierthy JF. Stimulation of 17 -
estradiol metabolism on MCF-7 cells by bromochloro- and chloromethyl-
substituted dibenzo-p-dioxins and dibenzofurans: Correlations with antiestrogenic
activity. J Tox Environ Health 1994;41:451-466.

53

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

Jansen HT, Cooke PS, Porcelli J, Liu TC, Hansen LG. Estrogenic and
antiestrogenic actions of PCBs in the female rat: in vitro and in vivo studies.
Reproduct Toxicol 1993;72237-248.

Santodonato J. Review of the estrogenic and antiestrogenic activity of polycyclic
aromatic hydrocarbons: relationship to carcinogenicity. Chemosphere
1997;34:835-848.

Birnbaum LS. Developmental effects of dioxins. Environ Health Perspect
1995;103 (suppl. 7):89-94.

 

Mobly T.A, Moore RW, Peterson RE. In utero and lactational exposure of male
rats to 2,3,7,8-Tetrachloro-dibenzo-p-dioxin. 1. Effects on Androgenic status.
Toxicology 1992;] 14:97-107.

Olea N, Pulgar R, Perez P, et al. Estrogenicity of resin-based composites and
sealants used in dentistry. Environ Health Perspec 1996;104:298-305.

Sharpe RM, Fisher J S, Millar MM, Jobling S,Sumpter JP. Gestational and
lactational exposure of rats to xenoestrogens results in reduced testicular size and
sperm production. Environ Health Perspec 1995;103:1136-1143.

White R, Jobling S, Hoare SA, Sumpter JP, Parker MG. Environmentally
persistent alkylphenolic compounds are estrogenic. Endocrinology
1994;135:175-182.

Sumpter JP, Jobling S, Tyler CR.. Toxicology of Aquatic Pollution. In: Taylor
EW, ed Physiology and Molecular Approaches. 1996;205-224.

Aheme GW, Briggs R. The relevance of the presence of certain synthetic steroids
in the aquatic environment. J Pharm Pharmacol 1989;41:735-736.

Jobling S, Sheahan D, Osborne JA, Matthiessen P, Sumpter JP. Inhibition of
testicular growth in rainbow trout (Oncorhyncus mykiss) exposed to estrogenic
alkylphenolic chemicals. Enivron Toxicol Chem 1996; 15:194-202.

Sheehan DM, Young M. Diethylstilbestrol and estradiol binding to serum

albumin and pregnancy plasma of rat and human. Endocrinology 1979; 104: 1442-
1446.

 

Solmssen UV. Synthetic estrogens and the relation between their structure and
their activity. Chem Reviews 1945;37:481-598.

Korach KS, Levy LA, Sarver PJ. Estrogen receptor stereochemistry: receptor
binding and hormonal responses. J Steroid Biochem 1987;27:281-290.

Soto AM, Lin TM, Justicia H, Silvia RM, Sonnennschein C. An "in culture"
bioassay to assess the estrogenicity of xenobiotics (E-screen), In: Colbum T,

54

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

Clement C, eds, Chemically induced alterations in sexual and firnctional
development: The wildlife/humpn connection. Princeton, NJ :Princeton Scientific
Publishing, 1992;295-309.

McLachlan JA.Functional Toxicology: A new approach to detect biologically
active xenobiotics. Environ Health Perspec 1993;101 :386-3 87.

Kumar V, Chambon P. The estrogen receptor binds tightly to its responsive
element as a ligand induced homodimer. _(_3_efl 1988;55:145-156.

Bondy KL, Zacharewski TR. ICI 164,384: A control for investigating estrogen
responsive genes. Nucleic Acids Res 1993;21:5277-5278.

Fawell SE, White R, Hoare S, Sydenham M, Page M, Parker MG. Inhibition of
estrogen receptor DNA binding by the "pure" anti-estrogen ICI 164,384 appears

to be mediated by impaired receptor dimerization. Proc Nptl Acad Sci USA
1990;87:6883-6887.

Jordan VC, Kock R, Lieberman ME. Structure-activity relationships of
nonsteroidal estrogens and anti-estrogens. In: Estrogen/Anti-estrogen action and

breast cancer therapy. In: Jordan VC, ed. Madison, leUniversity of Wisconsin
Press, 1986;19-41.

Farooqi ZH, Aboul-Enein HY. Conformational flexibility of
cyclohexylaminoglutethimide: a potent aromatase inhibitor. Comparison of the
three configurations of the cyclohexyl moiety. Cancer Lett 1994;87:121.

Giorgi EP, Stein WD. The transport of steroids in animal cells in culture.
Endocrinology 1981;108:688-697.

Price TM, Blauer KL, Hansen M, Stanczyk F, Lobo R, Bates GW. Single-dose
pharamacokinetics of sublingual versus oral administration of micronized 17 B-
estradiol. Obstet Gvnec 1997;89:340-345.

Dunn JF, Nisula BC, Rodbard D. Transport of steroid hormones: binding of 21
endogenous steroids to both testosterone-binding globulin and corticosteroid-
binding globulin in human plasma. J Clin Endocrinol Metab 1981;53:58-68.

Abramson F P, Miller HC. Bioavailability, distribution and phannacokinetics of
diethylstibestrol produced from stilphostrol. J Urol 1982;128:1336-1339.

 

Tew B, Xu X, Wang H, Murphy PA, Hendrich S. A diet high in wheat fiber
decreases the bioavailability of soybean isoflavones in a single meal fed to
women. J Nutr 1996;126:871-877.

55

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

Martin PM, Horwitz KB, Ryan DS, McGuire WL. Phytoestrogen interaction with

estrogen receptors in human breast cancer cells. Endocrinology 1978;103:1860-
1 867.

 

World Health Organization. Environmental Health Criteria for DDT and it’s
Derivatives-Environmental Aspects. Geneva : WHO, 1989.

Skalsky HL, Guthrie FE. Binding of insecticides to human serum proteins.
Toxicol Appl Pharmacol 1978;43:229-235.

Arnold SF, Klotz DM, Collins, BM, Vonier PM, Guillette Jr., LJ, McLachlan, JA.
Synergistic activation of estrogen receptor with combinations of environmental
chemicals. Science 1996;272:1489-1492.

 

Nagel SC, vom Saal F S, Thayer KA, Dhar MG, Boechler M, Welshons WV.
Relative binding affinity-serum modified access (RBA-SMA) assay predicts the
relative in vivo bioactivity of the xenoestrogens bisphenol A and octylphenol.
Environ Health Perspect 1997;105:70-76.

Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schutz G, Umesono K,
Blumberg B, Kastner P, Mark M, Chambon P, Evang RM. The nuclear receptor
superfamily: The second decade. §e_ll 1995;83:835-839.

Katzenellenbogen BS. Estrogen receptors: bioactivities and interactions with cell
signaling pathways. Biol Reprod 1996;54:287-293.

Kuiper GG, Carlsson B, Grandien K, Enmark E, Haggblad J, Nilsson S,
Gustanson JA. Comparison of the ligand binding specificity and transcript tissue
distribution of estrogen receptors or and 0. Endocrinology 1997;138(3):863-870.

Gorski J, Furlow JD, Murdoch FE, Fritsch M, Kanenko K, Ying C, Malayer JR
Perturbations in the model of estrogen receptor regulation of gene expression.
Biol Reprod 1993;48:8-14

Smith DF, Toft DO. Steroid receptors and their associated proteins. Mo_l
Endocrinol l993;7(1):4-1 1.

Landel CC, Kushner PJ, Greene GL. Estrogen receptor accessory proteins:
effects on receptor-DNA interactions. Environ Health Perspect 1995;103(suppl
7):23-28.

Pons M, Gagne D, Nicolas JC, Mehtali M. A new cellular model of response to
estrogens: A bioluminescent test to characterize (Anti) estrogen molecules.
Biotechniques 1990;9z450-456.

Aronica SM, Katzenellenbogen BS. Stimulation of estrogen receptor-mediated
transcription and alteration in the phosphorylation state of the rat uterine estrogen

56

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

receptor by estrogen, cyclic adenosine monophosphate, and insulin-like grth
factor. Mol Endocrinol 1993;7(6):743-752.

Le Goff P, Montano MM, Schodin DJ, Katzenellenbogen BS. Phosphorylation of
the human estrogen receptor. J Biol Chem l994;269(6):4458-4466.

Arnold SF , Oboum JD, Jaffe H, Notides AC. Serine 167 is the major estradiol-
induced phosphorylation site on the human estrogen receptor. Mol Endocrinol
1994;8(9):1208-1214.

Arnold S, Vorojeikina, Notides AC. Phosphorylation of Tyrosine 537 on the

human estrogen receptor is required for binding to an estrogen response element.
J Biol Chem l995;270(50):30205-30212.

Kuiper GG, Brinkrnan AO. Steroid hormone receptor phosphorylation: Is there a
physiological role? Mol Cell Endocrinol 1994:100(1-2): 103-107.

Meek DW, Street AJ. Nuclear protein phosphorylation and growth control.
Biochem J 1992;287:1-15.

Ignar-Trowbridge DM, Pimentel M, Parker MG, McLachlan JA, Korach KS.
Peptide growth factor cross-talk with the estrogen receptor requires the NB

domain and occurs independently of protein kinase C or estradiol. Endocrinology
1996;137(5):1735-1744.

Ingnar-Trowbridge DM, Nelson KG, Bidwell MC, et al. Coupling of dual
signaling pathways: Epidermal growth factor action involves the estrogen
receptor. Proc Natl Acad Sci USA 1992;89:4658-4662.

Cho H, Katzenellenbogen BS. Synergistic activation of estrogen receptor-
mediated transcription by estradiol and protein kinase activators. Mol Endocrinol
1993;7(3):441-452.

Archuleta MM, Schieven GL, Ledbetter JA, Deanin GG, Burchiel SW. 7,12-
Dimethylbenz[a]anthracene activates protein-tyrosine kinases Fyn and Lck in the
HPB-ALL human T-cell line and increases tyrosine phosphorylation of
phospholipase C-yl, Formation of inositol 1,4,5-trisphosphate, and mobilization
of intracellular calcium. Proc Ngul Acad Sci USA 1993;90:6105-6109.

Enan E, Matsumura F. Activation of phosphoinositide/protein kinase C pathway
in rat brain tissue by pyrethroids. Biochem Pharmacol l993;45(3):703-710.

Safe S, Astroff B, Harris M, et al. 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)
and related compounds as antiestrogens: Characterization and mechanism of
action. Pharmacol Toxicol 1991;69:400-409.

57

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

Romkes M, Piskorska-Pliszczynska J, Safe S. Effects of 2,3,7,8-
Tetrachlorodibenzo-p-dioxin on hepatic and uterine estrogen receptor levels in
rats. Toxicol Appl Pharmacol 1987;87:306-314.

Pappas, TC, Gametchu B, Watson CS. Membrane estrogen receptors identified by
multiple antibody labeling and impeded-ligand binding. FASEB J 1995;9:404-
410.

Morley P, Whitfield JF, Vanderhyden BC, Tsang BK, Scwartz JL. A new,
nongenomic estrogen action: the rapid release of intracellular calcium.
Endocrinology l992;l31(3):1305-1313.

Migliaccio A, Pagano M, Auricchio F. Immediate and transient stimulation of
protein tyrosine phosphorylation by estradiol in MCF-7 cells. Oncogene
1993;8(8):2183-2191.

Migliaccio A, Di Domenico M, Castoria G, de Falco A, Bontempo P, Nola E,
Auricchio F. Tyrosine kinase/p21ras/MAP kinase pathway activation by
estradiol-receptor complex in MCF-7 cells. EMBO J 1996;15(6): 1292-1300.

Kharat I, Saatcioglu F. Antiestrogenic effects of 2,3,7,8-tetrachlorodibenzo-p-
dioxin are mediated by direct transcriptional interference with the liganded
estrogen receptor. J Biol Chem 1996;271(18):10533-10537.

Kavlock RT, Daston GP, DeRosa C, et al. Research needs for the risk assessment
of health and environmental effects of endocrine disruptors: a report of the US
EPA sponsored workshop. Environ Health Perspect 1996; 104:715-740.

Ankley GT. et al.. Development of a research strategy for assessing the ecological
risk of endocrine disruptors. Rev Toxicol Series B -Environmental Toxicology
1996;(1n Press).

Brooks SC, Wappler NL, Corombos JD, Doherty LM, Horwitz JP. Estrogen
structure-receptor function relationships. In: Moudgil VK, ed. Recent Advzfles
in Steroid Hormone Action. New York: W. De Gruyer, 1987;443-366.

 

VanderKuur JA, Wiese T, Brooks SC. Influence of estrogen structure on nuclear

binding and progesterone receptor induction by the receptor complex.
Biochemistry 1993;32:70002-7008.

Dodds EC, Lawson W. Molecular structure in relation to oestrogenic activity.
Compounds without a phenanthrene nucleus. Proc Royal Soc London B Biology
1937;125:222-232.

Waller CL, Oprea TI, Chae K, Park HK, Korach KS, Laws SC, Wiese TE, Kelce
WR, Gray LE. Ligand based identification of environmental estrogens. Chem
Res Toxicol 1996;9: 1240-1248.

58

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

110.

Waller CL, Booker WJ, Gray LE, Kelce WR. Three-dimensional quantitative
structure- activity relationships for androgen receptor ligands. Toxicol Appl
Pharmacol 1996;137:219-227.

Hendry LB, Chu CK, Copland JA, Mahesh VB. Antiestrogenic piperidinediones
designed prospectively using computer graphics and energy calculations of DNA-
ligand comlexes. J Steroid Biochem Molec Biol 1994;48:495-505.

Hendry LB, Chu CK, Rosser ML, Copland JA, Wood JA, Mahesh VB. Design of
novel antiestrogens. J Steroid Biochem Molec Biol 1994;49:269-280.

Anon. Endocrine Screening Methods Workshop: Meeting Report. July 15-16
Nicholas School of the Environment, Duke University, Durham, NC. 1997. pp. 1-
84.

Kramer VK, Helferich WG, Bergman A, Klasson-Wehler E, Giesy JP.
Hydroxylated polychlorinated biphenyl metabolites are anti-estrogenic in a stably

transfected human breast adenocarcinoma (MCF7) cell line. Toxicol Appl
Pharmacol 1997;144:363-376.

Ireland JS, Mukku VR, Robinson AK, Stance] GM. Stimulation of uterine
deoxyribonucleic acid synthesis by 1,1,1-trichloro-2-(p-chloropheny1)-2-(O-
chlorophenyl)ethane(o,p’-DDT). Biochem Pharmacol 1980;24:1469-1474.

 

Scatchard G. The attraction of proteins for small molecules and ions. Ann NY
Aca. Sci l949;5 1 :660-672.

Cressie NAC, Keightly DD. Analyzing data from hormone receptor assays.
Biometrics 1981;37:235-249.

Pakdel F, Le-Gac F, Le-Goff P, Valotaire Y. F ull-length sequence and in vitro
expression of rainbow trout estrogen receptor cDNA. Mol Cell Endocrinol
1990;71:195-204.

Mani SK, Allen JMC, Clark JH, Blaustein JD, O’Malley BW. Convergent
pathways for steroid hormone and neurotransmitter induced rat sexual behavior.
Science 1994;265: 1246- 1249.

Zysk JR, Johnson B, Ozenberger BA, Bingham B, Gorski J. Selective uptake of
estrogenic compounds by Saccharomyces cerevesiae: a mechanism for

antiestrogen resistance in yeast expressing the mamallian estrogen receptor.
Endocrinology 1995;136:1323-1326.

Hwang KJ, Carlson KE, Anstead GM, Katzenellenbogen JA. Donor-acceptor
tetrahydrochrysenes, inherently fluorescent, high affinity, ligands for the estrogen
receptor: binding and fluorescence characteristics and fluorometric assay of
receptor. Biochemistry 1992;3 1: 1 1536-1 1545.

59

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

McDonnell DP, Vegeto E, Gleeson MA. Nuclear hormone receptors as targets
for new drug discovery. Biotechnology 1993;11:1256-1261.

Bulger WH, Kupfer D. Estrogenic activity of pesticides and other xenobiotics on
the uterus and male reproductive tract. In: Thomas J A, Korach KS, McLachlan
JA, eds. Endocrine Toxicology New York: Raven Press, 1985;1-33.

Wakeling AE. Anti-horrnones and other steroid analogues. In: Green B, Leake
RE, eds. Steroid Hormones, A Practical Approach. Washington, DC:IRL Press,
1987;219-236.

Raynaud JP, Ojasoo T, Bouton MM, Bingnon E, Pons M, Craaastes de Paulet A.
Structure-activity relationships of steroid hormones. In: McLachlan JA, ed.
Estrogens in the Environemnt. Amsterdaszlsivier, 1985;24-42.

Meyers CY, Kolb VM, Dandliker WB. Doisynolic acids: potent estrogens with

very low affinity for the estrogen receptor. Res Commun Chem Pathol Pharmacol
1982;35:165-168.

Rosser ML, Muldoon TG, Hendry LB. Computer modeling of the fit of 11-beta-
acetoxy-estradiol into DNA correlates with potent uterotropic activity but not
with binding to the estrogen receptor. 73 Annual Meeting of the Endocrine
Society, Washington, DC. 1991;Abstract No. 368: 122.

Polossek T, Ambros R, Von Angerer S, Brand] G, Mannschreck A, Von Angerer
E. 6-A1kyl-12-formylindolo[2,1a]isoquinolines. Synthesis, estrogen receptor
binding affinities and stereospecific cytostatic activity. J Med Chem
1992;35:3537-3547.

 

Thompson EW, Katz D, Shima TB, Wakeling WE, Lippman ME, Dickson RB.
ICI 164,3 84 a pure antagonist of estrogen-stimulated MCF-7 cell proliferation
and invasiveness. Cancer Res 1989;49:6929-6934.

Bitrnan J, Cecil HC, Harris SJ, F eil VJ. Estrogenic activity of o,p ’-DDT
metabolites and related compounds. J_A_gric Food Chem 1978;26:149-151.

I Martin MB, Saceda M, Garcia-Morales P, Gottardis MM. Regulation of estrogen

receptor expression. Breast Cancer Res and Treat 1994;31:183-189.

McDonnell DP, Clevenger B, Dana S, Santiso-Mere D, Tzukerrnan MT, Gleeson
MAG. The mechanism of action of steroid hormones: a new twist to an old tale.
J Clin Pharmacol 1993;33:1165-1172.

Hendry LB, Mahesh VB. A putatitve step in steroid hormone action involves
insertion of steroid ligands into DNA facilitated by receptor proteins. J Steroid
Biochem Molec Biol 1995;55:173-183.

60

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

133.

134.

Soto AM, Sonnenschein C. Mechanism of estrogen action on cellular
proliferation: Evidence for indirect and negative control on cloned breast tumor
cells. Biochem Biophys Res Comm 1984;122:1097-1103.

Soto AM, Sonnenschein C, Chung KL, Fernandez MF, Olea N, Serrano F0. The
E-screen Assay as a tool to identify estrogens: An update on estrogenic
environmental pollutants. Environ Health Perspect 1995;103 (suppl. 7):1 13-122.

Soule HD, Vazquez ., Long A, Alberts S, Brennan MJ. A human cell line from a
pleural effusion derived from a breast carcinoma. J Nat. Cancer Inst
1973;51:1409-1413.

 

Brooks SC, Locke ER, Soule HD. Estrogen receptor in a human cell line (MCF-7)
from breast carcinoma. J Biol Chem 1973;248:6251-6253.

 

Soto AM, Justicia H, Wray JW, Sonnenschein C. p-Nonyl phenol: An estrogenic
xenobiotic released from "modified" polystyrene. Environ Health Perspect 1991;
92:167-173.

Welshons WV, Rottinghaus GE, Nonneman DJ, Dolan-Timpe M, Ross PF. A
sensitive bioassay for detection of dietary estrogens in animal feeds. J Vet Diag
Investig 1990;2z268-273.

Soto AM, Sonnenshein C. The role of estrogens on the proliferation of human
breast tumor cells (MCF-7). J Steroid Biochem 1985;23:87-94.

Arnold SF, Robinson MK, Notides AC, Guillette LJ, McLachlan JA. A yeast
estrogen screen for examining the relative exposure of cells to natural and
xenoestrogens. Environ Hgglth Perspect 1996;104:544-548.

Ramamoorthy K, Wang F, Chen 1, Norris JD, McDonnell DP, Leonard LS, Gaido
KS, Bocchinfuso WP, Korach KS, Safe S. Estrogenic activity of a
dieldrin/toxaphene mixture in the mouse uterus, MCF-7 human breast cancer

cells, and yeast-based estrogen receptor assays: no apparent synergism.
Endocrinology 1997;138:1520-1527.

Flouriot G, Vaillant C, Salbert G, Pelissero C, Guiraud JM, Valotaaire Y.
Monolayer and aggregate cultures of rainbow trout hepatocytes: long-term and
stable liver-Specific expression of aggregates. J Cell Sci 1993;105:407-416.

Anderson MJ, Miller MR, Hinton DE. In vitro modulation of 17-B-estradiol-
induced Vitellogenin synthesis: effects of cytochrome P4501A1 inducing

compounds on rainbow trout (Oncorhynchus mykiss) liver cells. Aguat Toxicol
1996;34:327-350.

Tzukerrnan MT, Esty A, Santiso-Mere D, Danielen P, Parker MG, Stein RB, Pike
JW McDonnell DP. Human estrogen receptor transactivational capacity is

61

135.

136.

137.

138.

139.

140.

141.

142.

143.

144.

145.

146.

determined by both cellular and promoter context and mediated by two
functionally distinct intrrnolecular regions. Mol Endocrinol 1994;9:21-30.

Berry M, Metzger D, Chambon P. Role of the two activating domains of the
oestrogen receptor in the cell-type and promoter-context dependent agonistic
activity of the anti-oestrogen 4-hydroxytomoxifen. EMBO J l990;9:2811-2818.

Metzger D, White JH, Chambon P. The human oestrogen receptor functions in
yeast. Nature 1988;334:31-36.

 

McDonnell DP, Nawaz Z, Densmore C, Weigel NL, Pham TA, Clark JH,
O’Malley BW. High level expression of biologically active estrogen receptor in
Sacharomyces cerevisiae. J Steroid Biochem Mol Biol 1991;39:291-297.

Connor K, Howell J, Chen 1, Liu H, Berhane K, Sciarretta C, Safe S, Zacharewski
T. Failure of chloro-S-triazine-derived compounds to induce estrogen receptor-
mediated responses in vivo and in vitro. Fund Appl Toxicol 1995;30:93-101.

Routledge EJ, Sumpter JP. Estrogenic activity of surfactants and some of their
degradation products assessed using a recombinant yeast screen. Environ Toxicol
Chem 1996;15:241-248.

Gaido KW, Leonard LS, Lovell S, Gould JC, Babai D, Portier CJ, McDonnell DP.
Evaluation of chemicals with endocrine modulating activity in a yeast-based
steroid hormone receptor gene transcription assay. Toxicol Appl Pharmacol
1997;143:205-212.

Sathyamoorthy N, Wang TTY, Phang M. Stimulation of p82 expression by diet-
derived compounds. Cancer Res 1994;54:957-961.

Roelant CH, Burns DA, Scheirer W. Accelerating the pace of luciferase reporter
gene assays. Biotechnigues 1996;20:914-917.

Demirpence E, Pons M, Balaguer P, Gagne D. Study of an antiestrogenic effect
of retinoic acid in MCF-7 cells. Biochem Biophys Res Comm 1992;183:100-106.

Villeneuve DL, Crunkilton RL, DeVita WM. Aryl hydrocarbon receptor-
mediated toxic potency of dissolved lipophilic organic contaminants collected
from Lincoln Creek, Milwaukee, Wisconsin, USA, to PLHC-l (Poeciliopsis
lucida) fish hepatoma cells. Environ Toxicol Chem 1997;16:977-984.

Tillitt DE, Ankley GT, Verbrugge D, Giesy J .P, Ludwig JP, Kubiak TJ. H4IIE rat
hepatoma cell bioassay derived 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents in

colonial fish-eating waterbird eggs from the Great Lakes. Arch Environ Contam
Toxicol 1991;21:91-101.

 

Jones PD, Giesy JP, Newsted JL, Verbrugge DA, Beaver DL, Ankley GT, Tillitt
DE, Lodge KB, Niemi GJ. Determination of 2,3,7,8-tetrachlorodibenzo-p-dioxin
62

147.

148.

149.

150.

151.

152.

equivalents in tissues of birds at Green Bay, Wisconsin, USA. Arch Environ
Comm Toxicol 1993;24:345-354.

 

Sanderson JT, Aarts JMMJG, Brouwer A, Froese KL, Denison MS, Giesy JP.
Comparison of Ah receptor-mediated luciferase and ethoxyresorufin O-deethylase
induction in H4IIE cells: Implications for their use as bioanalytical tools for the

detection of polyhalogenated aromatic hydrocarbons. Toxicol Appl Pharmacol
1996;137:316-325.

Huckins JN, Tubergen MW, Manuweera GK. Semiperrneable membrane devices
containing model lipid: a new approach to monitoring the bioavailability of
lipophilic contaminants and estimating their bioconcentration potential.
Chemosphere 1990;20:533-552.

Sumpter JP, Jobling S. Vitellogenesis as a biomarker for estrogenic
contamination of the aquatic environment. Environ Health Perspect
1995;103:(supp1. 7) 173-178.

Miles-Richardson SR, Fitzgerald SD, Render J, Barbee S, Giesy JP, Kramer VJ.
Effects of waterborne exposure of l7-B-estradiol on secondary sex characteristics
and gonads of fathead minnow (Pimephales promelas). 1997;(submitted)

Crews D, Bergeron JM, McLachlan JA. The role of estrogen in turtle sex
determination and the effect of PCBs. Environ Health Perspect 1995;103(suppl.
7):73-77.

Gimeno S, Gerritsen A, Bowmer T, Komen H. F eminization of male carp.
Nature 1996;384:221-222.

63

Chapter 2

RAINBOW TROUT CELL BIOASSAY-DERIVED RELATIVE POTENCIES FOR
HALOGENATED AROMATIC HYDROCARBONS: COMPARISON AND

SENSITIVITY ANALYSIS

Daniel L. VilleneuveT, Catherine A. RichterI, Alan L. Blankenshipf, John P. Giesyt

T Department of Zoology, National Food Safety and Toxicology Center, and Institute for

Environmental Toxicology, Michigan State University, East Lansing, Michigan 48824,

USA

I Biochemistry Department, M121 Medical Sciences, University of Missouri-Columbia,

Columbia, Missouri 65212, USA

Published In: Environ. Toxicol. Chem. 1998. 18: 879-888.

64

Abstract-Rainbow trout hepatoma cells, stably transfected with a luciferase reporter
gene under control of dioxin responsive elements (RLT 2.0 cells) were used to derive
relative potencies (RPS) for a variety of halogenated aromatic hydrocarbons (HAHs) that
are structurally similar to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). This in vitro
bioassay utilizes 96-well microplates which provides high sample throughput and assay
efficiency without affecting sensitivity. RLT 2.0-derived potencies for dioxin and furan
congeners, relative to 2,3,7,8-TCDD, ranged from 0.917 for 1,2,3,4,7,8-
hexachlorodibenzofirran (HxCDF) to 0.208 for 1,2,3,7,8-pentachlorodibenzofuran
(PeCDF). All mono- and di-ortho polychlorinated biphenyls (PCBS) tested had RPS
which were orders of magnitude less than TCDD, but point estimates could not be
determined. RLT 2.0-derived RPS were found to be comparable to both other rainbow
trout-specific RPS and RPS based on mammalian bioassay. Sensitivity analysis suggested
that the range of uncertainty associated with TCDD equivalents (TEQ) estimates based
on RLT 2.0-derived RPS is approximately 10 fold. Within this degree of uncertainty and
the context of this study, the RLT 2.0 bioassay showed no definitive biases or
inaccuracies relative to similar mammalian- or fish-specific in vitro bioassays. Thus, the
RLT 2.0 bioassay appears to be a useful tool for evaluating dioxin-like potency of HAHs

to fish.

INTRODUCTION

Halogenated aromatic hydrocarbons (HAHs) are a diverse group of chemicals

which include polychlorinated dibenzo-p-dioxins (PCDDS), dibenzofurans (PCDFS),

65

biphenyls (PCBS) and others. Some of these chemicals are ubiquitous, persistent, and
toxic environmental contaminants [1]. These compounds have been shown to exert
adverse effects on fish, including mortality, wasting syndrome, fin and gill lesions,
hepatotoxicity, immunotoxicity, embryotoxicity, and reproductive impairment [2-5].
Such toxic effects have been observed at concentrations currently found in eggs of Great
Lakes salmonids [6].

Rapid, sensitive, and economical methods for assessing the biological potency of
environmental mixtures of HAHS are needed to facilitate risk assessment and formulate
objective policy regarding HAHS. Instrumental analysis is difficult and expensive, and
provides only limited information on the biological relevance of HAHS found in the
environment [7]. Because HAHS are thought to exert many of their adverse effects
through a common aryl hydrocarbon receptor (AhR)-mediated mechanism [1,8], in vitro
bioassays which measure AhR-mediated enzyme induction have been used to screen for
HAHS [7,9,10]. Correlations between AhR-mediated enzyme induction in vitro and toxic
potency in vivo make such assays useful for estimating AhR mediated toxic potency of
complex mixtures of HAHS [1,7]

Although it is a well established method, in vitro ethoxyresorufin-O-deethylase
(EROD) assay with H4IIE rat hepatoma cells [9], has been criticized as not being entirely
suitable for assessing the risk of AhR-active xenobiotics to fish and other aquatic
organisms. Differences in sensitivity to AhR-active compounds between fish and
mammals have been reported [4,11,12]. In particular, fish have been shown to be less
sensitive to certain PCBS than mammals. As a result, a variety of both in vivo and in

vitro approaches have been used to derive fish-specific relative potencies (RPS) [4,13,14].

66

Although they provide valid sets of fish-specific RPS, these methods are not particularly
amenable to environmental monitoring and risk assessment of unknown compounds. In
vivo approaches used to derive fish-specific RPS are costly and time consuming and thus
not very useful as a rapid screening tool. In vitro EROD assays using fish cell lines such
as RTL-Wl [13] and PLHC-l [15,16] are more reasonable for this purpose since they are
rapid, inexpensive, and sensitive fish-specific bioassays.

Another approach, developed recently, involves use of a novel recombinant
rainbow trout cell line, RLT 2.0 [17]. RLT 2.0 cells are rainbow trout hepatoma cells
stably transfected with a luciferase reporter gene under control of dioxin responsive
elements (DRES) [10,17]. Binding of the AhR-ligand complex to the DREs results in an
upregulation of luciferase transcription [17], which upon addition of a substrate, luciferin,
catalyzes a light producing reaction. The luminescent end-point can be measured with a
luminometer to provide a sensitive measure of AhR-mediated gene expression potency.
Greater sensitivity, selectivity, and dynamic range have been cited as potential
advantages of using luciferase-transfected cell lines rather than cytochrome P4501A1
(CYP1A1) expression in wild type cells as an endpoint [7]. Luciferase assays with
recombinant H4IIE cells were demonstrated to be more sensitive than in vitro EROD
assay using wild type H4IIE cells [7]. The RLT 2.0 bioassay was reported to have a
detection limit of 4 pM TCDD [17], while no significant EROD activity was detectable in
the parent cell line, RTH-149, at concentrations less than 100 pM TCDD [17]. Thus, the
RLT 2.0 cell bioassay is fish (salmonid) specific and has advantages conferred by the

reporter gene construct.

67

Historically, the terms toxic equivalency factor (TEF) and relative potency (RP)
have been used synonymously. Most literature RP values cited in this paper are referred
to as TEFS in the primary source. Recently, however, increasing efforts are being made
to distinguish the two terms. In the context of this paper, RPS are defined as Species-,
endpoint-, and assay-specific determinations of potency expressed relative to some
standard, such as TCDD. TEFS are defined as consensus values based on RP
determinations across multiple species and/or endpoints. TEFS are commonly used for
risk assessment purposes and are generally order of magnitude estimates. RPS are
generally more precisely defined and are needed for bioassay directed mass balance
analysis of complex mixtures.

The purpose of this study was to adapt the RLT 2.0 assay to a 96-well plate
format and increase assay efficiency without significant loss of sensitivity or resolution.
Such modifications Should increase the utility of the assay for screening, and bioassay
directed fractionation applications. In addition, a previously reported list of RLT 2.0-
derived RPS [17] was expanded to aid mass-balance evaluations. New congeners tested
included several mono- and di-ortho PCB congeners, 1,2,3,4,7,8-hexachlorodibenzo-p-
dioxin (HxCDD), 2,3,7,8-tetrachlorodibenzofuran (TCDF), 1,2,3,7,8-
pentachlorodibenzofuran (PeCDF), and 1,2,3,4,7,8-hexachlorodibenzofuran (HxCDF).
RLT 2.0 derived RPS were compared to both fish-specific and mammalian RPS and TEFS
commonly used in risk assessment and characterization of environmental samples. A
sensitivity analysis was performed to characterize the uncertainty associated with

application of RLT 2.0-derived values. Results presented should help guide the use of

68

RLT 2.0-derived potency to characterize environmental samples and evaluate aquatic

exposure to HAHS.

MATERIALS AND METHODS

RLT 2.0 cell culture

RLT 2.0 cells were cultured in 100-mm disposable petri plates (Corning, Corning
NY, USA) containing 12 ml Basal Medium Eagle (Life Technologies, Grand Island, NY,
USA) supplemented with 10% defined fetal bovine serum (Hyclone, Logan UT, USA)
and 292 mg/L L-glutamine (Sigma, St. Louis, MO, USA). Plates were incubated at 21°C
in a 95:5 airzCO2 (v/v) atmosphere with 80% relative humidity. Cells were passaged
when plates became confluent (generally once per week). Every 5 passages, 1000 mg/L
geneticin (G-418; Sigma G9516) was added to the culture medium to maintain selection
pressure on the recombinant cells. New cultures were started from frozen stocks after 20

passages.

Chemicals and Reagents

2,3,7,8-TCDD, 1,2,3,4,7,8-HxCDD, 2,3,7,8-TCDF, 1,2,3,7,8-PeCDF, 1,2,3,4,7,8-
HxCDF, and PCBS 105, 118, 138, 153, 156, and169 were purchased from Accustandard
(New Haven, CT, USA). Working solutions and dilution series were prepared in
pesticide residue analysis grade isooctane (Burdick & Jackson, Muskegon, MI, USA).

69

Luciferase assay reagent (LAR) was composed of 20 mM tricine (Life Technologies),
1.07 mM Mg(C03)4Mg(OH)2-5H2O (Sigma), 2.67 mM MgSO4-7H2O (Sigma), 0.1 mM
EDTA-disodium salt (J .T. Baker, Phillipsburg, NJ, USA), 33.3 mM dithiothreitol (DTT;
Sigma), 270 pM coenzyme A (Sigma), 530 pM ATP (Sigma), and 470 pM beetle
1uciferin(Promega, Madison, WI, USA). Viability reagent consisted of 1.0 pM calcein-
AM (Molecular Probes, Eugene, OR, USA) and 1.0 pM ethidium bromide (Sigma)

dissolved in non-supplemented Basal Medium Eagle.

Exposure of RLT 2. 0 cells

RLT 2.0 cells, grown in medium without geneticin, were trypsinized from petri
plates containing 80-100% confluent monolayers and resuspended in media before
counting. Cell number per ml was estimated using a hemocytometer. Cells were diluted
in medium to a concentration of approximately 7.5x104 cells/ml and seeded into the 60
interior wells of 96-well flat-bottom microplates (Corning 25860 for flash method as
defined below, Packard Instruments 6005181 for glow method, Meriden, CT, USA) at
250 pl per well (15,000-20,000 cells per well) using a repeating pipettor. To assure
homogeneity, the cell solution was vortexed continuously during seeding. The 36
exterior wells of each microplate were filled with 250 pl culture media. Cells were dosed
after an overnight incubation to allow for cell attachment. Test and control wells were
dosed with 2.5 pl of the appropriate test compound in isooctane or isooctane alone,
respectively. Blank wells received no dose. Each plate tested included a minimum of

three control wells, three blank wells, and one to three dilution series. Dilution series
70

consisted of Six concentrations (3 replicate wells per concentration) of test compound.
Dilution series ranged from 2-fold to logarithmic (IO-fold). Dosed cells were exposed

for 72 h at standard incubation conditions.

Luciferase assay

Luciferase assay methods used in this study were modifications of the previously
published RLT 2.0 assay method [17]. The modifications were designed to adapt that
method for use with a 96-well plate reading luminometer. Results from two methods are
reported. The flash method was the first modification developed and most closely
parallels the original assay method [1 7]. The glow method was developed to simplify

the luciferase assay method to increase assay efficiency and sample throughput.

Luciferase assay: flash method

Culture medium was removed and each well was rinsed twice with phosphate
buffered saline (PBS) using an eight channel vacuum manifold. 50 pl PBS was added to
each well and plates were inspected for cell loss due to washing. Following inspection, a
viability assay was conducted [17]. 50 p1 of viability reagent was delivered to each well
and plates were incubated for 10 min at room temperature. Fluorescence was measured
with a Cytofluor 2300 (Millipore, Bedford, MA, USA) (excitation 485 and 530 nm,
emission 530 and 645 nm). The viability solution was then removed by vacuum

manifold and wells were rinsed three times with PBS. Plates were again inspected for

71

 

cell loss during washing, then lysed by addition of 30 p1 reporter lysis buffer (Promega)
per well. After a 10 min incubation at room temperature, 20 p1 of lysate was transferred
from each well of the transparent Corning microplate to the corresponding well of an
opaque Microlite 2 plate (Dynatech Laboratories, Chantilly, VA, USA). The opaque
plate was scanned using a ML 3000 microtiter plate reading luminometer (Dynatech
Laboratories) set to enhanced flash mode. In this operation mode, 100 p1 LAR is injected
directly into each well by an ML 3000 dispenser system (Dynatech Laboratories) while
the well is directly under the photomultiplier tube. Peak light production and the time at
which it occurs is recorded. Integration begins 5 s after LAR injection, with subsequent
readings taken every 10 ms and integrated over 30 s before proceeding to the next well.
Using this mode, total instrument run-time is approximately 35 min per plate.

A protein assay [18] was performed on the lysates remaining in the transparent
Corning plate. This involved adding 90 p1 ultra pure water and 50 pl 1.08 mM
fluorescamine (Sigma) in acetonitrile to the 10 pl lysate remaining in each well,
incubating the plate for 15 min at room temperature, and scanning the plate with a
Cytofluor 2300 (excitation 400 nm, emission 460 nm). A protein standard curve
consisting of 6 concentrations of bovine serum albumin (BSA) (Sigma) ranging from 50-
1.5 pg per well, 4 replicates per concentration, was prepared and analyzed.

All data was collected electronically and imported into a spreadsheet program for
analysis. Protein per p1 lysate was determined for each well by regression of relative
fluorescence units (RFU) measured for that well against the BSA standard curve. Dose-
response relationships using both unadjusted RLU (relative luminescence units) and RLU

adjusted for protein (RLU/ug) versus log-dose were plotted. In general, correcting for
72

 

protein concentration did not significantly affect the variability of the data or the shape of
the dose-response relationships. Only unadjusted data was used in deriving the results
reported herein. A viability index (calcein fluorescence/ethidium fluorescence) was
calculated for each well and this data was inspected for gross differences in viability
between wells. Wells with a significantly different viability index were not included in
data analysis. In general, significant differences in this ratio were observed only where

differences in cell number were apparent by simple visual inspection of the wells.

Luciferase assay: glow method

Each test plate was inspected visually and differences in cell numbers and
condition relative to control wells and conditions normally observed during routine
culture were noted for each well. Culture medium was then removed and each well was
rinsed twice with phosphate buffered saline (PBS) using an eight channel vacuum
manifold. Plates were inspected for cell loss during washing. Cells were lysed for 5 min,
at room temperature, with reporter lysis buffer (25 pl per well) (Promega). After lysis,
75 p1 LAR was added to each well. Each plate was incubated for 10 min at 30° C then
scanned with an ML 3000 microplate reading luminometer (Dynatech Laboratories) set
to cycle mode, 4 cycles, 2 s pause between cycles. In this mode of operation, LAR is
added to all wells prior to inserting the plate into the instrument. Relative light output of
each well is scanned 4 times (4 cycles) and mean response, standard deviation, and
coefficient of variation (CV) over the 4 cycles are reported for each well. Total run-time

in the instrument is approximately 2 min per plate. Following the luminometer scan 125

73

 

p1 1.08 mM fluorescamine in acetonitrile was added to each well and plates were assayed
for protein after a 15 min incubation at room temperature. Plates were scanned using a
Cytofluor 2300 (excitation 400 nm, emission 460 nm) and responses were compared to a
standard curve similar to that used for the flash method, but with appropriate volumes of
glow method reagents.

All data was collected electronically and imported into a spreadsheet program for
analysis. Protein content per well was calculated by regression against the BSA standard
curve. Dose-response relationships depicting mean RLU (over three replicate wells)
versus log-dose were prepared. Protein data was used as an index of cell number to
detect outliers that were not apparent by visual inspection, but individual responses were

not adjusted for protein.

Calculation of RPs

Responses expressed as mean RLU (3 replicate wells) were converted to a
percentage of the mean maximum response observed for TCDD standard curves
generated on the same day (%-TCDD-max). This was done to normalize responses for
normal day-to-day variability in response magnitude. Responses expressed as %-TCDD-
max were converted to probits, such that 50 %-TCDD-max yielded a probit value of
5.00. Probit units were plotted against log dose and EC-lOs (probit = 3.718), 205 (probit
= 4.159), and 505 (probit = 5.00), were calculated based on regression through the
straight-line portion of each probit plot. Mean EC-lOs (not reported), 208, and 503, were

calculated for replicate dose-response relationships for each compound. RPS were

74

 

calculated for 1,2,3,4,7,8-HxCDD, 2,3,7,8-TCDF, 1,2,3,7,8-PeCDF, and 1,2,3,4,7,8-
HxCDF as mean EC-20 and EC-50 values for TCDD / mean EC-20 and EC-50 values for
the test compound.

The probit method used for estimating absolute and relative potencies in this
study is a parallel line method which assumes the following conditions: 1) all
compounds tested have equal efficacy (maximum response) if tested at sufficiently great 'P‘
concentration; and 2) the dose-responses being compared are parallel (have equal slopes).
Not all dose-responses being compared in this study met these conditions. Replicate

curves for TCDD and 1,2,3 ,4,7,8-HxCDD (Fig. 1-a,c) represent the ideal situation for use

 

fi".ln
A.

of a parallel lines method. Approximately equal efficacy was apparent and the slopes
were nearly equal. Replicate curves for 1,2,3,7,8-PeCDF (Fig. l-b) represent the worst—
case situation encountered in this study. The efficacy observed for 1,2,3,7,8-PeCDF was
less than 50% that of TCDD (Fig. l-b). Thus, the first condition was violated. The
slopes were reasonably similar considering the variation in observed efficacy, but
differences may cause some inaccuracy in potency estimates. Although assumptions
were violated in a few cases, reasonable parallel line approximations of relative potency
were attainable for most dose-responses generated in this study. Over 90% of the dose-
responses generated included responses 3 50%-TCDD-max. and all included values _>_
20%-TCDD-max. Thus, it was unnecessary to extrapolate beyond the data to achieve
potency estimates. Inaccuracies caused by violation of assumptions may impact the
variability of some estimates and subsequent uncertainty, but should not significantly

alter study conclusions.

75

b.

 

 

 

 

 

 

 

 

 

 

 

 

3L-
x- 140 TCDD so 1,2,3,7,a-PoCDF
Q
g 90 540
a 40 §
.‘7’ 320
=2 : i3

- o
0.1 10 1000 a! L ' p 1.
'2°'I' 10 100 1000
Concentration (pM in well)

(2. 4L

120 ch00 so PCB 153
moo x.
g so a 40
a so f
o 40 g 20
i3 20 g

I . o
3! o 1 I a!

.20 I I

100 10000 '20 . 10 1000

 

 

 

 

 

 

 

Concentration (pM in well)

Fig 1. Examples of replicate dose-response curves generated using the RLT 2.0 bioassay
(glow method). (a.) Replicate 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) standard
curves (n = 4). (b.) Replicate dose-responses for 1,2,3,7,8-pentachlorodibenzofirran
(PeCDF) (n = 3). Worst-case encountered for application of probit analysis for
estimating EC-20s and EC-SOS. (c.) Replicate dose-responses for 1,2,3,4,7,8-
hexachlorodibenzo-p-dioxin (HxCDD) (n = 3). (d.) Example of replicate dose-responses
obtained for mono- and di-ortho PCB congeners (n=3). No responses were significantly
different from background. Note, negative values for percentage of the mean maximum
response observed for TCDD standard curves generated on the same day (%-TCDD-max)
are due to subtraction of mean solvent control response prior to calculation of %-TCDD-

max.

76

Sensitivity Analysis

Replicate RP determinations for TCDD, 2,3,7,8-TCDF, 1,2,3,7,8-PeCDF,
1,2,3,4,7,8-HxCDD, and 1,2,3,4,7,8-HxCDF were fit to a selection of continuous
frequency distributions (normal, log-normal, and uniform) using Crystal
Ball®(Decisioneering , Boulder, CO, USA). Chi square goodness of fit was determined
for each distribution. A log-normal distribution provided the best fit for all congeners
except 2,3,7,8-TCDF which was best described by a normal distribution. Distributions
were defined for all congeners using the arithmetic mean and standard deviation of all
replicate RP determinations. Where replicate determinations were not available
(congeners tested in a previous study [17], and PCBS) log-normal distributions with a
mean and standard deviation equal to the value reported (or maximum possible RP for
PCBS) were assigned. The RP of each congener was varied independently over it’s
frequency distribution for 2000 separate trials and a TEQ for each sample was calculated
for each trial as TEQ; = 2 (Rpij * Conc.g), where i represents a different congener and j
represents one of n = 2000 trials. A sensitivity analysis was performed on each
frequency distribution of TEQs generated in this manner in order to determine the

relative contribution of each RP to the total variation in the TEQ estimate.

RESULTS

77

RLT 2. 0-specific RPs

None of the PCB congeners tested in this study were active in the RLT 2.0 assay
system. Finite RPs for PCBS 77 and 126 (Table 1), both non-ortho substituted congeners,
were reported previously [17]. PCBS tested in this study were predominantly mono- and
di-ortho substituted congeners. Dose-response relationships were not obtained for any of
the PCB congeners tested in this study and none of the concentrations tested produced a
response Significantly different from background (Fig. l-d). Consequently, point
estimates of RP could not be calculated for these compounds. RPs were reported as less
than the value obtained by dividing the maximum concentration tested by the mean EC-
20 or EC-SO for TCDD (Table 1). Actual relative potency values for these mono- and di-
ortho PCB congeners are probably much lower than the values presented. Not all
congeners were tested to their limit of solubility, consequently it may still be possible to
generate a response in the cell line and calculate a point RP estimate for some of these
congeners.

Point estimates of RPS were generated for all dioxin and furan congeners tested
(Table 1). All congeners had potencies similar to that of 2,3,7,8-TCDD. Their
approximate order of potency in RLT 2.0 cells was 1,2,3,4,7,8-HxCDF z 2,3,7,8-TCDD
> 1,2,3,4,7,8-HxCDD >> 1,2,3,7,8-PeCDF z 2,3,4,7,8-PeCDF z 1,2,3,7,8-PeCDD z
2,3,7,8-TCDF. RPs based on results from the flash method were lower than those

generated from glow method results. There was no consistent relationship in the relative

78

 

Table 1. RLT 2.0 bioassay-derived relative potencies (RPS) for halogenated aromatic
hydrocarbons presented for two different bioassay methods. RPs based on both EC-SOS
and EC-ZOS are presented.

 

 

Compounda Flash Method Glow Method Flash Method Glow Method
(ECSO RP) (ECSO RP) paczo RP) (ECZO RP)
2,3,7,8-TCDD 1.0 1.0 1.0 1.0
1,2,3,7,8-PeCDDb 0.225 0.279
1,2,3,4,7,8-HxCDD 0.781 0.743 0.41 1 0.877
2,3,7,8-TCDF 0.228 0.243 0.0970 0.361
1,2,3,7,8—PeCDF 0.1 12 0.304 0.0582 0.287
2,3,4,7,8-PeCDPb 0.296 _ 0.347
1,2,3,4,7,8-HxCDF 0.735 1.10 0.328 1.04
PCB 77",c 0.00595 _ 0.00173 _
PCB 105d <0.00005 _ <0.000006 _—
PCB 118d <0.006 _ <0.0007 _
PCB 126° 0.00628 __ 0.00717 _
PCB 138d <0.036 _ <0.004 _—
PCB 153 _ <0.00015 _ <0.00004
PCB 156d <0.036 _ <0.004 _
PCB 169 <0.005 __ <0.0006 _

 

a TCDD = tetrachlorodibenzo-p-dioxin; PeCDD = pentachlorodibenzo-p-dioxin;
HxCDD = hexachlorodibenzo-p-dioxin; TCDF = tetrachlorodibenzofuran; PeCDF =
pentachlorodibenzofuran; HxCDF = hexachlorodibenzofuran.

Reported previously [17].
° Estimated from incomplete dose-response curves.
b Showed no response in flash method bioassay. Not analyzed using glow method.

79

 

magnitude of the E020 versus EC-50 based RP estimates. The absolute difference
between the EC-20-RP and EC-50-RP tended to be less for the glow method. The
coefficient of variation (CV) across the four values reported for each congener in Table 1
was greatest for 1,2,3,7,8-PeCDF (65%) and least for 1,2,3,4,7,8-HxCDD (29%). The
average CV across the four RP values for each congener presented in Table 1 was 46%.
EC-SO and EC—20 estimates derived for the active compounds tested (Table 2)
provide additional information that is not readily apparent from RP values alone. EC-SOs
and 203 based on the glow method were approximately half the magnitude of those based
on flash method. Estimates based on the glow method were also less variable, with an
average CV of 71% for EC-SOS and 47% for EC-2OS compared to 103% and 72% for the
flash method. Neither the difference in magnitude nor the difference variability was
statistically significant (p = 0.05, two-tailed), however. EC-20 estimates were less

variable than EC-50 estimates.

DISCUSSION

Comparison of methods

As part of this study, two methods for performing the RLT 2.0 bioassay were

developed and tested. Both methods were designed for use with a 96-well plate reading

80

Table 2. RLT 2.0 bioassay-derived potency of halogenated aromatic hydrocarbons.
EC50 and EC20 estimates (pM in well) generated using flash and glow methods are
presented. Estimates are the mean of n replicates :t one standard deviation (SD).

 

_ ‘ .dt-

 

 

Compound”I Flash Method Flash method Glow method Glow method
EC50 (SD, n) EC20 (SD, n) ECSO (SD, n) EC20 (SD, n)
2,3,7,8-TCDD 299 (435, 14) 34.5 (31.6, 14) 147 (115, 4) 38.9 (20.6, 3)
2,3,7,8-TCDF 1310 (980, 5) 355 (204, 5) 603 (192, 3) 108 (23.5, 3)
1,2,3,7,8-PeCDF 2650 (2320, 5) 593 (321, 5) 1860 (2760, 3) 321 (377, 3)
1,2,3,4,7,8-HxCDF 406 (392, 4) 105 (70.4, 4) 133 (66.7, 3) 29.4 (7.13, 3)
1,2,3,4,7,8-HxCDD 382 (420, 5) 83.8 (74.5, 5) 197 (89.6, 3) 44.4 (9.48, 3

 

“ TCDD = tetrachlorodibenzo-p-dioxin; TCDF = tetrachlorodibenzofuran; PeCDF =

pentachlorodibenzofuran; HxCDF = hexachlorodibenzofuran; HxCDD =

hexachlorodibenzo-p-dioxin.

81

luminometer and thus have significant advantages over the original assay method [17] in
terms of sample throughput and assay efficiency. Results generated using the 96-well
plate methods were more variable than results generated using the original method [17].
Using the method published previously [17], the average CV across EC-50 estimates for
TCDD standard curves was 12.4% (n=4). The average CV across EC-SO estimates for
TCDD standard curves was 146% (n=14) and 78.5% (n=4) for the flash and glow
methods, respectively. Replicate to replicate variability (among replicate wells on a
plate, or replicate determinations with cuvet reading luminometer [17]) was similar for
all methods, however. Average CVs among replicates were 15.24%, 17.35%, and 15.6%,
for glow, flash, and original [17] methods, respectively. The increased variability among
standard curves (indicated by greater variability in EC-50 determinations) may be due to
changes in the stability cell line and its response to HAHS, or differences in the method.
The instrumentation needed to conduct a direct comparison of the methods was not
available, therefore, the source of the increased variability could not be definitively
determined. Assuming differences in the method are the cause, the 96-well plate
methods remain effective for screening large numbers of samples, but the RLT 2.0 assay
method published previously [17] may be more suitable for precise determination of
relative potencies.

The glow method appears to be a more suitable method for use in routine
bioassays than the flash method. The glow method greatly increases assay efficiency and
sample throughput. Using the glow method, the limiting factor on the number of plates

that can be analyzed on a given day is the time it takes to dose the plates (3 (1 prior to

82

analysis). One person can dose 10 to 20 plates in 8 h. Analysis of ten plates (3 d later)
can be accomplished in less than 2 h. It is possible to analyze over 50 plates per
instrument in a single 8 h day. Using the flash method, it takes an entire 8 h day to
analyze 10 plates. Thus instrumental analysis-time becomes the limiting factor on the
number of samples a laboratory can process.

In addition to decreasing instrument analysis-time, the glow method greatly
streamlined the bioassay procedure and subsequent data analysis. Experience with the
flash method indicated that performing a viability assay prior to the luciferase assay did
not increase the quality of bioassay results. A simple visual inspection of each well on
the test plates, prior to analysis, provided a more informative and equally reliable method
for detecting potential cytotoxicity or contamination that could yield spurious data.
Elimination of the viability assay from the procedure reduces the number of washing
steps during which cell loss can occur. The need for lysate transfer was eliminated in the
glow method by using opaque 96-well plates with a transparent bottom. Wells could be
inspected visually throughout the exposure period, then a sticker could be applied to the
bottom of the plate to render it opaque for luciferase assay, and subsequently removed to
facilitate a protein assay. Elimination of the lysate transfer step was advantageous since
it was a step where errors in technique could easily generate considerable variability
and/or inaccurate results. It is reasonable to assume that the flash method could benefit
from similar changes in procedure.

Glow method simplifications were not detrimental to assay sensitivity or
variability. The average variability between replicate wells on a single plate and

variability between replicate dose-response curves was less for the glow method,

83

although the difference in average CVs between methods was not statistically Significant
(p = 0.05, two-tailed) in either case. Mean EC-SOS and EC-ZOS for the glow method
were, on average, 47% less than estimates based on the flash method. This suggests that
the glow method may be slightly more sensitive than the flash method. Again, this
difference was not statistically significant, however (p = 0.05, two-tailed). Thus, the
glow method represents a significant improvement in assay efficiency without loss of

sensitivity and resolving power.

Comparison of relative potencies

RLT 2.0 RPs were reasonably well correlated with early life stage mortality
(ELSM) RPS but poorly correlated with in vivo and in vitro EROD RPs in rainbow trout
(R2 = -0.34 and -0.07, respectively; Table 3). The greatest correlation between any of the
rainbow trout specific RPS was between in vitro and in vivo EROD induction (R2 = 0.83;
Table 3). RPS based on ELSM were most closely correlated with World Health
Organization (WHO) TEF S [19] for fish (R2 = 0.95; Table 3). This is probably because
WHO fish-TEFS were largely based on ELSM-RPS [19]. The lack of strong correlation
(Table 3) between the various sets of rainbow trout specific RPS suggest that within the
range of uncertainty associated with these assays, there is no consensus rank order of
potency for the dioxins, furans, and non-ortho PCBS among rainbow trout specific
bioassays.

In terms of magnitude, RLT 2.0 derived RPs compare favorably with other sets of

rainbow trout-specific RPS (Table 3). Within the range of uncertainty, there were

84

essentially no differences between rainbow trout RPS for TCDD, 1,2,3,4,7,8-HxCDD,
1,2,3,4,7,8-HxCDF and PCB congeners (Table 3). No differences greater than 10-fold
were observed between RPs for any specific compound, despite the range of endpoints
encompassed.

RLT 2.0-RPs for 1,2,3,7,8-PeCDD and 2,3,4,7,8-PeCDF were approximately 10-
fold greater than both EROD based RP estimates (Table 3). They were relatively close to
ELSM-RPS, however. This may suggest that RLT 2.0 may be a better model for
predicting the in vivo potency of these compounds than those based on CYP1A1
induction.

RLT 2.0-RPS for 2,3,7,8-TCDF and 1,2,3,7,8-PeCDF were Similar to EROD based RPs
but are approximately 10-fold greater than ELSM-RPS (Table 3). There is some evidence
to suggest metabolic inactivation and/or excretion in vivo may account for such a
difference. Half-lives reported for 2,3,7,8-TCDF and 1,2,3,7,8-PeCDF in mammals and
fish in vivo are shorter than those for 2,3,7,8-TCDD, 1,2,3,4,7,8-HxCDD, and 2,3,4,7,8-
PeCDF in both fish and mammals (only mammalian half-lives were available for
1,2,3,7,8-PeCDF) [20]. This points out a potential limitation of RLT 2.0 and EROD
based assays for determining the potency of PCDFS to rainbow trout.

Overall, the RLT 2.0 bioassay appears to be a reasonable method for screening
samples for potential to cause dioxin-like biological responses in rainbow trout. RLT 2.0
RPs may slightly over or under-estimate the potency of individual congeners but no
definitive biases or inaccuracies were apparent based on comparison to other rainbow

trout RPS.

85

Table 3. Comparison of RLT 2.0 bioassay derived relative potencies (RPs) to rainbow
trout-Specific RPs based on other in vitro and in vivo endpoints and World Health
Organization (WHO) toxic equivalency factors (TEFS) for fish. A correlation matrix is
presented. Correlations were calculated using only those congeners for which point

estimates of RP were available for both columns being compared.8

 

 

 

Compound RLT 2.0 In vitro In vivo In vivo WHO
EROD b EROD ° ELSM d TEF °
2,3,7,8 TCDD 1.0 1.0 1.0 1.0 1
1,2,3,7,8 PeCDD 0.225f 2.6 1.8 0.73 1
1,2,3,4,7,8 HxCDD 0.7628 1.1 0.4 0.319 0.5
2,3,7,8 TCDF 0.2358 0.2 0.5 0.028 0.05
1,2,3,7,8 PeCDF 0.2083 0.2 0.4 0.034 0.05
2,3,4,7,8 PeCDF 0.296f 1.9 2.0 0.359 0.5
1,2,3,4,7,8 HxCDF 0.917g 1.1 0.4 0.280 0.1
PCB 77 (3,3’,4,4’) 5.95 x 10'3 fl" —- -- 0.00016 0.0001
PCB 105 (2,3,3’,4,4’) <5x10'5 8 -- -- <7x10‘5 <5x10'6
PCB 118 (234,435) <6x10’3 8 -- -- <7x10'5 <5x10'6
PCB 126 (3,3’,4,4’,5) 6.28x10‘3 " -- -- 0.005 0.005
Correlation Matrix R2 R2 R2 R2 R2
RLT 2.0 1.0 -0.072 -034 0.61 0.45
In vitro EROD 1.0 0.83 0.58 0.70
In vivo EROD 1.0 0.49 0.63
In vivo ELSM 1.0 0.95

 

a EROD = ethoxyresorufin O-deethylase; ELSM = early life stage mortality; WHO =
World Health Organization; TEF
tetrachlorodibenzo-p-dioxin; PeCDD

hexachlorodibenzo-p-dioxin;

TCDF

= toxic

pentachlorodibenzofuran; HxCDF = hexachlorodibenzofuran.

[131°1141 " 141

° [19] f[17] gthis study

I EC-50 estimates based on incomplete dose-response curves

86

equivalency factor;
pentachlorodibenzo-p-dioxin;
tetrachlorodibenzofuran;

TCDD
HxCDD
PeCDF

RLT 2.0-RPS are similar to RPs generated using a mammalian in vitro bioassay
(H4IIE rat hepatoma cells) (Table 4). RLT 2.0-RPS were similar to the limited set of
H4IIE-luc (recombinant H4IIE cells) -RPS reported in the literature (Table 4). They were
also approximately equal to H4IIE-wt (wild type) -RPs for TCDD, 1,2,3,7,8-PeCDD,
2,3,4,7,8-TCDF, and 1,2,3,7,8 PeCDF (Table 4). The RLT 2.0-RP for 2,3,4,7,8-PeCDF
was nearly five-fold less than the corresponding H4IIE-wt estimate (Table 4). It was
similar to H4IIE-luc-RPs and international TEFS for this congener, however. This
suggests that, the H4IIE-wt value may be an overestimate. The H4IIE-wt value for
2,3,4,7,8-PeCDF (Table 4) agrees with EROD based estimates in rainbow trout (Table 3).
Thus, the difference may be endpoint-dependent such that EROD assays may
overestimate the in vivo potency of this compound. The greatest differences between
H4IIE and RLT 2.0 RPS were observed for 1,2,3,4,7,8-HxCDD, and 1,2,3,4,7,8-HxCDF
(Table 4). RLT 2.0-RPS were 9- and 45-fold greater, respectively. All rainbow trout-RPS
(Table 3) were at least 4-fold greater than H4IIE-wt estimates (Table 4). The differences
suggest that endpoints in fish and fish cell lines, may be more responsive to
hexachlorinated PCDD and PCDF congeners than mammalian cells.

Due to the inability to generate point estimates of RP for the mono- and di-ortho
PCBS tested in this study, it is not yet clear whether the RLT 2.0 cell line is a better

model for predicting the in vivo potency of these compounds to fish. Differences in

87

Table 4. Comparison of RLT 2.0 bioassay derived relative potencies (RPS) to RPS based
on in vitro bioassay with H4IIE-rat liver cells (recombinant [Inc] and wildtype [wt]) and
international toxic equivalency factors (TEFS).

 

 

Compounda RLT 2.0 H4IIE-luc b H4IIE-wt° International
RP RP RP TEFS °
2,3,7,8 TCDD 1.0 1.0 1.0 1.0
1,2,3,7,8 PeCDD 0.225d 0.79 0.420 0.5
1,2,3,4,7,8 HxCDD 0.762° -- -- .0830 0.1
2,3,7,8 TCDF 0.235e -— -- 0.200 0.1
1,2,3,7,8 PeCDF 0.208° -- -- 0.200 0.05
2,3,4,7,8 PeCDF 0.296d 0.69 1.40 0.5
1,2,3,4,7,8 HxCDF 0.917c -- -- 0.020 0.1
PCB 77 (3,334,4’) 5.95 x 10'3 d" 0.00071 1.8 x 10'5 5 x 10“
PCB 105 (2,3,334,4’) < 5 x 105" <1x10'6 8 x 10‘6 1 x 10*1
PCB 118 (234,435) <6x10’3" <1r1045 3.5 x10'7 1x10'4
PCB 126 (3,334,435) 6.28 x 10'3 d 0.017 2.2 x 10‘2 0.1
PCB 156 (23,334,435) < 3.6 x 10'“ -- -- 5.5 x 10'5 5 x 10“
PCB 169 (3,334,435,5’) < 5 x 10'3 ° 0.00055 4.7 x 10“1 1 x 10'2

 

3‘ TCDD = tetrachlorodibenzo-p-dioxin; PeCDD = pentachlorodibenzo-p-dioxin; HxCDD
= hexachlorodibenzo-p-dioxin; TCDF = tetrachlorodibenzofuran; PeCDF =
pentachlorodibenzofuran; HxCDF = hexachlorodibenzofuran.
[7] ° [21] d[17] ° this study
f EC-50 estimates based on incomplete dose response curves

88

responsiveness of mammals and fish to mono-ortho PCBS have been reported [4,11,16].
This supports the hypothesis that a fish-specific in vitro bioassay model should be more
accurate in predicting the in vivo potency of samples containing mono-ortho PCBS to

fish, than a mammalian in vitro bioassay.

Sensitivity analysis

A common use for RPs and TEFS is in deriving a single value to characterize the
biological potency of a sample based on instrumental analysis. This is generally done by
multiplying the concentration of each compound detected by it’s RP or TEF and
summing the total for all compounds. The value attained is known as a TCDD equivalent
(TEQ). TEQs can be used to characterize potential biological potency based on
instrumental analysis alone or they can be applied in conjunction with bioassay. TEQs
calculated for risk assessment purposes tend to be based on TEFS. TEQs used in a mass
balance context, in conjunction with bioassay, tend to be based on assay specific RPs.
Comparison of RP-TEQS with bioassay derived TCDD-equivalents (TCDD-EQS)
provides a method to evaluate mass balance and possible interactions between
compounds. In the case of an unknown sample, if bioassay-derived TCDD-EQS are
approximately equal to a calculated TEQ, one can assume that the active compounds in
the sample have been detected and accounted for. Large differences between TCDD-EQS
and TEQS suggest either non-additive interactions between compounds or failure to

detect and account for all active compounds in the sample. Bioassay directed

89

fractionation and further instrumental analysis can be applied to elucidate which of the
above circumstances apply.

A sensitivity analysis was conducted to determine the uncertainty contributed to
assessment of complex environmental mixtures of HAHS by uncertainties in RPS
determined for individual congeners and by differences in RPS between in vitro and in
vivo responses and among species. The sensitivity analysis was conducted by calculating
a set of TEQS from representative sets of RPs and TEFS. Instrumental analyses reported
for samples of three species of fish (carp, walleye, and alewife) collected fiom Saginaw
Bay, MI, USA [22] were used [22-Tables 6-10, small walleye, alewife, carp]. Congener
concentrations and relative distributions in these samples should be representative of fish,
from a variety of trophic levels, exposed to HAHS in situ.

Sensitivity analysis was used to determine which RLT 2.0-RPS are likely to
contribute most to uncertainty in RLT 2.0-derived TEQS. A frequency distribution of
TEQS was generated for each species (Fig. 2) and a sensitivity analysis was performed to
determine the relative contribution of each RP to the total variation in the TEQ estimate
(Table 5). Values within the TEQ distributions for carp, walleye, and alewife varied up
to 15-fold (Fig. 2). Within the 95% confidence range, carp and alewife TEQS varied up
to six-fold and walleye TEQS varied up to eight-fold. This suggests that a ten-fold
uncertainty factor (i five-fold) is probably appropriate for TEQS derived from the current

set of RLT 2.0-RPs.

90

Forecast: Total TEQ-Ale
2.000 Trials Frequency Chart 27 Outliers

 

 

 

 

 

 

 

 

 

 

Forecast: Total TEQa-earp

 

 

 

    
 

   

AMI“!!! Emma." 37 DIM

1m 51

.021 ”m ”I. 417

, .71

g on I IIIIIIII .. 135 g 3
.3 1 IIIIIIIIII III III I g
3 am C-- mmwlw”, JLWHIII '\ “III'II'QJ .. ‘ 14:

000 I I I 7 ., 1 o

0.00 4) 75 run 1111: 175 on

 

 

 

Fig. 2. Frequency distributions for alewife (top), walleye (middle), and carp (bottom)
tetrachlorodibenzo-p-dioxin (TCDD) equivalents (TEQS) generated by Monte Carlo
Simulation (n = 2000 independent trials). RLT 2.0-relative potencies (RPs) for all
compounds listed in Table 1 were allowed to vary independently over empirically
defined, congener specific, frequency distributions.

91

Table 5. Sensitivity analysis: percent contribution to total variance in
tetrachlorodibenzo-p-dioxin (TCDD) equivalent (TEQs), calculated for three separate
samples, caused by congener specific uncertainties in RLT 2.0 relative potency (RP)
estimates. Determined by Monte Carlo Simulation (n = 2000 independent trials). RLT
2.0-RPs for all congeners listed in Table 1 were allowed to vary independently over
empirically defined, congener specific, frequency distributions“.

 

 

 

Carp-TEQB Walleye-TEQb Alewife-TEQb
Congener %c Congener %c Congener %c
2,3,7,8 TCDD 46 PCB 77 74 PCB 77 61
2,3,4,7,8 PeCDF 24 2,3,7,8 TCDF 13 2,3,7,8 TCDD 14
2,3,7,8 TCDF 13 2,3,7,8 TCDD 7.5 PCB 126 10
PCB 138 3.5 PCB 126 2.5 2,3,7,8 TCDF 6.5
1,2,3,7,8 PeCDF 3.5 PCB 138 1.0 PCB 138 3.5
All others 10 All others 2 All others 5

 

 

 

a PeCDF = pentachlorodibenzofuran; TCDF = tetrachlorodibenzofuran.
b Based on instrumental analysis of Saginaw Bay fish tissue samples [22].
° Refers to % contribution to total TEQ variance due to uncertainty in the RP estimate

for that congener.

92

Dioxin and furan congeners contributed most of the uncertainty to carp-TEQS,
while walleye and alewife TEQS were more sensitive to non-ortho PCBs (Table 5). In
carp, RPs for TCDD, TCDF, and PCDF were the major source of variability in the TEQ
estimate, with TCDD accounting for 46% of the variance (Table 5). Walleye- and
alewife-TEQS were more sensitive to uncertainty in the RPs for PCB 77 and 126 (Table
5). TEQs for all three species were sensitive to the RP estimate for PCB 138. This
suggests that simply using the maximum possible RP for this congener could lead to
significant overestimation of potency. This could confound mass balance analyses based
on such an estimate. Consequently, the RP of PCB 138 should be determined more
precisely. Even at their maximum possible RPs no other mono- or di-ortho PCBS
contributed significantly to the calculated concentrations of TEQ. This suggests that use
of the maximum limit for these congeners may not significantly bias the TEQ estimates
for fish with a similar exposure/ accumulation profile. Overall, the sensitivity analysis
suggests that more precise estimates of RP for TCDD, PCB 77, 2,3,7,8-TCDF, 2,3,4,7,8-
PeCDF, PCB 126, and PCB 138, in that order, would best decrease the uncertainty of
RLT 2.0 based TEQ estimates for similar samples.

Sensitivity of TEQ estimates to in vitro, in vivo, and between species differences
in reported RPs was evaluated. RLT 2.0, H4IIE-wt, ELSM, and international-TEQS were
calculated for carp, walleye, and alewife using the values presented in tables 3 and 4. In
order to avoid uncertainty associated with RP estimates for mono- and di-ortho PCB

congeners, TEQS were based solely on dioxin [22-Table 9], and furan [22-Table 10]

93

congeners presented in table 3, and PCBs 77 and 126 [22-Table 8]. The resulting TEQS
are presented (Fig. 3).

Within the range of certainty determined for RLT 2.0-TEQS there were no
statistically significant differences between RLT 2.0-TEQS and those generated using
other sets of RPs. TEQS based on RLT 2.0 bioassay were, on average, 50% greater than
those based on in vivo early life stage mortality in fish. H4IIE and international TEQS
were, on average, 140%, and 250% greater than ELSM-TEQS, respectively. All values
were within the same order of magnitude and all four sets of TEQS were significantly
correlated with one another (R2 = 0755-0998). These results suggest that, based on
dioxin, furan, and non-ortho PCB congeners, RLT 2.0-based characterization of a
complex mixture would yield essentially the same conclusions as in vivo ELSM
characterization despite absolute differences in RP estimates (Table 3) discussed
. previously. Furthermore, RLT 2.0 characterization would also be comparable to that
based on H4IIE-wt RPs and international TEFS. Relatively large differences in the RPs
for specific congeners like 1,2,3,4,7,8-HxCDF and -HxCDD between RLT 2.0 and
mammalian RPs did not result in equally large differences in TEQS. These results
support the hypothesis that the RLT 2.0 bioassay is a valid method for evaluating the
potency of AhR-active compounds. They do not, however, support the hypothesis that a
fish-specific cell line is necessary to estimate in vivo potency of dioxins, furans, and non-

ortho PCBS to fish.

94

 

120
1 00
80
60
40
20

TEQ

Carp

 

 

Sm. Alewife
Walleye

 

 

Fig. 3. Comparison of tetrachlorodibenzo—p-dioxin (TCDD) equivalents (TEQ) estimates
based on instrumental analyses of Saginaw Bay carp, walleye, and alewife samples [22],
calculated using RLT 2.0 (RLT), H4IIE-wild type (H4IIE), or in vivo rainbow trout early
life stage mortality (ELSM) relative potencies or international toxic equivalency factors
(INT) (Tables 3,4). TEQ estimates shown here are based on concentrations of PCBs 77
and 126, and the dioxin and furan congeners listed in Table 1.

95

Conclusions

RLT 2.0 in vitro bioassay is a useful tool for characterizing AhR mediated
biological potency, comparable to other assays currently used for this purpose. The
original RLT 2.0 assay method [17] was modified to increase efficiency and sample
throughput without marked loss in precision or sensitivity. RLT 2.0-RPs were reported
for a number of environmentally relevant HAHS but precise RP estimates for mono- and
di- ortho PCB congeners were not achieved. Sensitivity analysis indicated that RLT 2.0-
RPs could be applied with about 10 fold uncertainty to calculate TEQS based on
instrumental analysis. This is comparable with the uncertainty around most TEFs used
for risk assessment purposes. The utility of RLT 2.0-RPs could be increased by more
precise determinations of RPs for specific congeners identified by sensitivity analysis.
For dioxins, furans, and non-ortho-PCBs, RLT 2.0 bioassay does not appear to be more
accurate than analogous mammalian bioassays (like H4IIE assays) for estimating in vivo
potency in fish. The results presented do not support or reject the hypothesis that RLT
2.0 bioassay may more accurately predict the in vivo potency of mono- and/or di-ortho
PCB congeners, to fish, than in vitro bioassay with mammalian cells, however. Further

studies are needed to test this hypothesis.

96

ACKNOWLEDGEMENT

This work was supported by US EPA Biology Exploratory Grants Program, Grant
No. R8537l-01-0, cooperative agreement No. CR 822983-01-0 between Michigan State
University and the US. EPA - Office of Water Quality, the NIEHS - Superfund Basic
Research Program (NIH-ES-04911), and a Michigan State University Distinguished
Fellowship to D. Villeneuve. We thank Emily Nitsch for her assistance in maintaining
the cells. Finally we acknowledge the members of MSU’s Aquatic Toxicology

Laboratory (1995-1998) for their assistance and comments on the manuscript.

REFERENCES

1. Safe S. 1990. Polychlorinated biphenyls (PCBS), dibenzo-p-dioxins (PCDDS),
dibenzofurans (PCDFS), and related compounds: environmental and mechanistic
considerations which support the development of toxic equivalency factors (TEFs).

Crit Rev Toxicol. 21:51-88.

2. Spitsbergen JM, Kleeman JM, Peterson RE. 1988. Morphologic lesions and acute
toxicity in rainbow trout (Salmo gairdneri) treated with 2,3,7,8 tetrachlorodibenzo-p-

dioxin. J T oxicol Environ Health. 23:333-358.

3. Spitsbergen JM, Schat KA, Kleeman JM, Peterson RE. 1988. Effects of 2,3,7,8

tetrachlorodibenzo-p-dioxin (TCDD) or Arochlor 1254 on the resistance of rainbow
97

trout (Salmo gairdneri) to infectious haematopoietic necrosis virus. J Fish Dis.

11:73-83.

. Walker MK, Peterson RE. 1991. Potencies of polychlorinated dibenzo-p-dioxins,
dibenzofilrans, and biphenyl congeners for producing early life stage mortality in

rainbow trout (Oncorhynchus mykiss). A quat T oxicol. 21 :219-238

. Peterson RE, Theobald HM, Kimmel GL. 1993. Developmental and reproductive
toxicity of dioxins and related compounds: cross-species comparisons. Crit Rev

Toxicol. 23:283—335.

. Giesy JP, et al. 1993. Uptake, disposition, and effects of dietary 2,3,7,8-
tetrachlorodibenzo-p-dioxin on the survival, growth, reproduction, histology,
biochemistry, and haematology of rainbow trout. In Fiedler H, Frank H, Hutzinger
O, Parzfall W, Riss A, Safe S, eds, Organohalogen compounds: emission control,
transport and fate, environmental levels, and ecotoxicologv. Vol. 12. Federal

Environmental Agency, Austria. pp 235-238.

. Sanderson JT, Aarts JMMJG, Brouwer A, Froese KL, Denison MS, Giesy JP. 1996.
Comparison of Ah receptor-mediated luciferase and ethoxyresorufin O-deethylase
induction in H4IIE cells: implications for their use as bioanalytical tools for the

detection of polyhalogenated aromatic hydrocarbons. Toxicol Appl Pharmacol.

137:316-325.

98

10.

11.

12.

13.

Poland A, Knutson JC 1982. 2,3,7,8-tetrachlorodibenzo-p-dioxin and related
halogenated aromatic hydrocarbons: examination of the mechanism of toxicity.

Annu Rev Pharmacol Toxicol. 22:517-554.

Tillitt DE, Giesy JP, Ankley GT. 1991. Characterization of the H4IIE rat hepatoma
cell bioassay as a tool for assessing toxic potency of planar halogenated hydrocarbons

in environmental samples. Environ Sci Technol. 25:87-92.

Garrison PM, Tullis K, Aarts JMMJG, Brouwer A, Giesy JP, Denison MS. 1996.
Species-specific recombinant cell lines as bioassay systems for the detection of
2,3,7,8-tetrachlorodibenzo-p-dioxin-like chemicals. F undam Appl T oxicol. 30:194-

203.

Gooch JW, Elskus AA, Kloepper-Sams PJ, Hahn ME, Stegeman JJ. 1989. Effects of
ortho and non-ortho substituted polychlorinated biphenyl congeners on the hepatic

monoxygenase system in scup (Stenotomous chrysops). T axicol Appl Pharmacol.

98:422-433.

Denison MS, Wilkinson CF, Okey AB. 1986. Ah receptor for 2,3,7,8-
tetrachlorodibenzo-p-dioxin: Comparative studies in mammalian and nonmammalian

species. Chemosphere. 15: 1665- 1667.

Clemons JH, van den Heuvel MR, Stegeman JJ, Dixon DG, Bols NC. 1994.
Comparison of toxic equivalent factors for selected dioxin and furan congeners

derived using fish and mammalian liver cell lines. Can J Fish Aquat Sci. 51:1577-

1584.
99

14.

15.

16.

17.

18.

19.

Parrot JL, Hodson PV, Servos MR, Huestis SL, Dixon DG. 1995. Relative potency
of polychlorinated dibenzo-p-dioxins and dibenzofurans for inducing mixed function

oxygenase activity in rainbow trout. Environ Toxicol Chem. 14:1041-1050.

Villeneuve DL, Crunkilton RC, DeVita WM. 1997. Aryl hydrocarbon receptor-
mediated toxic potency of dissolved lipophilic organic contaminants collected from
Lincoln Creek, Milwaukee, Wisconsin, USA, to PLHC-l (Poeciliopsis lucida) fish

hepatoma cells. Environ T oxicol Chem. 16:977-984.

Hahn ME, Lamb TM, Schultz ME, Smolowitz RM, Stegeman JJ. 1993. Cytochrome
P4501A induction and inhibition by 3,3 ’,4,4’-tetrachlorobiophenyl in an Ah receptor-

containing fish hepatoma cell line (PLHC-l). Aquat T oxicol. 26:185-208.

Richter CA, Tieber VL, Denison MS, Giesy JP. 1997. An in vitro rainbow trout cell
bioassay for aryl hydrocarbon receptor-mediated toxins. Environ Toxicol Chem.

16:543-550.

Kennedy SW, Jones SP. 1994. Simultaneous measurement of cytochrome P4501A
catalytic activity and total protein concentration with a fluorescence plate reader.

Anal Biochem. 222:217-223.

Van den Berg M, et al. 1999. Toxic equivalency factors (TEFS) for PCBS, PCDDs,

PCDFS, for humans and wildlife. Environ Health Perspect. 106:775-792.

100

20. Van den Berg M, Jongh JD, Poiger H, Olson JR. 1994. The toxicokinetics and
metabolism of polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans

(PCDFS) and their relevance for toxicity. Crit Rev T oxicol. 2421-74.

21. Tillitt DE, et al. 1996. Dietary exposure of mink to carp from Saginaw Bay. 3.
Characterization of dietary exposure to planar halogenated hydrocarbons, dioxin

equivalents, and biomagnification. Environ Sci T echnol. 30:283-291.

22. Giesy JP, et al. 1997. Concentrations of polychlorinated dibenzo-p-dioxins
(PCDDS), polychlorinated dibenzofurans (PCDF) and polychlorinated biphenyls
(PCBS) in fishes from Saginaw Bay, Michigan, USA. Environ Toxicol Chem.

16:713-724.

101

Chapter 3

DERIVATION AND APPLICATION OF RELATIVE POTENCY ESTIMATES

BASED ON IN VITRO BIOASSAY RESULTS

Daniel L. Villeneuve*, Alan L. Blankenship, and John P. Giesy
Dept. of Zoology, National Food Safety and Toxicology Center, and Institute for

Environmental Toxicology, Michigan State University, East Lansing, MI, USA, 48824

Submitted to Environ. Toxicol. Chem. 10/99; accepted with minor revisions 03/00

102

Abstract- Relative potency (RP) estimates are widely used to characterize and compare
the potency of a wide variety of samples analyzed using in vitro bioassays. The most
commonly reported RP estimate is a single ratio of potency estimates calculated at a
defined level of response such as the EC-SO. RPs based on a single ratio of point
estimates are only valid when the slope and efficacy (maximum achievable response) of
the sample are equivalent to those of the standard to which they are being compared.
Often, these conditions are either violated or cannot be demonstrated. As a result, there
is a need to calculate and present relative potencies in a manner which addresses the
potential uncertainties caused by violation of the assumptions of parallel slopes and equal
efficacy. Uncertainty due to non-parallel slopes can be evaluated mathematically, but
uncertainty due to efficacy differences cannot. Use of multiple point estimates to derive
ranges of RP estimates, termed RP-bands, is recommended. RP-bands can provide a
discrete characterization of relative potency without sacrificing accuracy. Furthermore,
they provide a means to test the assumption of parallel slopes in situations where
statistical tests are not applicable. A systematic method for evaluating sample efficacy
has been developed into a framework to guide the derivation and application of RP
estimates based on in vitro bioassay results. Use of the systematic framework and
multiple point estimates was illustrated using three sample data sets. It is hoped that the
framework and discussion presented will facilitate the use of bioassay-derived RP
estimates to characterize samples of both known and unknown composition without

sacrificing accuracy.

103

INTRODUCTION

In vitro bioassays can be useful tools for environmental monitoring. They
provide rapid and cost-effective methods to screen large numbers of samples for their
ability to elicit a biological response through a specific mechanism of action. Though
instrumental analyses are essential for the identification and quantitation of compounds
in complex environmental mixtures, instrumental results are not well suited for predicting
and/or understanding potential effects of complex environmental mixtures on biota [1].
Instrumental analyses can miss compounds that are biologically active at concentrations
below analytical detection limits or compounds for which there are no established
methods or analytical standards [1]. Furthermore, even when compounds are detected,
instrumental analyses provide no information regarding their biological potencies,
particularly in conjunction with the other components of the mixture. In vitro bioassays
can integrate the overall potency of samples containing complex mixtures of compounds
inactive compounds, agonists, and antagonists which may be interacting both additively
and non-additively. Because they are based on biologically relevant mechanisms of
action, in vitro bioassays can provide an indicator of the mechanism-specific, biological
potency of a sample.

To facilitate quantitative risk assessment and simplify data interpretation, a
complex mixture’s potency to cause a defined biological response is ofien expressed
relative to that of a well-characterized standard or prototypical compound (e. g. 2,3,7,8-
tetrachlorodibenzo-p-dioxin [TCDD] and 17-B-estradiol [E2]) [2-4]]. In theory,

expression of sample potency in terms of equivalents of a standard compound allows for

104

comparison of diverse samples and may provide a basis for approximating risk to
biological organisms with analogous biochemical processes [1,2]. Additionally,
bioassay-derived equivalents can be compared to equivalents calculated from congener-
specific instrumental analyses and established toxic equivalency factors (TEFS) or assay
specific relative potencies (REPS) in a mass balance analysis [1,5-7]. This kind of
analysis can be used to assess whether the known composition of a sample (based on
instrumental analyses), can account for the bioassay responses observed [1,5-10]. Lack
of agreement between bioassay-derived and instrumentally-based equivalents can suggest
either the presence of unidentified, biologically active, compounds in a sample, or non-
additive interactions between components of the sample [1,5-10]. Thus when coupled
with chemical fractionation and instrumental analysis, in vitro bioassays become very
powerful tools for characterizing environmental samples when accurate relative potency

estimates can be derived.

Relative potency estimation

Relative potency estimation or the determination of TCDD- or Ez-equivalents (-
EQ) can be considered a type of indirect bioassay. An indirect bioassay has been
described as one in which an estimate of equally effective doses of standard and sample
is determined, such that the inverse ratio of their equally effective doses describes the
potency of the sample relative to a standard [11]. The underlying assumption of an
indirect assay is that doses of standard and sample which yield some selected magnitude

of response in a bioassay can be defined as equally effective doses [11]. Thus, at any

105

given magnitude of response, the potency of a sample relative to a standard is the ratio of

the doses of standard and sample needed to elicit that response [11] (Equation 1).

Relative potency = dose std.- / dose sample,-

where i = a defined magnitude of response (1)

The concentration needed to elicit a 50% response is widely regarded in
toxicology as the standard point estimate for describing the potency of a chemical to
elicit a specified response in an organism or test system [12]. EC-SOS and other point
estimates (ECX) can be calculated using a wide variety of methods including probit or
logit analysis [12-14], Scatchard and Woolf analysis [15,16], graphical interpolation,
linear regression, and non-linear regression [17,18]. As a result, relative potencies have
traditionally been calculated as the ratio of the EC-SOs of the two compounds being

compared (Equation 2).

relative potency = EC-50A / EC-503 (2)

typically where A = a well characterized standard compound and B= the test

compound

Although this is a logical approach which seems compatible with the nature of
indirect bioassay, relative potencies based on a ratio of point estimates are only valid

under very limited conditions [11,14,19,20]. Indirect bioassay assumes that the sample

106

being analyzed responds as if it were a dilution (or more concentrated form) of the
standard compound [11]. This implies that the dose-response curves being compared are
parallel (Fig. 1A) [11,14,19]. They should be effectively identical except for their
position along the x-axis when log dose is plotted against response (Fig. 1A) [20]. For
these conditions to be met, the dose-responses must have a common slope [11,14,19] and
the maximum achievable response (efficacy) for the standard and sample must be
identical [20] (Fig. 1A).

For non-parallel dose-response relationships (Fig. 1B) relative potency is a
function of dose [11,20]. The relationship described at a single level of response, such as
the EC-50, is not constant over the entire range of responses for the compounds being
compared [14] (Fig. 1B). Relative potency estimates based on a ratio of EC-205, EC-SOs,
and EC-803, for example, would be quite different (Fig. 1B). Thus, a point estimate of
relative potency may only represent a single point along a broad range of potential
relative potencies. Use of a single point estimate can lead to misleading and/or
inaccurate interpretations. When dose responses cross in the positive quadrant even the
rank order of potency may change [20]. Thus, it has been argued that point estimates of
relative potency do not provide an accurate characterization of relative potency for
compounds whose dose-responses are not statistically parallel [14,20]. Parallelism
should be demonstrated before a relative potency estimate is calculated [11,14,19].

Statistical methods for testing parallelism are available [11,14,19]. Such tests are
of little value in attempting to compare complex mixtures or unknowns to a standard
compound, however. Due to their complex or unknown composition environmental

samples and other unknowns cannot be assigned a meaningful set of dose units which can

107

 

120
100 + “Minimum 33§B9£§€.(E‘T' ac

   

6° I 13050
40 ‘F‘

20 ~-

......

 

°/o-TCDD-max.

 

 

 

 

 

 

100 A .3
90 /
80 _ _

—o—A

 

 

B 1

 

 

 

 

 

150 200

Fig. 1. A. Illustration of relative potency estimation for dose response relationships
which conform to the assumptions of indirect bioassay (i.e. equal efficacy and parallel
slopes). Relative potency is constant over the effective response range. B. Illustration of
relative potency estimation for dose-response relationships which do not conform to the
assumption of parallel slopes. Relative potency varies over the effective response range.

108

be statistically compared to those of the standard compound. Even if they can be plotted
on a common scale such as volume of extract, mass of sample, percent dilution, etc., a
statistical test for parallelism is not biologically meaningful. As a result, statistical tests
for parallelism cannot be applied to unknowns and complex mixtures.

Furthermore, there is little reason to assume that dose-response relationships for
complex mixtures analyzed by the same in vitro bioassay will be parallel or show equal
efficacy. Even for single compounds of known concentrations, effects such as partial
agonism, differences in binding affinities, in vitro toxicokinetics, solubility, sensitivity to
environmental conditions, etc. can produce non-parallel dose-responses and limit the
magnitude of response that can be achieved [20]. In a complex mixture, the likelihood of
violating the assumptions of indirect bioassay is greatly increased. Interactions between
compounds in the sample could be expected to produce variations in the shape of the
dose-response relationship. For example, as the concentration of agonists in the sample
increases, so too might the concentration of antagonists, such that a more gradual slope
and lower maximum response is produced. Alternatively, compounds in a complex
mixture may affect other aspects of cell function which cross-talk with the pathway of
interest to modulate the magnitude and/or rate of change in response relative to dose [21].
Furthermore, for environmental samples, lack of control over sample composition,
limited aqueous solubility, limited sample volume, and the need to conserve sample for
other analyses can often limit the ability to achieve a maximal level of response. As a
result, sample efficacy is often unknown. Such factors could be expected to produce

significant variation in dose-response relationships from sample to sample or relative to a

109

standard. Thus, parallelism and equal efficacy can neither be assumed nor, in the case of
parallelism, demonstrated statistically, for complex mixtures and unknowns.

While numerous authors have discussed the problems associated with the use of
point estimates to characterize relative potencies, there remains a need to reduce complex
dose-response data to simple quantitative estimates, in order to simplify data
interpretation and risk assessment. The purpose of this paper was to develop a
framework to facilitate the use of bioassay-derived relative potency estimates to
characterize samples of known and unknown composition without sacrificing accuracy of
data interpretation.

Relative potency estimation methods which incorporate multiple point estimates
(MPE) are recommended as an alternative to single point estimation techniques in
situations where adherence to the assumptions of indirect bioassay cannot be
demonstrated. These estimates are presented in a manner which identifies uncertainties
in the relative potencies derived. A systematic framework for evaluating in vitro
bioassay results was developed to facilitate evaluation of indirect bioassay assumptions,
direct selection of appropriate relative potency estimation techniques, and guide the use
and interpretation of the estimates generated. Application of MPE methods and use of

the systematic framework is demonstrated using several example data sets.

110

METHODS

Multiple Point Estimates

In order to be meaningful for risk assessment, relative potency estimates must be
representative of the relationship between sample and standard over their entire effective
range of response. The problem with non-parallel dose-responses is that a single ratio of
point estimates does not provide a characterization that is representative of all positions
along the curves. Relative potency estimates will vary with the response level selected
[11,14,20] (Fig. 1B). Estimates based on the 50% level of response (EC-50) may give
very different results from those based on the 20% or 80% response [14,20,22] (Fig. 1B).
One solution to the problem stated above would be to report relative potencies as a
function, rather than a quantitative estimate [20]. Although more accurate, this would
tend to make risk assessment and data interpretation a rather cumbersome and laborious
process. As an alternative, when the dose-responses being compared are non-parallel or
cannot be demonstrated to be parallel, statistically, relative potency can be determined for
multiple points along the effective range of responses. The range of relative potency
values generated, which we have termed a relative potency band (RP-band), can provide
useful information about the dose-responses being compared without sacrificing
accuracy.

MPE techniques provide an empirical method for testing the parallel slopes
assumption of indirect bioassay. For parallel curves, RP is independent of response (Fig.

1A) (or dose), thus, the RP-band for parallel curves should be, essentially, a single value.

111

The wider the RP-band, the greater the curves deviate from parallelism. Thus, coupled
with empirical observations of sample efficacy, the calculation of an RP-band allows the
assumptions of indirect bioassay to be evaluated for complex mixtures and unknowns.

MPE methods also provide a means to derive useful and accurate relative potency
information for non-parallel dose-responses. RP-bands provide a measure of the
potential range of uncertainty in relative potency values that is generated by selection of
the response level at which a point estimate is taken (uncertainty due to non-parallel
slopes). As long as the sample and standard dose-responses have the same efficacy and
have been modeled over the entire range of effective response, the RP-band generated
provides a quantitative estimate of relative potency that is valuable for comparing among
samples as well as establishing TEFS or RPs, conducting mass balance analyses, guiding
risk assessments, etc. The only limitation is whether or not the range of uncertainty in
the RP estimate is sufficiently small for the desired application. For example, an RP-
band which ranges from 1-10 pg TCDD-EQ/ g sample may be usefiil for risk assessment
purposes, but may not be sufficient for mass balance analysis or comparing it’s potency
to that of similar samples.

One limitation to the RP-band approach is that the width of the band is sensitive
to the range of responses selected. As a result, in order for RP-bands to be directly
comparable and give an independent measure of the uncertainty due to non-parallel
slopes, it is necessary to standardize the range of response over which they are calculated.
This standard range has arbitrarily been defined as 20%-80% of the maximum response
achieved for the standard compound (20-80%-std.-max.). For samples with different or

unknown efficacies, it may be necessary to extrapolate beyond the known range of

112

response in order to calculate an RP-band over this range. In such cases, the

uncertainties involved in such an extrapolation should be reported and discussed.

Relative Potency Bands

RP-bands are fairly simple to calculate. The first step involves fitting an
appropriate regression model to the dose-response data. Once a regression model has
been defined for each of the dose-responses being compared, the regression equation for
each curve is used to solve for the doses associated with multiple levels of response (Y,)
over a standard range (20-80%-std.-max.) and equation 1 is used to calculate a relative
potency value (RP,) for each level of response (Y,) selected. The maximum and

minimum RP, values define the limits (or range) of the RP-band (Equation 3).

RP-band = minimum RP,- to maximum RP,- (Equation 3)
where RP,- = the relative potency determined at a defined level of response Y,-
and multiple values forY, are tested such that they encompass the standardized

range of 20—80%-std. -max..

The simplest approach for calculating the RP-band involves selecting response levels (Y,)
which correspond to 20%-, 50%-, and 80%-std.-max. and calculating an RP, value for
each response value (Y,). In situations where slopes vary over the standardized range of
response, Monte-Carlo simulation may be used to calculate RP, over a uniform

distribution whose limits are defined as 20-80%-std.-max. Using this approach RP, can

113

be calculated rapidly for hundreds of points along the standardized range of response, and
the maximum and minimum RP, generated can be reported. This can be done using
computer programs such as Crystal Ball® (Decisioneering, Boulder, CO, USA). The
principle is the same, only the number of response levels (Y,) tested differs. For cases in
which equal efficacy cannot be demonstrated, one of the points (Y,) selected should be
the response level corresponding to the maximum response observed for the sample in
question. The RP, determined at this point should be identified when presenting the RP-
band, and all RP, calculated at responses greater than the maximum observed
(extrapolated RP,) should be highlighted to indicate that they were based on extrapolation

beyond the range of empirical data.

Interpretive framework

A systematic framework to guide derivation, critical evaluation, and use of
relative potency estimates based on in vitro bioassay results was developed (Fig. 2). The
framework was designed as a dichotomous decision tree that guides critical evaluation of
the dose-responses being compared. A series of yes/no questions regarding the
properties of the dose-responses guides the user to a reasonable method for calculating

and interpreting relative potency using MPE methods.

Fit regression model (Fig. 2-A). The first step in any attempt to estimate relative potency

is to fit an appropriate regression model to the dose-response relationship. Dose

114

 

 

 

 

 

 

 

 

 

 

62:.me “.9800 05 a. .u

 

 

EL.

 

 

 

W J
.350. 85093.6
.o :88 8a c0388
83 S 2.888 88.3
8a... . .58880 x»:
W J .2 23.3 3 .6... 83...:
.3358!
.0: 3: 8:33 .8368 9.9:: 58858
89.. . 8:838 .3: .o. .355. 83...:
.8188 9.0:... cont—3&8 .6860
.o. 2.5.8 92.8.. a: 29:8 c. 253.85
2. 38.8. .820
0m .2 88%828 8.22.8.6 a. .2. 83 o...
x- .
E888. 85.82 .o :88 2. 28.8. 2.86
.8 v 22.3 83d: 88:
.08 co 823M828
98.2 .o 92.8.. .50: 85-99898 52. ma: 8:88 .38.: .a 8..
f U f .xk ._.
a o... _ all» E.
.xg..§-$o~
A 03mg... 05 .2 8:008. wm>

fl

 

 

 

6:33:88 .2005
.0 £30 .8. 30> :8 .o

 

 

.0: 8:93 80:.
28.588: x»... .2 8:
629:8 9.06: 28:09:00
.0. 035.3 89:80 um

.099..qu
wa .2 25:82
:38». 505.002

.8 v 29.;
xdm S 8.3 vwage...

28.2 .o 32.8 .58

0:509:00 8:658 05
.0.“ @0520: ocean... 8:88. 82:38:. 05 m0 owfiaoobm : w: 38898 8:38. 29:8 n .88-.Em-..\o .A.x:E-.Em-o\oow-oNV
88030. «0 own... Faucfim a .0>0 003.0: 82? 8:82. 03.28 «0 omen. 05 u :53 mm .33): 83838 .500 0322.. mam:

88 .3885 05> 5 80¢ 838:8 adv 5:80.“ 0330. m0 83:239. can meagre: 02% 0. x8305 032.85% .N .3..—

 

 

 

 

W J
.8860 c. 08:82::
9 on: 03588::
.0288 .36 8:33 8!...
25:88.30 .3: .2 83

8.068
9.080 28:38 .0.
035.3» 8958 good”.

9:09am .0 2.3.0:

 

 

 

 

 

28...... 28.8... .688
0.068 002030 on; .u

38.5.... 29.8 .806
230000 a. 3:208. .. on: 802.0 809.8»
8:33 89: . E06888 60.. 80:6 .6850
.3: .o. 28.8 22.8w 0168.2. 38...... 2.86
0:09 am :08! 0:3 mm :83...
38:. .3098 En... ma! .x9:.9?$8.0~ EB. ma!
.2 I .o\
* Oz — mw>b
$82.» 2:85... EB.

 

 

 

 

 

 

a 2

 

 

 

 

30 M
9...qu o. 68 83 .0 mm».

 

 

05 .9 090200 3:088. mm».

 

 

 

 

 

 

 

 

115

responses from in vitro bioassay will generally not be linear when plotted in the original
dose and response units. In some cases, however, a linear transformation can be used to
simplify data analysis without biasing the conclusions [11]. Commonly used
transformations include log-dose, probit, logit, and logistic [11,14,18,23-25]. In addition
to linear transformations and linear regression, non-linear models may be applied to
characterize dose-response relationships [17,18,24,25]. In particular, generalized linear
models and other non-linear regression techniques have been recommended for modeling
relatively low responses since they utilize the inherent S shape of the dose-response
relationship [18,22,24-26]. Choice of an appropriate model to describe a concentration-
response curve is often arbitrary, but may be quite critical, particularly when the
estimation technique involves extrapolation from the observed data [22]. Readers are
directed to the references provided above for detailed descriptions on fitting regression

models to dose response data.

Evaluate Sample Efi‘icacy. For indirect bioassay to be valid, the samples and standard
being compared must have the same efficacy [11,20]. The assumption of equal efficacy
can be tested empirically by examining whether the maximum responses observed for the
samples and standard were approximately equal (or statistically equal if replicate dose-
responses are available). This requires that the maximal response of the sample is known
(Fig. 2-B). For the purpose of this discussion a maximal response is defined as the
magnitude of response at the point where the S-shaped dose—response curve reaches the

upper plateau.

116

When a maximal response is achieved, the assumption of equal efficacy can be
evaluated (proceed to Fig. 2-C). If the maximal response of the sample is less than 20%-
std.-max. the assumption of equal efficacy is clearly violated and MPE will generally not
be suitable for deriving a reasonable RP estimate (Fig. 2-1). If necessary, a point estimate
of RP can be made at a level of response less than 20%-std.-max., but such an estimate is
not suitable for risk assessment or mass balance purposes (Fig. 2-1). Non-linear
regression methods are recommended for deriving point estimates of RP at response
levels less than 20%-std.-max. (Fig. 2-I). Such estimates may be cautiously applied for
comparing among samples (Fig. 2-1), but the marked differences in efficacy between
sample and standard should be clearly noted. If the maximal response of the sample is
greater than 20%-std-max., it is generally feasible to derive at least an approximate
estimate of relative potency using MPE methods (proceed to Fig. Z-E). If replicate dose-
response curves are available, statistical methods may be used to determine whether the
observed efficacy of the sample and standard are statistically equivalent (Fig. 2-E). In
cases where the observed maximal response for the sample and standard differ markedly
(Fig. 2-G), MPE methods can still be employed to derive an approximate estimate of
sample RP. Such estimates are not completely valid, however. Differences in efficacy,
and extrapolation beyond the empirical range of sample response should be clearly noted
and considered when using the approximations for comparative, risk assessment, or mass
balance purposes (Fig. Z-G). In cases where the observed maximum response of the
sample and standard do not differ markedly, RP-bands provide a valid quantitative

estimate of relative potency (Fig. Z-H). Such estimates are suitable for risk assessment or

117

mass balance applications, provided the width of the RP-band and the resultant range of
uncertainty in the estimate is small enough to provide adequate resolution.

If a maximal response was not achieved, equal efficacy cannot be demonstrated
empirically or tested statistically (proceed to Fig. 2-D). If possible, the sample should be
tested at greater concentrations (Fig. Z-J). In some cases, this may not be feasible,
however, due to logistical constraints. In situations where the maximal response of the
sample is unknown, a RP estimate can be derived, but the uncertainties due to unknown
efficacy must be considered (proceed to Fig. 2~F). If the observed maximum response for
the sample is greater than 20%-std.-max., MPE methods can be used to calculate an RP-
band approximation of the sample RP (Fig. 2-G). Again, the extrapolated region of the
band and potential uncertainty due to unknown efficacy must be identified and discussed
when presenting and applying the estimates (Fig. 2-G). If the observed maximum
response of the sample is less than 20%-std.-max., use of MPE to calculate an RP-band is
not feasible (Fig. Z-H). If an estimate is needed, non-linear regression may be used to
derive a point estimate of RP at a level of response less than 20%-std.-max (Fig. l-L).
Such an estimate should be interpreted and applied conservatively, however, due to the

inability to evaluate the assumptions of indirect bioassay.

Example data sets

Several example data sets were used to demonstrate the use of the systematic

framework (Fig. 2) and RP-bands to evaluate relative potency based on in vitro bioassay

118

a. data set I

120 TCDD
+
100 —a— PCDF

80 + HxCDF x—x
60 —-—x-- TCDF /

 

 
 
  
 
 
 
 

%-TCDD-max
.b
O

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

   
  
 

 

    

 

 

 

 

 

 

20
0 ' f
-20
'2 0 log fmol 2 4 6
b. data set 11 TCDD standard curves
200 _ 120
: —a—1
x 150 -; +9 § 100 -
CU : +12 E 80 -
g 100 -=- "*-‘5 ‘ 0'
D : —D—16 D 60 "l
D I 0
f3 50 1E '7 4° '
oi” o 5 "\° 2° ‘
:- o -
-50 0
-3 -2 l1 0 1 log fmol
og pl
3‘CDD standard curves
120 c. data set 111 12
100 -Er+3.d >é 100'
é +3-p cu 30 .
E 80 _E +2-d g J
o' 60 :fi 8 6°
8 40 " —o— 1-p g 40 -
t; 20 -; & kg, 20 -
°\ 0 o -
-20 : -2o
-2 - 1 1 0 1 -2 o 2 4
0g p log fmol

Fig. 3. Example data sets. (a.) data set I - H4IIE-luc responses to 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD), 1,2,3,7,8-pentachlorodibenzofi1ran (PCDF),
1,2,3,4,7,8-hexachlorodibenzofuran (HxCDF), and 2,3,7,8-tetrachlorodibenzofuran
(TCDF). (b.) data set 11 - H4IIE-luc responses to extracts of sediment from Masan
Bay, Korea and corresponding TCDD standard curves. (c.) data set 111 - RLT 2.0
responses to extracts of sediment collected from a superfund site contaminated with
Aroclor 1268 and corresponding TCDD standard curves. Response magnitudes
presented as a percentage of the mean maximum response observed for the
corresponding TCDD standard (%-TCDD-max.)

119

results. Example data sets I (Fig. 3a) and 11 (Fig. 3b) represent the response of H4IIE-
luc cells [27] to individual halogenated aromatic hydrocarbons (HAHS) and sediment
extracts, respectively. H4IIE-luc cells are rat hepatoma cells stably transfected with a
luciferase reporter gene under control of dioxin responsive enhancers (DRES) [27]. In
vitro H4IIE-luciferase assays were conducted as described previously [10,28]. 2,3,7,8-
tetrachlorodibenzo-p—dioxin (TCDD), l,2,3,7,8-pentachlorodibenzofuran (PCDF),
l,2,3,4,7,8-hexachlorodibenzofuran (HxCDF), and 2,3,7,8-tetrachlorodibenzofuran
(TCDF) analyzed for data set I (Fig. 3a) were purchased from Accustandard (New
Haven, CT, USA). Working solutions and dilution series were prepared in pesticide
residue analysis grade isooctane (Burdick and Jackson, Muskegon, MI, USA). Dose
responses for PCDF and HxCDF consisted of six 3-fold dilutions, with the maximum
concentration being equal to 8.8 and 8.0 nM for PCDF and HxCDF, respectively. TCDF
was tested at the following six concentrations: 98, 33, 1.2, 0.40, 0.13, and 0.045 nM.
Sample dose response curves were compared to a TCDD standard curve (IO-fold dilution
series, 6 concentrations, 10-0.0001 nM). Sediment extracts (data set 11) were prepared
from sediments collected from Masan Bay, Korea. Sample collection, extraction,
fractionation, and analysis was performed using methods detailed elsewhere [28,29].
Dose-responses consisted of six 3-fold dilutions of each extract or fraction. Sample dose-
response curves were compared to a TCDD standard curve (5-fold dilution series, 6
concentrations, 0.67 - 2100 pM).

Example data set III (Fig. 30) was generated using the RLT 2.0 bioassay which
utilizes rainbow trout hepatoma cells transfected with a luciferase reporter gene under

control of DREs to screen for Ah-receptor-active compounds [30,31]. Extracts from

120

sediments collected from a superfund site contaminated with Aroclor 1268 were
analyzed. Detailed methods for the collection, preparation, fi'actionation, and
instrumental analysis of the sediment samples were reported elsewhere [32-34]. RLT 2.0
bioassay was performed according to the glow method presented by Villeneuve et al. [31]
which is a modification of a procedure presented by Richter et al. [30]. Dose-responses
consisted of six 3—fold dilutions of each extract or fraction. Sample dose-response curves
were compared to a TCDD standard curve (5-fold dilution series, 6 concentrations, 0.67 -

2100 pM).

Data analysis

Example data sets I, II, and III were analyzed using the systematic framework
outlined above. Sample responses expressed in relative luminescence units (RLU) were
converted to a percentage of the mean maximum response observed for the TCDD
standard (%-TCDD-max.). The mean solvent control response was subtracted from both
sample and standard responses, prior to conversion to a percentage, in order to scale the
values from 0 to 100%-std.-max. Responses expressed as %-TCDD-max. were plotted as
a function of either log pl (sediment extracts) or log frnol (standards and known
compounds) (Fig. 3). For simplicity, the linear portion of each example dose response
was defined by dropping points from the tails until a R2 2 0.95 was obtained and a linear
regression model was fit. Although this technique may not be appropriate for all
situations, particularly for evaluation of relative potency at low levels of response, it was

sufficient for the purposes of this paper. Regression equations for the samples and

121

standards were then used to calculate RP, values for Y, = 20, 50, and 80%-TCDD-max.
(Equation 1). For samples whose observed maximum response were less than 80%-
TCDD-max. an additional RP, estimate was calculated at Y, = observed maximum
response expressed as %-TCDD-max. RP-bands were then generated based on the
multiple RP, point estimates. For comparative purposes, RP-bands were also calculated
by Monte Carlo analysis (2000 iterations, uniform response distribution ranging from 20-
80%-TCDD-max.) using Crystal Ball® (Decisioneering). For all the example data sets
used, Monte Carlo analysis gave the same RP-band as the simple 3-point or 4—point

calculation.

RESULTS and DISCUSSION

Applying the Framework

Data set 1. A linear regression model was fit to each dose-response curve (Fig. 2-
A). In this case, the samples consisted of single compounds of known concentration and
three replicate dose-response curves were available for each compound tested. As a
result, statistical methods could be used to test the assumptions of indirect bioassay.
Maximal responses were achieved for all samples (Fig. 2-B; Fig. 3a) and all those
responses were greater than 20%-TCDD-max (Fig. 2-C; Fig. 3a). Sample efficacy was
not statistically different from that of the TCDD standard curve (t-test, two-tailed, 2 df,

p < 0.05). As a result, multiple point estimates provided RP-bands that were valid for the

122

samples and suitable for risk assessment, mass balance, etc. (Fig. 2-H). The width of the
RP—bands for the samples in data set I was very small (Fig. 4a). This is in agreement
with the fact that the slope of the regression lines for the sample dose responses were not
significantly different from that of the TCDD standard curve (t-test, two-tailed, 2 df, p <
0.05).

Data set 11. The linear regression model described above (methods) was fit to the
dose-responses (Fig. 2-A). Sample 15 showed no significant activity. Changes in the
slopes of the dose-responses for samples 9, 12, and 21 suggest that the responses were
approaching a maximum (Fig. 2-B; Fig. 3b), but in all cases the maximum observed was
significantly greater than the efficacy of the TCDD standard i it’s 95% confidence
interval (Fig. 2-C,E; Fig. 3b). Samples 1 and 16 did not appear to have reached a
maximal response (Fig. 23; Fig. 3b). They could not be tested at a greater concentration
(Fig. 2-D), but their observed maximum responses were near 100%-TCDD-max (Fig. 2-
F; Fig. 3b). Multiple point estimates were used to generate RP-bands for samples 1, 9, 12,
16, and 21 (Fig. 4b; Table 1). Because all the active samples showed a maximum
observed response greater than 80%-TCDD-max., only Y, = 20-, 50-, and 80%-TCDD-
max. were considered. All the RP-bands generated were suitable for comparing among
samples (Fig. 2-G,K) and no extrapolation was needed to generate the standard RP-band
estimate. Unfortunately, uncertainty due to deviations from parallelism to the TCDD
standard curve limited the ability to resolve and definitively rank the relative potency of
the samples (Fig. 4b). The fact that the efficacy of the samples was either unknown or

greater than that of the TCDD standard suggests that these estimates should not be used

123

RP

fmol TCDD-EQ/ul

fmol/pl

0.12

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a. data set I
0'1 4' E: - RP-20
0'08 " [:53 RP-50
0.06 -- CI RP-80
0.04 --
0.02 +-
o - . 1:621 .
I I _—
PCDF HxCDF TCDF
160
140 ‘_.b. data set 11 —o— - RP-20
‘20 r _0_ RP-50
£3 ;C a RP-80
4o --
Lo—
20 .. —I—
1 9 12 16 21
sample
1000 _
c. data set 111 x RP max
100 D RP-80
+ CEO [3'3 RP-50
10 - RP-20
T [in
0.1 | = : = I
3-d 3-p samplez'd 2'9 1-d

Fig. 4. Relative potency (RP) bands for three example data sets. RP-20, RP-50,
and RP-80 refer to RPs calculated as a ratio of potency estimates (Equation 1)
where the defined level of response (Y,) was 20—, 50-, and 80%-TCDD-max.
respectively. RP-max refers to the RP calculated at Yi = the maximum
magnitude of response observed for the sample expressed as %-TCDD—max.
Bars indicate regions of the RP band within the range of empirical data. A line
extending beyond the bar indicates the region of the RP band which is based on
extrapolation beyond the range of the empirical data.

124

 

Table 1. Relative potency estimates for example data sets (F ig. 2).

 

 

 

 

 

 

 

Data set-sample RP—banda RP-50b Extrapolatedc Sampled
(RP-20 to RP-80) Region Efficacy
Data set I unitless unitless unitless
PCDF (n=3) 0.0445 i 0.0303 0.0353 :5 0.0323 NA equal
to
0.0283 1: 0.0392
HxCDF (n=3) 0.0505 1: 0.0409 0.0847 3: 0.00616 NA equal
to
0.143 i 0.0493
TCDF (n=3) 0.00992 2% 0.00333 0.00625 i 0.00087 NA equal
to
0.00397 i 0.00048
Data set 11 fmol / pl fmol Q11 finol / pl
1 9.64 - 51.2 22.2 NA unknown
9 23.2 - 143.6 57.7 NA greater
12 27.0 - 111 54.8 NA greater
15 NA NA NA NA
16 6.32 - 28.8 13.5 NA unknown
21 16.5 - 63.8 32.4 NA greater
Data set 111 fmol/pl finol/pl fmol/pl
3-d 41.0 - 35.0 37.9 < 35.9 less
3-p 4.16 - 0.179 0.863 < 2.70 unknown
2-d 61.6 - 107 81.2 NA equal
2-p 7.00 - 0.872 2.47 < 3.99 unknown
l-d 77.1 - 37.8 54.0 < 41.4 less
l-p NA NA NA NA

 

“ RP-band = the range of relative potency estimates generated from multiple point

estimates made for responses ranging from 20-80%-std.-max.

b RP-50 = the point estimate of relative potency derived using equation 1, where 50%-
std.-max. is the selected magnitude of response.
° Refers to relative potency estimates generated at magnitudes greater than the observed
efficacy for the sample. Uncertainties due to extrapolation apply.

d Describes the observed efficacy of the sample relative to that of the standard.

NA = not applicable

125

in a quantitative mass balance which assumes response additivity. Use of the estimates
for risk assessment purposes should consider the fact that the complex mixtures of
contaminants found in these sediments may be able to elicit more efficacious responses
than TCDD.

Data set 1111. A linear regression model was fit to the dose-responses (Fig. 3c) as
described previously. A maximal response was observed for samples l-d, 2-d, 3-d, and
the TCDD standard (Fig. 2-B; Fig. 3c). In all cases, the maximal responses observed
were greater than 20%-TCDD-max. (Fig. 2-C; Fig. 3c). The efficacy of samples l-d and
3-d were markedly different from the efficacy of TCDD (Fig. 2-E; Fig. 3c). Sample 2-d
had the same observed efficacy as the TCDD standard (Fig. 2-E; Fig. 3c). Sample l-p
exhibited no significant activity in the RLT 2.0 bioassay (Fig 3c). Samples 2-p and 3-p
showed significant activity but did not reach a maximal response (Fig. 2-B; Fig. 3c). Due
to limited sample volumes, samples 2-p and 3-p could not be tested at a greater
concentrations (Fig. 2-D), but in both cases the observed maximum responses were
greater than 20%-TCDD-max.

Multiple point estimates were used to calculate RP-band for samples l-d, 2-d, 3-
d, 2-p and 3-p (Fig. 40., Table 1). The points used (Y,) were 20-, 50-, and 80%-TCDD-
max. as well as the maximum observed response for each sample if 'it was less than 80%-
TCDD-max. The RP band generated for sample 2-d (Fig. 4c; Table 1) is valid and
suitable for risk assessment (Fig. 2-H), but the uncertainty due to non-parallel slopes
restricts the resolution of the estimate to somewhere between 62 and 107 frnol TCDD-
EQ/ pl (Table 1), which may or may not limit the utility of the estimate for mass balance

analyses or comparing among samples. Samples l-d and 3-d did not reach 80%-TCDD-

126

max., therefore some extrapolation was necessary to generate an RP-band for these
samples. The extrapolated portion was minimal relative to the size of the bands,
however. Thus, the RP-bands for l-d and 3-d provide a reasonable approximation of
their relative potency, but risk assessment and mass balance applications should consider
the fact that they did not cause the same magnitude of response as the TCDD standard
(Fig. 2-G). The efficacies of samples 2-p and 3-p were unknown and significant
extrapolation was required to generate a RP-band (Fig. 4c). The broad width of the
extrapolated band suggests that the extrapolated slope was quite different from that of the
TCDD standard curve (Fig. 40). As a result, the RP-band estimates for 2-p and 3-p
should be applied cautiously. While they may be suitable for comparative purposes, they
should not be used for risk assessment or mass balance analysis (Fig. 2-K). Based on the

RP-bands calculated, the rank order of potency among the samples was 2-d z l-d > 3-d >

2-p z 3-p > l-p.

Conclusions

Relative potencies based on a single ratio of point estimates are only valid when
the assumptions of indirect bioassay have been met [11,14,19,20]. The example data sets
illustrate that violation of these assumptions do occur, however. Sample efficacy is often
either unknown or shown to be different from that of the standard. Slopes commonly
vary among samples, particularly for complex mixtures of varying compositions.
Overall, the situations in which the assumptions of indirect bioassay are met may be

fairly rare.

127

The extent to which use of a single ratio of point estimates may lead to erroneous
conclusions is dependent on both the degree of deviation from the assumptions of indirect
bioassay and the application for which the estimate is used. In some cases, a single point
estimate provides a reasonable characterization of relative potency which is suitable for
the intended application. There are, however, situations in which reliance on a single
point estimate can lead to erroneous conclusions. Deviations from parallelism for sample
2-p, for example, resulted in approximately 8-fold uncertainty in the relative potency
estimate (Table 1). This degree of precision may be suitable for a risk assessment
involving order of magnitude differences between exposure and effect concentrations
and numerous safety factors. Use of a single point estimate such as the RP-SO could lead
to an incorrect conclusion in a situation where the exposure and effects concentration
were less than 8-fold different. Similar examples could be drawn for use of single point
estimates in mass balance analyses. Thus, although a single point estimate may be
suitable for certain applications even when the assumptions of indirect bioassay have
been violated, they may not be suitable for all applications.

The systematic framework and relative potency estimation methods presented in
this paper are not complex. They provide a simple and consistent means to evaluate the
assumptions of indirect bioassay for samples of both known and unknown composition.
They provide discrete relative potency estimates which can be more readily interpreted
than a function. The discrete estimates, represented as a relative potency band, can be
portrayed in figures and tables in such a way to make uncertainties in the estimates

apparent to a reader. In this manner, it is hoped that the use of multiple point estimates

128

and a systematic framework for evaluating the assumptions of indirect bioassay can

accommodate the need to characterize relative potency without sacrificing accuracy.

Acknowledgment— This work was supported by a Michigan State University
Distinguished Fellowship to D. Villeneuve, US EPA Biology Exploratory Grants
Program, Grant No. R85 371-01-0, cooperative agreement No. CR 822983-01-0 between
Michigan State University and the US EPA - Office of Water Quality, and the NIEHS -
Superfund Basic Research Program (NIH-ES-04911). Thanks to R. Cory, C. Cory, N.

J ohanson, all the members of MSU’s Aquatic Toxicology Laboratory, past and present,

and numerous other colleagues for the discussions and debates that led to this manuscript.

REFERENCES

l. Sanderson JT, Giesy JP. 1998. Wildlife toxicology, functional response assays. In
Meyers RA ed. Encyclopedia of Environmental Analysis and Remediation. John
Wiley and Sons. pp. 5272-5297.

2. Safe, S. 1990. Polychlorinated biphenyls (PCBS), dibenzo-p-dioxins (PCDDS),
dibenzofurans (PCDFS), and related compounds: environmental and mechanistic
considerations which support the development of toxic equivalency factors (TEFs).
Crit. Rev. Toxicol. 51-88.

3. Tillitt DE, Ankley GT, Verbrugge DA, Giesy JP, Ludwig JP, Kubiak TJ. 1991.
H4IIE rat hepatoma cell bioassay-derived 2,3,7,8-tetrachlorodibenzo-p-dioxin
equivalents in colonial fish-eating waterbird eggs from the Great Lakes. Arch

Environ Contam T oxicol 21:91-101.

129

10.

Zacharewski, TR, Berhane K, Gillesby BE, and Bumison BK. 1995. Detection of
estrogen- and dioxin-like activity in pulp and paper mill black liquor and effluent
using in vitro recombinant receptor/reporter gene assays. Environ Sci T echnol
29:2140-2146.

Safe, 8, Davis D, Romkes M, Yao C, Keyes B, Piskorska-Pliszczynska J, Farrell K,
Mason G, Denomme MA, Safe L, Zmudzka B, and Holcomb M. 1989. Development
and validation of in vitro bioassays for 2,3,7,8-TCDD equivalents. Chemosphere.
19:853-860.

Tillitt et al. 1996. Dietary exposure of mink to carp from Saginaw Bay. 3.
characterization of dietary exposure to planar halogenated hydrocarbons, dioxin
equivalents, and biomagnification. Environ Sci T echnol 30:283-291.

Giesy et al. 1997. Polychlorinated dibenzo-p-dioxins, dibenzofurans, biphenyls and
2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents in fishes from Saginaw Bay,
Michigan. Environ T oxicol Chem 16:713-724.

Hilscherova K, Kannan K, Kang YS, Holoubek I, Machala M, Masunaga S,
Nakanishi J, Giesy JP. 2000. Characterization of dioxin-like activity of riverine
sediments from the Czech Republic. Environ T oxicol Chem (submitted).

Khim J S, Villeneuve DL, Kannan K, Koh CH, Giesy JP. 1999. Characterization and
distribution of trace organic contaminants in sediment from Masan Bay, Korea. 2. in
vitro gene expression assays. Environ Sci T echnol (in press).

Khim J S, Kannan K, Villeneuve DL, Hu WY, Giesy JP, Kang SG, Song KJ, Koh CH.

2000. Instrumental and bioanalytical measures of persistent organochlorines in blue

130

ll.

12.

13.

14.

15.

16.

17.

18.

19.

mussel, Mytilus galloprovinciallis, from Korean coastal waters. Arch Environ
Contam T oxicol (submitted).

F inney DJ. 1978. Statistical Method in Biological Assay. Charles Griffin and
Company Ltd., London, England

Rand GM, ed. 1995. Fundamentals of Aquatic Toxicology: Effects, Environmental
Fate, and Risk Assessment. Second Edition. Taylor and Francis, Washington, DC,
USA.

Klaassen CD ed. 1996. Cassarett and Doull 's Toxicology: The Basic Science of
Poisons. Fifth Edition. McGraw Hill, New York, NY, USA.

Oris JT and Bailer AJ. 1997. Equivalence of concentration-response relationships in
aquatic toxicology studies: testing and implications for potency estimation. Environ
T oxicol Chem 16:2204-2209.

Scatchard, G. 1949. The attraction of proteins for small molecules and ions. Ann.
NY Acad. Sci. 51:660-672.

Cressie, NAC, Keightly DD. 1981. Analyzing data from hormone receptor assays.
Biometrics. 37:235-249.

Bruce RD, Versteeg DJ. 1992. A statistical procedure for modeling continuous
toxicity data. Environ T oxicol Chem 11:1485-1494.

Kerr DR, Meador JP. 1996. Modeling dose response using generalized linear
models. Environ T oxicol Chem 15:395-401.

Lausanne, G.P. , ed. 1973. Biostatistics in Pharmacology, lst ed., Vol. 1,2.

Permagon Press, New York, NY, USA.

131

20. Putzrath RM. 1997. Estimating relative potency for receptor-mediated toxicity:

21.

22.

23.

24.

25.

26.

27.

28.

reevaluating the toxic eqivalence factor (TEF) model. Regulat Toxicol Pharmacol
25:68-78

Villeneuve DL, Blankenship AL, and Giesy JP. 1998. Interactions between
environmental xenobiotics and estrogen receptor-mediated responses. In Denison
MS, Helferich WG, eds, T oxicant-Receptor Interactions, Modulation of Signal
Transduction and Gene Expression. Taylor and Francis, Washington, DC, USA.
Moore DRJ, Caux PY. 1997. Estimating low toxic effects. Environ T oxicol Chem
16:794-801.

Western Ecosystems Technology. 1994. Toxstat 3.4. Cheyenne, WY, USA

Bailer JA, Oris JT. 1997. Estimating inhibition concentrations for different response
scales using generalized linear models. Environ T oxicol Chem 16:1554—1559.
Nyholm N, Sorensen PS, Kusk KO, Christensen ER. 1992. Statistical treatment of
data from microbial toxicity tests. Environ T oxicol Chem 11:157-167.

Caux PY, Moore DRJ. 1997. A spreadsheet program for estimating low toxic
effects. Environ T oxicol Chem 16:802-806.

Sanderson, JT, Aarts J MMJG, Brouwer A, Froese KL, Denison MS, and Giesy JP.
1996. Comparison of Ah receptor-mediated luciferase and ethoxyresorufin O-
deethylase induction in H4IIE cells: implications for their use as bioanalytical tools
for the detection of polyhalogenated aromatic hydrocarbons. Toxicol Appl
Pharmacol 137:316-325.

Khim JS, Villeneuve DL, Kannan K, Lee KT, Snyder SA, Koh CH, Giesy JP. 1999.

Alkylphenols, polycyclic aromatic hydrocarbons (PAHs), and organochlorines in

132

29.

30.

31.

32.

33.

34.

sediment from Lake Shihwa, Korea: instrumental and bioanalytical characterization.
Environ T oxicol Chem (in press).

Khim J S, Kannan K, Villeneuve DL, Koh CH, Giesy JP. 1999. Characterization and
distribution of trace organic contaminants in sediment from Masan Bay, Korea. 1.
instrumental analysis. Environ Sci T echnol (in press).

Richter CA, Tieber VL, Denison MS, and Giesy JP. 1997. An in vitro rainbow trout
cell bioassay for aryl hydrocarbon receptor—mediated toxins. Environ T oxicol Chem
16:543-550.

Villeneuve DL, Richter CA, Blankenship AL, and Giesy JP. 1999. Rainbow trout
cell bioassay derived relative potencies for halogenated aromatic hydrocarbons:
comparison and sensitivity analysis. Environ T oxicol Chem 18:879-888.

Kannan K, Maruya KA, and Tanabe S. 1997. Distribution and characterization of
polychlorinated biphenyl congeners in soil and sediments from a superfund site
contaminated with Arochlor 1268. Environ Sci Technol 31:1483-1488.

Kannan K, Watanabe I, and Giesy JP. 1998. Congener profile of
polychlorinated/brominated dibenzo-p-dioxins and dibenzofurans in soil and
sediments collected at a former chlor-alkali plant. T oxicol Environ Chem 67: 135-
146.

Kannan K, Imagawa T, Blankenship AL, and Giesy JP. 1998. Isomer-specific
analysis and toxic evaluation of polychlorinated napthalenes in soil, sediment, and
biota collected near the site of a former chlor-alkali plant. Environ Sci T echnol

32:2507-2514.

133

Chapter 4

EFFECTS OF WATERBORNE EXPOSURE TO 4-NONYLPHENOL ON PLASMA
SEX STEROID AND VITELLOGENIN CONCENTRATIONS IN SEXUALLY

MATURE MALE CARP (C YPRIN US CARPIO)

D.L. Villeneuve'”, S.A. Vinalobos”, T.L. Keith”, E.M. Snyder”, S.D. Fitzgerald1#,J.P.

Giesy”

lDepartment of Zoology, 1‘National Food Safety and T oxicologz Center, 1Institute for
Environmental Toxicology, # Animal Health Diagnostic Lab, #Department of Pathology,
Michigan State University, East Lansing, MI, USA, 48824
§ Present Address: Novartis Crop Protection, Environmental Safety, Greensboro, NC,

USA, 27419

Submitted to Aquat. Toxicol. March, 2000.

134

Abstract

4-Nonylphenol (NP) has been shown to elicit estrogenic responses both in vivo and in
vitro. The mechanism by which NP exerts estrogenic and other endocrine-modulating
effects in vivo remains unclear, however. The goal of this study was to evaluate the
ability of NP to elicit estrogenic responses through indirect mechanisms of action
involving the modulation of endogenous steroid hormone concentrations. Sexually
mature male common carp (Cyprinus carpio) were exposed to aqueous NP
concentrations ranging from 0.1-10 pg NP/L (nominal) for 28-31 (1. Approximately 0.5-
3.5 ppm of NP was detected in pooled plasma samples or tissue samples from the carp
studied. NP exposure did not significantly increase plasma concentrations of 170-
estradiol (E2), testosterone (T), or Vitellogenin (VTG). Excluding outliers, plasma E2
concentrations ranged from <175 pg E2/ml to 700 pg E2/ml. T concentrations ranged
from 940-24,700 pg T/ml plasma. The greatest VTG concentration detected was 52
pg/ml. One third of the plasma samples tested contained <1 pg VTG/ml. Overall, the
results of this study did not support the hypothesis that exposure to waterborne NP can
modulate concentrations of steroid hormones in the plasma of sexually mature male carp.
The results did, however, raise a number of questions regarding the utility of estradiol
equivalent (EEQ) estimates as a means of predicting in vivo effects of estrogenic
substances. Furthermore, they provide information regarding the concentrations and
variability of E2, T, and VTG in the plasma of sexually mature male carp, which may aid

in design and interpretation of future studies.

135

1. Introduction

In recent years concern has emerged over xenobiotic chemicals which may
adversely affect humans or wildlife by modulating endocrine functions through either
direct or indirect mechanisms (Tyler et a1. 1998, Kendall et al. 1998, Colbom et al.
1993). Alkylphenols (AP), namely nonylphenol (NP) and octylphenol (OP), have been
widely regarded as suspect endocrine disrupting compounds (Nimrod and Benson 1996,
Servos 1999). NP has been shown to bind to the estrogen receptor (ER) with a affinity
approximately 10“1 to 10’5 times that of 17B—estradiol (E2; Lutz and Kloas 1999,
Tremblay and Van Der Kraak 1998, Villeneuve et a1. 1998, White et al. 1994). NP has
also been shown to elicit a number of estrogenic responses in vitro. NP induced ER-
mediated gene transcription in several recombinant cell lines (Legler et al. 1999,
Villeneuve et al. 1998, Gaido et al. 1997) and significantly increased proliferation of
MCF-7 human breast carcinoma cells (Soto et al. 1995). Exposure of primary
hepatocytes from rainbow trout (Oncorhynchus mykiss), common carp (Cyprinus carpio),
and South African clawed frog (Xenopus laevis), to NP was found to promote
Vitellogenin (VTG) expression relative to controls (Smeets et a1. 2000, Kloas et al. 1999,
Tremblay and Van der Kraak 1998, Jobling and Sumpter 1993). Relative potencies for in
vitro responses were also around 10'4 to 10'5 relative to E2. In addition to estrogenic
responses in vitro, exposure to NP has also been linked to estrogenic responses in vivo
(Kloas et al. 1999, Tremblay and Van der Kraak 1998, Christiansen et a1. 1998, Madsen
et al. 1997, Nimrod and Benson 1997). Thus, there is ample evidence to suggest that NP

can act as an endocrine modulating compound.

136

The mechanism by which NP exerts estrogenic and other endocrine-disrupting
effects in vivo remains unclear. Based on its ability to bind to the ER and induce ER-
mediated reporter gene and VTG expression in vitro, it has been postulated that NP
produces its estrogenic effects in vivo through a direct-acting mechanism. The proposed
direct-acting mechanism involves NP binding to the ER, promoting dimerization of the
ER-ligand complex, binding to genomic DNA at estrogen responsive elements (ERES),
and upregulating the transcription of estrogen-responsive genes whose products are then
responsible for down-stream estrogenic effects (Villeneuve et al. 1998). Recent evidence
suggests that the estrogenic effects of NP may occur through indirect mechanisms. A
recent study with fathead minnows (Pimephales promelas) linked NP exposure to
increased plasma E2 concentrations in both male and females (Giesy et al. 2000). Based
on a mass-balance analysis, it was hypothesized that the increased plasma E2, rather than
the direct action of NP, was responsible for the increased concentrations of plasma
Vitellogenin which were observed (Giesy et al. 2000). An opposite effect was found in a
study using Atlantic salmon (Salmo salar), in which exposure to NP caused a 24-43%
decrease in plasma estradiol levels (Arukwe et al. 1997). Furthermore, NP exposure was
linked to changes in the activities of steroid metabolizing enzymes (Arukwe et al. 1997).
Other studies have associated increased ER levels with NP exposure (Nimrod and
Benson 1997, White et al. 1993). Increased ER could be explained through a direct-
acting effect of NP on ER gene expression, which would tend to support a direct-acting
mechanism for NP. It also suggests, however, a mechanism by which NP may simply
enhance the activity or effects of endogenous E2. Recent studies with daphnids have

suggested that NP may also interfere with metabolic elimination of testosterone (T;

137

LeBlanc et al. 1999). This suggests the potential for both indirect androgenic and
estrogenic effects either through the actions of elevated T or through increases in E2
facilitated by the increased concentrations of aromatizable substrate. Thus, it is unclear
whether NP exerts its effects by acting directly as an xenoestrogen, or rather by
promoting physiological alterations which alter either the levels or effectiveness of
endogenous hormones.

Although it may seem somewhat academic, the question of mechanism is an
important one. Currently, risk assessments related to the endocrine modulating effects of
NP are based, primarily, on relative potencies derived for ER binding and ER—mediated
responses in vitro. Environmental monitoring and screening of new compounds for
estrogenic and other endocrine modulating activities is also at least partially based on in
vitro responses (Ankley et a1. 1998). Although in vitro approaches provide a measure of
the direct-acting potency of NP, they cannot easily model the spectrum of potential
indirect effects on endocrine homeostasis in a whole organism. The complex regulation
of steroid hormone levels through the hypothalamic-pituitary-gonadal axis, via
gonadotropins and both positive and negative feedback among multiple tissues has not
been adequately modeled in vitro. Even at the level of a single tissue, few in vitro
approaches can address responses mediated through indirect effects on the synthesis,
metabolism, and/or activities of endogenous steroids. In vitro approaches have obvious
advantages. They are generally lower in cost, shorter in duration, and avoid many of the
ethical questions associated with the use of whole animals (Purchase 1999, Stokes and
Marafante 1998). Thus, they can be very useful and powerful tools for research,

monitoring, and risk characterization. In order for in vitro assays to provide viable and

138

accurate information, however, it is imperative to establish that the mechanism of action
modeled in vitro is, indeed, the primary mechanism through which adverse effects may
be manifested in vivo and/or calibrate in vitro responses to in vivo effects, particularly
adverse ones.

The goal of this study was to examine some potential mechanism(s) of action for
NP in fish and evaluate the results as they relate to the use and design of in vitro
bioassays to characterize the endocrine modulating potency of NP and similar
compounds. The specific objectives of the study were to 1) test the hypothesis that
exposure to NP can cause significant increase in plasma E2; 2) test the hypothesis that
exposure to NP can cause significant increases in plasma T; 3) calibrate in vivo endpoints
such as plasma VTG induction and/or histopathological lesions with in vitro potencies
reported for NP; 4) examine NP accumulation in fish tissues and plasma; and 5)
characterize any indirect mechanisms suggested by the results of objectives 1 and 2.

Sexually mature, male, common carp (Cyprinus carpio) were chosen as the test
organism for this study. Carp were commercially available, relatively easy to maintain in
the laboratory, and large enough to provide tissue and plasma volumes sufficient for
multiple analyses. Additionally, laboratory studies with carp would support on-going
field work involving the use of VTG in feral and caged carp as a biomarker of exposure
to xenoestrogens (Snyder 2000, Goodbred et al. 1997, Bevans et al. 1996, F olmar et a1.
1996). Previous studies have observed both feminization and demasculinization of male
carp exposed to 4-tert-pentylphenol (Gimeno et al. 1998a, Gimeno et al. 1998b). Finally,
because previous studies with fathead minnows and Atlantic salmon provided

contradictory results regarding NP’s potential effect on endogenous E2 levels (Giesy et

139

al. 2000, Arukwe et al. 1997) it seemed prudent to examine the effect in additional
species. A single gender and developmental stage was used for this study to minimize

the natural variability among individual fish, allowing for greater statistical power.

2. Materials and Methods

2.] Exposure

Sexually mature, male, common carp (Cyprinus carpio; 2-3 years old; 50-150 g),
were obtained from J&J Aquafarms (Sanger, CA, USA). Upon arrival, 20 fish were
randomly assigned to each of seven 600 L fiberglass tanks (Frigid Units, Toledo, OH,
USA) and acclimated for 3 wk. Throughout acclimation and exposure, well water was
delivered to each tank at a flow rate of 3.5-4 L/ min. Water temperature was maintained
between 10 and 13° for the duration of the acclimation and exposure. Temperature was
monitored daily using electronic Hi/Lo thermometers (Fisher Scientific, Pittsburgh, PA,
USA) and electronic thermometers were calibrated weekly. Water quality parameters
including ammonia, nitrite, dissolved oxygen (DO), pH, hardness, and conductivity were
monitored weekly. The fish were fed daily with approximately 200 g trout chow
(Silvercup, Murray, UT, USA) per tank. Waste and uneaten food were removed daily.

Fish were exposed to 4-nonylphenol (Schenectady International, Freeport, TX,
USA; 95% pure) for 28-31 d using a flow through system. Concentrated solutions of NP
(18.5, 5.55, 1.85, 0.555, or 0.185 mg/L) in 3.75% ethanol were delivered to treatment

tanks using a variable speed peristaltic pump (MasterFlex/Cole-Parmer, Vernon Hills, IL,

140

USA) equipped with an 8-channel pump head (MasterFlex). Solutions were delivered at
a flow rate of approximately 2 mein through Teflon tubing (MasterFlex). Thirty cm
sections of silicon/platinum tubing (MasterFlex) were used at the pump head and were
changed every 10 d due to wear. Nominal concentrations in the NP treatment tanks were
10, 3, 1, 0.3, and 0.1 pg/L. Solvent control (SC) and control tanks received 3.75%
ethanol and well water, respectively, at a flow rate of approximately 2 ml per min using
the same peristaltic pump and tubing. At steady state, the ethanol concentration in the
treatment and SC tanks was approximately 0.002%.

NP concentrations in the exposure tanks were measured weekly. A 500 ml
sample was collected from each tank using a graduated cylinder. Samples were spiked
with 3.22 pg octylphenol (50 pl of 64.4 pg/ml in methanol) as an internal standard.
Samples were then extracted three times (liquid/liquid extraction) with 100 ml high
purity dichloromethane (DCM; Burdick and Jackson, Muskegon, MI, USA). DCM
fractions were passed through 30 g anhydrous Na2S04 and stored overnight. Each 300
ml DCM extract was concentrated to approximately 5 ml by rotary evaporation, and
evaporated to 0.5 ml under a gentle stream nitrogen. Extracts were then transferred to
high purity acetonitrile (ACN; Burdick and Jackson) by adding 1.0 ml ACN, vortexing,
and evaporating again to 0.5 ml. Final extracts were analyzed by reverse phase HPLC

with fluorescence detection (Snyder et al. 1999).

2.2 Sample Collection

141

In order to minimize variability due to diurnal fluctuations in plasma steroid
concentrations (Zohar and Billard 1984) all samples were collected between the hours of
9:00 and 10:30 AM. Five fish were collected from each treatment group, each day, over
the four day period from exposure day 28-31. To avoid confounding results with
sampling time, individual fish were taken sequentially, from each treatment group. Fish
were netted then anesthetized in 8 L of well water containing 200 mg/L MS-222 (3-
aminobenzoic acid ethyl ester methane sulfonate salt; Sigma, St. Louis, MO, USA, A-
5040). Immediately after opercular movement ceased, blood was drawn from the caudal
vein using a 3cc disposable syringe equipped with a 21 gauge needle (Becton Dickinson,
Franklin Lakes, NJ, USA) which had been rinsed with 10 mg/ml heparin (Sigma H-3393)
in 0.9% NaCl. The volume of blood collected ranged from 100 pl to 4100 pl with a
mean volume of 1700 d: 600 p1. Afier collection, blood samples were immediately
transferred to 15 ml disposable centrifuge tubes containing 20 pl of 2.55 TIU/ml
aprotinin (Sigma A-1153; 5.2 TIU/mg) and placed on ice. The length and weight of each
fish was measured and individual fish were placed into separate plastic bags, and stored
on ice until dissection.

After blood was collected from 5 fish from each treatment group, tissue samples
were obtained. The gonads and hepatopancreas were removed and weighed. The brain
and a gill arch were also collected, but were not weighed. A small section of each tissue
was immersed in a 15 ml centrifuge tube containing 10% neutral buffered formalin and
stored at room temperature. The remaining mass of each tissue (excluding gill arch) was
placed in 1.0 m1 cryovials and frozen immediately in liquid nitrogen. The remaining

carcass was placed back in its individual plastic bag and stored at -200 C.

142

After collection, blood samples were stored on ice for approximately 4 hrs. Two
60 pl samples of whole blood were taken for each fish. The remaining blood was
centrifuged at 2000 rpm for 10 min at 4° C. Plasma was aliquotted into 400 pl eppendorf

tubes, 120 pl per aliquot, and stored at -80°C along with the whole blood samples.

2.3 Estradiol Radioimmunoassay (RIA)

Plasma E2 concentrations were measured for all fish in each treatment group
(except where plasma volume was insufficient). Plasma samples were extracted to
separate E2 from binding proteins (McMaster et al. 1992). Two hundred pl of plasma
was combined with 800 pl Nanopure H20 and extracted three times with 5 ml high purity
diethyl ether (Burdick and Jackson). After separation of ether and aqueous phases,
samples were snap-frozen by immersion in a bath of dry ice and acetone. Ether phases
from each of the three extractions were combined in a new test tube and evaporated to
dryness in a 45°C water bath under a gentle stream of nitrogen. Samples were then
reconstituted with 750 pl Phosgel (McMaster et al. 1992) and stored at -80° C.
Extraction efficiency, as determined by spike/recovery with tritiated estradiol (3H-E2),
was 78 fl: 4% (n = 4).

E2 concentrations were quantified by radioimmunoassay (RIA). The RIAs were
conducted according to a protocol published by McMaster et a1. (1992) with a few
modifications. Tritiated estradiol (72 Ci/mmol, 3.78 pg/ml, NENTM Life Science, Boston,
MA, USA) was diluted to 1 x 10'5 pCi / pl (0.002 pCi per tube) in Phosgel for use in the

RIA. This dilution was found to yield approximately 700 cpm per 200 pl. Polyclonal

143

antibody to l7B-estradiol (Biogenesis, Brentwood, NH, USA; AR1702) was diluted
l:20,000 for use in the RIA. This antibody dilution yielded approximately 50% binding
of the added 3H-E2 in the absence of competitor. Cross-reactivity of the antibody was
reported to be 100% for l7B-estradiol, 14% for estrone, 5% for estriol, and <0.01% for
other steroid hormones (Biogenesis, AR1702, batch no. P-l). E2 standards were
prepared by 2-fold serial dilution to yield 9 standards ranging from 800 to 3.125 pg.
Standards, samples, non-specific binding, and total counts tubes were all run in triplicate.
A total of 80 tubes were run in each assay, allowing 19 samples to be run per assay. A
pooled plasma extract was run in triplicate for each assay, to provide a measure of inter-
assay variation. Parallelism was tested by analyzing serial dilutions of plasma extracts (5
dilutions, n=3). Accuracy was tested by spiking extracts with known amounts of

unlabelled E2 (5 concentrations, n=3).

2.4 Testosterone Enzyme Immunoassays (EIA)

Plasma T concentrations were measured for 10 fish, randomly selected from each
treatment group, using enzyme immunoassay (EIA). Plasma samples were extracted to
separate T from binding proteins (McMaster et al. 1992). One hundred pl of plasma was
combined with 400 pl nanopure H20 and extracted three times with 5 ml high purity
diethyl ether (Burdick and Jackson). After separation of ether and aqueous phases,
samples were snap-frozen by immersion in a bath of dry ice and acetone. Ether phases
from each of the three extractions were combined in a new test tube and evaporated to

dryness in a 45°C water bath under a gentle stream of nitrogen. Samples were then

144

reconstituted with 500 pl EIA buffer (Cayman Chemical, Ann Arbor, MI, USA) and
stored at -80° C. Extraction efficiency, as determined by spike/recovery with tritiated T
(3H-T; 96 Ci/mmol, 3.00 pg,/ml,NEN1m Life Science, Boston, MA, USA), was 90 i 3%
(n = 5).

Testosterone EIAs were conducted in a 96 well plate format using commercial
testosterone enzyme immunoassay kits (Cayman Chemical, Cat. # 582701). T standard
was provided in each kit and diluted to generate T standard curves consisting of 9
concentrations ranging from 250 to 2 pg/ml. Standards were run in duplicate on each 96
well plate. Four blank and maximum binding wells were run on each plate. Non-specific
binding and total activity wells were run in triplicate. Samples were run at two dilutions,
1:30 and 1:90, and each dilution was run in triplicate. Two dilutions (1 :30 and 1:90) of a
pooled plasma extract were run in triplicate for each assay, to provide a measure of inter-
assay variation. Parallelism was tested by analyzing serial dilutions of plasma extracts (8
dilutions, n=3). Accuracy was tested by spiking extracts with known amounts of
unlabelled T (4 concentrations, n=3). Cross-reactivity of the EIA antibody was reported
to be 100% for testosterone, 21% for 5a-dihydrotestosterone, 10% for 513-
dihydrotestosterone, 3.6% for androstenedione, 1.2% for llB-hydroxytestosterone, and
less than 0.5% for all other steroids tested (Cayman Chemical). Cross-reactivity with 11-

ketotestosterone was not reported.

2.5 Vitellogenin ELISA

145

Plasma VTG concentrations were measured by ELISA for 12 fish randomly
selected from each treatment group. The VTG ELISA protocol used and its
characterization has been detailed elsewhere (Snyder 2000). The polyclonal rabbit anti-
goldfish antiserum used for the VTG ELISA was developed and characterized by Nichols
(Nichols 1997, Nichols et al. 1999). ELISAs were conducted using a 96 well plate
format. VTG standard curves consisted of 10 serial dilutions ranging from 2730 to 5.3
ng/ml. Duplicate standard curves were run on each plate. Maximum binding and non-
specific binding wells were also run in duplicate. Plasma samples (unextracted) were run
at three dilutions; 1:33, 1:99, and 1:297. No significant plasma interferences were
observed for samples diluted 1:25 or greater (Appendix C). Each sample dilution was
run in triplicate. Accuracy, parallelism, and specificity of the VTG ELISA have been

reported elsewhere (Snyder 2000).

2. 6 Histological Examination

Testes, hepatopancreas, brain, and gill arch tissue from 10 fish per treatment
group were examined for histological lesions. Tissues fixed in 10% neutral buffered
formalin were trimmed to <0.5 cm thick sections and embedded in paraffin. Embedded
tissues were sectioned at 6 pm and stained with Hematoxylin and Eosin. Tissues were
examined by a certified veterinary pathologist. All tissues were evaluated for
microscopic changes including degeneration, inflammation, and neoplasia. Lesions were
scored on a 0-3+ scale; 0=no lesion, l+=mild, 2+=moderate, 3+=severe. Testes were

evaluated using three different criteria; 1) active versus inactive germinal epithelium, 2)

146

stage of maturation, and 3) Sertoli cell proliferation. Stages of maturation were defined
as follows: stage 1- thick germinal eptithelium, early spermatogenesis; stage 2-
moderately thick germinal epithelium, moderate spermatogenesis; stage 3-thin germinal
epithelium, scattered areas of sperrnatogenesis. Sertoli cell proliferation was scored on a
0-3+ scale, with 0 corresponding to no proliferation and 3+ corresponding to marked

proliferation.

2. 7 Tissue and plasma NP concentrations

Tissue NP concentrations were quantified for four fish from the highest treatment
group (10 pg/L nominal), one fish from the solvent control treatment, and one control
fish. The extraction and quantification procedures used are detailed elsewhere (Snyder et
al. 2000). Briefly, 20 g samples were extracted by steam distillation. NP in the extracts
was quantified by normal phase HPLC with fluorescence detection.

NP concentrations in pooled plasma samples from each treatment group were also
determined. Plasma from 10-15 fish from each treatment group was pooled to provide a
total volume greater than 2 ml per treatment group. Two thousand pl of pooled plasma
from each treatment group was spiked with 1.08 pg butylphenol (BP; 10 pl of 108 pg/ml
in methanol) as an internal standard. Pooled plasma samples were then liquid/liquid
extracted three times with diethyl ether. Ether layers from each extraction were pooled in
a graduated test tube and evaporated to dryness at room temperature under a gentle
stream of nitrogen. Samples were reconstituted in 200 pl high purity acetonitrile

(Burdick and Jackson). The extraction procedure used was an adaptation of a plasma

147

extraction method presented by Fang et al. (2000). NP and BP concentrations in the
extracts were quantified by reverse phase HPLC with fluorescence detection (Snyder et
al. 1999). Nanopure water and normal goat serum (Sigma 9023) extracted and analyzed

using the same procedure served as procedural blanks.

3. Results

3.] Exposure

During the 28—31 (1 exposure to NP, there were few differences in exposure
conditions among treatment tanks which might be expected to confound the effects of NP
treatment. Water quality conditions were stable throughout the exposure period (Table 1)
and no marked differences among treatment tanks were observed. Actual treatment
concentrations were generally around 50% of nominal (Table 2). Less than 20% of the
nominal NP concentration was detected in the 0.3 pg/L treatment tank, however (Table
2). NP concentrations were below the method detection limit (MDL) for the two lowest
treatment groups as well as the control tanks (Table 2).

Although there was substantial variability in the size of the fish exposed, there
were no significant differences in mean fish length or mass among treatment tanks.

Mean length was 16.0 cm i 1.5. The minimum and maximum fish lengths were 12.8 and
20.6 cm, respectively. The mean fish mass was 94.8 g i 28.8 g, with minimum and

maximum masses of 43.3 g and 207 g, respectively. Post-exposure examination of the

148

Table 1. Average water quality conditions during exposure.

 

 

 

 

Water Quality Parameter Meana i Standard Deviation
Temperature° (°C) 11 :t 0.5
pH° 7.57 i 0.03
Ammonia° (mg/L) < 0.02
Nitrite° (mg/L) < 0.02
Dissolved Oxygenf (mg/L) 10.5 i 0.4
Hardnessg (mg/L) 402 i 13

Conductivityh Spmho) 546 :1: 15

a Averaged over all tanks and all measurements taken over the course of the exposure.
° Monitored daily with electronic Hi/Lo thermometers (Fisher Scientific, Pittsburgh, PA,
USA).

c Monitored weekly using a Pinpoint pH meter (Aquatic Eco-systems, Apopka, FL,
USA).

d Monitored weekly using a LaMotte Ammonia Nitrogen Test Kit - low range
(Aquaculture Supply, Dade City, FL, USA)

c Monitored weekly using a LaMotte Nitrite Test Kit (Aquaculutre Supply)

f Monitored weekly using a YSI Model 57 Dissolved Oxygen Meter (YSI, Yellow
Springs, OH, USA).

8 Monitored weekly using a LaMotte Hardness Test Kit (Aquaculture Supply)

° Monitored weekly using a Pinpoint Conductivity Meter (Aquatic Eco-systems)

Table 2. Measured concentrations of 4-nonylphenol (NP) in treatment tanks.

 

 

Treatment Group Measured Concentrationa
(nominal NP concentration, pg NP/L) (pg NP /L; mean i st. dev.)

10 5.36 :t 0.37

3.0 1.51 :t 0.17

1.0 0.58 i 0.07
0.3 < 0.05
0.1 < 0.05
Solvent Control < 0.05
Control < 0.05

 

a Averaged over four separate sampling periods. Based on 500 ml samples liquid/liquid
extracted with dichloromethane. Quantified by reverse phase HPLC with fluorescence
detection. Concentrations were adjusted for recovery as determined by recovery of an
octylphenol internal standard. Method detection limit = 0.05 pg NP/L.

149

 

carp gonads revealed that one female was exposed in the 0.3 pg/L exposure tank. The
presence of a female may have confounded results from the 0.3 pg/L treatment group.
All other fish exposed were male. There was one mortality during the exposure period.
A fish from the 10 pg/L treatment group was observed to be swimming on its side for
several days and was found lying on its side, immobile, on exposure day 16. The fish
was euthanized and examined. Cranio-facial abrasions and a broken pectoral fin were
detected and the gall bladder was found to be nearly black. All other fish appeared
healthy throughout the exposure. No differences in gonado-somatic index (GSI) or
hepatosomatic index (HSI) were detected among treatment groups. GSI ranged from 0.5-
10.2%, with a mean of 3.32% :1: 1.31%. HSI ranged from 0.38-2.99%, with a mean of
1.62% :t 0.47. G81 and HSI were calculated relative to whole body weight after blood

collection.

3.2 Plasma Estradiol

Accuracy of the plasma E2 RIA was demonstrated by analyzing plasma extracts
spiked with E2 standard. Measured concentrations were plotted as a function of expected
concentration (Figure 1). The slope of a regression line drawn through the points was not
significantly different from 1.0 (p<0.749). Dose-response curves generated by RIA
analysis of several dilutions of plasma extract did not deviate markedly from parallelism

to the E2 standard curve (Figure 1). The coefficient of variation (CV) among replicate

150

200
180 .t Accuracy Test

160 ,, y = 1.2402x + 23.89
R2 = 0.9996

 

140 «-

pg 52 measured
8
O

H‘Zmfi

 

 

 

0 f l l t r
O 20 40 60 80 100 120
pg E2 expected ;

 

 

100
90 .
80 -.
7o -.
so ..
so «
40 «
30
20
10

 

   

Parallelism Test

%-bound

 

 

—o— P-1
-o—P-2
-o— P-3
-0- E2 std.

 

 

 

 

 

1 0.1 0.01 0.001

dilution factor

Fig. 1. Accuracy and parallelism tests for 17B-estradiol radioimmunoassay. Accuracy
test was determined over the range of 20 to 80%-bound. The slope of the accuracy test
plot was not significantly different from 1.0 (p<0.749). The parallelism test was
conducted using five serial dilutions of three separate plasma extracts (P-1, P-2, P-3). A
dilution factor of 1.0 corresponds to 200 pl of plasma extract or 800 pg of 17B-estradiol
standard per sample tube.

151

determinations for samples was generally less than 10%, prior to correction for plasma
dilution. The CV among assays was 25% (n=8) for the final calculated concentration of
plasma E2.

No significant differences in plasma E2 concentrations were detected among
treatment groups (Figure 2). Approximately 50% of the plasma samples contained E2
concentrations less than the MDL of 175 pg E2/ ml plasma (Figure 2). The greatest
plasma E2 concentrations detected were 3370 pg E2/ml for one fish from the 10 pg/L
treatment group and 2935 pg E2 / ml for the female from the 0.3 pg/L treatment. Both
were considered outliers and were not included in the statistical analyses nor shown in
Figure 3. Among non-outliers, the greatest plasma E2 concentration observed was 700

pg E2 /ml (Figure 2).

3.3 Plasma Testosterone

Accuracy of the T EIA was demonstrated by analyzing plasma extracts spiked
with T standard. Measured concentrations were plotted as a function of expected
concentration (Figure 3). The slope of a regression line drawn through the points was not
significantly different from 1.0 (p<0.969). EIA analysis of several dilutions of plasma
extract demonstrated that extract responses were parallel to the T standard curve (Figure
3). The CV among replicate determinations for samples was generally less than 10%.

The CV among plates was 9.11% (n=7) for the final calculated concentration of plasma

T.

152

 

 

800
700 -- 0 o
600 -—
500 -r
400 -- °
300 1*-
200 —-
100 -—

pg E2/ml plasma
0.

 

 

 

 

 

 

10 3 1 0.3 0.1 SC C

 

 

 

Fig. 2. Plasma estradiol (E2) concentrations, measured by radioimmunoassay, for
sexually mature male carp exposed to 4-nonylphenol (NP). X-axis = nominal exposure
concentrations in pg NP/L. SC = solvent control exposure. C= control exposure.
Circles represent the values measured for individual fish. Columns represent the mean
concentration for each treatment group. Error bars represent 1 standard deviation above
and below the mean. The method detection limit (175 pg E2/ml plasma) is represented
by a horizontal line.

153

90

 

   
 

 

 

 

80 “Accuracy Test
‘0 70 -—
9
g 60 +
o y = 0.8985x + 6.5728
E 50 -- 2
'E' R =0.9728
P 40 -~
8“

30 -»

a

20 -~

10 -»

0 t 4 .L t

0 20 40 60 80 100

pg Tlml expected

 

Parallelism Test

%-bound

 

—o—Plasma
+T std.

 

 

0 1 # fi‘ 1
1 0.1 0.01 0.001
dilution factor

Fig. 3. Accuracy and parallelism tests for testosterone enzyme immunoassay. Accuracy
was determined over the range of 20-80%-bound. The slope of the accuracy test plot was
not significantly different from 1.0 (p<0.969). The parallelism test was conducted using
8 serial dilutions of a pooled plasma sample (n=3). A dilution factor of 1.0 corresponds
to 50 pl of plasma extract or 250 pg/ml of testosterone standard per test well.

154

There were no significant differences in plasma T concentrations among
treatments (Figure 4). All T concentrations were greater than the MDL. Plasma T
concentrations ranged from 940 pg T/ml to 24.7 ng T/ml. Plasma T levels were not
significantly correlated with either testes mass or GSI. There were no significant

differences in E2/T ratios among treatment groups.

3.4 Plasma Vitellogenin

Accuracy and parallelism of the VTG ELISA have been demonstrated elsewhere
(Snyder 2000). There were no significant differences in plasma VTG concentrations
among treatment groups (Figure 5). One third of the plasma samples tested contained
VTG concentrations less than the 1 pg/ml MDL (Figure 5). VTG concentrations ranged
over at least two orders of magnitude (Figure 5). The greatest VTG concentration
detected was 52 pg/ml. VTG concentrations were not significantly correlated with
concentraions of E2 or T in the plasma, E2/T ratios, or morphological indices. The CV
among replicate determinations (n=3) for samples was generally less than 10%, prior to
correction for plasma dilution. The CV among plates was 44% (n=12). The pooled
plasma sample used for determination of among-plate variability had a mean
concentration of 2.3 $1.0 pg VTG/m1, which was near the MDL. Estimated

concentrations VTG in the pooled plasma sample ranged from 0.8 pg/ml to 3.9 pg/ml.

155

 

 

30000

25000

I
I
O

20000 a—

15000 «- 0

pg Tlml plasma
0
O
O 0

10000 -r-

5000 4

I

 

 

 

 

 

 

 

10 3 1 0.3 0.1 SC C

 

 

 

Fig. 4. Plasma testosterone (T) concentrations, measured by enzyme immunoassay, for
sexually mature male carp exposed to 4-nonylphenol (NP). X-axis = nominal exposure
concentrations in pg NP/L. SC = solvent control exposure. C= control exposure.
Circles represent the values measured for individual fish. Columns represent the mean
concentration for each treatment group. Error bars represent 1 standard deviation above
and below the mean.

156

 

 

 

 

 

100 . ‘3 r
C
- o
o
9 10 “E .
E : 0
9 5 ' .
(D r
*>' _ _ L—
1“: _T
C o o
I o o o o
: o o o o o
f o o o o
0.1 1 1 1 F l l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 3 1 0.3 0.1 SC C

 

 

 

Fig. 5. Plasma Vitellogenin (VTG) concentrations, measured by ELISA, for sexually
mature male carp exposed to 4-nonylphenol (NP). X-axis = nominal exposure
concentrations in pg/L. SC = solvent control exposure. C= control exposure. Circles
represent the values measured for individual fish. Columns represent the mean
concentration for each treatment group. Error bars represent 1 standard deviation above
and below the mean. The method detection limit (1.0 pg/ml plasma) is represented by a
horizontal line.

157

3.5 Histopathology

No treatment-related lesions were observed in testes, hepatopancreas, brain, or
gill. A11 fish examined were sexually mature. Testes were stage 2-3, with most fish
being in the late sperrnatogenic stage. No treatment-related trends in Sertoli cell
proliferation were detected. Microgranulomas, a sign of inflammation, were observed in
gonad tissues from some fish, but their occurrence was not related to NP treatment.
Similarly, lipidosis and/or cholestasis was observed in the hepatopancreas tissue from

some fish, but again, there was no relationship between treatment group and occurrence.

3.6 Tissue and Plasma NP Concentrations

NP concentrations in tissues from four randomly selected fish exposed to a
nominal concentration of 10 pg NP/L ranged from 2.1-3.8 pg NP/g tissue (Table 3). A
fish randomly selected from the solvent control group contained approximately 0.016 pg
NP/g tissue (Table 3). The NP concentration in the tissue of a randomly selected control
fish was less than the MDL of 5 ng NP/g tissue (Table 3). NP was detected in all pooled
plasma samples tested, but was not found in the procedural blanks (Table 3). Plasma
pooled from fish exposed to the greatest nominal NP concentration (10 pg/L) contained
3.6 pg NP/ml plasma (Table 3). This concentration was similar to the tissue
concentrations detected for that treatment group. Among the three highest treatment
groups, NP concentrations in pooled plasma samples appeared to be correlated with

exposure concentrations (Table 3). Among exposures of 1.0 pg/L or less, including

158

Table 3. Nonylphenol (NP) concentrations in tissue and pooled plasma samples.

 

 

 

 

Sample Nominal Exposure NP concentration°
Conc. (pg NP/g tissue)
(pg NP/L)
Tissue (a1) 10 2.33
Tissue (a4) 10 2.11
Tissue (a8) 10 3.46
Tissue (a10) 10 3.81
Tissue (c6) Solv. Control 0.016
Tissue (d3) Control < 0.005
Sample Nominal Exposure NP concentration°
Conc. (pg NP/ml pooled plasma)
(pg NP/L)
Pooled plasma 10 3.63
Pooled plasma 3.0 1.33
Pooled plasma 1.0 0.478
Pooled plasma 0.30 NAc
Pooled plasma 0.10 0.452
Pooled plasma Solv. Control 0.524
Pooled plasma Control 0.693
Nanopure H20 Blank < 0.0033
Normal Goat Serum Blank < 0.0033

 

a Concentrations not adjusted for recovery. Method recoveries z 80% (Snyder et al.

2000)

° Concentrations adjusted for recovery of butylphenol internal standard. Recoveries
were highly variable. Results should be considered semi-quantitative.
° Sample lost during extraction procedure.

159

control groups, there was no marked difference pooled plasma NP concentrations (Table
3). Pooled plasma from the control and solvent control groups contained approximately
0.7 and 0.5 pg NP/ml, respectively (Table 3), despite the fact that water concentrations in
these tanks were found to be less than 50 ng NP/L. Recoveries of the internal standard
(BP) were highly variable, ranging from 80% (10 pg NP/L exposure group) to 8% (3 pg
NP/L exposure group) with a median recovery of 44%. Thus, pooled plasma

concentrations of NP were considered semi-quantitative.

4. Discussion

4.1 Nonylphenol exposure

The sexually mature male carp examined in this study were exposed to
environmentally relevant concentrations of NP. Freshwater NP concentrations ranging
from <0.010 pg/L to 180 pg/L have been reported (Bennie 1999). A study of 30 rivers in
the United States reported an average NP concentration of 0. 12 pg/L (Naylor et al. 1992).
Actual water concentrations in this study ranged from approximately < 0.05 pg NP/L to 5
pg NP/L (Table 2). Similar studies have reported actual concentrations which were
approximately 1/3 to 1/5 of the nominal concentrations (Nimrod and Benson 1998,
Nichols et al. 2000). Factors such as volatilization of NP, adsorption of NP to tubing or
the walls of the tanks, adsorption to organic matter present in the tanks, degradation, etc.

probably account for the difference between nominal and actual exposure concentrations.

160

 

Analysis of tissue and pooled plasma extracts suggested that the carp used in this
study had accumulated body burdens of NP. Concentrations as great as 3.8 pg NP/ g
tissue and 3.6 pg NP/ml pooled plasma were detected (Table 3). Based on tissue
concentrations of NP detected in four randomly selected fish exposed to 10 pg NP/L and
the aqueous concentration of NP measured in the exposure tank (Table 2), the
bioconcentration factor (BCF) for NP, over the course of the 28-31 d exposure period
was approximately 550. BCFs reported for NP in fish range from 0.9 to 1250 (Servos
1999). BCFs determined for fathead minnows exposed for 14 and 28 d were reported to
be 586 and 741, respectively (Brooke 1993). Thus, the accumulation of NP observed was
similar to that reported for other studies.

The presence of NP in pooled plasma samples from the control and solvent
control groups suggests either an uncontrolled exposure of the study fish to NP, or the
presence of an artifact in the pooled plasma samples. NP was detected in the tissue of at
least one solvent control fish. There were not sufficient resources for this study to
analyze tissues from additional fish, however. There was no detectable NP in the
procedural blanks for the pooled plasma extractions, suggesting that artifacts were not
introduced during the sample extraction or analysis procedure. Procedural blanks were
not collected or stored under the same conditions as the plasma samples, however. Thus,
it is possible that NP artifacts were introduced during either the sample collection or
storage phases, although there was no direct evidence to support that hypothesis.
Alternatively, the fish may have been exposed to an uncontrolled source of NP, either
during the laboratory exposure or during holding in the laboratory or at the commercial

aquafarm. Possible sources of NP during the exposure include leaching from tubing,

161

laboratory plumbing, or tanks. Waterbome concentrations were below 50 ng/L, however.
Assuming a BCF of 550, waterborne NP could account for up to 27.5 ppb NP in tissue or
plasma. This is over an order of magnitude lower than the pooled plasma concentrations.
This suggests that the most likely potential source of uncontrolled NP exposure was

either the food used, or exposure prior to the laboratory study.

4.2 Morphological/Histological Endpoints

The exposure concentrations used in this study were not expected to cause
significant effects on survival and/or growth. LC-SOS reported for acute toxicity of NP to
fish range from 125-400 pg/L (Servos 1999, Staples et al. 1998). Concentrations used in
this study were at least one order of magnitude less. Furthermore, the actual water
concentrations used in this study were less than NOECs reported for chronic effects of
NP exposure on fish (Servos 1999). Given their stage of development and the water
temperature used, the exposed carp were not expected to grow markedly during exposure.
Thus, the lack of effects on morphology and survival was consistent with expectations.

Based on several reports in the literature, investigation of potential effects of NP
exposure on HSI, G81, and gonad histology seemed warranted. Injection of Atlantic
salmon (Salmo salar) with approximately 0.1 mg NP/g caused a significant increase in
HSI over a period of 10-30 (1 (Madsen et al. 1997). Such doses are pharmacological,
however, and were nearly two orders of magnitude greater than the tissue and pooled
plasma concentrations observed in this study. Currently, there are no known reports of

significant changes in HSI caused by exposure to NP at environmentally relevant doses.

162

Exposure to 1.1 or 3.4 pg NP/L was reported to cause significant changes in numbers and
size of Sertoli cells and germ cell syncytia in breeding male fathead minnows (Miles-
Richardson 1999). Intraperitoneal injection of 10 or 100 pg NP/g into sexually mature
male eelpout (Zoarces viviparus) was found to cause a significant reduction in GSI, as
well as severe effects on testicular structure (Christiansen et al. 1998).

Demasculinization was observed in sexually mature male carp exposed to 4-tert-
pentylphenol (TPP) for up to 3 months (Gimeno et al. 1998b). Similar effects were not
observed for sexually mature male carp exposed as part of this study, however.
Differences among species may, at least in part, explain differences between results
observed for fathead minnow or eelpout and the results of this study. Concentrations at
which effects were observed in eelpout dosed with NP and carp exposed to TPP exceeded
the concentrations used in this study (Christiansen et a1 1998, Gimeno et al. 1998b).
Furthermore, some of the effects of TPP on G81 and testes histology were not observed
until the third month of exposure (Gimeno et al. 1998b). Exposure concentrations in the
fathead minnow study were similar to those used in this study, but the use of breeding
pairs and a longer exposure period (42 d vs. 31 d) confounds direct comparison of study
results. Thus, differences in study design could easily account for the lack of significant

effects of NP exposure on GSI and gonad histology observed in this study.

4.3 Estrogenic Potency

Exposure to aqueous concentrations of NP as great as 5 pg NP/ml did not elicit

significant VTG induction in sexually mature male carp (Figure 5). This is in agreement

163

with other studies which have demonstrated significant VTG induction only at greater
exposure concentrations. Ten pg NP/L was reported to be the threshold exposure
concentration required to cause VTG induction in 2 year old rainbow trout (Jobling et al.
1996). Another study observed significant induction of VTG in rainbow trout exposed to
greater than 25 pg NP/L (Tremblay and Van der Kraak 1998). A previous study with
fathead minnows observed no significant induction of VTG in males at exposure
concentrations up to 2.4 pg NP/L, but significant VTG induction was observed in
females (Giesy et al. 2000). Thus, compared to other reports in the literature it seems
plausible that the exposure concentrations used for this study were insufficient to induce
VTG in sexually mature male carp.

The lack of VTG induction, as a model estrogenic response, is a bit more
surprising when one considers the estimated tissue and plasma concentrations of estradiol
equivalents (EEQ) contributed by NP. The relative estrogenic potency of NP has been
estimated based on estrogen response element (ERE) mediated reporter gene expression
(Legler et al. 1999, Gaido et al. 1997, Villeneuve et al. 1998, White et al. 1994), VTG
induction in rainbow trout hepatocytes (Tremblay and Van der Kraak 1998, White et al.
1994, Jobling and Sumpter 1993), and VTG induction in viva (Tremblay and Van der
Kraak 1998, Madsen et al. 1997). Relative potencies were in the range of 10'4 to 10'5 .
Based on this relative potency range, and the concentrations of NP measured in tissue
and pooled plasma samples, fish exposed to approximately 5 pg NP/L were estimated to
contain between 30 and 300 ng NP-derived EEQ (EEQNp) / ml plasma or g tissue.
Assuming the pooled plasma concentrations (Table 3) were representative, the minimum

plasma concentration of EEQNp was approximately 4.5-45 pg EEQNp/ml plasma. With

164

the exception of outliers, plasma E2 concentrations, ranged from < 175 pg E2/ml to 700
pg E2/m1 (Figure 2). Thus, for fish from the highest exposure group, EEQNp may have
exceeded endogenous EEQs by a factor of 40-400. Even in fish for the control group,
NP may have contributed 1-10% of the total EEQs in the plasma. Assuming that NP
functions as if it were simply a less potent form of E2, one might reasonably expect that a
40-400 fold increase in EEQ would be sufficient to yield a significant estrogenic
response. For example, over the spawning cycle of female carp, plasma E2
concentrations may vary by as little as 10-fold (Aida 1988).

These observations raise a number of important questions related to the use EEQs
for predicting estrogenic effects. What concentration or change in EEQs will produce a
physiological effect? Are physiological responses controlled by absolute concentrations,
changes in concentration, the timing or pulsatility of the changes, or combinations of all
the above? What factors control the uptake and distribution of EEQs among target
tissues? Unfortunately, there do not appear to be simple answers to these questions. Due
to the endogenous nature of estradiol, its effects, both physiological and adverse, are
intimately confounded with other variables. For example, up to 20-fold increases in E2
secretion may be associated with a temperature change as small as 5°C (Manning and
Kime 1984). Positive and negative feedback controls, tissue interactions, steroid
synthesis and metabolism pathways, carrier proteins, etc. all act together to modulate the
activity of endogenous E2. Thus, it would be extremely difficult to develop clear dose-
response (cause-effect) relationships between endogenous estradiol concentrations and

estrogenic effects in vivo. Currently, the inability to effectively characterize such

165

relationships dramatically limits the utility of EEQ estimates as a basis for predicting

biological effects of estrogenic substances.

4.4 Effect of NP Exposure on Plasma Steroid Concentrations

Exposure to concentrations of NP similar to those used in this study were reported
to produce statistically significant increases in plasma E2 in both male and female
fathead minnows (Giesy et al. 2000). In one experiment, a 10-fold increase from 2ng
E2/ml to 20 ng E2/ml was observed (Giesy et al. 2000). In a second experiment, a 2-3
fold increase from 2-3 ng E2/ml to 6 ng E2/ml was detected (Giesy et al. 2000). In this
study, plasma E2 concentrations varied at least 7-fold among fish (Figure 2).
Furthermore, approximately 50% of the samples had E2 concentrations which were
below the MDL of 175 pg/ml. Thus, although the results of this study do not support the
hypothesis that exposure to NP induced significant increases in plasma E2 in sexually
mature male carp under the exposure conditions used, they do not completely reject it
either. Fish to fish variability and the inability to accurately quantitate the full range of
plasma E2 concentrations may have simply limited the ability to resolve the effect.

Concentrations of E2 in plasma from carp exposed in this study were markedly
less than those detected in the fathead minnow study (Giesy et al. 2000). The E2
concentrations observed were very similar to those reported for feral carp, however
(Goodbred et al. 1997, F olmar et al. 1996). Values reported for male carp were generally
in the range of 100-800 pg/ml (Goodbred et al. 1997, Folmar et a1. 1996). The mean

concentration of E2 levels in serum from mature male channel catfish was reported to be

166

950 pg E2/ml i: 90 (Schlenk et al. 1997). Based on the results available, it was not clear
whether the fact that plasma E2 concentrations in this experiment were less than plasma
E2 concentrations in fathead minnows was due to species differences, differences in
breeding status or developmental stage, or differences in the exposure conditions.
Relative to literature values for carp, as well as those for catfish, however, the
concentrations detected in this study seem reasonable.

ll-ketotestosterone (l l-KT) is generally regarded as the most important form of
androgen for spermatogenesis in male fish (Goodbred et al. 1997). Several studies have
reported a relationship between contaminant body burdens and ll-KT (Fitzsimmons
1990, Leatherland 1992). E2/11-KT or E2/T ratios have been cited as a sensitive
biomarker of abnormal sex steroid concentrations (Bevans et al. 1996, F olmar et. al.
1996). In at least some fish species, balance between E2 and ll-KT concentrations
affects phenotypic sexual characteristics, brain and behavioral differentiation, and
development of reproductive organs (Hunter and Donaldson, 1983). Male carp collected
near a major metopolitan sewage treatment plant were found to have depressed plasma T
concentrations (F olmar et a1. 1996). Furthermore, NP was reported to interfere with
metabolic elimination of T in daphnids (LeBlanc et al. 1999). Thus, there were reasons
to investigate the potential for NP exposure to affect both T and ll-KT. Plasma
testosterone concentrations have been shown to be closely correlated with ll-KT
(Goodbred et a1. 1997), and methods for measuring T were more readily available than
those for ll-KT. Thus, this study focused on the effects of NP exposure on plasma T in

sexually mature male carp.

167

Exposure to NP did not elicit significant increases in plasma testosterone in
sexually mature male carp (Figure 4). Unlike for E2, all plasma samples analyzed had T
concentrations which were greater than the MDL. The range of plasma T variation
among fish was greater than 25-fold. This suggests that either much greater changes in
concentration or much greater sample sizes would be needed to detect a significant
change in plasma T resulting from NP exposure. Overall, however, the results of this
study do not support the hypothesis that exposure to NP significantly altered plasma T

concentrations in sexually mature male carp.

4.5 Study Limitations

Plasma sex steroids and the physiological processes they regulate can be
influenced by a number of factors. Temperature is known to have dramatic effects on
steroidogenesis in carp (Manning and Kime 1984). Within the optimal range for
steroidogenesis, around 24-29 °C, steroid concentrations may vary by as much as 20 fold
over 5°C (Manning and Kime 1984). As a result, careful control of water temperatures in
the exposure tanks was critical for this study. An exposure temperature around 10°C was
chosen for two reasons. First, this was a water temperature that could easily be
maintained in the laboratory without the need for hot and cold water mixing which would
have made it difficult to regulate flow-rates. Second, at temperatures less than 15°C there
is relatively little change in plasma steroid concentrations per °C change in temperature.
Thus, the study conditions were designed to minimize the effect of temperature as a

confounding factor in the study. At the same time, however, it was recognized that

168

temperatures less than 15° C are not ideal for steroidogenesis. Thus, it was unclear
whether NP could modulate steroidogenesis or steroid metabolism at the temperatures
used in this study, even if it were able to do so at greater temperatures. Thus, the
conclusions of this study are restricted to the range of 10-13°C temperatures maintained
during the exposure.

Plasma steroid concentrations also vary with gender, stage of development,
reproductive status, season, time of day, stress, etc (Goodbred 1997, Pankhurst and
Dedual 1994, Down et al. 1990, Aida 1988, Pickering et al. 1987, Zohar and Billard
1984). For the purposes of this study, these factors were controlled as much as possible
to reduce the variability of the endpoints and to minimize confounding factors which
would obscure the potential relationship between NP exposure and changes in the
endpoints measured. Single sex, sexually mature fish, were utilized to minimize
variation due to differences in gender, developmental stage, or reproductive state. In
order to minimize the effects of diurnal fluctuations, sample collection was restricted to
the same 1.5 hr period on each sampling day. Thus, these factors should not have
markedly affected study results. For comparative purposes, however, results of this study
will only be directly comparable to other studies using sexually mature male carp
exposed at 10-13° C for 28-31 (1 at approximately the same time of year. Numerous
studies using a variety of species and conditions are needed to fully elucidate the

potential mechanisms by which NP may elicit endocrine disrupting effects on fish.

4. 6 Conclusions

169

Under the conditions used for this study, exposure to NP did not elicit a
statistically significant increase in plasma E2, plasma T, or plasma VTG concentrations.
Furthermore, no-treatrnent related changes in gonad histology or morphological
parameters were observed. The results of this study do not provide support for the
hypothesis that NP can elicit endocrine modulating effects in fish through an indirect
mechanism of action. They do not, however, provide evidence to reject that hypothesis.
Exposure to NP was confirmed by the detection of approximately 0.5-3.5 ppm of NP in
pooled plasma or tissue samples from the carp studied. The lack of significant VTG
induction, despite the fact that NP may have contributed between 30 and 300 ng EEQNP /
ml plasma or g fish tissue, raises a number of questions regarding the utility of EEQ
estimates for predicting biological responses. Additionally, because of the lack of a
significant estrogenic response, it was not possible to calibrate the in vivo potency of NP
for producing an estrogenic effect in sexually mature male carp with in vitro potencies
reported for NP. The study does, however, provide useful information regarding the
concentrations and variability of plasma steroids and plasma VTG in sexually mature

male carp, which may aid in the design or interpretation of future studies.

5. Acknowledgements

This work was supported by US. Environmental Protection Agency (US. EPA)
Biology Exploratory Grants Program, grant R8537l-01-0; cooperative agreement CR
822983-01-0 between Michigan State University and the US. EPA; the National Institute

of Environmental Health Sciences Superfund Basic Research Program NIH-ES-049l 1;

170

and a Michigan State University Distinguished Fellowship to D.L. Villeneuve. The
authors thank C. Duda, S. Pastva, W. Hu, K. Kemler, K. Staffova, M. Nie, Dr. P. Jones,
R. Cory, and E. Nitsch for their assistance with sample collection and processing; J.
Olivero, Dr. P. Ganey, and Dr. J. Tiedje for access to instruments for performing RIAs;
Dr. J. Wade for technical advice regarding the RIAs; and S. Snyder and K. Kannan for

technical advice regarding extraction of NP from water and plasma.

6. References

1. Aida, K., 1988. A review of plasma hormone changes during ovulation in cyprinid
fishes. Aquaculture. 74: 1 1-21.

2. Ankley, G., Mihaich, E., Stahl, R., Tillitt, D., Colbom, T., McMaster, S., Miller, R.,
Bantle, J., Campbell, P., Denslow, N., Dickerson, R., Folmar, L., Fry, M., Giesy, J .,
Gray, L.E., Guiney, P., Hutchinson, T., Kennedy, 8., Kramer, V., LeBlanc, G.,
Mayes, M., Nimrod, A., Patino, R., Peterson, R., Purdy, R., Ringer, R., Thomas, P.,
Touart, L., Van der Kraak, G., Zacharewski, T., 1998. Overview of a workshop on
screening methods for detecting potential (anti-) estrogenic/androgenic chemicals in
wildlife. Environ. Toxicol. Chem. 17:68-87.

3. Bennie, D.T., 1999. Review of the environmental occurrence of alkylphenols and
alkylphenol ethoxylates. Water Qual. Res. J. Canada. 34279-122.

4. Bevans, H.E., Goodbred, S.L., Miesner, J .F ., Watkins, S.A., Gross, T.S., Denslow,
N.D., Schoeb, T., 1996. Synthetic organic compounds and carp endocrinology and

histology in Las Vegas Wash and Las Vegas and Callville Bays of Lake Mead,

171

10.

Nevada, 1992 and 1995. Water Resources Investigations Report 96-4266, US.
Department of the Interior—US. Geological Survey, Carson City, NV, USA.
Brooke L.T., 1993. Accumulation and lethality for two freshwater fishes (fathead
minnow and bluegill) to nonylphenol. Report to the US. EPA, Duluth, MN (68-C l -
0034). Lake Superior Research Institute, University of Wisconsin-Superior,
Superior, WI.

Christiansen, T., Korsgaard, B., Jespersen, A., 1998. Effects of nonylphenol and 17-
B-estradiol on Vitellogenin synthesis, testicular structure and cytology in male eelpout
Zoarces viviparus. J. Exp. Biol. 201 : 179-192.

Colbom, T., vom Saal, F.S., Soto, A.M., 1993. Developmental effects of endocrine-
disrupting chemicals in wildlife and humans. Environ. Health Perspect. 101:378-
384.

Down, N.E., Peter, R.E., Leatherland, J .F., 1990. Seasonal changes in serum
gonadotropin, testosterone, ll-ketotestosterone, and estradiol-17B levels and their
relation to tumor burden in gonadal tumor-bearing carp x goldfish hybrids in the
Great Lakes. Gen. Comp. Endocrinol. 77: 192-201.

Fang, J., Kannan, K., Giesy, J .P., Clemens, L.C., Nunez, A.A., 2000. Effects of
bisphenol A on energy balance and accumulation in rats. Arch. Environ. Contam.
Toxicol. (in press)

F itzsimmons, J ., 1990. Steroid hormones in male lake trout. Proceeding, Round

Table on Contaminant and Reproductive Problems in Salmonids: Windsor, Ontario,

Canada (Apr. 24-25, 1990). pp. 29-35.

172

11.

12.

l3.

14.

15.

16.

F olmar, L.C., Denslow, N.D., Rao, V., Chow, M., Crain, D.A., Enblom, J., Marcino,
J ., Guillette, L.J., 1996. Vitellogenin induction and reduced serum testosterone
concentrations in feral male carp (Cyprinus carpio) captured near a major
metropolitan sewage treatment plant. Environ. Health Perspect. 104:1096-1101.
Gaido, K.W., Leonard, L.S., Lovell, S., Gould, J.C., Babai, D., Portier, C.J.,
McDonnell, DP, 1997. Evaluation of chemicals with endocrine modulation activity
in a yeast-based steroid hormone receptor gene transcription assay. Toxicol. Appl.
Pharmacol. 143:205-212.

Giesy, J .P., Pierens, S.L., Snyder, E.M., Miles-Richardson, S.M., Kramer, V.J.,
Snyder, S.A., Nichols, K.M., Villeneuve, D.L., 2000 Effects of 4-nonyl phenol on
fecundity and biomarkers of estrogenicity in fathead minnows (Pimephales
promelas). Environ. Toxicol. Chem. (in press)

Gimeno, S., Komen, H., Gerritsen, A.G.M., Bowmer, T., 1998a. Feminisation of
young males of the common carp, Cyprinus carpio, exposed to 4-tert-pentylphenol
during sexual differentiation. Aquat. Toxicol. 43:77-92.

Gimeno, S., Komen, H., Jobling, S., Sumpter, J ., Bowmer, T., 1998b.
Demasculinisation of sexually mature male common carp, Cyprinus carpio, exposed
to 4-tert-pentylphenol during spermatogenesis. Aquat. Toxicol. 43:93-109.
Goodbred, S.L., Gilliom, R.J., Gross, T.S., Denslow, N.P., Bryant, W.B., Schoeb,
T.R., 1997. Reconnaissance of 17-B-estradiol, ll-ketotestosterone, vitellogenin, and
gonad histopathology in common carp of United States streams: potential for
contaminant-induced endocrine disruption. United States Department of the

Interior—US. Geological Survey, Sacramento, CA, USA, pp.48

173

17.

18.

19.

20.

21.

22.

23.

Hunter, G.A., Donaldson, E.M., 1983. Hormonal sex control and its application to
fish culture. In: Hoar, W.S., Randall, D.J., Donaldson, E.M., (Eds), Fish
Physiology, v. IX, Reproduction. Academic Press, Orlando, FL, USA, pp.223-303.
Jobling, S., Sumpter, J .P., 1993. Detergent components in sewage effluent are
weakly oestrogenic to fish: an in vitro study using rainbow trout (Oncorhynchus
mykiss) hepatocytes. Aquat. Toxicol. 27:361-372.

Jobling, S., Sheahan, D., Osborne, J.A., Matthiessen, P., Sumpter, J .P., 1996.
Inhibition of testicular growth in rainbow trout (Oncorhynchus mykiss) exposed to
estrogenic alkylphenolic chemicals. Environ. Toxicol. Chem. 15:194-202.

Kendall, R.J., Dickerson, R.L., Giesy, J .P., Suk, W.A., (Eds.), 1998. Principles and
Processes for Evaluating Endocrine Disruptors in Wildlife. SETAC Press, Pensacola,
FL, USA.

Kloas, W., Lutz, 1., Einspanier, R., 1999. Amphibians as a model to study endocrine
disruptors: II. Estrogenic activity of environmental chemicals in vitro and in vivo.
Sci. Total Environ. 225159-68.

Leatherland, J .F ., 1992. Endocrine and reproductive function in Great Lakes salmon.
In: Colborn, T., Clement, C., (Eds), Chemically Induced Alterations in Sexual and
Functional Development: The Wildlife/Human Connection. Princeton Scientific
Publishing, Princeton, NJ. p. 129-146.

Legler, J ., van den Brink, C.E., Brouwer, A., Murk, AJ., van der Saag, P.T., Vethaak,
A.D., van der Burg, B., 1999. Development of a stably transfected estrogen receptor-
mediated luciferase reporter gene assay in the human T47-D breast cancer cell line.

Toxicol. Sci. 48:55-66.

174

24.

25.

26.

27.

28.

29.

30.

Lutz, I., Kloas, W., 1999. Amphibians as a model to study endocrine disruptors: 1.
environmental pollution and estrogen receptor binding. Sci. Total Environ. 225:49-
57.

Madsen, S.S., Mathiesen, A.B., Korsgaard, B., 1997. Effects of l7B-estradiol and 4-
nonylphenol on smoltification and vitellogenesis in Atlantic salmon (Salmo salar).
Fish Physiol. Biochem. 17:303-312.

Manning, N.J., Kime, DE, 1984. Temperature regulation of ovarian steroid
production in the common carp, Cyprinus carpio L., in vivo and in vitro. Gen. Comp.
Endocrinol. 56:376-388.

McMaster, M.E., Munkittrick, K.R., Van Der Kraak, GJ., 1992. Protocol for
measuring circulating levels of gonadal sex steroids in fish. Can. Tech. Rept. Fish
Aquat. Sci. 1836: 29p.

Miles-Richardson, S.R., Pierens, S.L., Nichols, K.M., Kramer, V.J., Snyder, E.M.,
Snyder, S.A., Render, J .A., Fitzgerard, S.D., Giesy, J .P., 1999. Effects of waterborne
exposure to 4-nonylphenol and nonylphenol ethoxylate on secondary sex
characteristics and gonads of fathead minnows (Pimephales promelas). Environ. Res.
Section A. 80:8122-S137.

Naylor, C.G., Mieure, J.P., Adams, W.J., Weeks, J .A., Castaldi, F .J ., Ogle, L.D.,
Romano, R.J., 1992. Alkylphenol ethoxylates in the environment. JAOCS. 69:695-
703.

Nichols, K.M., 1997. Effects of suspect environmental endocrine disrupters on the
reproductive physiology of fathead minnows, Pimephales promelas. MS thesis.

Michigan State University, East Lansing, MI, USA.

175

31.

32.

33.

34.

35.

36.

37.

38.

Nichols, K.M., Miles-Richardson, S.R., Snyder, E.M., Giesy, J.P., 1999. Effects of
exposure to municipal wastewater in situ on the reproductive physiology of the
fathead minnow (Pimephales promelas). Environ. Toxicol. Chem. 18:2001-2012.
Nimrod, A.C., Benson, W.H., 1996. Environmental estrogenic effects of alkylphenol
ethoxylates. Crit. Rev. Toxicol. 26:335-364.

Nimrod, A.C., Benson, W.H., 1997. Xenobiotic interaction with and alteration of
channel catfish estrogen receptor. Toxicol. Appl. Pharmacol. 147:381-390.
Pankhurst, N.W., Dedual, M., 1994. Effects of capture and recovery on plasma levels
of cortisol, lactate, and gonadal steroids in a natural population of rainbow trout
Oncorhynchus mykiss. J. Fish Biol. 45:1013-1025.

Pickering, A.D., Pottinger, T.G., Carragher, J ., Sumpter, J .P., 1987. The effects of
acute and chronic stress on the levels of reproductive hormones in the plasma of
mature male brown trout Salmo trutta L. Gen. Comp. Endocrinol. 68:249-259.
Purchase, I.F.H., 1999. Ethical review of regulatory toxicology guidelines involving
experiments on animals: The example of endocrine disrupters. Toxicol. Sci. 52: 141-
147.

Servos, M.R., 1999. Review of the aquatic toxicity, estrogenic responses and
bioaccumulation of alkylphenols and alkylphenol polyethoxylates. Water Qual. Res.
J. Canada. 34: 123-177.

Smeets, J .M.W., van Holsteijn, I., Giesy, J.P., Seinen, W., Van den Berg, M., 2000.
Estrogenic potencies of several environmental pollutants, as determined by

vitellogenin induction in a carp hepatocyte assay. Toxicol. Sci. (in press).

176

39.

40.

41.

42.

43.

44.

45.

Snyder, E.M., 2000. Use of fish as bioassay organisms to assess wastewater effluents
for reproductive endocrine modulating chemicals. Ph.D. Dissertation. Michigan
State University, East Lansing, MI, USA (in preparation).

Snyder, S.A., Keith, T.L., Naylor, C.G, Staples, C.A., Giesy, J .P., 2000. Reliable
method for determining concentrations of nonylphenol and lower oligomer
nonylphenol ethoxylates in fish tissues. (submitted to Environ. Sci. Technol.) n
Snyder, S.A., Keith, T.L., Verbrugge, D.A., Snyder, E.M., Gross, T.S., Kannan, K.,
Giesy, J .P., 1999. Analytical methods for detection of selected estrogenic compounds

in aqueous mixtures. Environ. Sci. Technol. 33:2814—2820.

 

 

Soto, A.M., Sonnenschein, C., Chung, K.L., F emandes, M.F., Olea, N., Serrano, F .O.,
1995. The E-Screen assay as a tool to identify estrogens: an update on estrogenic
environmental pollutants. Environ. Health Perspect. 103(suppl. 7):l 13-122.

Staples, C.A., Weeks, J ., Hall, J .F ., Naylor, CG, 1998. Evaluation of the aquatic
toxicity and bioaccumulation of C8- and C9-alkylphenol ethoxylates. Environ.
Toxicol. Chem. 17:2470-2480.

Stokes, W.S., Marafante, E., 1998. Introduction and summary of the 13th meeting of
the Scientific Group on Methodologies for the Safety Evaluation of Chenricals
(SGOMSEC): Alternative testing methodologies. Environ. Health Perspect.
106(Suppl. 2):405-412.

Tremblay, L., Van Der Kraak, G., 1998. Use of a series of homologous in vitro and
in vivo assays to evaluate the endocrine modulating actions of B-sitosterol in rainbow

trout. Aquat. Toxicol. 43: 149- 162.

177

46.

47.

48.

49.

Tyler, C.R., Jobling, S., Sumpter, J.P., 1998. Endocrine disruption in wildlife: a
critical review of the evidence. Crit. Rev. Toxicol. 28:319-361.

Villeneuve, D.L., Blankenship, A.L., Giesy, J .P., 1998. Interactions between
environmental xenobiotics and estrogen receptor-mediated responses. In: Denison,
M.S., Helferich, W.G., (Eds), Toxicant Receptor Interactions. Taylor and Francis,
Philadelphia, PA, USA, pp. 69-99.

White, R., Jobling, S., Hoare, S.A., Sumpter, J.P., Parker, M.G., 1994.
Environmentally persistent alkylphenolic compounds are estrogenic. Endocrinology.
135:175-182.

Zohar, Y., Billard, R., 1984. Annual and daily changes in plasma gonadotropin and
sex steroids in relation to teleost gonad cycles. Trans. Amer. Fish. Soc. 1132444-

451.

178

SUMMARY AND CONCLUSIONS

The studies described in this dissertation centered around the application of
mechanism-specific in vitro bioassays in the field of environmental toxicology. The first
study (Chapter I) examined interactions between environmental xenobiotics and estrogen
receptor-mediated responses. Two types of in vitro assays were used to screen and rank
the potential estrogenicity of 11 chemicals based on their affinity to bind to the estrogen
receptor (ER) and/or induce estrogen response element (ERE)-mediated reporter gene
expression in MCF-7-luc cells. Six of the 11 compounds tested were able to displace
tritiated l7B-estradiol from the ER. Their rank order of affinity was 17B—estradiol (E2) >
coumestrol > l7B-ethynyl-estradiol (EE2) > nonylphenol (NP) > octylphenol (OP) >
bisphenol A (BPA). Indole-3-carbinol, atrazine, o,p ’-DDE, p,p ’-DDE, and 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD) had no measurable affinity for the ER. The ability
of each compound to induce estrogen response element (ERE)-mediated gene
transcription was measured using the MCF-7-luc in vitro bioassay. The rank order of
potency for producing luciferase expression in the MCF—7-luc bioassay was E2 > EE2 >
NP > OP > coumestrol > atrazine > BPA > indole-3-carbinol. TCDD, o,p ’-DDE, and
p,p ’-DDE did not induce luciferase expression. TCDD was shown to antagonize the
effect of E2 in the MCF-7-luc bioassay. In addition to screening and ranking, the assay-
specific potencies reported in Chapter 1 were later applied in several studies involving
the use of mass-balance analysis to aid in the identification of compounds or classes of
compounds responsible for the estrogenicity of various environmental samples

(Appendix A). Finally, comparison of receptor binding affinity and gene expression

179

assay results was used to generate several hypotheses related to the mechanism of action
of the compounds tested. Relative binding affinity for calf ER was correlated with the
potency for gene expression in MCF-7-luc cells, but was not predictive of efficacy. The
“ligand insertion hypothesis” was cited as one potential explanation for the lack of
correlation between receptor binding affinity and efficacy. It was also noted that
atrazine, which showed no affinity for the ER, was able to induce ERE mediated gene
expression, suggesting a non-ER-mediated mechanism for the in vitro estrogenic effects
of atrazine. Thus, Chapter 1 provided a useful overview of potential uses for in vitro
bioassays in environmental toxicology research.

The second study presented in this dissertation (Chapter 2) focused on the
development and characterization of a recombinant, rainbow trout cell line-based, assay
for assessing dioxin-like potency. The RLT 2.0 bioassay developed by Richter et a1.
(1997) was adapted to a 96-well plate format to increase assay efficiency. RLT 2.0-
specific relative potencies (REPS) were derived for a number of halogenated aromatic
hydrocarbons (HAHS) and compared to REPS based on other fish and mammalian
bioassays, both in vitro and in vivo. Overall, the REPS and rank ordering of HAHS based
on the RLT 2.0 assay was similar to those based on other rainbow trout-specific
bioassays in vitro and in vivo. This helped establish the utility of the RLT 2.0 bioassay
for evaluating dioxin-like potency of HAHS to fish. Because rank ordering and REPS
were relatively similar to those for mammalian assays, however, it was not clear whether
the fish cell line-based RLT 2.0 assay is more relevant for predicting effects in fish than
analogous mammalian cell-based assays. Sensitivity analysis indicated that variability in

RLT 2.0-based REP estimates yields approximately 10-fold uncertainty in TCDD

180

equivalents (TEQ) estimates calculated using RLT 2.0-derived values. This uncertainty
estimate was important for future application of the RLT 2.0 assay in studies involving
mass-balance analysis. Differences of less than one order of magnitude between
instrumentally derived TEQ and RLT 2.0 bioassay-derived TCDD equivalents (TCDD-
EQ) may not be relevant. Thus, Chapter 2 illustrates the type of method development
and characterization needed to establish and enhance the utility of a mechanism-specific
in vitro bioassays as tools for environmental toxicology research and risk
characterization.

The study described in Chapter 3 focused on the current debate surrounding the
derivation, use, and misuse of in vitro bioassay-based REP estimates in risk assessment
and environmental toxicology research. Chapter 3 presented a systematic approach
which was developed for evaluating the assumptions underlying REP estimation.
Furthermore, it described the use of multiple point estimates and relative potency bands
to characterize REPS in situations where parallelism between sample and standard dose-
response relationships cannot be demonstrated. The methods described were
demonstrated using three example data sets and have been subsesquently applied in a
number of other studies in our laboratory (Appendix A)

The fourth study described (Chapter 4) was an in vivo study designed, in part, to
address the relevance of current in vitro models for predicting estrogenic potencies in
fish, in vivo. Previous studies suggested that 4-nonylphenol (NP) may elicit estrogenic
effects in fish in vivo by modulating plasma steroid hormone concentrations. The study
described in Chapter 4 detected no significant increases in concentrations of E2,

testosterone (T), or vitellogenin (VTG) in plasma from sexually mature male common

181

carp exposed to waterborne NP for 28-31 (1 at 10-13°C. Thus, the study did not provide
evidence to support the hypothesis that NP may mediate estrogenic effects through an
indirect mechanism involving elevation of steroid hormone concentrations in plasma.
The lack of a detectable estrogenic effect in vivo as a result of exposure to NP, hindered
the ability to calibrate the known in vitro potency of NP to its potency for producing
estrogenic effects in sexually mature male carp. The lack of estrogenic response also
raised a number of questions regarding the utility of estimating plasma or tissue
concentrations of 17B-estradiol equivalents (EEQ) as a means of predicting the potential
for estrogenic effects in vivo.

There is clearly a need and demand to develop and establish the utility of
mechanism-specific in vitro bioassays as tool in environmental toxicology research and
risk characterization. The studies presented in this dissertation exemplify the type of
research that is needed to meet this demand. To a large extent, the technology and
understanding of molecular and cellular processes needed to develop effective in vitro
tools is already in place. The array of in vitro assays developed to address the emerging
issue of environmental estrogens (Chapter 1) provides evidence for this. What has
lagged behind, however, is research aimed at correlating and calibrating in vitro
responses to effects in vivo. The paucity of effective studies correlating responses in
vitro with effects in vivo is currently the greatest hindrance to the widespread use and
utility of in vitro bioassays as risk assessment tools. The utility of in vitro bioassay
results could also be increased through the development of standardized, widely

accepted, and reliable methods for analyzing and interpreting in vitro bioassay results.

182

Given such developments, however, mechanism-specific in vitro bioassays can be

important and powerful tools for environmental toxicology research and risk assessment.

Reference:

Richter CA, Tieber VL, Denison MS, Giesy JP. 1997. An in vitro rainbow trout cell

bioassay for aryl hydrocarbon receptor-mediated toxins. Environ Toxicol Chem. 16:543-

550.

183

APPENDIX A

RELATED PUBLICATIONS

184

Villeneuve, D.L., J.S. Khim, K. Kannan, J.P. Giesy. 2000. Response of fish and
mammalian in vitro bioassays to complex mixtures of polychlorinated

napthalenes, polychlorinated biphenyls, and polycyclic aromatic hydrocarbons.
(Submitted to Aquat. T oxicol. 1/5/00)

Villeneuve, D.L., J .S. Khim, K. Kannan, J. Falandysz, A.L. Blankenship, J .P.
Giesy. 2000. Relative potencies of individual polychlorinated napthalenes to
induce dioxin-like responses in fish and mammalian in vitro bioassays.
(Submitted to Arch. Environ. Contam. T oxicol. 12/1/99)

Villeneuve, D.L., W.M. DeVita, and R.L. Crunkilton. 1998. Identification of
cytochrome P4501A inducers in complex mixtures of polycyclic aromatic
hydrocarbons (PAHs). In Litte, E.E., A.J. DeLonay, and BM. Greenberg, eds.
Environmental Toxicology and Risk Assessment: Seventh Volume, AST M ST P
1333. American Society for Testing and Materials.

Villeneuve, D.L., R.L. Crunkilton, and W.M. DeVita. 1997. Aryl hydrocarbon
receptor-mediated toxic potency of dissolved lipophilic organic contaminants
collected from Lincoln Creek, Milwaukee, Wisconsin, USA, to PLHC-l
(Poeciliopsis lucida) fish hepatoma cells. Environ. Toxicol. Chem. 16:977-984.

Snyder, S.A., D.L. Villeneuve, E.M. Snyder, J .P. Giesy. 2000. Toxicity
identification and evaluation of estrogenic and dioxin-like compounds in
wastewater effluents. (Submitted to Environ. Sci. T echnol. 2/00).

Kannan, K., D.L. Villeneuve, N. Yamashita, T. Imagawa, S. Hashimoto, A.
Miyazaki, J .P. Giesy. 2000. Vertical profiles of dioxin-like and estrogenic
activities associated with a sediment core from Tokyo Bay, Japan. (Submitted to
Environ. Sci. T echnol. 2/21/00).

Yamashita, N., K. Kannan, T. Imagawa, D.L. Villeneuve, S. Hashimoto, A.
Miyazaki, J .P. Giesy. 2000. Vertical profile of polychlorinated dibenzo-p-
dioxins, -dibenzofurans, -napthalenes, -biphenyls, polycyclic aromatic
hydrocarbons, and alkylphenols in a sediment core from Tokyo Bay, Japan.
(Submitted to Environ. Sci. T echnol. 2/21/00)

Khim, J .S., D.L. Villeneuve, K. Kannan, W.Y. Hu, J .P. Giesy, S.G. Kang, K.J.
Song, C.H. Koh. 2000. Instrumental and bioanalytical measures of persistent
organochlorines in blue mussel (Mytilus edulis) from Korean coastal waters.
(Submitted to Arch. Environ. Contam. T oxicol 11/2/99).

Giesy, J .P., S.L. Pierens, E.M. Snyder, S. Miles-Richardson, V.J. Kramer, S.A.
Snyder, K.M. Nichols, and D.L. Villeneuve. 2000. Effects of 4-nonyl phenol on

185

fecundity and biomarkers of estrogenicity in fathead minnows (Pimephales
promelas). Environ Toxicol Chem (in press)

Khim, J .S., D.L. Villeneuve, K. Kannan, , C.H. Koh, and J .P. Giesy. 1999.
Characterization and distribution of trace organic contaminants in sediment from
Masan Bay, Korea: 2. in vitro gene expression assays. Environ Sci Technol.

33 :4206-421 1 .

Khim, J .S., K. Kannan, D.L. Villeneuve, C.H. Koh, and JP. Giesy. 1999.
Characterization and distribution of trace organic contaminants in sediment from

Masan Bay, Korea: 1. instrumental analysis. Environ Sci Technol. 33:4199-
4205.

Snyder, S.A., E. Snyder, D. Villeneuve, K. Kannan, A. Villalobos, A.
Blankenship, and J .P. Giesy. 1999. Instrumental and bioanalytical measures of
endocrine disruptors in water. In: Analysis of Environmental Endocrine
Disruptors, L. Keith, L.L. Needham, and T. Jones Eds. ACS Symposium Series.
American Chemical Society, Washington DC, USA.

Senthilkumar, K., C.A. Duda, D.L. Villeneuve, K. Kannan, J. Falandysz, and J .P.
Giesy. 1999. Butyltin compounds in sediment and fish from the Polish coast of
the Baltic Sea. Environ Sci Pollut Res 6:200-206..

Khim, J .S., D.L. Villeneuve, K. Kannan, K.T. Lee, S.A. Snyder, C.H. Koh, and
J .P. Giesy. 1999. Alkylphenols, polycyclic aromatic hydrocarbons (PAHS) and
organochlorines in sediment from lake Shihwa, Korea: instrumental and
bioanalytical characterization. Environ Toxicol Chem 18:2424-2432.

Blankenship A.L., K. Kannan, S. Villalobos, D. Villeneuve, J. Falandysz, T.
Imagawa, E. Jakobsson, A. Bergman, and J .P. Giesy. 1999. Relative potencies of
halowax mixtures and individual polychlorinated napthalenes (PCNs) to induce
Ah receptor-mediated responses in the rat hepatoma H4IIE-luc cell bioassay.
Organohalogen Compounds 42:217-220 (short paper). (Also submitted).

Kannan K., N. Yamashita, D.L. Villeneuve, S. Hashimoto, A. Miyazaki, and JP.
Giesy. 1999. Vertical profile of dioxin-like and estrogenic potencies in a
sediment core from Tokyo Bay, Japan. Organohalogen Compounds 42:33-38
(short paper).

Kannan K., D.L. Villeneuve, A.L. Blankenship, and J .P. Giesy. 1998. Interaction
of tributyltin with 3,3’,4,4’,5-pentachlorbiphenyl - induced ethoxyresorufin 0-

deethylase (EROD) activity in rat hepatoma cells. J. Toxicol. Environ. Health.
55:373-384.

186

APPENDIX B

RAW DATA FROM WATERBORNE EXPOSURE OF SEXUALLY MATURE MALE

COMMON CARP (C YPRINUS CARPIO) TO 4-NONYLPHENOL

187

Fig. 8.1. Temperature profiles over the course of the exposure of sexually mature male carp to 4-nonylphcnol (NP).

 

 

 

+Amin
-o—Amax

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+Cmin
1.
+Cmax
0
9.rtr.:..:.¢.:..::::::::+:t+ttt
012 3 4 5 6 7 8 91011121314d1516171a192021222324252627282930
all
13
“12*
g +Dmin
0.11..
g —a—Dmax
l—104» j
9t+4tltttr1:4,,ttttlrlttlttttlt.
012 3 4 5 6 7 8 9101112131435161718192021222324252627282930
aY
13
424»
%1 +Emin
g _ —u—Emax
*‘10--
9 fi+itifLLit"#¢%t#*‘#4%%%ii*i#

 

012 3 4 5 6 7 8 91011121314dfiy161718192021222324252627282930

 

 

 

min = minimum temperature max = maximum temperature
Temperatures monitored daily with electronic Hi/Lo thermometers (Fischer Scientific, Pittsburgh, PA, USA).
A = 10 uyL; C = solvent control; D = control; E = l ug/L; F = 0.3 ug/L; G = 0.1 ug/L H = 3.0 ug/L; nominal NP concentration in tank

188

Fig. 8.1. (continued)

 

 

 

 

 

 

13
(:12 . +Fmin
3111 -a—Fmax
11°10l

9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I I
I T Y I I T

 

I T I I

012 3 4 5 6 7 8 91011121314Ja§161718192021222324252627282930

 

 

 

min = minimum temperature max = maximum temperature
Temperatures monitored daily with electronic Hi/Lo thermometers (Fischer Scientific, Pittsburgh, PA, USA).
A = 10 ug/L; C = solvent control; D = control; E = 1 ug/L; F = 0.3 ug/L; G = 0.1 ug/‘L H = 3.0 ug/L; nominal NP concentration in tank

189

Table 8.1. Results of weekly water quality monitoring over the course of exposure of
sexually mature male carp to 4-nonylphenol (NP).

 

Date Tank pH NH-3 NO-2 DO Hardness Conduct.
mg/L my]. mg/L mg/L umho

 

 

 

 

3/16/99 A 7.57 <0.02 <0.02 10 400 546
B 7.58 <0.02 <0.02 10 390 537

C 7.57 <0.02 <0.02 9.8 390 542

D 7.58 <0.02 <0.02 9.8 380 546

E 7.56 <0.02 <0.02 9.8 380 546

F 7.56 <0.02 <0.02 9.8 390 543

O 7.57 <0.02 <0.02 9.8 390 546

H 7.57 <0.02 <0.02 9.8 390 539

3/23/99 A 7.6 <0.02 <0.02 10.6 400 533
B 7.61 <0.02 <0.02 10.7 400 530

C 7.6 <0.02 <0.02 10.8 400 531

D 7.6 <0.02 <0.02 10.8 400 527

E 7.59 <0.02 <0.02 10.8 410 509

F 7.59 <0.02 <0.02 10.8 408 531

G 7.6 <0.02 <0.02 10.8 420 563

H 7.59 <0.02 <0.02 10.8 400 557

3/30/99 A 7.58 <0.02 <0.02 10.9 400 577
B 7.65 <0.02 <0.02 10.9 425 567

C 7.58 <0.02 <0.02 10.9 410 567

D 7.59 <0.02 <0.02 10.8 410 565

E 7.57 <0.02 <0.02 10.7 425 567

F 7.58 <0.02 <0.02 10.7 425 559

G 7.58 <0.02 <0.02 10.7 410 564

H 7.57 <0.02 <0.02 10.4 410 557

4/6/99 A 7.53 <0.02 <0.02 10.4 400 544
B 7.53 <0.02 <0.02 10.6 400 537

C 7.51 <0.02 <0.02 10.6 nm 538

D 7.55 <0.02 <0.02 10.6 nm 540

E 7.51 <0.02 <0.02 10.6 nm 543

F 7.52 <0.02 <0.02 10.6 nm 539

O 7.52 <0.02 <0.02 10.6 nm 544

H 7.5 <0.02 <0.02 10.3 nm 539

Mean 7.57 --- --- 10. 5 402 546
SD 0.03 --- --- 0.4 12.6 14.7

CV 0.44 -- -- 3.8 3.1 2.7

 

A = 10 ug/L; C = solvent control; D = control; B = 1 ug/L; F = 0.3 ug/L;
G = 0.1 ug/L H = 3.0 ug/L; nominal NP concentration in tank
Tank B was not part of this study.

pH measured with a Pinpoint pH meter (Aquatic Eco-systems, Apopka, FL, USA)

NH-3 = Ammonia; measured with a LaMotte Ammonia Nitrogen Test Kit - low range
(Aquaculture Supply, Dade City, FL, USA)

NO-2 = Nitrite; measured with a LaMotte Nitrite Test Kit (Aquaculture Supply)

D0 = Dissolved oxygen; measured with a YSI Model 57 Dissolved Oxygen Meter
(YSl, Yellow Springs, OH, USA)

Hardness measured with a LaMotte Hardness Test Kit (Aquaculture Supply)

nm = no measurement; ran out of kit reagents

Conduct. = Conductivity; measured with a Pinpoint Conductivity Meter (Aquatic Eco-systems)

190

Table B.2. Blood volume collected, length, mass, gonad mass, hepatopancreas mass. gonado-somatic index
(051), hepato-somatic index (HSI), and duration of exposure (time d) for individual sexually mature male

carp exposed to 4-nonylphcnol at the nominal concentrations indicated below.

 

 

 

 

 

 

 

 

LD. time (d) NP (tag/L) blood vol (til) lggtflem) mass and It CS] (7.) H81 (7.)
D1 27 control 800 13.7 53.5 1.757 0.21057 3.28 0.39
d2 27 control 1800 14.6 80.9 0.63531 1.30662 0.79 1.62
d3 27 control 2400 17.6 121.8 4.385 2.1346 3.60 1.75
d4 27 control 2300 17.2 125 .6 4.69882 1.78915 3.74 1.42
d5 27 control 2200 15 83.5 2.8637 1.47 3.43 1.76
d6 28 control 2400 17.5 135.7 4.214 2.8133 3.11 2.07
d7 28 control 1100 15.8 90.9 3.0777 1.5763 3.39 1.73
d8 28 control 1400 15.3 74.7 1.9734 1.5849 2.64 2.12
d9 28 control 2000 17.6 128.5 3.7452 3.3926 2.91 2.64
d10 28 control 1800 16.1 93.4 2.9781 1.6256 3.19 1.74
d1 1 29 control 2200 16.7 103.5 2.253 1.1624 2.18 1.12
d12 29 control 1500 16.1 81.2 3.9125 1.2384 4.82 1.53
dl3 , 29 control 3000 18.1 125.6 3.4927 1.7862 2.78 1.42
dl4 29 control 800 15 73.7 2.6472 1.5177 3.59 2.06
d15 29 control 1200 15.3 91.5 1.9245 1.7097 2.10 1.87
d16 30 control 1500 13.5 55 1.3849 0.8402 2.52 1.53
d]? 30 control 1800 16.5 101.9 3.79803 1.576 3.73 1.55
d18 30 control 1200 14.9 74.5 3.07143 1.53093 4.12 2.05
dl9 30 control 1800 14 66.9 1.4034 1.467 2.10 2.19
d20 30 control 2200 15.6 90.5 2.3891 1.68671 2.64 1.86
Mean 1770 15.8 92.6 2.83025 1.62094 3.03 1.72
SD 583 1.36 24.4 1.11356 0.64957 0.87 0.46
CV _ _ . ..32-_..9_._- __ 8-60 _. 26.-4-" ._ - .39-_3_ _ _.__._.‘19.-_1 _, .-__28-35_ ___._26.77__
W) blood vol (III) len II em mass onad h GSI (7.) [181 (7.)
cl 27 SC 2000 18.5 150.4 3.97341 4.22042 2.64 2.81
c2 27 SC 800 14.3 68.4 2.7392 0.762237 4.00 1.11
c3 27 SC. 1700 17 114.5 4.4003 1.74 3.84 1.52
c4 27 SC. 2300 16.5 112.2 5.08457 0.82609 4.53 0.74
c5 27 SC. 1000 14 62.4 2.3753 0.90514 3.81 1.45
c6 28 SC. 2000 16 97 2.2854 2.0756 2.36 2.14
c7 28 SC. 1000 15.3 79.4 1.1448 0.9985 1.44 1.26
c8 28 SC 1400 15 53.5 1.9069 1.3563 3.56 2.54
e9 28 SC 2300 17.1 112.7 5.1165 1.6246 4.54 1.44
c10 28 SC 1600 14.5 65.4 2.1758 1.0003 3.33 1.53
cl 1 29 SC 1100 14.1 61.2 4.4283 0.3861 7.24 0.63
C12 29 SC. 3200 19.2 172.4 6.7389 2.0452 3.91 1.19
CB 29 SC. 1700 14.6 77.7 1.8398 1.3633 2.37 1.75
CM 29 SC. 2300 16.8 104.9 2.5843 1.963 2.46 1.87
c15 29 SC. 2900 17.5 123.6 3.1813 2.176 2.57 1.76
c l 6 30 SC. 1700 15 81.1 3.95022 0.8852 4.87 1.09
cl7 30 SC. 1900 16 87.9 3.91827 1.37 4.46 1.56
cl 8 30 SC. 700 13 52 1.90752 0.94295 3.67 1.81
c19 30 SC 2500 15.9 95.8 2.6012 1.31301 2.72 1.37
c20 30 SC. 2100 15.5 81 1.07739 1.29003 1.33 1.59
c2] 30 SC. 1700 15.1 74.8 3.02034 0.83616 4.04 1.12
Mean 1805 15.8 91.8 3.16427 1.43239 3.51 1.54
SD 664 1.55 31.2 1.44600 0.80735 1.32 0.53
CV 36.8 9.82 34.0 45.7 56.4 37.5 34.2

 

191

Table B.2. (continued)

 

 

 

 

 

 

 

 

 

 

 

192

ID. time (d) NP (ugly blood vol (ul) lenW GSI (%) HSI (%)
g1 27 0.1 2400 110. 7 3. 2312 1.93503 2. 92 1. 75
g2 27 0.1 1500 17.9 133.6 3.48636 3.07345 2.61 2.30
g3 27 0.1 1800 16.9 106.1 2.2535 1.6724 2.12 1.58
g4 27 0.1 800 15.7 96.5 3.22408 1.1198 3.34 1.16
g5 27 0.1 2200 17.3 110.1 2.91226 2.19941 2.65 2.00
g6 28 0.1 1600 14.5 165.8 1.5246 0.6309 0.92 0.38
g7 28 0.1 2000 16.7 101.3 4.6432 1.7809 4.58 1.76
g8 28 0.1 1500 15.4 76 3.2346 1.0339 4.26 1.36
g9 28 0.1 2200 17.3 118.2 4.0266 1.506 3.41 1.27
g10 28 0.1 1000 15.8 91.2 2.9243 1.3885 3.21 1.52
gl 1 29 0.1 1300 16.3 93.1 3.8056 1.5088 4.09 1.62
g12 29 0.1 1700 15 70.3 2.9256 0.9968 4.16 1.42
g13 29 0.1 1700 14.5 64 1.369 1.3714 2.14 2.14
gl4 29 0.1 1800 15.7 78.4 1.0325 0.8981 1.32 1.15
g15 29 0.1 1600 16 95.6 3.6046 1.8872 3.77 1.97
g16 30 0.1 2000 15.8 89.1 3.1352 1.2258 3.52 1.38
g17 30 0.1 1800 17.1 104.8 3.3001 1.85825 3.15 1.77
g18 30 0.1 2000 14.6 72.2 1.75212 0.56474 2.43 0.78
gl9 30 0.1 1600 14.1 70 2.1501 0.72502 3.07 1.04
Mean 1711 16.0 97.2 2.87029 1.44086 3.03 1.49
SD 398 1.1 1 24.8 0.95790 0.61404 0.97 0.48
_CV 23.3 6.96 25.5 33.4 42.6 32.1 31.9

LD. time (d) NP (ug/L) blood vol (ul) lenflhSem) mass“) gonad (g) be (g) GSI (°/o) HSL%)
f1 27 0.3 1200 17.7 140.6 6.89015 4.20506 4.90 2.99
12 27 0.3 2200 16.1 103.4 5.09631 2.35151 4.93 2.27
13 27 0.3 1700 17.5 112.5 4.37504 1.79755 3.89 1.60
f4 27 0.3 3000 18.5 143.6 5.43126 2.88449 3.78 2.01
f5 27 0.3 2000 15.5 90.8 4.1596 1.71012 4.58 1.88
f6 28 0.3 2800 18 131.2 5.7918 1.5984 4.41 1.22
17 28 0.3 1800 18 130.8 5.3373 2.1175 4.08 1.62
18 28 0.3 2100 16.2 82.1 2.2708 0.8699 2.77 1.06
f9 28 0.3 1100 14 66.9 2.7776 1.1525 4.15 1.72
110 28 0.3 1500 16 90.7 2.632 1.1645 2.90 1.28
f1 1 29 0.3 1600 15.1 77.4 3.4591 0.9156 4.47 1.18
f1 2 29 0.3 1700 15.7 94.8 9.6727 2.1954 10.20 2.32
f] 3 29 0.3 1700 15.2 79.1 0.9295 1.5803 1.18 2.00
f14 29 0.3 1700 15.5 78.5 3.6662 1.5357 4.67 1.96
f1 5 29 0.3 1700 16.2 94.2 1.1289 2.0749 1.20 2.20
116 30 0.3 1800 15.7 88.9 4.17535 0.65057 4.70 0.73
117 30 0.3 1800 16.5 99.5 2.954 1.43282 2.97 1.44
f] 8 30 0.3 2000 16.6 98 2.75294 1.02731 2.81 1.05
f19 30 0.3 500 14.1 61 1.5541 1.21414 2.55 1.99
Mean 1784 16.2 98.1 3.95024 1.70938 3.95 1.71
SD 552 1.27 23.9 2.13430 0.83249 1.90 0.55
CV_ 30.9 7.82 24.4 54.0 48.7 48.1 __ 32.4

Table B.2. (continued)

 

LD. time (d) NP (us/L) blood vol Q11) lenEhScm) mass“! gonad (5! he“) GSlg/o) 1181 (%L

 

 

 

 

 

 

 

 

e1 27 1 1400 15 82.1 3.79961 1.47332 4.63 1.79
e2 27 1 400 18.3 137.4 3.85092 2.65732 2.80 1.93
e3 27 1 1800 15.1 80.9 4.24401 1.43 5.25 1.77
e4 27 1 2900 18.6 143 4.47401 2.50955 3.13 1.75
e5 27 1 3000 18.6 151.4 3.14216 3.21218 2.08 2.12
e6 28 1 1100 14.5 62.4 0.776 0.9395 1.24 1.51
e7 28 1 1600 14.4 56.8 0.521 1.032 0.92 1.82
e8 28 1 1300 14.7 43.3 1.6992 0.9944 3.92 2.30
e9 28 1 4100 20.6 206.9 7.9936 5.3495 3.86 2.59
e10 28 1 1500 17.7 120.2 4.1787 1.0992 3.48 0.91
e11 29 1 100 12.8 48.7 2.0337 0.8294 4.18 1.70
e12 29 l 1600 14.6 70.5 2.6908 0.9482 3.82 1.34
e13 29 1 1700 15.7 89.7 2.9794 1.1068 3.32 1.23
e14 29 l 2900 16.3 99.3 3.382 2.29 3.41 2.31
e15 29 1 1800 15.8 83.7 2.7934 1.294 3.34 1.55
e16 30 1 2100 17 111.3 2.08 1.7265 1.87 1.55
e17 30 1 1300 16.7 108 4.0576 1.905 3.76 1.76
e18 30 1 2400 17.6 130.7 7.53862 1.7458 5.77 1.34
e19 3O 1 2600 18.5 136.1 4.25165 2.3104 3.12 1.70
e20 30 1 3000 17.9 134.8 4.67539 2.6518 3.47 1.97
Mean 1930 16.5 104.9 3.55809 1.87524 3.37 1.75
SD 960 1.95 41.0 1.86575 1.07806 1.20 0.40
CV 49.8 11.8 39.1 52.4 57.5 35.6 22.7
1.0. time (d) NP (lug/L) blood vol (ul) len2h(cmz mass(g) gonad (g! 112 15) CS! (7.) HS! (:6)
h] 27 3 1700 18.4 126.8 5.62726 1.86302 4.44 1.47
h2 27 3 2000 18.5 132.6 5.14829 2.37267 3.88 1.79
h3 27 3 3000 18.1 135.4 1.31 2.25401 0.97 1.66
M 27 3 800 15.5 83.1 2.701 1.13207 3.25 1.36
h5 27 3 500 16.3 97.8 2.90587 1.939 2.97 1.98
h6 28 3 300 15 72.7 2.7285 0.5114 3.75 0.70
M 28 3 1800 15.5 84.6 2.4848 1.8501 2.94 2.19
h8 28 3 2400 17.3 112.8 0.6043 1.7675 0.54 1.57
h9 28 3 1200 14 61.4 2.5772 0.7022 4.20 1.14
MO 28 3 1200 14.1 67.1 2.1011 0.5604 3.13 0.84
1111 29 3 1700 15.3 77.9 2.5545 1.0251 3.28 1.32
h12 29 3 1300 15.3 81.2 2.5702 1.0704 3.17 1.32
h13 29 3 2500 17.3 118.5 4.0724 1.4348 3.44 1.21
h14 29 3 2900 17.9 128.7 3.1004 1.3896 2.41 1.08
ms 29 3 1300 14.5 65.6 1.4886 0.9214 2.27 1.40
h16 30 3 1400 14.5 58.9 1.4653 0.4582 2.49 0.78
h17 30 3 800 13.2 56.4 1.54761 0.37485 2.74 0.66
h18 30 3 2000 14.6 75.5 3.43564 1.27311 4.55 1.69
h19 3O 3 1200 13.9 55.8 1.82523 0.79142 3.27 1.42
h20 30 3 2500 17.5 120 4.66924 2.16846 3.89 1.81
Mean 1625 15.8 90.6 2.74587 1.29299 3.08 1.37
SD 766 1.69 28.1 1.31578 0.63812 1.02 0.42
52y 47.2 10.1.6. 31.9_.,_-_-47-9 49.4 .33.; 30;9_

 

193

 

Table B.2. (continued)

 

I..D time (d) NPu 1gag/L) blood vol (ulLlengthcm) mass(g) gonad (g) liver (5.,

 

GSI (%) H3117.)

 

 

a1 15 500 76.7 4. 5078 1.0109 5.88 1.32
a2 27 10 1800 16.2 86.5 1.53261 1.45122 1.77 1.68
33 27 10 200 13.2 56.9 1.5434 0.9658 2.71 1.70
a4 27 10 2600 17 108.9 4.47443 1.70399 4.1 1 1.56
a5 27 10 1800 17.5 120.1 2.25501 3.03653 1.88 2.53
36 27 10 1700 16 83.3 1.963 1.57929 2.36 1.90
217 28 10 1600 17 110.6 5.8313 2.0188 5.27 1.83
218 28 10 1100 16 93.3 5.035 1.9561 5.40 2.10
39 28 10 2000 16.3 89.9 0.685 1.7822 0.76 1.98
2110 28 10 2100 16.4 93.4 2.3099 1.397 2.47 1.50
all 28 10 3100 18.1 139.7 2.2068 2.5225 1.58 1.81
a12 29 10 1500 14.7 71.3 1.777 1.2115 2.49 1.70
a13 29 10 2000 15.5 75.1 3.143 0.836 4.19 1.11
314 29 10 3100 18.5 143 4.4995 3.034 3.15 2.12
315 29 10 2000 16.3 97.3 4.1581 1.413 4.27 1.45
a16 29 10 1800 14 59.8 1.0879 1.0664 1.82 1.78
a17 30 10 2100 15.2 74.5 3.46291 not weighed 4.65 NA
a18 30 10 1400 13.9 58 0.8713 0.91787 1.50 1.58
319 30 10 1100 14.4 66.4 3.31897 1.2805 5.00 1.93
a20 30 10 700 14.9 74.6 3.227 1.488 4.33 1.99
Mean 1710 15.8 89.0 2.89450 1.61429 3.28 1.77
SD 757 1.42 25.0 1.49408 0.65662 1.54 0.32
CV 44.3 8.96 28.1 51.6 40.7 46.9 18.2

 

194

Table B.3. Concentrations of ”beta-estradiol (52) in plasma

from individual sexually mature male carp exposed to 4-nonylphenol (NP)
Mean, standard deviation (SD), and coefficient of variation (CV)

across three replicate determinations is presented.

Method detection limit was 175 pg E2/ml. Values less than 175 pg E2/ml
may not be accurate.

 

 

 

 

 

 

 

 

 

LD. E2 (pg/m1) SD CV
A10 112 12.3 11.0
All 116 9.8 8.4
A12 [ 3374 144.4 4.3JOutlier
A13 191 26.0 13.6
A14 217 41.9 19.3
A15 142 2.3 1.6 Nominal Exposure Conc.
A16 114 8.1 7.1 10ugNP/L
A17 277 9.4 3.4
A18 302 67.6 22.4
A19 140 35.4 25.2
A20 168 31.1 18.5
A4 188 21.1 1 1.2
A5 133 11.7 8.8
A6 158 14.0 8.8
A7 199 14.9 7.5
A9 68 18.4 27.2
Mean 168
SD 63.2
n 16
LB. E2 (pg/ml) SD CV
C1 88 76.7 87.5
C10 272 25.8 9.5
C11 179 27.3 15.3
C 12 128 29.2 22.7 Nominal Exposure Conc.
C13 185 39.8 21.5 Solvent Control
C14 420 19.4 4.6
C15 181 12.9 7.1
C16 196 21.7 1 1.1
C19 301 20.5 6.8
C2 142 100.5 71.0
C20 1 19 11.0 9.2
C21 177 18.6 10.5
C3 221 103.9 47.1
C4 687 47.4 6.9
C5 136 11.9 8.7
C6 249 35.4 14.2
C7 107 5.6 5.2
C8 184 22.0 12.0
C9 174 23.9 13.7
Mean 218
SD 137.5
n 19

 

195

Table B.3. (continued)

 

1.1). E2 (pg/m1) so CV

 

 

 

 

 

 

D1 138 128.2 92.8
D10 145 29.4 20.3
D11 235 57.6 24.6
D12 133 30.8 23.1
D13 287 19.8 6.9
D14 209 29.2 14.0
D15 204 25.9 12.7
D17 156 23.3 14.9
D19 206 29.4 14.2
D2 91 45.1 49.3
D20 152 9.8 6.5
D3 188 86.8 46.2
D4 198 52.6 26.6
D5 102 22.6 22.2
D6 165 32.3 19.6
D8 158 27.4 17.3
D9 105 17.3 16.4
D9 238 42.5 17.8
Mean 173

SD 52.1

n 18

LB. E2 (pg/m1) SD CV
E1 172 129.2 74.9
E10 110 9.3 8.5
E12 186 21.9 11.8
E13 228 45.2 19.8
E14 179 29.0 16.2
E15 314 17.7 5.6
E17 159 31.3 19.7
E18 269 20.4 7.6
E19 123 30.7 24.9
E20 160 8.4 5.2
E3 218 82.8 38.1
E4 161 22.3 13.9
E5 136 10.3 7.6
E6 298 5.7 1.9
E7 163 19.0 11.6
E8 217 17.9 8.2
E9 78 21.2 27.0
Mean 187

SD 64.4

n 17

 

196

Nominal Exposure Conc.
Control

Nominal Exposure Conc.
1.0 ug NP / L

Table B.3. (continued)

 

1.1). 132 (pg/ml) s1) cv

 

 

 

 

 

 

 

 

F1 107 96.6 90.5
F10 208 16.4 7.9
F11 79 8.0 10.1
[F12 2935 240.4 8.2
F13 255 36.9 14.5
F14 241 48.2 20.0
F15 163 12.4 7.6
F16 150 5.1 3.4
F17 327 21.5 6.6
F18 170 20.6 12.1
F2 113 95.5 84.7
F3 301 161.7 53.7
F4 136 8.3 6.1
F5 229 20.0 8.7
F6 232 26.8 11.6
F8 193 46.5 24.1
F9 174 41.3 23.7
F9 173 2.8 1.6
Mean 191
SD 67.4
n 17
1.13. E2 (pg/m1) so CV
(31 115 91.1 79.2
(310 166 10.1 6.1
(311 134 34.1 25.5
(312 261 30.4 11.7
(313 299 40.5 13.5
014 180 25.6 14.2
(315 249 20.3 8.2
G16 364 10.6 2.9
(318 337 51.6 15.3
(319 171 28.1 16.5
(32 115 88.5 76.9
(33 147 12.1 8.2
(33 380 68.3 18.0
(34 331 15.3 4.6
(35 97 14.9 15.4
G6 155 51.2 33.0
(37 402 85.8 21.3
(38 122 24.5 20.0
(39 88 17.8 20.2
Mean 216
SD 105.9
n 19

 

Female

197

Nominal Exposure Conc.
0.3 ug NP / L

Nominal Exposure Conc.
0.1 ug NP / L

Table B.3. (continued)

 

1.13. E2 (pg/ml) s1) CV

 

 

 

H1 202 158.6 78.6
H10 100 3.3 3.3
H11 103 28.2 27.3
H12 276 19.3 7.0
H13 700 88.8 12.7
H14 159 23.0 14.5
H15 301 54.1 18.0
H16 262 10.1 3.9
H17 191 11.4 6.0
H18 298 60.1 20.2
H2 136 94.5 69.4
H20 357 43.4 12.1
H3 241 26.9 11.2
H4 169 13.1 7.8

H5 174 28.6 16.5
H7 306 21.9 7.2
H8 109 7.6 7.0

H9 108 10.1 9.4

Mean 233

SD 142

n 18

 

198

Nominal Exposure Conc.
3.0 ug NP / L

Table B.4. Concentrations of testosterone (T) in plasma

from individual sexually mature male carp exposed to 4-nonylphenol (NP)
Mean, standard deviation (SD), and coefficient of variation (CV)

across three replicate determinations is presented.
a = 1:30 dilution of plasma extract; b = 1:90 dilution of plasma extract

 

 

 

 

 

 

 

LD. T (pg/ml) SD CV a,b mean
A10a 3613 245 6.8 3780
A 10b 3948 399 10.1
Alla 9194 715 7.8 9519
A1 1b 9844 977 9.9 Nominal Conc.
A12a 2222 329 14.8 2489 10 ug NP/L
A12b 2756 339 12.3
Al4a 2616 272 10.4 2692
A14b 2768 72 2.6
A15a 5711 258 4.5 5948
Ale 6184 924 14.9
A 1 7a 8673 933 10.8 9038
A17b 9404 290 3.1
Al8a 4120 276 6.7 4240
A18b 4361 305 7.0
A53 3372 113 3.4 4176
A5b 4981 560 l 1.2
A6a 11610 4847 41.7 13624
A6b 15638 717 4.6
Mean 6167 6167

SD 3745 3777

CV 61 61

1.0. T ml) SD CV a,b mean
C 10a 1 1648 748 6.4 1 1246
C10b 10844 1576 14.5
C12a 6742 222 3.3 6643
C12b 6544 186 2.8 Nominal Cone.
C 1 3a 4983 224 4.5 6359 Solvent control
C13b 7736 930 12.0
C 14a 8384 40 0.5 8776
C14b 9168 150 1.6
C 1 53 6187 394 6.4 6854
C15b 7521 399 5.3
C17a 13555 890 6.6 13858
C17b 14161 1378 9.7
C19a 8440 309 3.7 9001
C19b 9563 454 4.7
Cla 10137 493 4.9 10492
C 1 b 10847 2968 27.4
C4a 10342 1244 12.0 11010
C4b 11678 1292 11.1
C9a 21227 341 1.6 24669
C9b 281 10 2859 10.2
Mean 10891 10891

SD 5403 5394

CV 50 50

 

199

Table B.4. (continued)

1.D. ngml) SD CV a,b mean

D10a 9866 384 3.9 10990

D10b 12114 276 2.3

D1 1a 3800 410 10.8 3709

D1 1b 3617 664 18.4 Nominal Conc.
D13a 4138 135 3.3 3897 Control
D13b 3655 315 8.6

D15a 3419 338 9.9 3199
D15b 2980 324 10.9

D16a 2049 181 8.8 2302

D16b 2555 423 16.5

D19a 7247 43 1 5.9 7496
01% 7744 288 3.7

D20a 8884 361 4.1 9807
D20b 10730 1036 9.7

D2a 3229 638 19.8 3434

D2b 3640 520 14.3

D33 9122 166 1.8 8830

D3b 8538 885 10.4

D9a 12651 303 2.4 12087

D9b 11523 419 3.6

 

 

 

Mean 6575 6575
SD 3612 3669
CV 55 56

 

 

1.D. TSEg/ml) SD CV a,b mean
E12a 12705 676 5.3 14256
E12b 15807 324 2.0

E14a 1 1698 343 2.9 12679
E14b 13659 533 3.9

E15a 5700 320 5.6 5325 Nominal Conc.
E15b 4950 963 19.4 1.0 ug NP/L
E16a 9865 882 8.9 8545
E16b 7225 201 2.8

E18a 16274 992 6.1 18727
E18b 21180 1724 8.1

E19a 11815 1038 8.8 13941
E19b 16067 664 4.1

E2a 20741 608 2.9 241 18
E2b 27495 3316 12.1

ESa 10265 655 6.4 1031 1
E5b 10357 907 8.8

E7a 3973 237 6.0 3961
E7b 3948 828 21.0

E8a 1293 145 1 1.3 1767
E8b 2241 191 8.5

 

Mean 1 1363 1 1363
SD 6909 6880
CV 61 61

 

200

Table B.4. (continued)

 

 

 

 

 

 

 

LD. T (pg/m1) SD CV a,b mean
F10a 8329 428 5.1 8189
F 10b 8049 339 4.2

F1 la 3639 335 9.2 3757
F1 1b 3874 392 10.1

F12a 6060 466 7.7 5601
F12b 5142 445 8.7

F13a 3435 1 10 3.2 3466
F13b 3497 398 11.4

F15a 1558 359 23.1 1888
F 15b 2219 203 9.2

F16a 2443 208 8.5 2573
Fl6b 2704 163 6.0

F17a 4758 89 1.9 4876
F17b 4993 666 13.3

F3a 14078 1146 8.1 15991
F3b 17904 1541 8.6

F6a 18393 583 3.2 20896
F6b 23398 682 2.9

F7a 16777 846 5.0 18763
F7b 20748 1633 7.9

Mean 8600 8600
SD 7090 7175
CV 82 83
LB. T1301“) SD CV a,b mean
612a 1544 98 6.3 1775
Gle 2007 351 17.5

613a 5172 400 7.7 4802
Gl3b 4433 494 11.1

615a 7465 435 5.8 7857
Gle 8250 426 5.2

G18a 6000 148 2.5 6084
Gle 6168 259 4.2

01% 6875 380 5.5 6713
019a 6551 923 14.1

Gla 14060 988 7.0 13986
Glb 13912 1767 12.7

65a 6290 945 15.0 5650
05a 5010 477 9.5

GSb 7229 377 5.2 5955
05b 4681 224 4.8

6721 13592 418 3.1 17478
G7b 21364 733 3.4

6% 5260 528 10.0 4866
6% 4472 404 9.0

Mean 7517 7517
SD 4755 4681
CV 63 62

 

201

Nominal Conc.
0.3 ug NP/L

Nominal Conc.
0.10 ug NP/L

Table B.4. (continued)

1.1). T (pg/m1) SD

CV a,b mean

 

 

Hlla 2642 312 11.8 2802
H1 1b 2961 230 7.8
H123 7224 116 1.6 6498
H12b 5772 853 14.8 Nominal Conc.
Hl3a 2196 142 6.5 2387 3.0 ug NP/L
H13b 2578 398 15.4
Hl6a 943 250 26.5 1711
H16b 2479 185 7.5
H18a 12860 167 1.3 14760
H18b 16661 1568 9.4

Hla 3281 280 8.5 3223
Hlb 3166 133 4.2

H2a 10999 212 1.9 11318
H2b 11638 555 4.8

H3a 5303 379 7.1 5672
H3b 6041 146 2.4

H8a 4727 388 8.2 4767
H8b 4807 162 3.4
Mean 5904 5904

SD 4355 4419

CV 74 75

 

202

Table B.S. Concentrations of vitellogenin (VTG) in plasma

from individual sexually mature male carp exposed to 4-nony1phenol (NP)
Mean across three replicate determinations is presented. Value reported is
for the dilution which yielded a %-bound closest to 50%.

The method detection limit (MDL) was 1 ug VTG/ml.

Values less than 1.0 ug VTG/m1 may not be accurate.

 

 

 

 

 

 

 

 

 

l.D. VTG (uglml) I.D. VTG (lug/ml) I.D. VTG (uglml)
A10 0.2 E10 0.7 H10 0.2

A11 0.7 E12 0.1 H12 0.8

A12 3.7 E13 0.7 H13 0.5

A13 1.6 E15 0.2 H14 0.4

A14 0.5 E17 0.9 H15 52

A15 0.5 E18 5.8 H16 0.2

A16 2.7 E19 1 H18 12.3

A18 0.9 E2 0.2 H2 0.4

A2 0.1 E20 10.8 H20 0.1

A4 0.5 E3 0.2 H3 0.3

A6 0.8 E5 0.1 H7 3.2

A9 0.7 E8 1.1 H8 0.3

C10 1.4 F11 1.4

C13 0.3 F12 >> 50

C14 2.2 F13 3.6

C16 0.3 F15 1.8

C17 0.6 F16 1.9 Nominal Conc.
C19 1.6 F18 0.2 A = 10 ug NP/L
CZ 0.1 F2 0.4 C = Solvent control
C20 0.7 F3 7.3 D = Control
C3 0.5 F4 3.6 E = 1.0 ug NP/L
08 0.3 F5 0.9 F = 0.3 ug NP/L
09 0.3 F6 0.5 G = 0.1 ug NP/L
D10 0.5 F8 0.3 H = 3.0 ug NP/L
D12 8.9 G1 0.5

D13 0.7 G11 2.2

D15 0.5 G12 1.7

D17 50 G13 4.5

D19 14.2 G15 50

DZ 1.6 G17 3.2

020 1.3 G18 0.1

D3 0.3 G2 1

D4 1.1 G4 0.1

05 0.2 G6 3.8

08 0.6 G8 0.3

09 0.8 G9 0.6

 

 

203

APPENDIX C

VITELLOGENIN ELISA PLASMA INTERFERENCES EXPERIMENT

204

Background

At high concentrations, plasma proteins have the potential to interfere with the
vitellogenin (VTG) ELISA. Plasma samples are routinely diluted to avoid such
interferences. When VTG concentrations are relatively high, such dilution does not
restrict accurate quantitation. When trying to detect relatively low concentrations of
VTG, however, however, it is desirable to dilute plasma samples as little as possible in
order to lower the method detection limit (MDL). The objective of this experiment was
to determine the minimum amount of plasma dilution which could be used while still

avoiding interference from other plasma constituents.

Experimental Design

Plasma samples (C9 and H8), which had been previously shown to have non-
detectable concentrations of VTG, were diluted in buffer containing a constant
concentration of VTG standard (268 ng VTG/m1). This was selected as a concentration
expected to yield approximately 50% binding in the VTG ELISA. Nine dilutions of each
sample were prepared using 2-fold serial dilution. This yielded final plasma dilutions of
1:125 (0.08), 1:25 (0.04), 1:50 (0.02), 1:100 (0.01), 1:200 (0.005), 1:400 (0.0025), 1:800
(0.00125), 1:1600 (0.000625), and 1:3200 (0.000313). Each dilution was analyzed in
triplicate using the VTG ELISA procedure described by Snyder (2000). Results
expressed in both %—bound and final calculated concentration of VTG were compared to

determine at which dilutions plasma interferences may occur.

205

Results

Fig. C.1. VTG ELISA results for plasma samples (C9 and H8) diluted in buffer
containing 268 ng VTG/ml. Mean of three replicate determinations i one standard
deviation is presented. Plots A and C depict responses expressed as %-bound for
dilutions of samples C9 and H8, respectively. Plots B and D depict responses expressed
as final calculated concentration of VTG for dilutions of samples C9 and H8,
respectively. Dilution refers to the ratio of whole plasma to the total volume of dilute

plasma sample (ex. 0.08 = 1 111 plasma/ 12.5 111 dilute sample).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A. C.
70 0911163” H8.mean
gag :1 § § 11 fi 1: :1 g 60 a 6 i :1 a u a
338 § 40
°\° 10 I; L I l l l I ale 20
° ' 1 ' o . 1 1 1 1 1 1 1
<8 b- q, x (o 95 (o '5
Q Q Q Q Q ‘1 ’1 (1r '\ ‘b b N ‘0 ‘3 ‘5 ‘0 ’5
_o o o- 0. o go go
9 9 .9 Q9
dilution dilution
B. D.
CQ.mean H8.mean
600 - A 600
€333 1 E 388 1
\ \ 1:
513188g 11111 3188” U ..
c”100 5,? 100
g o l I a 1 1 n L O ' § ‘ 1 : : = 4'
I I I U
6 1- '1. N 6 <0 ‘0 <6 "b 6‘5 6“ 6" 6" 6° '1," q." q," <5
09 0.0 09 09 QQQ 61' 6“} 061 69 Q Q- Q Q- 06 9% 00's Q60 065
‘ Q 00 Q0 Q Q Q Q
Q Q' Q' 0' 0‘
dilution dilution

 

 

 

 

 

Among the dilutions tested, only the 0.08 dilution (1: 12.5) appeared to give an ELISA
response which was markedly different from that of the other dilutions tested. Dilutions

greater than 1:20 appeared to be sufficient to prevent marked plasma interferences.

206

 

 

Reference:

Snyder, E.M., 2000. Use of fish as bioassay organisms to assess wastewater effluents for

reproductive endocrine modulating chemicals. Ph.D. Dissertation. Michigan State

University, East Lansing, MI, USA (in preparation).

207

 

 

111111119131111111121111111111