.4

Urdu. ..(
..V .

.J

.t.)

 

 

 

 

 

 

 

xv

fidnuwflram{m=1mnwvmhu
. u. 1.“. L?!)
7 nun .

59...?
.
, ,umw

.
I .
\‘v. n...
ma.

4»,

, VHF-I . .
"Emfiuéé
vflufiflfir.
. 2..
3 w... ‘
spa l‘ «:11
Jflmmmfin

. “w”:

flat

:3

3%:

fig? .2.
.1152594 2.5;.
Ifar.

‘5be

. Y Wu!
+5

a

s! u
“If...

Mm”

.1
I.

fix
fimw

f
‘ ,1. 3.
. n. an;

5:

 

| J
r ‘3; $3
, '3

tr.

 

LIBRARY
Michigan State

UDIVG rsuty This is to certify that the
dissertation entitled

 

 

 

MEMBRANE ASSOCIATED EFFECTS AND ALTERATIONS
IN GENE EXPRESSION RESULTING FROM
PERFLUOROOCTANE SULFONIC ACID EXPOSURE

presented by

WENYUE HU

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in ZOOLOGY / IET

 

 

Major Professor nature
/7/1§103C7

Date

MSU is an Affirmative Action/Equal Opportunity Institution

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p. 15

MEMBRANE ASSOCIATED EFFECTS AND
ALTERATIONS IN GENE EXPRESSION RESULTING FROM
PERFLUOROOCTAN E SULFONIC ACID EXPOSURE

By

Wenyue Hu

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology
Institute of Environmental Toxicology

2003

ABSTRACT
MEMBRANE ASSOCIATED EFFECTS AND
ALTERATIONS IN GENE EXPRESSION RESULTING FROM
PERFLUOROOCTANE SULFONIC ACID EXPOSURE
By

Wenyue Hu

The recent detection of perfluorinated fatty acids (PFFAS) in wildlife from even remote
locations has spurred interest in the environmental occurrence and effects of these
chemicals. Among them, perfluorooctane sulfonic acid (PFOS) is the end metabolite of a
number of perfluorinated fatty acid analogues extensively used in industrial materials and
commercial applications. Few studies have been conducted on this novel compound, and
its mechanism of action still remains unclear. The amphiphillic nature of PFOS suggests
its cell membrane related effects. In the current study, effects of PFOS on membrane
fluidity and mitochondrial membrane potential were examined using flow cytometry and
effects on membrane permeability were tested using cell bioassay procedures (H4IIE,
MCF-7, PLHC-l). PFOS increased plasma membrane fluidity and decreased
mitochondrial membrane potential in fish leukocytes in a dose-dependent fashion. The
lowest effective concentrations for both membrane fluidity and mitochondrial membrane
potential effects of PFOS were 5 to 15 mg/L. This suggests that membrane properties
could be used as sensitive biomarkers for PFOS related adverse effects. Besides
membrane related effects, studies were also designed to examine modulations of gene
expression by PFOS exposure using states of art molecular toxicology techniques. In the

current study, the restriction fragment differential display (RFDD-PCR), a mRNA

fingerprinting technique, and Affymetrix rat genome U34A chips, the high throughput
genomic technique with 8790 genes, were used to identify alterations in gene expression
level due to PFOS exposure in vitro and in viva. RNA samples were extracted from
H4IIE rat hepatoma cells and rat livers exposed to PFOS, and prepared for subsequent
analysis. Following the RFDD-PCR procedure, 55 bands on sequencing gel were
identified as different across treatment groups. All these candidate genes were sub-cloned
and sequenced. Gene chip analysis was conducted by hybridizing U34A chips with
experimental samples. Approximately 5% of the genes on the chip were identified as
differentially expressed in response to PFOS exposure, and clustered as genes coding for
fatty acid metabolizing enzymes, drug and xenobiotic metabolizing enzymes, and
proteins involved in signal transduction pathways and hormone regulation. Consistent
results were obtained from replicate exposures, however, expression profiles of samples
prepared in vitro and in vivo showed only limited similarity. The major pathway affected
by PFOS is postulated to be peroxisomal fatty acid beta-oxidation, which could be
explained by the structural similarity between PFOS and endogenous fatty acids.
Comparisons were made between differential display and gene chip techniques based on
their specificity, sensitivity, and scope of applications. The mechanistic interpretation
derived from these two methods was in agreement, although the results were not directly

comparative.

To my dear husband Michael Gu

and my parents, Hua Hu and XiaoRu Liu

iv

ACKNOWLEDGEMENT

Funding for this project was provided by the 3M Company at St. Paul, MN.

I sincerely thank my academic supervisors Dr. John Giesy and Dr. Paul Jones for their
continuous support and tremendous help. Only with their never ceased encouragement
could I make it through all the difficulties during the process of doing this project and
writing the dissertation. I am honored to have Dr. Don Hall, Dr. Will Kopachik, and Dr.
Robert Roth on my graduate committee, and to receive invaluable advices from them.
Special thanks are owed to Dr. Wim DeCoen and Dr. Trina Celius for their help on the
differential display and gene chip experiments. The assistance from staffs at the
Michigan State University Macromolecular Structure Facility is gratefully acknowledged.
The assistance of Dr. Chris Lau and associates from the Reproductive Toxicology
Laboratory at US-EPA, Research Triangle Park, NC, in carrying out the in viva exposures
was invaluable to this project. I am also thankful to Dr. Louis King from the
Biochemistry department at Michigan State University for kindly providing the flow
cytometry instrument and helping out with sample measurement. I greatly appreciate the
friendship and supports from all my friends at Michigan State University, especially my

colleagues at the Aquatic Toxicology Laboratory.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... viii

LIST OF FIGURES ............................................................................... ix

LIST OF SYMBOLS OR ABBREVIATIONS .............................................. xiv

INTRODUCTION ................................................................................... 1
Acute Toxicity .............................................................................. 2
Subchronic and Chronic Toxicity ......................................................... 3
Tissue Distribution, Metabolism and Excretion ........................................ 3
Induction of Fatty Acid and Lipid Metabolizing Enzyme Activities ............... 5
Effect on Hepatic Microsomal Cytochrome P450 Enzyme Activity ............... 7
Effect on Leydi g Cell Function ............................................................ 7
Non-genotoxic Tumor Promoter ......................................................... 8
Previous Findings ........................................................................... 8
Study Objectives and Rationale ........................................................... 10

CHAPTER 1.

MEMBRANE EFFECTS OF PFOS ............................................................. 14
Abstract ..................................................................................... 15
Introduction ................................................................................ 16
Material and Methods ..................................................................... 17

Chemicals ......................................................................... 17
Cell Biaassays ................................................................... 17
F law Cytametry Membrane F luidity Assay .................................. 20
Results ....................................................................................... 24
PF 0S Effects on Membrane F luidity and Membrane Potential ............ 24
Eflects of PF OS on EROD and Luciferase Activities ........................ 25
Discussion .................................................................................. 27

CHAPTER 2.

GENE PROFILING IN RESPONSE TO PFOS ............................................... 41
Abstract .................................................................................... 42
Introduction ................................................................................ 43
Material and Methods ..................................................................... 45

Chemicals ........................................................................ 45
Cell Culture and Treatment ................................................... 45
In viva Treatment ................................................................ 46
Gene Chip Array Experimental Procedure .................................. 46
Gene Chip Data Analysis ........................................................ 48
Results ....................................................................................... 50
Overall Changes ................................................................. 50

vi

Scatter Plats ....................................................................... 52

Gene Tree .......................................................................... 53
Gene List .......................................................................... 53
Discussions .................................................................................. 54

CHAPTER 3.

COMPARISON OF GENE EXPRESSION TECHNIQUES ................................. 72
Abstract .................................................................................... .73
Introduction ................................................................................ 74
Material and Methods ..................................................................... 77

Chemicals ......................................................................... 77
Cell Culture and Treatment ................................................... 77
RNA Extraction and Purification .............................................. 78
Differential Display Procedure ............................................... 78
Subclaning of PA GE bands .................................................... 80
Gene Array Analysis ............................................................ 80
Results ....................................................................................... 81
Identifi/ and Subclone Differentially Expressed Genes ...................... 81
Comparison with Gene Chip Results .......................................... 83
Discussions ................................................................................. 84

CONCLUSION ..................................................................................... 94

APPENDICES ..................................................................................... 98
Appendix A ................................................................................. 99
Appendix B ............................................................................... 101
Appendix C .............................................................................. 112

REFERENCES ................................................................................... 120

vii

LIST OF TABLES

Table 2.1. List of genes induced significantly (p<0.0025) by PFOS in viva exposure.
Long-term values are the mean fold-change and standard deviation (s.d.) for the four

possible control/exposed comparisons ............................................................ 62

Table 2.2. List of genes suppressed significantly (p<0.0025) by PFOS in viva exposure.
Long-term values are the mean fold-change and standard deviation (s.d.) for the four

possible control/exposed comparisons ........................................................... 63

Table 3.1. Genes identified to be differentially expressed due to PFOS exposure in
H4IIE cells in vitra. The “Clone seqs” indicates the number of clones sequenced from
each original gel band. The “majority sequence”, if present, is the sequence that was
present in 50% or more of the sequenced clones. The “proportion majority” is the

number of clones out of the total that were the majority sequence ........................... 91

Table 3.2. Alterations in gene expression (as fold change from control) assessed by

differential display and Genechip analysis. “dd” and “gc” refer to differential display or

GeneChip analysis at 2 or 50 mg/L as indicated ................................................. 92

Table C.l. Comparison of differential display and gene chip methods .................. 97

viii

LIST OF FIGURES

Figure 0.1. Molecular structure of perfluorooctane sulfonic acid (PFOS), a synthetic,

fully fluorinated, eight-carbon chain fatty acid analogue ...................................... 13

Figure 1.1. Density Plot of Fluorescence Emission of Pyrene Labeled Leukocytes that
were treated with Pentanol and PFOS. Normal membrane fluidity is represented by
‘cells+pyrene’. The potent membrane affecting reagent pentanol increases cellular
fluorescence at 450nm as it increases membrane fluidity. PFOS at 33 and 100 uM

(equivalent to 16.5 and 50 mg/L) also increase membrane fluidity .......................... 34

Figure 1.2. Effects of 1% pentanol (positive control) and PFOS on fish leukocyte
membrane fluidity. Cells were labeled with pyrene decanoic acid and then exposed to
test chemicals as described in the text. Cells were analyzed by flow cytometry as
described in text. Error bars represent standard deviation of 3 determinations. Results
were analyzed by ANOVA with Dunnett's test using DMSO exposed cells as control; *

p<0.05; ** p< 0.01; *** P< 0.001 ................................................................. 35

Figure 1.3. Effects of PFOS on mitochondrial membrane potential. Cells were labeled
with the mitochondrial membrane potential dye J C-1 and were exposed to test chemicals.
Cells were analyzed by flow cytometry as described in the text. Error bars represent
standard deviation for three determinations. Valinomycin (100 nM), a reagent known to

decrease mitochondrial membrane potential, was used as a positive control. Values were

ix

compared to the WBC control by ANOVA followed by Dunnett's test * p<O.l, **

p<0.05 .............................................................................................. 36

Figure 1.4. Direct effects of PFOS (squares) on H4IIE-luc cell and PLH C-l cell EROD
activity, and on H4IIE-luc cell luciferase activity compared with the effects of TCDD
(diamonds). A) EROD activity of H4IIE-Inc cells dosed with PFOS or TCDD; B)
EROD activity of PLH 01 cells dosed with PFOS or TCDD; C) Luciferase activity of
H4IIE-luc cells dosed with PFOS or TCDD. EROD activity was expressed as % of
control, and luciferase activity was expressed as relative luminescence units (RLU).
Control cells were dosed with 0.1% (v/v) solvent (methanol) only. Error bars represent

standard deviation of three determinations ...................................................... 37

Figure 1.5. (A) Interactive effects of PFOS and TCDD on H4IIE-luc cell EROD
activity. H4IIE-luc cells were exposed to TCDD alone (squares) or to TCDD in the
presence of PFOS at concentrations of 0.1 (triangles) or 10 (diamonds) mg/L. EROD
activity is expressed as activity relative to control (no TCDD or PFOS exposure). Error
bars represent standard deviation of three measurements. (B). Summary of interactive
effects of PFOS and TCDD on H4IIE-luc EROD activity. Significance levels are: *

p<0.05 ** p<0.01 .................................................................................... 38

Figure 1.6. Interactive effects of PFOS and TCDD on PLHC-l cell EROD activity.

Cells were exposed to 0.2 ug/L TCDD and various concentrations of PFOS, general

linear model pair-wise comparisons were conducted (* p<0.05; ** p<0.01). Control was

cells exposed to 0.2 ug/L TCDD alone .......................................................... 39

Figure 1.7. Interactive effects of E2 and PFOS or E2 and TCDD on MVLN cell
luciferase activity. Cells were exposed to E2 alone; E2 in the presence of 0.1 mg/L
PFOS; or E2 in the presence of 0.5 ng TCDD /ml as marked. Each point represents the

mean of three measurements; error bars are one standard deviation ........................ 40

Figure 2.1. Scatter plots of gene expression comparisons due to exposure to PFOS.
Each point represents a single gene or EST; the middle line represents a 1:1 regression
line and the outerlines represent the 95% confidence interval. 1A) Comparison of gene
expression in two in viva control rats (Controll, Control2). 13) Comparison of gene
expression in two individual rats exposed in viva to PFOS for 21 d (PFOSl, PFOSZ).
1C) Comparison gene expression in a rat exposed in viva to PFOS for 21 d (PFOSl) to a
control rat (Controll). 1D) Comparison of samples from in vitra and in viva control

exposures (In-vitro-Control, In-vivo-Control respectively) ............................... 64-65

Figure 2.2. Gene tree dendrogram comparison of all samples from rats and cell lines
exposed to PFOS in viva and in vitro. Samples with a similar gene expression patterns
were clustered under the same node. Genes with similar expression patterns across
samples were clustered together. The Dendrogram links samples based on the gene

expression pattern .................................................................................. 66

xi

Figure 2.3. Gene tree dendrogram comparison of samples from rats exposed to PFOS in
viva. Genes with similar expression patterns across samples were clustered together.

Samples with a similar banding pattern show similar gene expression pattern ............ 67

Figure 2.4. Identification and fold change of genes whose expression was significantly
(p<0.0025) increased by PFOS exposure in viva. For long term exposure values are
means for the 4 possible control/exposed pairings, error bars are one standard deviation.

For the short term exposure one control/exposed pair was analyzed ........................ 68

Figure 2.5. Identification and fold change of genes whose expression was significantly
(p<0.0025) decreased by PFOS exposure in viva. For long term exposure values are
means for the 4 possible control/exposed pairings, error bars are one standard deviation.

For the short term exposure one control/exposed pair was analyzed ........................ 69

Figure 2.6. Diagram of pathways for mitochondrial and peroxisomal fatty acid [3-
oxidation and the relative fold of induction of the enzymes caused by long term exposure

to PFOS in viva ...................................................................................... 70

Figure 2.7. Comparison of effects of PFOS, phenobarbital and Wyeth 14,643 on gene
expression at different time periods. Data for phenobarbital and Wyeth 14,643 are from
(Hamedah et al., 2002a, 2002b). Abbreviations are: CYP2Bl, cytochrome P450 2B1;
CPT = Carnitine palmitoyl transferase 1; ALD RED = Aflatoxin aldehyde reductase;

p55cdc = P55cdc; CYT PHOS = cytocolic phosphoprotein (p19); E-COA-ISO = enoyl

xii

CoA isomerase; thiolase = thiolase; AhR = Ah Receptor; Ces = carboxylesterase

precursor ............................................................................................. 71

Figure 3.1. Example of an RT-PCR differential display gel showing both increased and
decreased gene expression. Diagrams to the right indicated interpretation of the observed
bands. Lanes are BC, Blank Control - no exposure; SC, Solvent Control; P2, cells

exposed to 2 mg/L PFOS; P50, cells exposed to 50 mg/L PFOS ............................. 93

xiii

INTRODUCTION

Perfluorinated fatty acids (PFFAs) (CF 3 (CF2)x COOH) are fatty acid analogues in which
the carbon backbones are fully fluorinated. The high—energy of the carbon-fluorine bond
renders these compounds resistant to hydrolysis, photolysis, microbial degradation, and
metabolism, making them environmentally persistent (Giesy and Kannan, 2002).
Perfluorinated compounds have been manufactured for over 50 years and are commonly
used in industrial materials such as wetting agents, lubricants, corrosion inhibitors, stain
resistant for leather, paper and clothing, as well as in foam fire extinguishers (Shinoda
and Nomura, 1980; Sohlenius et al., 1994; Giesy and Kannan, 2002). PFFAs also possess
unique biological characteristics that make them suitable for red blood cell substitutes
and hepatic drugs (Ravis et al., 1991). The global environmental distribution,
bioaccumulation, and biomagnification of several perfluorinated compounds have
recently been studied (Hansen et al., 2001; Giesy and Kannan, 2001, 2002; Kannan et al.,
2001a, b). These studies indicate that perfluorooctane sulfonic acid (PF OS) (Figure l) is
the most persistent and widely dispersed in the environment. PFOS has been identified at
low concentration in human, wildlife, and environmental media samples. Perfluorooctane
sulfonamide (PFOSA), perfluorooctanoic acid (PF 0A), and perfluorohexane sulfonate
(PFHS) have also been detected in the tissues of several species (Giesy and Kannan,
2001). Because of their growing list of applications and increasing potential for exposure
to humans and wildlife, more mechanistic toxicological studies are now needed for

assessing the potential toxicity of PFFAs at environmentally relevant concentrations.

Since PFFAS are chemically stabilized by strong covalent bonds between carbon and
fluorine, historically they had been considered to be metabolically inert and non-toxic
(Sargent et al., 1970). However, it has recently been found that some PFFAS are
biologically active and can cause peroxisome proliferation, increased lipid metabolizing
enzyme activity, altered xenobiotic metabolizing enzyme activity, and modulations in
other important biochemical processes in exposed organisms (Obourn et al., 1997;
Kawashima et al., 1995; Sohlenius et al., 1994; Seacat et al., 2003). Several PFFAs have
also been associated with the production of liver tumors in rodents and are classified as
nongenotoxic hepatocarcinogens (Youssef and Badr, 1998). PFOS has been shown to
accumulate primarily in blood and liver (Giesy and Kannan, 2001) and the major target
organ for PFOS is therefore believed to be the liver. However, this does not exclude

other possible target organs such as the pancreas, testis and kidney.

Acute Toxicity

The most thoroughly studied compounds in the PFFAs family are perfluorooctanoic acid
(PFOA) and perfluorodecanoic acid (PFDA). The acute toxicities of these two
compounds were evaluated in male Fisher rats, and the LD50/30 days for PFOA and
PFDA were determined to be 189 mg/kg body weight and 41 mg/kg body weight,
respectively (Olson and Andersen, 1983). Rats treated with a lethal dose of PFOA
exhibited incipient death within the first five days; however, those exposed to PFDA
showed a delayed lethality after two weeks. This difference is probably due to different
rates of accumulation and elimination of these two compounds in male rats (Olson and

Andersen, 1983).

Subchranic and Chronic Toxicity

Sub-chronic toxicity study of PF OS in rats for 90 days and chronic toxicity study in rats
for 2 years both demonstrated decreases in body weight, hepatocellular hypertrophy,
elevation in blood glucose and serum alkaline phosphatase level, hepatic vacuolation and
death at the high dose (Goldenthal et al., 1978; Seacat et al., 2002b). Similar responses
were observed in primates, when cynomolgus monkey were treated repeatedly with
PFOS up to 0.75 mg/kg/day for 16 weeks (Seacat et al., 2002a). Major effects occurring
at the high dose were a marked reduction in serum cholesterol concentrations, decreased
body weight, lipid vacuolation and hepatocellular hypertrophy. The reproductive and
developmental toxicity of PFOS was tested in pregnant female rats at 1, 2, 3, 5, and 10
mg/kg/day from gestation day 2 to day 21 (Lau et al., 2003; Thibodeaux et al., 2003).
PFOS treatment resulted in suppressed maternal weight, decreased serum thyroxin and
triglycerides level. In the rat fetuses, PFOS was detected in the liver at nearly half of the
maternal concentration. PF OS severely compromised postnatal survival of neonatal rats,

and caused delay in the grth and development in the surviving rat pups.

Tissue Distribution, Metabolism and Excretion

When Wistar rats were treated with a single intraperitoneal dose (20 mg/kg body weight)
of PFDA, approximately 15% of the administered PFDA were found in the serum, with
more than 99% bound to serum proteins. In the liver, 5% of PFOA was found to be either
in the free anionic form or bound to the lipid fraction (Y linen and Aurivla, 1990). PFOS
was found to accumulate in the liver and blood of exposed organisms (Giesy and Kannan,

2001). The binding of PFOS to serum proteins was investigated by assessing its ability to

displace a variety of steroid hormones from specific binding proteins in the serum of
birds and fishes (Jones et al., 2002). PFOS had only a weak ability to displace estrogen or
testosterone from carp serum steroid binding proteins. Displacement of cortisone in avian
serum occurred at relatively low PFOS concentrations. Corticosterone displacement
potency increased with PFFA chain length, and sulfonic acids were more potent than
carboxylic acids (Jones et al., 2002). Direct protein binding study was also conducted,
where PFDA and PFOA were found to ‘covalently’ bind proteins when administered to
rats in viva (V anden Heuvel eta1., 1992b). In cytosolic and microsomal incubation, there
was no effect of the addition of CoA, ATP or NADPH on the magnitude of the covalent
binding. Therefore it was not necessary for PFDA and PFOA to be metabolically
activated to form the covalent adduct. In fact, most of the PFFAs administered via the
diet were unaffected by metabolic enzymes. Elimination of PFFAs was primarily through
urinary excretion, with little biliary or fecal excretion, and the rate of elimination was
determined by the carbon chain length of the PFFA molecules (Kudo et al., 2001). In
male rats, PFHA was rapidly eliminated in urine by 92% within 120 hr of an
intraperitoneal injection, whereas PFOA and PFDA were eliminated in urine by 55% and
2%, respectively, over the same time period. There was also a marked sex difference in
the whole-body elimination rate of PFOA in rats, with female excreting PFOA more
rapidly than males (Ylinen et al., 1989; Hanhijarvi et al., 1982; Vanden Heuvel et al.,
1992a). The renal elimination rate of PFOA in female Wistar rats was ten-fold greater
than in male rats. Castration of male rats greatly increased the elimination rate of PFOA.

Castration plus testosterone treatment reduced the rate of elimination to the original level.

It was therefore suggested that testosterone exerted an inhibitory effect controlling the

renal excretion rate of PFOA (Vanden Heuvel et al., 1992a).

Induction of F any Acid and Lipid Metabolizing Enzyme Activities

Although the mechanisms by which PFFAs elicit their toxic effects are not well
understood, the one consistent observation is that they act as peroxisome proliferators
(Just et al., 1989; Sohlenius and Reinfeldt, 1996; Wallace et al., 2001). Peroxisome
proliferators include a number of structurally diverse compounds. Regardless of their
dissimilarities in structure, these compounds all have one thing in common: they all
induce the proliferation of peroxisomes (a membrane-bound organelle that both generates
and breaks down hydrogen peroxide), which results in an increase in both the number of
peroxisomes and their corresponding enzyme activities (Ikeda et al., 1985). PFFAs can
interfere with lipid metabolism by increasing peroxisomal fatty acid B-oxidation, and
induce several hepatic enzyme activities (Sohlenius and Reinfeldt, 1996). Both in viva
and in vitra exposures to PFFAs result in increased activities of peroxisomal Acyl-CoA
oxidase, which is known to catalyze the first and rate-limiting step in fatty acid oxidation
(Sohlenius et al., 1994). The potency of the induction of peroxisomal B-oxidation was
compared among PFFAs with different carbon chain length in the liver of male and
female rats (Kudo et al., 2000). Perfluorohexanoic acid (PFHA) had little effect, while
PFOA and PFDA caused substantial induction of peroxisomal B-oxidation. This
differential induction potency was strongly correlated with the actual dose of PFFAs in
the liver regardless of the chemical structure. Treatment with PFFAS exerted a coordinate

induction of acyl-CoA binding protein, fatty acid binding protein and peroxisomal B-

oxidation (Vanden Heuvel et al., 1993). Fatty acid oxidation is also a process that can
produce hydrogen peroxide, an oxidative radical, which can cause oxidative stress and
may result in DNA damage (Sohlenius et al., 1994). Administration of PF OA and PFDA
in male rats significantly increased the amount of 8-hydroxydeoxyguanosine in liver
DNA, but not in kidney DNA. Thus, PFOA and PFDA induced peroxisome proliferation
was proven to be associated with organ specific oxidative DNA damage (Takagi et al.,
1991). The peroxisome proliferator activated receptor (PPAR), a member of steroid
hormone receptor family, can be activated by peroxisome proliferators and then binds to
the peroxisome proliferator responsive elements (PPRE). Previous studies identified
several PPREs located upstream from a battery of structural genes, including acyl-CoA
oxidase, peroxisomal camitine octanoyltransferase, and lipoprotein lipase (Sohlenius and
Reinfeldt, 1996; Braissant et al., 1996). A good correlation had been observed between

PPAR activation and peroxisome proliferation potency (Green, 1992).

PF FAs have been shown to regulate tissue fatty acid composition and content. PF FAs can
reduce cholesterol and triacylglycerol concentrations in serum, increase liver
triacylglycerol concentration, and reduce hepatic lipid output (Haughom and Spydevold,
1992). Chronic exposure of primates to PF OS has also been demonstrated to significantly
alter blood lipid concentrations (Seacat et al., 2002a). It has also been found that
treatment with PFFAs can inhibit Acyl-CoA synthetase activity and result in an increase
in the level of free fatty acids (Reo etal., 1996). Free fatty acids are known to be able to
activate protein kinase C (PKC), which leads to a signaling cascade that is important for

normal cell function, cell proliferation and gene expression.

Effect on Hepatic Microsomal Cytochrome P45 0 Enzyme Activity

Cytochrome P450 enzymes (CYPs) are a group of primary oxidative enzymes involved in
phase I metabolism, 3 process that detoxifies xenobiotics by making them more polar so
that they can be conjugated and excreted more easily. Microsomal cytochrome P450
enzymes were induced in rats treated with PFFAS (Permadi et al., 1992). This induction
was sex-related and organ-specific, based on the fact that male rats were more sensitive
than female rats, and liver was the major target organ compared to the kidney. For
example, administration of PFOA to male rats induced CYP4A1 enzyme activity by 6.8
fold in liver and 2.] fold in kidney (Diaz et al., 1994). The CYP4A sub-family is a group
of nine enzymes that are specific for fatty acid w-hydroxylation. Which of the
isoenzymes may be induced depends on the testing species, the administration pathways

and the duration of exposures.

Effect on Leydig Cell Function

The Leydig cells or interstitial cells are a group of cells located outside the seminiferous
tubules in mammalian testes. Leydig cell is the primary site of testosterone synthesis,
which regulates spermatogenesis, the growth and secretory activity of accessory sex
organs, male behavior, and various metabolic processes (Boujrad et al., 2000). PFFAs
were founds to affect Leydig cell function and produce Leydig cell adenomas (Liu, Hahn,
and Hurtt, 1996). So far most information available was for the effects of ammonium
perfluorooctanoate (C8) on Leydig cells of adult male rats. Three levels of effects have
been observed: 1) overall depression of Leydig cell function in vitro (Cook et al., 1992);

2) decreased testosterone release and increased serum estradiol concentration in viva

(Biegel et al., 1995); 3) elevation of aromatase (CYP 19) activity by 16 fold in viva (Liu

et al., 1996 a, b).

Non-genotoxic Tum or Promoter

Treatment with PFF As has been associated with the induction of hepatic necrosis,
hepatocyte carcinomas, Leydig cell adenomas, and pancreatic tumors (Obourn et al.,
1997). It has been postulated that the increase in oxidative stress and alteration in protein
kinase C levels are responsible for the possible carcinogenic activities of PFFAs (Reo et
al., 1996). Recently the alternative hypothesis has been suggested that these effects may
be non-genotoxic and caused by the disruption of hormone regulation (Cook et al., 1992),

or blocking of intercellular communication (Upham et al., 1998; Hu et al., 2002b).

Previous Findings

Several aspects of the biochemical toxicity of perfluorooctane sulfonic acid (PFOS) were
investigated using in vitra cell culture systems (Hu, 2000). The effects of PFOS on aryl
hydrocarbon receptor (AhR) mediated cytochrome P4501A1 (CYP1A1) activity were
tested using in vitro cell bioassays. PFOS had no adverse effects on cell viability nor did
it directly effect CYP1A1 activity. However when cells were dosed with PFOS and
TCDD in combination, interactive effects on both CYP1A1 induction and AhR activation
were observed at environmentally relevant concentrations. PFOS increased the effects of
TCDD by 30-40% (Hu et al., 2002a). It was hypothesized that these effects were due to

alterations ion membrane fluidity and permeability.

Effects of PFOS and related sulfonated fluorochemicals, on gap junctional intercellular
communication (GJIC) were studied using a rat liver epithelial cell line (WB-F344) and a
dolphin kidney epithelial cell line (CDK). In viva effects of PFOS on GJIC were studied
in Sprague-Dawley rats orally exposed to PFOS for 3 days or 3 weeks (Hu et al., 2002b).
PFOS, PFOSA and PFHS were found to inhibit GJIC in a dose-dependent fashion, and
this inhibition occurred rapidly and reversibly. Perfluorobutane sulfonate (PF BS) showed
no significant effects on GJIC in vitro within the concentration range tested. A structure
activity relationship was established among the four tested compounds, indicating that the
inhibitory effect was determined by the length of the fluorinated tail and not by the nature
of the functional group. The results from the two cell lines and the in viva exposure were
comparable suggesting that the inhibitory effects of the selected perfluorinated
compounds on GJIC were neither species- nor tissue—specific, and can occur both in vitro

and in viva.

Aromatase (CYP19) is the enzyme that catalyzes the last step in the estradiol biosynthesis
pathway, which converts testosterone to estradiol (Simpson et al., 1994). Aromatase
activity was determined in a human adenocarcinoma cell line (NCI-H295R) treated with
PFOS. Results from this study indicated that PFOS at a concentration of 50 mg/L induced
aromatase activity by 1.5 fold at 24 hrs, and by 1.7 fold at 48 hrs exposure. This is a
relatively modest induction compared to ammonium perfluorooctanoate, which induced
aromatase activity to a much greater degree (liu et al., 1996). The specific mechanism by
which PFOS may elicit its effect on aromatase activity is still under investigation. It has

been found that treatment with PFOA and PFDA can affect the level of protein kinase C,

which mediates an important signaling pathway that may modulate steroidogenesis (Reo

et al., 1996).

Study Objectives and Rational
To continue with the investigation on the biochemical effects of PFOS, studies were
designed to examine the membrane-associated effects of PFOS and the modulations of

gene expression by PF OS exposure at whole genome level.

Previous studies had suggested effects of PFOS on plasma membrane fluidity and
permeability (Hu et al 2002; Hu et a1 2003). The molecular structure (Figure 0.1) and the
physical-chemical properties of PFOS also suggest possible membrane related effects.
Therefore, potential effects of PFOS on fish leukocytes cell membrane fluidity and
mitochondria membrane potential were investigated using fluorescent probe labeling and
flow cytometry measurement. Membrane fluidity is a measurement of the relative
mobility of the phospholipid bilayer in cell plasma membrane. Mitochondria membrane
potential provides the driven power for mitochondria energy chain production. Three
perfluorinated compounds with similar structure but different chain length were tested,
including PFOS, PFHS, and PFBS, in hope to establish a structure-activity relationship

between testing compounds and membrane associated effects.

Besides membrane related effects, alteration at gene expression level represents another

important aspect of cellular regulatory mechanism in response to xenobiotic exposure.

Eukaryotic organisms contain approximately 100,000 different genes, of which only a

10

small subset, around 15%, are expressed in any particular cell type. It is the profile of
genes expressed and the level of expression that determine all cellular process, including
differentiation and proliferation, maintenance of homeostasis, response to insults,
regulation of cell cycle, aging and even programmed cell death (Maniatis and Reed,
2002). Furthermore, alteration of gene expression lies at the heart of regulatory
mechanisms, which result in pathological changes such as in cancer or exposure to
xenobiotics and environmental toxicants. Therefore a thorough examination and
comparison of gene expression profiles in control and treated animals can provide critical

information regarding the molecular mechanism of the exposure.

There is a wide choice of methods available for comparing RNA samples and identifying
differentially expressed transcripts (Lockhart and Winzeler, 2000). At the single gene
level, reverse transcriptase polymerase chain reaction and Northern blotting are the
classic methods for measuring relative transcript abundance. At the whole genome level,
there are a variety of technical approaches used for differential screening of RNA
transcripts. For example, subtractive cloning involves building a cDNA library, and
subtracting the number of clones of one sample from the other. Since the early 1990’s, a
variety of protocols have been successfully developed and widely applied to measure
relative mRNA abundance, such as nuclease protection, cDNA sequencing, subtractive
hybridization, and serial analysis of gene expression (SAGE) (Lockhart and Winzeler,
2000). In 1992, an alternative approach to subtractive hybridization was proposed which
was generally known as mRNA finger printing or differential display (Matz and

Lukyanov, 1998; Liang and Pardee, 1992). Differential display involves randomly

11

generating cDNA fragments from mRNA samples and resolving them side-by- side on
polyacrylamide gel. The obvious advantages of differential display over subtractive
cloning are that a large number of transcripts can be examined simultaneously and two or
more RNA samples can be compared at once. In the following years, differential display
was used extensively and resulted in hundreds of publications reporting its successful
application (Matz and Lukyanov, 1998). At the end of the last century, driven by the
massive grth of genome sequence information and the rapid development of
computational biology, the revolutionary DNA microarray or oligonucleotide genechip
technology was developed (Lockhart and Winzeler, 2000). Nucleic acid arrays work by
hybridization of labeled RNA derived from testing samples to the DNA molecules
attached to a specific surface. One of the major advantages of array technology is the
extremely high density of genes that can be contained on a chip and thus a great amount
of information can be obtained from one such experiment. Based on this aspect, array
technology is more powerful and high-throughput than any of the conventional method

for evaluating mRN A abundance.

In the current study the effects of PFOS on gene expression were determined using two
states of art molecular toxicology methods: differential display and high—density
oligonucleotide genechip arrays. The purpose for using these genome wide screening
approaches was to identify novel mechanisms of PFOS induced toxicity, and to establish
mRNA expression profiles and gene markers specific to the exposure of PFOS. We also
aimed to compare two of the most popular gene expression analysis methods: differential

display and gene chips. The experiments were designed with both in vitro and in vivo

12

exposure systems to allow comparison of gene expression profiles between in vitro and in

viva models, and long— term verses short-term exposure.

Fig. 0.1. Molecular structure of perfluorooctane sulfonic acid (PFOS), a synthetic, fully

fluorinated, eight-carbon chain fatty acid analogue.

FCF\/ FCFV FF\/ \C/
F\ \ CC\/c\ C\:O//
F/R/F:/C<H:CF/\ F/C<F OsO//\

13

CHAPTER ONE

Hu, W.Y., Jones, P.D., DeCoen, D., King, L.E., Fraker, P., Newstead, J .L., and Giesy,
J .P. (2002a) Alterations in Cell Membrane Properties Caused by Perfluorinated
Compounds. Comp. Biochem. Physiol. 135: 77-88.

14

ABSTRACT

The recent detection of perfluorinated fatty acids (PFFAS) in wildlife from even remote
locations has spurred interest in the environmental occurrence and effects of these
chemicals. While the global distribution of PFFAs is increasingly understood, there is
still little information available on their effects on wildlife. The amphiphillic nature of
PFFAS suggests that their effects could be primarily on cell membranes. In this study we
measured the effects of PFFAS on membrane fluidity and mitochondrial membrane
potential using flow cytometry analysis in fish leukocyte, and effects on membrane
permeability using cell bioassay procedures in H4IIE, MCF-7 and PLHC-l cell lines.
Three PFFAs were tested in the membrane fluidity assay: perfluorooctane sulfonic acid
(PFOS), perfluorohexane sulfonic acid (PFHS), and perfluorobutane sulfonic acid
(PFBS). Of the PFFAS tested, only PFOS increased the permeability of cell membranes
to the hydrophobic ligands used. PFOS increased membrane fluidity in fish leukocytes in
a dose-dependent fashion, while PFHS and PFBS had no effect in the concentration range
tested. Threshold concentrations for the membrane fluidity effects of PFOS were 5 to 15
mg/L. Effects on mitochondrial membrane potential occurred in the same concentration
range as effects on membrane fluidity. This suggests that effects of PF OS on membrane

properties occurred at concentrations below those associated with other adverse effects.

15

INTRODUCTION

Previous studies of the effects of perfluorinated compounds on gap junctional
intercellular communication (GJIC) suggested that at least some of the observed effects
might be due to alterations in membrane fluidity (Hu et al., 2002b). Membrane fluidity is
a measurement of the relative mobility of the phospholipid bilayer of the cell membrane.
The fluidity of membranes allows movement of molecules within the plane of the
membrane, providing the basis for lipid-lipid, lipid-protein, and protein-protein
interactions. PFOS has also been observed to moderately affect the potency of ligands
such as dioxin and estradiol used in in vitra cell culture bioassays (Hu, 2000). These
observations suggested possible effects of PFOS on membrane permeability. The
selectively permeable cell membrane forms the first barrier that separates the cell from
exogenous exposures. Effects on the permeability status of the cell membrane could play
an important role in mediating the adverse effects of a number of environmental
contaminants, especially surface acting compounds. Perfluorinated compounds are of
special interest because of their structural similarity to endogenous fatty acids, their
surface-acting physico-chemical property, and the previously shown membrane-related
effects (Upham et al., 1998). The ability of PFOS to affect membrane permeability and
membrane fluidity suggests that the effects observed may be due to relatively non-

specific detergent like effects on the membrane.

The experiments described in this paper were aimed at better describing and
understanding the effects of perfluorinated fatty acids on specific membrane properties.

The effects of PFOS and related chemicals on membrane fluidity were investigated using

16

flow cytometry. Fish blood cells were used as a model membrane system and membrane
fluidity was measured using an excimer-forming lipid technique with pyrenedecanoic
acid (Pownall and Smith, 1989). In addition, we used the cationic carbocyanine dye JC-
1, which accumulates in mitochondria, as a sensitive marker for mitochondrial membrane
potential (Cossarizza et al., 1993; Zoeteweij et al., 1992). To investigate fiirther of the
possible effects on membranes, specifically membrane permeability, the effects of several
perfluorinated fatty acids in several cell line/ligand bioassay models were investigated.
While these assay systems are generally used to investigate the direct receptor mediated
effects of the target ligands, E2 and TCDD, in these studies we used the assays as means
to measure the ability of the perfluorinated compounds to alter the permeability of the

cell membranes to the target ligands.

MATERIALS AND METHODS
Chemicals
Perfluorooctane sulfonic acid (PFOS), perfluorohexane sulfonate (PFHS), and
perfluorobutane sulfonate (PF BS) were obtained from 3M company (St. Paul, MN).
Stock solutions were prepared by dissolving testing compounds in DMSO to a final

concentration of IOmM.

Cell Bioassays
Four bioassays were used to investigate the effects of PFOS on different biochemical
responses, which could be used indirectly to indicate effects on membrane permeability

to known substrates. H4IIE-luc cells are rat hepatoma cells that were stably transfected

l7

with a firefly luciferase reporter gene under direct control of the aryl hydrocarbon
receptor (Ah-R) and dioxin-responsive elements (DREs) in the DNA (Sanderson et al.,
1996). The H4HE-luc cell line can be assayed for both luciferase activity and
ethoxyresorufin O-deethylase (EROD, CYPlA) activity. To determine if the effects of
PFOS were directly on the expression of cytochrome P4501A1, results for the
endogenous AhR-mediated EROD activity were compared to the response of an
exogenous reporter gene (luciferase) under the direct control of the AhR. The analysis of
both endpoints increases confidence that any effects observed can be attributed to ligand
permeability rather than 'non-specific' effects of PFOS on the enzyme systems assayed.
PLHC-l cells were derived from a hepatocellular carcinoma of desert topminnow
(Paeciliapsis lucida), and were tested in the same way as H4IIE cells since previous
studies have indicated the presence of Ah-R and inducible cytochrome P450 1A1 activity
(Hahn et al., 1993; Hightower and Renfro, 1988; Hahn and Chandran, 1996; Richter et
al., 1997). The MVLN cell bioassay is based on a human breast cancer cell line MCF-7
stably transfected with a reporter gene under control of the estrogen receptor and
regulatory element, and it was used to assess effects of PFOS on membrane permeability

to estradiol ligand (Pons et al., 1990; Kramer et al., 1997).

Ah-Receptor Based Assays

H4HE-luc and PLHC-l cells were cultured in 100 mm disposable tissue culture dishes.
All cells were grown under sterile conditions (pH=7.4) in a humidified 5/95% COz/air
incubator. H4IIE-luc cells were cultured at 37°C, and the PLHC-l cells were grown at
30°C. H4IIE-luc cells were cultured in Dulbecco’s Modified Eagle Medium (Sigma, St.

Louis MO), supplemented with 10% fetal bovine serum (Hyclone, Logan UT). PLHC-l

18

cells were cultured in Eagle's Minimum Essential Medium (Sigma, St. Louis MO)
supplemented with 292 mg/L L-glutamine and 10% FBS (Hyclone, Logan UT). All cells
were passaged when cultures became confluent, and new cultures were started from
frozen stocks afier 30 passages. Cell bioassay procedures were conducted as previously
described (Sanderson et al., 1996) with additions of PF OS or other perfluorinated
compounds made as indicated for the different experiments. EROD and/or luciferase
assays with H4IIE-Inc and PLHC-l cells were performed following previously described
procedures (Sanderson et al., 1996). Luciferase Reporter Gene Assay Kit reagents
(Packard Instruments, Meriden CT) were reconstituted freshly before performing the
assay. Under subdued light, 75 ul per well of reconstituted substrate solution was added
and agitated, and the plates were incubated for 10 min at 30°C. Luminescence was
measured on a plate-reading luminometer (Dynatech, Laboratories, Chantilly, Virginia).
Before cells were assayed cell viability was determined by visual inspection and by use

of the live/dead cell viability assay kit (Molecular Probes, Eugene OR).

M VLN- 7 Biaasscy

MVLN cells were obtained from Dr. Michel Pons, Institut National de la Sante et la
Recherche Medicale, Montpelier, France (Pons et al., 1990). MVLN cells were grown in
Dulbecco’s Modified Eagle Medium with Hams F-l2 nutrient mixture (Sigma, St. Louis
MO) supplemented with NaHC03, 1 mM sodium pyruvate, 1 mg/ml insulin. For
culturing the cells on 100 mm plates 10% of PBS (Hyclone, Logan, UT) was added to
media. For bioassays in 96 well plates 5% charcoal stripped FBS (Hyclone, Logan, UT)

was used to reduce the amount of background due to l7B-estradiol (E2<5 pg/ml) present

19

in the serum. The cells were cultured at 37°C in humidified C02 incubator, 5/95 %
C02/air, > 90% humidity. For bioassays cells were plated in 96-well culture ViewPlates
(Packard Instruments, Meriden, CT) at a density of approximately 15,000 cells in 250 pl
media. Cells were dosed 24 hr after plating and were exposed for another 72 hr. E2 was
dissolved in acetonitrile and PFOS was dissolved in methanol. Each exposure
concentration was dosed in triplicate with 2.5 ul of extract solution, the final
concentration of solvents was 0.5 % v/v or less. At least three replicate standard
calibration curves ranging from 0.15 to 500 pM E2 were used with each assay. Each
sample was dosed in six serial dilutions (1:3 diluting step) with 3-4 replicates per
dilution. The exposure time for all bioassays was 72 hr. In competition experiments the
concentration of E2 added was 10 pM, equivalent to an EC20. There were at least three
blank and solvent control replicates on each plate. Cell viability for MVLN cells was
assessed using the same method as for the H4IIE and PLHC-l cells. Luciferase activity

was determined as described for the H4IIE cells.

F low Cytometty Membrane Fluidity Assay

Pyrenedecanoic acid (Molecular Probes, Eugene, OR) was dissolved in 0.03% ethanol
and 0.1 M phosphate buffer (pH=7.4) to a concentration of 300 pM. JC-l (5,5',6,6'-
tetrachloro—l,l',3,3'-tetraethylbenzimidazolylcarbocyanine iodide, Molecular Probes,
Eugene, OR) was dissolved in DMSO to a concentration of 150 pM. The stock solutions
were stored in the dark at —20°C after flushing the headspace with N2. Working solutions
of pyrenedecanoic acid and JC-l were prepared freshly on the day of assay by diluting

stock solution ten times to the concentration of 30 uM and 15 pM, respectively.

20

The effects of PFOS and related chemicals on membrane fluidity were investigated by
use of flow cytometry. Fish blood cells were used as a model membrane system and
membrane fluidity was measured using the excimer-forming lipid technique with
pyrenedecanoic acid (Pownall and Smith, 1989). Fish blood cells were chosen rather
than cultured cells as the membranes of cultured cells must be perturbed to release the
cells from the culture dishes. In addition most cultured cells have been 'immortalized'
and so cannot be considered normal. Fish blood cells represented a readily available
source of cells that could be easily manipulated in the laboratory without causing undue
stress to the cells. The excimer-forming lipid method is based on the formation of
excimers (dimer of excited monomer and ground state monomer) of fluorescent pyrene
molecules. The emission spectrum of pyrene is composed of two parts: one due to the
excited pyrene monomers, and the other, at longer wavelength, originating from excimers
formed upon collision of an excited pyrene with a ground-state pyrene. The rate of the
excimer formation is dependent on the translational diffusion rate of pyrene molecules,
which are incorporated into the cell membrane. Therefore, the ratio of excimer
fluorescence to monomer fluorescence intensities (IE/1M) is proportional to membrane
fluidity (Masuda et al., 1987). JC-l (5,5',6,6'-tetrachloro-1,1',3,3'-
tetraethylbenzimidazolylcarbocyanine iodide) is a novel cationic carbocyanine dye that
accumulates in mitochondria. The dye exists as a monomer at low concentrations and
yields green fluorescence, similar to fluorescein. At higher concentrations, the dye forms
J-aggregates that exhibit a broad excitation spectrum and an emission maximum at ~590
nrn. These characteristics make JO] a sensitive marker for mitochondrial membrane

potential (Cossarizza et al., 1993).

21

Pyrenedecanoic acid and JC-l were excited with 365nm and 488nm argon lasers
respectively. The fluorescence intensities of monomer and excimer pyrenedecanoic acid
were determined using a FACS Vantage flow cytometer (Becton Dickinson, San Jose,
CA) equipped with bypass filters of 400:15 nm and 450:30 nm, respectively. JC-l
fluorescence was determined at 530i30 nm and 590:42 nm for monomer and J-
aggregate respectively. At least 10,000 cells were examined in each sample. Cell
scattering was shown as contour plot for FCS and SSC. Fluorescence intensities were
recorded as histograms with event number (cell count) vs. channel number (fluorescence

intensity).

Preparation of Carp Leukocytes

Carp were anaesthetized with MS-222 (250 mg/L in water). Blood was collected from
the caudal vein into a heparinized syringe, an average 2.5 ml blood per fish can be
collected in this way. During the course of these experiments blood was collected on 3 to
4 occasions and the whole blood of 3 to 4 individual fish was collected and pooled on
each occasion. Three ml of Histopaque-1077 (Sigma, St Louis MO), was added to a 15
ml centrifuge tube and allowed to warm to room temperature. Three ml of the collected
fish blood was carefully laid on top of the histopaque before centrifugation at 400 x g for
30 min at room temperature. After centrifugation, the upper serum layer was removed
with a Pasteur pipette and discarded. The opaque interface (white blood cells and
histopaque) was transferred to clean centrifuge tube, 10 ml PBS was added and the

mixture was mixed gently. The cells were centrifuged at 250 x g for 10 min at room

22

temperature, the supernatant was discarded and the cell pellet was resuspended in 5 ml
PBS before another centrifugation at 250 x g for 10 min. The final cell pellet was
resuspended in 0.5 ml PBS (or McCoy’s 5A medium). Cell numbers were determined in
a hemocytometer and the final cell concentration was adjusted to 1 x 105 ~ 1.5 x 106

cells per 200 pl of suspension.

Labeling afCam Leukocytes
Labeling was performed by adding 100 pl of 30 uM pyrenedecanoic acid solution, 100 pl
of 15 DM JC-l solution and 200 pl of the cell suspension to a 5 ml round-bottom tube

and gently mixing for 15 min at 25°C, excess label was removed by two washes with
PBS before the final volume was adjusted to 1 ml with PBS. Chemical treatments
including blanks, solvent controls, and positive controls (1% pentanol for membrane
fluidity and 100 nM valinomycin for mitochondria membrane potential), and test
compounds in serial dilutions were carried out by incubating the labeled cells with test
chemicals for 15 min at 25°C. For PFOS each treatment was performed in triplicate, for
other chemicals single determination was sufficient to demonstrate their inactivity in the
assays as performed. The concentrations of positive controls were based on previously

published data (Pownall & Smith, 1989; Cossarizza et al., 1993).

F low Cvtametrv Data Analysis
Flow cytometry data were acquired and analyzed using CellQuest software (Becton
Dickinson, San Jose, CA) interfaced to the flow cytometer. The raw data from each

histogram were extracted, and copied to a Microsoft Excel spreadsheets for subsequent

23

analysis. Total fluorescence intensity for each wavelength was calculated as sum of
event number times channel number. Fluorescence ratios were calculated as the ratio of
the total fluorescence intensities. Appropriate statistics were performed on multiple

determinations of the fluorescence ratio.

Statistical Analysis
All cell bioassay data were collected electronically and converted into Excel spreadsheet
format. Dose response curves were analyzed using Microsoft Excel 98, ANOVAs and

Tukey’s test were conducted using SYSTAT 10 (SPSS, Chicago IL).

RESULTS

PF 0S Effects on Membrane F luidity and Membrane Potential

Exposure to PFOS significantly increased membrane fluidity of fish leukocytes (Figure
1.1) at 33 and 100 pM (16.5 — 50 mg/L). The degree of the maximal response observed
was similar to that observed for 1% pentanol, the positive control for the experiment. In
subsequent experiment the response was determined to be dose-dependent (Figure 1.2).
In two independent experiments the least dose significantly different (p<0.05) from the
control were 15 mg/L (30 pM) and 16.5 mg/L PFOS. Therefore, the threshold for effects

on membrane fluidity in vitra is approximately 15 mg/L.

PFHS and PFBS, compounds that have similar structures to PFOS but with shorter
carbon chain lengths, had no effect on membrane fluidity in the same concentration range

used for the PFOS exposures (results not shown). Therefore, as with other effects

24

observed for perfluorinated compounds the response seems to be related to the length of

the carbon chain (Hu et al., 2002b).

The effects of PFOS on mitochondrial membrane potential were also determined by flow
cytometry (Figure 1.3). Mitochondrial membrane potential was inversely related to the
PFOS concentration. The maximum decrease observed was similar in magnitude to that
observed for 100 nM valinomycin, the positive control. The variability in determination
of mitochondrial membrane potential was greater than that for the membrane fluidity.
Statistical analysis of the membrane potential data revealed that the first dose
significantly different (p=0.0018) from control was 30 11M (15 mg/L), which is similar to

the threshold concentration for effects on membrane fluidity.

Effects of PF OS on EROD and Luciferase Activities

PFOS alone did not induce cytochrome P450 1A1 (CYP1A1), as measured by EROD
activity, compared to solvent-exposed cell culture controls (Figure 1.4A). TCDD
induced EROD activity in a dose-dependent manner, with the greatest induction being 17
fold with an EDSO of approximately 0.01 11ng (Figure 1.4A). To assess the interactive
effects between TCDD and PFOS, cells were exposed to the two chemicals in
combination. Cells were dosed with TCDD alone, or with TCDD in combination with
PFOS at concentrations ranging from 0.0001 to 10 mg PFOS/L. Co-exposure of cells to
PFOS and TCDD increased the CYP1A1 activity induced by TCDD (Figure 1.5A).
Compared to the TCDD standard dose-response curve, the addition of PFOS increased

both the slope of the curve and the magnitude of maximum response, with PFOS at 0.1

25

mg/L causing the greatest increase in the TCDD response. The interactive effects
observed were statistically significant at 0.2 ug/L TCDD plus 0.1 mg/L PFOS (p<0.05), 1
ug/L TCDD plus 0.01 mg/L PFOS (p<0.05), and 1 ug/L TCDD plus 0.1mg/L PFOS

(p<0.01) (Figure 1.5B). In the last combination, the addition of PFOS increased the

effect of TCDD by 40%.

Results were similar for PFOS exposure to PLHC cells. PFOS alone exhibited no
detectable effect on CYP1A1 induction (Figure 1.4B). For PLHC-1 cells the standard
TCDD dose-response curve had a slightly different shape compared to that of the H4IIE-
luc cells, however the general trend of interactive effects was similar to that of the H4IIE-
luc cells. The most significant interactive effects in the PLHC cells were observed at a
TCDD concentration of 0.2 ug/L and at a PFOS concentration of 0.1 mg/L (p<0.01),

which increased the effect of TCDD by approximately 40% (Figure 1.6).

To determine whether the PFOS related increase was specific to the CYP450 enzyme
assay used, the luciferase assay was also conducted with H4IIE-luc cells dosed with
PFOS and/or TCDD. In H4IIE-luc cells PFOS alone did not induce AhR-mediated
luciferase activity relative to that of the control. In contrast, TCDD induced luciferase
activity in a dose-dependent manner (Figure 1.4C). Exposure to l ug/L TCDD plus 0.1
mg/L PFOS (ANOVA, p<0.05), and 0.2 ug/L TCDD plus 0.1 mg/L PFOS (ANOVA,
p<0.05), significantly increased induction over TCDD alone, with the maximum of

increase by 40% (data not shown).

26

MVLN cell treated with PFOS showed no indication of induction of estrogen receptor-
controlled genes at concentrations as high as 10 mg/L. In contrast, 17B-estradiol (E2)

added to the cells strongly induced the production of luciferase in a dose dependent

manner with maximal activity observed between 25 and 100 nM E2 (Figure 1.7).

To determine whether the interactive effects observed between PFOS and TCDD in the
AhR reporter gene system were also acting in the ER reporter system, experiments were
conducted with mixtures of PFOS and E2 and TCDD (Figure 1.7). As in the AhR
bioassay system a moderate (approximately 40%) PFOS dependent increase in the E2-
mediated expression of luciferase was observed at the three higher doses of E2. The
additional increase in luciferase activity was dependent on the dose of PFOS, with a
PFOS concentration of 0.1 mg/L resulting in the greatest increase in expression at all E2
concentrations. In addition, in cells treated with PFOS alone, concentrations as high as
10 mg/L, did not adversely affect MVLN cell viability or the responsiveness of the ER-
mediated pathway (results not shown). In contrast TCDD at 0.5 ug/L caused a significant
decrease in the activity of the ER-mediated pathway and cell viability in the MVLN cells

(Figure 1.7).

DISCUSSION

These studies indicated that of the chemicals tested only PFOS significantly altered
membrane properties in the concentration range tested. The effects observed were subtle
and different effects occurred at distinctly different concentrations. It is our hypothesis

that these effects represent a series of concentration dependent changes in specific

27

membrane properties brought about by the detergent like effects of PF OS on membrane
lipids and/or proteins. It has become apparent over recent years that the physical
structure of all cellular membranes is tightly controlled, and that the physical properties
of different membranes are important for their function. There is evidence of extensive
differentiation of lipid components between the two sides of many biological lipid bilayer
membranes. As well as these ‘vertical’ differences in membrane composition, it has been
demonstrated that lateral domains exist within membrane layers. In particular,
cholesterol forms ‘raft’ like structures that are characterized by their low detergent
solubility (Galbiati et al., 2001). It has also been demonstrated that outside these rafts the
lateral movement of individual lipid molecules within the ‘bulk’ phase of the membrane
appears to be limited. Lateral movement of these molecules appears to progress as a
series of transitions between distinct lipid compartments within the ‘bulk’ membrane
phase (Kawasaki et al., 2001). It is clear fiom these observations that any alterations in
cellular membrane properties caused by xenobiotics could have a considerable impact on
the various fimctions of the membrane and its substructures. Alterations in membrane
fluidity have been measured as a consequence of decreased cell proliferation (Beguinot et

al., 1987) and obesity (Beguinot et al., 1985).

In addition to the direct physical effects of PFOS on cellular lipid components it is also
possible that alterations in cholesterol metabolism may have consequences for membrane
function. Alterations in membrane fluidity have been associated with alterations in
cellular or membrane cholesterol by a number of studies (Beguinot et al., 1985; Beguinot

et al., 1987; Jefferson et al., 1990). PFOS has been demonstrated to be

28

hypocholesterolaemic in primates during long-term sub-chronic exposure (Seacat et al.,
2002b). Therefore, the observed increases in membrane fluidity due to PFOS exposure
could be compounded during in viva exposures by decreases in the cholesterol content of
the membranes resulting in further increases in membrane fluidity. While alterations in
the permeability of the membrane may have more directly observable effects, our
understanding of the consequences of changes in specific aspects of membrane fluidity

and function is less clear.

The in vitra systems in these studies were used as means of probing effects of PFOS on
membrane permeability. While PFOS itself was inactive in AhR and ER receptor-
mediated pathways, it was able to increase the amount of TCDD and Estradiol reaching
the cell signaling pathways. Thus, PFOS was hypothesized to be able to increase the
permeability of the cell membranes to these two ligands. The fact that these responses
are neither ligand nor biochemical pathway specific yet occur at essentially the same
PFOS concentrations suggests that the effect is at the level of the cell membrane rather
than effects on specific transporter protein systems. In addition PFOS at concentrations
of 15 mg/L and greater was able to decrease mitochondrial membrane potential in
exposed cells. These results are in agreement with functional assessments of the effects
of PFOS on mitochondrial energy production, which demonstrate that at 10 uM PFOS
(equivalent to 5 mg/L) only weakly affected energy production (Starkov and Wallace,
2002). These alterations are attributed to increased membrane permeability resulting in
dissipation of the mitochondrial proton gradient. These effects were distinct from the

more potent effects of other perfluorinated chemicals that acted as either classical

29

protonophoric uncouplers (Starkov & Wallace, 2002) or chemicals capable of inducing

the mitochondrial membrane permeability transition (Sokol et al., 2001).

Since the optimal PFOS concentrations for the ligand permeability effects observed were
similar (approximately 0.1 mg/L), as were the extents of the increase (approximately
40%) in effect, we hypothesize that the increase in permeability is non-specific and so is
most likely a result of lipid/PFOS interactions. Alternatively, it is possible that the
amphiphilic nature of PF OS acts to improve penetration of hydrophobic ligands (TCDD
and E2) through the cell membrane or the ligands may be involved in a co-transportation
mechanism, although it is generally assumed that at least TCDD crosses the membrane

by simple diffusion.

These studies have also demonstrated the ability of PF OS to modulate membrane fluidity
in vitra. The threshold concentrations of PFOS which elicited these effects were in the
range of 5 to 15 mg/L which is similar to the threshold concentrations that have been
observed for cell permeability effects and for other PFOS-mediated cellular responses,
such as gap-junction intercellular communication (Hu et al., 2002b). Together these
results suggest a range of responses linked by a common mode of action. From these
studies on membrane fluidity it appears that the mechanism of action is the interaction of
PFOS with membrane lipids since the regulator of the association of the pyrene dimer is
the horizontal fluidity of the lipid bilayers that form the cellular membrane. This fluidity
should not be interpreted as indicating that the cell membranes are any more ‘leaky’ than

unexposed membranes. The transport or translocation of compounds across the cell

30

membrane is a different physiological process, not directly related to the horizontal

fluidity of the membrane.

The kinetics of occurrence for the effects observed in these experiments indicate that the
interaction of PFOS with cell membranes is rapid, with effects observed after only 15
minutes of incubation. Similarly, rapid effects and rapid recovery have previously been
described for the effects of PFOS on Gap Junction Intercellular Communication (GJIC)
(Hu et al., 2002b). The short time period before the onset of the effects preclude the
possibility of direct incorporation of these fatty acid analogues into membrane lipid,
which would normally require a timeframe of hours. Indeed, the relatively short chain
length of PFOS compared to the normal 16 and 18 carbon chain fatty acids present in
phospholipids would suggest that PFOS is unlikely to be covalently inserted into
phospholipids. While the effects observed here on membrane fluidity are suggestive of
lipid/PFOS interactions previous work on the effects of PFOS on GJIC suggest a
mechanism more related to protein/lipid interactions (Hu et al., 2002b). It seems most
probable given the highly hydrophobic nature of the fluorocarbon chain of PFOS that this
compound may be most active at lipid/protein interfaces within membranes. This
mechanism of action is supported by the highly surface active nature of PFOS and other
perfluorinated compounds. It is clear that the effects on membrane fluidity (15 mg/L) are

observed at different concentrations from those observed on membrane permeability (0.1

mg/L).

31

It is difficult to interpret what the observed changes in membrane fluidity and
permeability mean in viva. We are aware of no studies which have linked membrane
fluidity effects to other toxic endpoints. Those studies that are available suggest rather
that alterations in membrane fluidity are a consequence of diseased or abnormal
conditions (Beguinot et al., 1987; Beguinot et al., 1985). The experiments described here
and those of other investigators clearly demonstrate that the alterations in membrane
properties caused by PFOS do not result in the classical mitochondrial membrane
permeability transition, which leads to apoptotic cell death (Sokol et al., 2001). It is
therefore unclear what, if any, would be the likely consequences of the subtle membrane
fluidity alterations at the whole organism level. We have however previously
demonstrated that the effects of PFOS on GJIC observed in in vitra exposures also

occurred in viva (Hu et al., 2002b).

All the assay systems used here were in vitra and results cannot be expected to directly
reflect in vivo conditions. Notably PFOS has been shown to bind to a variety of proteins
both intracellular (Luebker et al., 2002) and extracellular (Jones et al., 2002). It is
possible that binding of PFOS to proteins could significantly ameliorate the membrane
related effects observed here if the affinity for protein is greater than that for membranes.
Additional studies will need to be conducted to determine whether the observed effects
actually occur in viva. Studies on the inhibition of gap junctions by PFOS have indicated
that effects observed in cell culture also occur in viva albeit at different concentrations
(Hu et al., 2002b). Given that the tissue concentrations of PFOS measured in some

organisms can reach 1-10 mg/kg (Giesy and Kannan, 2001; Kannan et al., 2001a; 2001b)

32

we would expect that to some extent alterations in membrane fluidity and permeability
might occur providing that there are no other factors, which might ameliorate these
effects. If the suggested alterations in membrane fluidity do occur there is little evidence

to indicate whether adverse whole organism effects are likely to occur.

33

Fig 1.1

 

 

 

 

   

A B
53 I.
V
2 ..
.2
S
'a
In
Emission at 400 nm Emission at 400 nm
Cells+Pyrene Cells+Pentanol
C D
O
V)
v
‘5
s:
.52
S
'5
LL}

 

 

 

 

 

 

Emission at 400 nm Emission at 400 nm

Cells+PFOS(15 mg/L) Cells+PFOS(50 mg/L)

34

Fluorescent Ratio 450/400

Figure 1.2

 

1.3

 

1.1~--— H, ___ w--. a W

 

 

 

 

I
I
I
I
I
r—l—I
Fl...
l--l
I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(WIT-I—M—{ITE‘M i. *-I- __
0.8 ~—
0.7 w
0.6 I T T T T 1 .
e . ¢
$©Qo$08D~r°§Doe°W$ (9&3) 55$ 69¢? 5w"?

‘0 e e
0

Exposure Type and Concentration

35

 

Figure 1.3

 

 

 

an:
al—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.95

"TIT
. T I
1 _
Til—HI.
i
*T||_H||.
— L
.I —i u
. T
I _
_ I
_ .
_ P _
m. m m .7..
O O O 0

Did“ oofioomqu—ATH

Exposure Type and Concentration

36

Figure 1.4

 

2000

A
1800 ~
1600 . 9mm)

1400 ~ E" PFOS

1200 -
1000 ‘
800 “
600 I
400 “
200 "
0 .

1200 13
1000 “

     
    

ty % control

1

EROD Activ

 

 

 

 

OO
O
O

600 ‘
400 “
200 “

r
0. . . . , i

1000
900
800 I e-TCDD
700 "
600
500
400
300

200
100

EROD Activity % control

 

 

 

 

 

 

Luciferase Activity % control

 

0 1 i 1
0.0001 0.001 0.01 0.1 1 10

Concentration PFOS (mg/l)/ TCDD (pg/l)

I l fi

37

Figure 1.5

 

 

 

S
mm
WP
A O I mmy
mom
DMD
.. . www
‘ n:
52/-
.A./
/:M
Ao-
A
5 4 3 2 1

_ob:oo 8 2528 b_>uom QOMm

 

10

1

0.001 0.01 0.1

0.0001

TCDD (ng/ml)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

38

Figure 1.6

EROD relative to control

 

 

*1:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.0001 0.001 0.01 0.1 1 10

PFOS (mg/L)

39

 

 

Figure 1.7

Luciferase Activity (RLU)

0.15

0.10

0.05

0.00

 

 

 

, ’1
z ' ’E2+PFOS

E2+TCDD

\

I I I I J

 

 

20 40 60 80 100 120

E2 (nM)

40

CHAPTER TWO

Hu, W.Y., Jones, P.D., Celius, T, and Giesy, JP. (2003) Identification of genes
responsive to perfluorooctane sulfonic acid using gene expression profiling. European
Journal of Toxicology and Pharmacology (submitted)

41

ABSTRACT

Perfluorooctane sulfonic acid (PF OS) is widely distributed in the environment including
in the tissues of wildlife and humans, however its mechanism of action remains unclear.
In the current study, the Affymetrix rat genome U34A gene chip was used to identify
alterations in gene expression due to PFOS exposure. Rat hepatoma cells were treated
with PFOS at 2 or 50 mg/L in culture medium for 96 hr, and Sprague-Dawley rats were
orally dosed with PFOS at 5mg/kg/day for 3 d or 3 wk. Genes that were significantly
(p<0.0025) induced were primarily genes for fatty acid metabolizing enzymes,
cytochrome P4505, or genes involved in hormone regulation. The significantly down
regulated genes were mostly involved in signal transduction pathways. Consistent
expression profiles were obtained for replicate exposures within treatment, and for short-
terrn and long-term in viva exposures. Limited similarity in gene expression profile was
observed between the in viva and in vitra exposure systems. The structural similarity
between PFOS and endogenous fatty acids may explain why the major pathway affected

by PFOS is postulated to be peroxisomal fatty acid beta-oxidation.

Key words: PFOS, gene expression, fatty acids metabolism

42

INTRODUCTION

Perfluorinated fatty acids (PFFAS) are synthetic, fully fluorinated, fatty acid analogues.
Recent studies indicate that perfluorooctane sulfonic acid (PFOS) is the most commonly
found compound in the tissues of wildlife, perfluorooctane sulfonamide (PFOSA),
perfluorooctanoic acid (PFOA), and perfluorohexane sulfonate (PFHS) have also been

detected in the tissues of several species (Giesy and Kannan, 2001).

To date, most toxicological studies have been conducted on PFFAs such as PFOA and
perfluorodecanoic acid (PF DA), rather than the more environmentally prevalent PFOS.
PFOS appears to be the ultimate degradation product of a number of commercially used
perfluorinated compounds (Giesy and Kannan, 2002). The concentrations of PFOS
found in wildlife are greater than other perfluorinated compounds (Giesy and Kannan,

2002; Kannan et al., 2001a, 2001b).

The mechanisms by which some PF FAS elicit their toxic effects are not well known. For
example several PF FAS have been reported to be peroxisome proliferators. PF FAs, such
as PFOA and PFDA can interfere with lipid metabolism by increasing peroxisomal fatty
acid B-oxidation, and inducing several hepatic enzyme activities (Sohlenius et al., 1996).
Both in viva and in vitra exposures to PFOA result in increased activities of peroxisomal
acyl-CoA oxidase, which catalyzes the first and rate-limiting step in fatty acid oxidation
(Sohlenius et al., 1994). Fatty acid oxidation is also a process known to produce
hydrogen peroxide, an oxidative radical, such that PFF As can lead to oxidative stress and

could possibly result in DNA damage (Sohlenius et al., 1994).

43

Some PFF As, including PFOA and PF DA, have been shown to be involved in regulating
tissue fatty acid composition and content. PFOA can reduce cholesterol and
triacylglycerol concentrations in serum, increase liver triacylglycerol concentration, and
reduce hepatic lipid output (Haughom and Spydevold, 1992). Treatment with PF DA can
inhibit acyl-CoA synthetase activity and result in an increase in the level of free fatty
acids, which are known to be able to activate protein kinase C (PKC), and lead to a
signaling cascade that is important for normal cell function, cell proliferation and gene
expression (Reo et al., 1996). Hepatic microsomal cytochrome P450 enzymes were
induced in rats treated with PFOA (Permadi et al., 1992). In the CYP4A sub-family, nine

enzymes specific for fatty acid w-hydroxylation, were significantly induced with

exposure to 500 11M PFOA for 7 d. Recent studies have also demonstrated effects of
PFOS in vitro and/or in viva on Gap Junctional Intercellular Communication (Hu et al.,
2002b), membrane fluidity (Hu et al., 2002a) and serum steroid binding globulins (Jones
et al., 2002). Chronic exposure of rodents and primates to PFOS resulted in significantly

altered concentrations of cholesterol in blood (Seacat et al., 2002a, 2003).

In the current study, the effects of PFOS on gene expression were determined using the
Affimetrix GeneChip array, a genome-wide expression analysis method based on the rat
genome. Our null hypothesis is that PFOS exposure does not cause mechanism specific
modulations of gene expression. This screening approach was used to identify genes
responsive to PF OS with the intent of identifying critical target pathways for the

biological effects of PFOS. In vitro and in viva exposures were used to allow comparison

of gene expression profiles between in vitro and in viva models, and long-term verses

short-term exposure.

MATERIALS AND METHODS

Chemicals

Perfluorooctane sulfonic acid (PFOS) was obtained from 3M company (St. Paul, MN).
The PFOS (potassium salt) used for in vivo experiments was purchased from F luka
Chemicals (Buchs, Switzerland), chemical analysis of the isomer patterns revealed that it

was essentially the same as the product obtained from 3M.

Cell Culture and Treatment

H4IIE rat hepatoma cells, between passages 5 and 15, were cultured in 100 mm
disposable tissue culture dishes at 37°C under sterile conditions (pH=7.4) in a humidified
5/95% C02/air incubator (Forma Scientific, Model 8173). Cells were grown in
Dulbecco’s Modified Eagle Medium (DMEM, Sigma D-2902, Sigma, St. Louis MO),
supplemented with 10% fetal bovine serum (FBS, Hyclone, Logan, UT). At confluence,
cells were removed from the dish with trypsin/EDTA (Hyclone, Logan, UT), and split
into four tissue culture plates. The cells were given 24 h after splitting to allow for
attachment, the medium was then replaced and cells were dosed with PFOS to achieve
final concentrations of 2 mg/L and 50 mg/L, methanol was used as solvent control, and

the blank control received no dose. Cells were incubated for 72 hr after exposure.

45

In vivo Treatment

Sixty-day old Sprague-Dawley rats (males 294:4 g; females 209:2 g) were obtained
from Charles River Laboratories (Raleigh, NC), and housed at 20-24°C in humidity-
controlled (40-60%) facilities at the US-EPA Reproductive Toxicology Laboratory
(Research Triangle Park NC). Estrous cycle was not determined in female rats and
breeding was not a targeted endpoint for these studies either. Rats were randomly
assigned to two blocks, block one with six males and six females, and block two with
four males and four females. Block one was exposed to PFOS for 21 d, block two was
exposed for 3 d. Within each block, half of the males and half of the females were
randomly assigned to treatment or control groups. Rats received PFOS (5 mg/kg) or
vehicle control (0.5% Tween-20) daily by oral gavage at a rate of 1 ml/kg body weight.
At the end of exposure, animals were sacrificed, livers were removed within 1-2 minutes
of sacrifice and portions of liver were removed and placed in TriReagent. Liver samples

were processed for RNA isolation on the same day as collection.

GeneChip Array Experimental Procedure

The Affymetrix rat genome U34A gene chip array was purchased from Affymetrix Inc.
(Santa Clara, CA). The oligonucleotide probes on U34A array cover approximately 8800
known genes and Expressed Sequence Tags (ESTS) in the Rat genome. Transcripts were

selected from Genebank Unigene build 34 and the dbEST database.

Total RNA from cell cultures and rat liver samples were extracted using TriPure Isolation

Reagent (Boehringer Ingelheim, Germany) using the manufacturers recommended

46

procedures. The optical densities of RNA samples were measured at 260 nm and 280
nm, and the 260 nm/ 280 nm ratio was evaluated as a measurement of nucleic acid purity.
RNA concentration was quantified using optical density at 260 nm. The quality of RNA
was evaluated by the appearance of distinct of 183 and 28s ribosomal RNA bands on 1%

agarose gel.

First and second strand cDNA was synthesized from total RNA samples using the
SuperScript Choice System (Gibco BRL life Technologies, Cheshire, UK). High quality
total RNA (16 ug) was used as the starting material and lul 100 pmol/ul T7-(dT)24
primer (5’ — GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG —
(dT)24 —3’; Genset Crop. San Diego, CA) was used to prime the reaction. After double
standed cDNA clean up and quality check, an in vitra transcription reaction was
conducted to produce biotin-labeled cRNA from the cDNA. The cRNA was then purified
and fragmented for hybridization analysis. Following hybridization and washing,
staining and scanning procedures were performed in the Genomics facility on Michigan
State University campus (Fluidics Station 400 and Hybridization Oven 640 from
Affymetrix, Santa Clara, CA). Briefly the biotin-labeled cRNA was combined with
probe array control, BSA, and herring sperm DNA into a hybridization cocktail, and
applied to the probe array after a cleanup procedure. It was then allowed to hybridize on
the array for 16 hr at 45°C. Following the hybridization, the arrays underwent an
automated washing, staining and scanning protocol on the fluidics station (Affymetrix,
Santa Clara, CA). Each complete probe array image was stored in a separate data file

identified by experimental name.

47

A total of nine chips were used in this study. Three were used to analyze samples from
the in vitro exposure (solvent control, PFOS at 2mg/L and 50mg/L), while four chips
were used for examining the long-term in viva samples (2 animals exposed to solvent
controls and 2 animals exposed to PFOS at 5 mg/kg/day for 21d). Finally, two chips
were used for the in viva short-term sample (1 animal treated with solvent control and 1
animal exposed to PFOS at 5 mg/kgd for 3 d). The PFOS concentration was measured
in each rat liver samples with an average of 600 mg/kg—body weight for 21d exposure,
and 90 mg/kg-body weight for 3 (1 exposure. Data collected from the nine chips was

transferred to a Microsoft Access database.

GeneChip Array Data Analysis

 

Each image file was analyzed and data was retrieved using the Affymetrix “data mining”
tool (Affymetrix Santa Clara, CA). Initial data normalization and filtering was also
conducted. The output file was then stored in a Microsoft Access database and the initial
data analysis was conducted using Microsoft Access query design. Cluster analysis,
Genetree construction and pathway analysis were conducted using GeneSpring software

(SiliconGenetics, Redwood City, CA).

Scatter plot

The scatter plot is useful for examining the expression levels of genes in two distinct
conditions, samples, or normalization schemes. To evaluate the reproducibility of results
for individual animals and cell cultures, scatter plots based on a correlation analysis were

prepared for the different samples.

48

Gene tree

Genes can be classified in a manner similar to classification of organisms into phylogenic
dendrograms or trees. As organisms sharing evolutionary properties tend to be clustered
together, genes sharing similar expression pattern can be used to determine the similarity
in responses of species, doses, or duration of exposure. The vertical distance along the
branches of such a tree represents a measure of degree of similarity. The Genetree
algorithm of GeneSpring was used to draw a hierarchical dendrogram of clustered genes
according to their expression profiles among treatments. The algorithm calculated the
correlation for each gene with every other genes in the set. Then pairs of genes
exhibiting the greatest correlations were merged and their expression profiles averaged.
The new composite 'gene' was then compared with all the unpaired genes. This was
repeated until all of the genes had been paired. Based on the way their expression was
altered across the nine samples, the 8800 genes were grouped on the horizontal axis with
the nine treatment group clustered based on how they affect gene profiles on the vertical

axis.

Pathway analysis and Target path way

Based on the genes that were identified as having their expression modulated by exposure
to PF OS, a pathway analysis was performed by linking genes via their Enzyme
Commision (EC) numbers to their positions in known metabolic pathways as described in
metabolic pathway maps in the Kyoto Encyclopedia of Genes and Genomes

(http://wwwgenomeadjp/keggI.

49

Statistical Analysis

Statistical analysis of gene array data is complicated by the large amount of data
generated, the different arraying technologies used by different manufacturers and the
generally small sample sizes used in experiments involving Gene Arrays. In addition,
many of the software programs used to analyze capture and analyze array data perform at
least initial background corrections and some statistical analysis using proprietary
algorithms. For example GeneChip probe arrays are manufactured such that each gene is
represented with a series of 11-20 probes pairs (each probe is 25 bp in length). Each
probe pair is composed of a perfect match probe and a mismatch probe, the mismatch
probe has almost the same sequence as that of the “perfect match”, except one nucleotide
difference. This so-called “tiling” design serves as an internal control for hybridization
specificity, which allows consistent discrimination between closely related target
sequences. While the number of chips used in the current study was limited to 9, these

chips represented exposures at a variety of concentrations both in vitra and in viva.

RESULTS
Overall changes
To determine the overall gene expression changes associated with PFOS exposure, genes
were classified by their fold change in expression. Expression of only about 5% of the
genes analyzed were significantly (p<0.0025) induced or suppressed beyond a 3-fold
change in expression in samples treated with PFOS relative to the control. However, this
still represents some 400 genes whose expression was altered by exposure to PF OS. The

use of a 3-fold cutoff for significance is based on a variety of previous studies (Wan and

50

Nordeen 2002; Gerhold et a1. 2002) and the desire to identify the most dramatically

alterations in gene expression.

The Affymetrix GeneChip system utilizes extensive statistical analysis of the array image
before a gene is reported as induced or suppressed. The detection of a single gene
product is based on analysis of between 11 and 20 oligonucleotide ‘probe pairs’. Each
probe pair consists of two 25-mer oligonucleotides, one ‘perfect match’ for the target
sequence and one with a base mismatch at nucleotide13. The relative spot intensities
between all perfect match and mismatch probe pairs give these chips extraordinary
sensitivity and specificity. In ‘comparison analysis’ mode the software compares the
arrays for two samples analyzed on two different chips. One array is designated as the
baseline (control) and the other the experiment (exposed). Before comparing the two
arrays scaling and normalization are carried out automatically to correct for variations in
overall signal intensity and heterogeneity between the two arrays. During a comparison
analysis, each probe set on the experiment array is compared to its counterpart on the
baseline array and, using Wilcoxon’s Signed Rank Test, a ‘Change p-value’ is calculated
indicating an ‘increase’, ‘decrease’, or ‘no change’ in gene expression (GeneChip®
Expression Analysis Technical Manual). The degree of significance for the change call is
user specified, for chip comparisons conducted in the current study default probability
cut-off values were used so in all cases when a gene expression is reported as altered,
p<0.0025. For the determination of the degree of alteration in gene expression the
software determines the “Signal Log Ratio” using a one-step Tukey’s Biweight method

by taking a mean of the log (base 2) ratios of probe pair intensities across the two arrays

51

(Affymetrix® Microarray Suite 5.0 User’s Guide). This approach helps to cancel out
differences in individual probe intensities, since ratios are derived at the probe level,
before computing the Signal Log Ratio. Since log base 2 is used to determine the signal
log ratio “Fold Change” = 25‘3"" “’8 km). From this discussion it is clear that although
the analysis system reports only a ‘fold change’ in gene expression there is a high degree

of statistical rigor in the determination of these changes in gene expression.

Scatter Plots

 

Scatter plots were first constructed by comparing the duplicates in control or 21 day in
viva PFOS exposed samples. In each case, the majority of the data points fell into the
reference space representing the 95% confidence interval for a correlation coefficient of
1.0 (Figures 2.1A and 2.18). This demonstrated the degree of reproducibility of the

biological responses as well as the reproducibility of the analysis method.

In contrast, when a sample from the long-term in viva exposure was plotted against a
control sample from the same experiment a greater degree of scatter was observed
(Figure 2.1C). Scatter plots comparing in vitra and in viva exposure systems showed
differences in gene expression between two controls from the rat liver cells in viva and
the genetically modified rat hepatoma cells in culture. A significant degree of data
scattering and deviation from the reference lines were observed across all expression
levels (Figure 2.1D). This demonstrated the relatively large difference in gene expression
between the in vitro and in viva systems even without considering the impact of chemical

exposure.

52

Gene Tree

When the patterns of responses for all samples were examined using a 'Gene Tree'
analysis, the in vitro samples exhibited a clearly different profile than did the in viva
samples (Figure 2.2), consistent with the result observed in the scatter plot analysis. In
all three in vitra samples, genes separated into two nodes according to their expression
level. No distinguishing patterns were recognizable among the six in viva samples, since
subtle differences among these samples were masked by the greater differences between

the in vitro and in viva exposure systems.

When a similar gene tree analysis was conducted using solely the in viva data, a group of
genes exhibiting a distinct pattern among treatments was discemable (Figure 2.3). All of
the genes in this group were expressed at a significantly (p<0.0025) greater level in the
two long-term PF OS treated samples than in the controls. The short-term sample
exhibited a pattern similar to that of the long-term exposure although lack of replicate

analyses in the short term exposure prohibited statistical comparison.

am

A list of genes whose expression level was significantly (p<0.0025) altered by PFOS
exposure was identified (Tables 2.1 and 2.2; Figures 2.4 and 2.5). For the long-term
exposure two exposed and two control rats were available for comparison. Since the
Affymetrix system is limited to comparing two chips at a time, each of the exposed rats
was compared to each of the control rats providing a total of four estimates of the fold

change in gene expression. This approach was taken to ensure that only consistent

53

alterations in gene expression were identified. The largest grouping of genes induced by
PFOS exposure in viva were the cytochrome P4508 and genes that coded for lipid
metabolizing enzymes. Several genes involved in hormone regulation and other
regulatory processes were also induced significantly (p<0.0025). Several genes encoding
factors involved in signal transduction pathways were suppressed by PFOS exposure as
were genes involved in regulating neuro-system functions. Of the pathways represented
by these altered genes the peroxisomal fatty acid 0 oxidation pathway seems to be most
affected by exposure to PFOS (Figure 2.6). The gene expression for the enzymes
involved in peroxisomal lipid metabolism were altered but those for the same pathway in

mitochondria were not.

DISCUSSION

The results of this study illustrate the utility of high-throughput toxicogenomics methods
to study the effects of a compound for which the mechanism(s) of action are still unclear.
Gene expression analysis is useful in identifying chemical-specific alterations in gene
expression to allow classification of toxicants and provide important insights into
mechanisms of action (Hamadeh et al., 2002a, 2002b). Alterations in expression profiles
were used to determine potential critical pathways affected by exposure to PFOS. It is
clear that alterations in the concentrations of mRNA species do not necessarily translate
to alterations in the corresponding enzyme concentration or activity. However,
coordinated alterations in mRN A concentrations for a particular biochemical pathway

provide strong evidence for an effect of PFOS on that pathway. Confirmation of

54

alterations at the protein/enzyme level is the next step in the assessment, but is beyond

the scope of this study.

Of the 8,800 functionally annotated genes and ESTs present on the array, only about 5%
responded to PFOS with a more than three-fold change in expression. Differences
between gene expression profiles were observed between the in vitro and in viva control
samples. All three in vitra exposures exhibited an expression pattern distinct from that of
the in viva samples. This could be explained by the differences in exposure system,
dosage, toxicokinetics, toxicodynamics, levels of organization and functional integration,
which make in viva exposure more complicated than in vitra exposure. Concentrations of
PF OS were measured in the rat liver samples (Hu et al., 2002b), however the exposure
concentration provided for the in vitro samples was the dose applied to the culture
medium not the dose internalized by the cells. The genetic profiles of the transformed rat
hepatoma cell line and the freshly isolated normal rat liver tissue would also be expected
to be different. Also, rat liver tissue is composed of a variety of cell types, including
hepatocytes, Kupffer cells, fibroblasts and stellate cells. Thus, the response of liver tissue
to chemicals is different from a monoculture of cells such as the H4IIE cells that are
composed solely of hepatocytes derived from a limited population of progenitor cells.
While this finding may not be surprising the data provided demonstrates just how
significant those differences, previously unknown, can be. Based on our results,
interpretation of in vitro data and implication on in vivo response should be conducted
with caution. No direct extrapolation of in vitro data should be made without further in

vivo testing.

55

Previous studies have indicated that PF OS can be incorporated into cell membranes and
elicits physical membrane effects both in vitra and in viva (Hu et al., 2002a; 2002b).
PFOS inhibited GJIC in a dose-dependent fashion, which occurred rapidly and reversibly.
PF OS increased fish leukocyte membrane fluidity and decreased mitochondria membrane
potential in a dose-dependent fashion. The results from these studies established that
PF OS could alter membrane structure and function, but they could not determine if these
physical effects were also accompanied by effects on gene expression. In this study we
have confirmed that the physical effects on membrane processes, such as alterations in

GJIC (Hu et al., 2002b), are accompanied by alterations in lipid metabolizing enzymes.

In mammalian cells, both mitochondria and peroxisomes are involved in the B-oxidation
of fatty acids and the substrate specificity of the two systems overlap. Mitochondria
oxidize mainly short, medium, and long, straight-chain fatty acids, while peroxisomes are
capable of oxidizing very long straight-chain and branched-chain fatty acids. Short-chain
fatty acids (2-6 carbons) are poor substrates for peroxisomes because of the low affinity
of the peroxisomal B-oxidation enzymes for short-chain substrates. Results obtained
through this study of gene expression indicated that PFOS specifically enhanced the
peroxisomal but not mitochondrial B-oxidation. Fatty acid B-oxidation in peroxisomes is
carried out in four consecutive steps. The enzymes involved in these processes were
increased from 2- tolO-fold by in viva exposure to PFOS. Enzymes responsible for the

equivalent functions in mitochondria were not significantly affected by PFOS exposure

(Figure 2.6).

56

Other PF FAs have been reported to cause peroxisome proliferation (Berthiaume &
Wallace, 2002). However, the response observed for PFOS was not characteristic of a
'classic' peroxisome proliferator. Genes which are indicative of peroxisome proliferation
and other xenobiotic responses have recently been identified (Hamedah et al., 2002a). In
the present study PFOS induced the gene expression level of camitine palmitoyl
transferase (CPT I) in a manner similar to the architypal peroxisome proliferator Wyeth
14,643. However PFOS also increased the activities of carboxyesterase and CYP2B1, a
response characteristic of phenobarbital inducible systems (Figure 2.7). PFOS exposure
resulted in increases in the activity of thiolase and enoyl-CoA isomerase, enzymes not
increased significantly (p<0.0025) by either of the above xenobiotics. Clofibrate, a
classic peroxisome proliferator, was found to modulate expression of genes involved in
fatty acid B-oxidation in both peroxisome and mitochondria. Finally, while CYP4A is
strongly induced by other peroxisome proliferators it was not increased by exposure to
PFOS in either the in viva or in vitro exposures. Therefore, it seems that while other
PFFAs function mainly through peroxisome proliferation, PFOS results in additional
alterations to gene expression and so may exert its biological effects via other
mechanisms of action. There is evidence of cross talk between peroxisome proliferator
and lipid metabolism pathways, however the exact mechanism for this cross talk is
unclear (Duplus et al., 2000). While there is evidence that free fatty acids are capable of
altering gene expression, the mechanism by which fatty acids can act as signaling
molecules is unknown (Duplus et al., 2000). It has been suggested that liver X receptors
(LXR) are involved in regulation of both fatty acid and sterol metabolism (Tobin et al.,

2002). Indeed the fact that LXR is responsive to fatty acids could provide a clear

57

mechanism for the cross talk observed between cholesterol and fatty acid metabolism
(Duplus et al., 2000). The implication of the LXR receptor in the mode of action of
PFOS would provide plausible explanation for the hypocholesterolaemic effects observed
in primates chronically exposed to PF OS (Seacat et al., 2002a). The possibility of PFOS
acting on more than one metabolic or regulatory pathway is plausible because
commercial mixtures of PFOS contain both straight chain and branched chain

homologues.

There are several possible ways that PFOS could alter peroxisome function. The simplest
explanation could be that, due to the structural similarity of PFOS to endogenous fatty
acids, PFOS could be mistaken by the fatty acid metabolism machinery as a substrate.
However, due to the non-degradable nature of this compound, the B-oxidation process
would fail to oxidize PF OS. To compensate, the major enzymes involved in this pathway
could be induced. However, this hypothesis does not explain the lack of increase in the
more energetically important mitochondrial pathway, which provides most of the cell’s
energy. Another possible explanation of the effects of PFOS on peroxisomal fitnction is
that PFOS alters peroxisomal membrane permeability in a way that allows fatty acid
influx, requiring greater activity of the oxidation enzymes. This mechanism of action
may be less relevant to the mitochondrial pathway since fatty acid entry into

mitochondria is a three-step enzymatic transport process.

It is also possible that the increase in peroxisomal metabolism is a response to partial

uncoupling of the mitochondrial membrane potential resulting in an increase in energy

58

production from peroxisomes. The fact that studies have indicated that PFOS can act as
weak protonophoric uncouplers of mitochondrial respiration (Starkov and Wallace, 2002)
and is able to alter mitochondrial membrane potential (Hu et al., 2002a) support this

hypothesis.

Peroxisomal B—oxidation is a process that generates hydrogen peroxide (H202) that can
cause oxidative stress and oxidative damage to proteins and DNA. While peroxisome [3-
oxidation enzymes were induced up to lO-fold, catalase and glutathione peroxidase, two
of the enzymes involved in detoxification of hydrogen peroxide, were relatively
unchanged. If catalase or GPX were limiting steps in the removal of peroxide this could
result in an increase in hydrogen peroxide, which could induce responses including lipid

peroxidation, membrane damage, accumulation of lipofuscin, and DNA damage.

The one cytosolic enzyme that was dramatically induced (90 fold) by PFOS treatment
was long chain acyl-CoA hydrolase, which cleaves acyl-CoA to free fatty acid and CoA.
The counterparts of acyl-CoA hydrolase found in microsomes and mitochondria are
carboxylesterase and long chain acyl-CoA thioesterase, in viva these enzymes were
induced 5.9 fold and 10.6 fold, respectively. Acyl-CoA hydrolases and related enzymes
are important in the regulation of fatty acid metabolism. They have been proposed to
maintain CoASH pools for both oxidation and synthesis of fatty acids and to regulate the
B-oxidation of fatty acids by controlling the level of acyl-CoA. Induction of acyl-CoA
hydrolase would increase cytosolic free fatty acid concentrations. Rodents and primates,

when exposed to PFOS exhibit hepatocellular hypertrophy and lipid vacuolation, which

59

could be caused by accumulation of free fatty acids (Seacat et al., 2003). Another
significant finding from those studies was a lowered serum total cholesterol level.
Cholesterol production is controlled by HMG CoA reductase, the expression of which
was reduced 2.5-fold in the current gene expression study. This is consistent with
previous studies that suggest that the hypolipemic effect of PFOS may, at least partly, be
mediated via a common mechanism; impaired production of lipoprotein particles due to
reduced synthesis and esterification of cholesterol together with enhanced oxidation of

fatty acids in the liver (Haughom and Spydevold, 1992).

The effects on peroxisome fatty acid B-oxidation does not seem to be receptor mediated,
since PPAR 0t mRNA expression was not affected. This is consistent with previous
studies conducted in our laboratory investigating alterations in expression of PPARor and
y in PFOS exposed fat head minnows (Celius et al., unpublished results). Another
observation that supports this hypothesis is that even though PFOS has been classified as
a peroxisome proliferator, it did not induce P450 4A1, as did most peroxisome
proliferators (Figure 2.7). Two groups of P450s that were up-regulated by PFOS
exposure were the P450 28 and P450 3A families. Both P450 2B1 and P450 232 were
significantly (p<0.0025) induced (9-fold and 22-fold respectively) by exposure to PFOS.
These two cytochrome P450 enzymes are phenobarbital inducible, which is a response
mediated by the constitutive androstane receptor (CAR). It is known that cytochrome
P450 gene expressions are regulated at gene transcriptional level and mRNA processing
and stabilization stages, which are mediated through various nuclear factors and co-

regulators. Each family of the major CYP genes is under control of distinct nuclear factor

60

pathway, however, cross-talk does occur between different pathways through substrate or
co-factor sharing. Therefore the induction of CYP ZB family by PFOS could be explained
as cross-talk between CAR and PPAR mediated pathways. In fact, one classic
peroxisome proliferator, Clofibrate, was found to induce both CYP 4A1 and CYP 2B1

genes.

Gene expression data is useful in identifying affected pathways and possible mechanism
of action, but should not be used to develop dose-response relationship. Furthermore the
degree of toxicity can not be inferred from these results. Once “significant” genes or
pathways have been identified, changes in more toxicologically relevant parameters such
as proteins or substrate should be measured subsequently. Also the comparison between
in vitro and in viva results indicates that while in vitra studies can be used to focus on
response of specific pathway to compounds such as PFOS, the in vitro system is not a
substitute of in viva studies. The greatest utility of the in vitra studies is to determine the
effect of different structured compounds on a specific pathway and their responses once

the critical pathway has been determined.

61

Table 2.1. List of genes induced significantly (p<0.0025) by PFOS in viva exposure.
Long-term values are the mean fold-change and standard deviation (s.d.) for the four

possible control/exposed comparisons.

 

 

Gene ID Long-term Long-term Short-
Mean s.d. term
cytochrome P450 2B15 gene 7.10 0.16 2.6
P450 6beta-2 6.15 0.79 -3.1
cytochrome p-450e 22.57 6.33 6.5
P-450(l) variant 9.09 2.72 5.5
cytochrome P-450b 21.55 5.03 6.7
acyl-COA hydrolase 90.25 36.80 1.7
carboxylesterase precursor 5.89 0.44 - l .l
mitochondrial acyl-CoA thioesterase 10.61 3.88 20.1
delta2-enoyl-COA isomerase 6.02 1.25 2.5
stearyl-COA desaturase 2 mRNA 12.88 2.43 5.3
peroxisomal 3-ketoacyl-COA thiolase 9.78 2.05 1.1
peroxisomal enoyl-CoA-hydrotase/3 -hydroxyacyl- 6.5 1.05 2.3
CoA dehydrogenase bifunctional protein
peroxisomal enoyl hydratase-like protein (PXEL) 5.11 0.34 2.7
aldehyde dehydrogenase (ALDH) 6.04 0.53 1
17-alpha-hydoxylase cytochrome P-450 19.30 0.36 -2.5
neuroendocrine-specific protein (RESP18) 6.44 0.99 2.6
androgen binding protein (ABP) 12.49 0.57 2.6
Tsx gene 3.50 0.63 2.9
testosterone 6-beta-hydroxylase (CYP3A1) 5.19 0.29 1.4
multidrug resistance-associated protein 5.50 2.22 1.6
DNA polymerase alpha 13.54 0.60 2.2
G-protein coupled receptor RAlc 3.96 0.45 2.8
cytochrome P450 PCNl, NADPH mono-oxygenase 7.04 1.93 17.1

 

62

Table 2.2. List of genes suppressed significantly (p<0.0025) by PFOS in vivo exposure.

Long-term values are the mean fold-change and standard deviation (s.d.) for the four

possible control/exposed comparisons.

 

 

Gene Name Long-Term Long Term Short-
Mean s.d. Term
B regulatory subunit of protein phosphatase 2A -5.81 0.79 -l .9
Ca++ independent phospholipase A2 -2.45 0.23 -3
protein tyrosine phosphatase -5.76 2.47 -2.7
postsynaptic protein CRIPT mRNA -3.26 0.46 -1.2
tyrosine kinase p72 -2.79 0.33 -3
phosphorylase kinase catalytic subunit -3.37 0.42 -2.8
rat CELF mRNA -O.51 2.03 -4.6
Na+/K+ ATPase alpha2 subunit -11.14 4.69 -2
liver Na+/Cl- betaine/GABA transporter -3.25 0.13 -1.5
R3109 (brain specific gene) -1.83 1.06 -4.1
synaphin 2 -0.16 2.61 -5.5
skeletal muscle selenoprotein W (SelW) -3. 12 0.52 -1.6
apolipoprotein A-IV mRNA -5.90 3.11 -1.2
peripherin mRNA -2.65 1.79 -2.3
cholesterol 7-alpha-hydroxylase -1 .83 0.97 -4
peptidylarginine deiminase type 111 -1.76 0.83 -11.9
mRNA for RT] 975 4.11 -1.1
MHC class II A-beta RTl -16.39 3.97 -2.8
DNA binding protein (GATA-GT2) -5.03 1.41 -1.2

 

63

Figure 2.1

 

A

Control 1

 

l llllll

 

 

 

 

PFOS l

 

.4.
+ +
1
2 + * +
q
d
+
.. + it:
1 4"” *4-
-1 +
I + + ++
1 + +
q + t+ 4-
-I + .2“. +
+
E + 'I'+
I ‘0’ 4+4.
4 + + +
r +
.. + + + +
++ +
++- +
1
1
.1
I V V 1 "fi' ' ' 'Y'7 W V V ' "Trr‘ Y Y I ' ""1 I' 1 ' MMTj‘
Control 2
II #
+ +1-
5 4' +
? +
a +
-1 +
at
4.
g + + +"*'
-4 .h- i
‘ + 3.7.. +:'
.. + + W
+ + ++ ‘t
.I fi+ ++
1 + + *
d
+ +
1 + + + +
~ +
2 1 + "' + +

a i 1 111111

 

 

 

PFOS 2

64

 

 

Figure 2.1

 

 

 

 

v vvv“

v vvvwv'

v v

v v rvrvrw' fir v vvvtvv'

vvv-I

vvvv

v v vvvvvv'

vrv v rvvvv‘

 

I

 

 

PFOS]

 

Contro 1

 

 

‘_ MA...
' vvv

I I'VVV‘I

A J_JLLAA-- -
'v'V“ V "Y'

1

V

 

 

 

D
In Vivo
Control

In Vitro-Control

65

Figure 2.2

Grouped by gene expression pattern

    
 

H in vitro control

III-I
e IIIIIIIIIIIIIIII I-IIII

       

  

in vitro PFOS at2 mg/L

- IIIIIIIII IIII III—I II “WWI
5:1 .
§_ 5' ‘ I; I I: II iEHENE ”H invivo long-term control
0 , I . I If; .1
III III" IIIIIIIIII III
—o I
-~ IIIIIIII IIIIIII IIIIIIII
. I” III IIIIIIIII IIIIIII III IIIIIIIIII

I I I

 

t
{I I H“ “U in vivo short-term control
vi

1.

II ,.
III III II

I
,1
"i

t

I
IIIIIIIII III
I ‘I

 

I IIIIIIII I invasiontampms

66

Figure 2.3

Grouped by gene expression pattern

 

long-term controll

long-term controlZ

 

 
  

  
   

 

 

 

 

 

g I ll long-term PFOSl
“O : I ~
0 .E‘, _'i' . ;' I
5 - II
: i I i
I III WEI I E . I I short-term PFOS
__ . I .
. - I I: - .I . '

67

I7ignlre:2254

 

 

 

A
honnone
regulator

lipid
metabolism

V

4

CYPs
V

41

cytochrome P450 PCN1. NADPH mono- _ i
oxygenase
DNA polymerase alpha

multidrug resistance-associated protein

testosterone 6-beta-hydroxylase (CYP3A1 )

androgen binding protein (ABP)

neuroendocrine-specific protein (RESP18) y

 

 

 

l
l

l

 

 

 

 

 

 

 

 

 

 

17-alpha-hydoxylase cytochrome P450 é . Short-term
{ I D long- -term :
aldehyde dehydrogenase (ALDH) 1;. ' ‘A '
peroxisomal enoyl hydratase-like protein b 3 ,
(PXEL) i
peroxisomal 3oketoacyl-CoAthiolase L'FH ? ; l
r
stearyl- -CoA desaturase 2 mRNA
detta2-enoy4~CoAisomerase
mitochondrial acyt-CoAthioesterase
carboxylesterase precursor
acyl-CoAhydrolase .fi
cytochrome P-4SOb _=
P-4SO(1 ) variant -_ 1;
cytochrome p-4SOe _.._. l :
P450 6beta-2
I I
cytochrome P45028159ene E , i
0 30 60 90

68

 

 

Figure 2.5

 

 

l

 

 

 

 

 

 

ll short-term l
D “9949'? ‘

___ l

 

9— k

 

 

 

rLIH-TLrlrll

 

 

 

 

minim .i‘i.

i1

 

 

1 r

 

 

DNA binding protein (GATA-GT2)

MHC class II A—beta RT1

mRNA for RT1

peptidylarginine deiminase type III

cholesterol 7-alpha—hydroxylase

peripherin mRNA

apolipoprotein A-IV mRNA

skeletal muscle selenoprotein W (SelW)

 

synaphin 2 A
R8109 brain s ific ene
( pee g ) neurosysten
Ii\er Na+/Cl- betaine/GABA transporter regulation
mRNA
Na+lK+ ATPase alpha2 subunit V
rat CELF mRNA

phosphorylase kinase catalytic subunit
tyrosine kinase p72

postsynaptic protein CRIPT mRNA signal
transduction

protein tyrosine phosphatase

Ca++ independent phospholipase A2

8 regulatory subunit of protein
phosphatase 2A *

 

69

Figure 2.6

fatty acid

CoASH, ATP, Mg2+

fatty acyl-CoA

synthetase .
D9 0 I 09 AMP+PP1
(I .5 fold)
acyl-CoA
membrane

 

 

 

fatty acid
CoASH,

fatty acyl-CoA ATP, Mg2+
synthetase

D90109 (1 .5 fold) AMP+PPi

acyl-CoA
outer memm

Carnitine { camitine
acyltransferase I
D43623 (1 fold) é. CoA-SH

 

 

 

i

acyl-CoA

FAD H
acyI-CoA oxidase
J02 753 (2 fold)
FADH2 O

enoyl-CoA

enoyl- C 0A hydratas
K03249 (6.5 fold)

hydroxyacyl-COA

hydroxyacyl—COA NAD+
dehydrogenase
K03249 (6.5 fold) NADH

ketoacyl-COA

3-Ket0acyl—COA
T hiolase

J02 749 (1 0 fold)

acetyl-CoA + acyl-CoA
shortened by 2 carbons

Peroxisomal Fatty Acid B-Oxidation

7O

Fatty acyl-carnitine

 

 

 

inner membrane
Carnitine r COA'SH
acyltransferase II
202 J054 70 ( I .2 fold) b‘ camitine
acyl-CoA
2 acyI-CoA FAD
dehydrogenases
M35826 (1.1f01d)
FADH? ATP
enoyl-CoA
enoyl-CoA hydratase
D16478 (I .2 fold)

hydroxyacyl-CoA
h -

ydroxyacyl CoA NAD+
dehydrogenase

DI 64 78 (1.2 fold) NADH

ket acyl-CoA

3 -Ket0acyl- C 0A
T hiolase

D164 79 (1.5 fold)

 

acetyl-CoA + acyl-CoA
shortened by 2 carbons

Mitochondrial Fatty Acid B-Oxidation

Figure 2.7

   
  

0 W

2
$0/ v.0 9045

71

CHAPTER THREE

Hu, W.Y., Jones, P.D., DeCoen, W., and Giesy, J.P. (2003b) Comparison of gene
expression methods to identify genes responsive to perfluorooctane sulfonic acid.
Environmental Toxicology and Pharmacology (Submitted)

72

ABSTRACT

Genome-wide expression techniques are being increasingly used to assess the toxic
effects of environmental contaminants. Oligonucleotide or cDNA microarray methods
make possible the screening of large numbers of known sequences for a given species,
while differential display analysis makes possible analysis of the expression of all the
genes from any species. We report a comparison of two currently popular methods for
genome-wide expression analysis in rat hepatoma cells treated with perfluorooctane
sulfonic acid. The two analyses provided 'complimentary’ information. Approximately
5% of the 8800 genes analyzed by the GeneChip array, were altered by a factor of three
or greater. Differential display results were more difficult to interpret since multiple gene
products were present in most gel bands. A probabilistic approach was required in this
analysis. The mechanistic interpretation derived from these two methods was in

agreement, both showed similar alterations in a specific set of genes.

Keywords: PFOS, Gene expression, Differential display, GeneChip array

73

INTRODUCTION

A variety of techniques have been developed to examine alterations in gene expression as
a result of exposure to chemical agents or other stressors (Higuchi et al., 2003; Mong et
al., 2002; Matsuba et al., 1998). Of particular significance are those techniques which
allow examination of the gene modulation in the entire genome. These techniques
represent new approaches in predictive toxicology and allow identification of unknown
modes of action, screening for potential toxicity and grouping of chemicals on the basis
of the mode of action. Each of the techniques used has specific advantages and

disadvantages and, more significantly, produces information of a different nature.

Differential display, first introduced by Liang and Pardee (1992), is a useful tool that
permits the identification of numbers of differentially expressed genes at the whole
genome level. The basic principle of differential display is the separation of total mRNA
into many subsets of approximately four hundred gene products each produced by
choosing appropriate primer pairs for reverse transcriptional polymerase chain reaction
(RT-PCR). Each of the subsets can then be ‘displayed’ using a denaturing
polyacrylamide gel. Differential display is an mRNA “finger-printing” technique that
facilitates the identification of altered functions resulting from changes in mRNA
transcription and/or degradation rates. Differential display has been successfully
employed by many research groups to compare the gene expression patterns in different
organisms and tissues, collected at different developmental stages, in normal and
diseased states, in exposed and unexposed individuals or under different in vitro cell

culture conditions (Liang and Pardee, 1992; Zhang et al., 1993, Douglass et al., 1995;

74

Green et al., 1996; Gao, 1998). In the field of environmental toxicology, differential
display is becoming a useful and powerful too], especially when the mode of action of a

stressor is unknown.

However, the original design of the differential display technique is not without
limitations. Specifically the technique is labor intensive, time consuming and
preferentially amplifies the 3’-ends of transcripts. Therefore, a modified version of the
original technique which is termed restriction fragment differential display PCR (RFDD-
PCR) was developed (www.displaysystems.com). Instead of directly amplifying the
first-strand cDNA, the RF DD-PCR approach adds specially designed adaptors to the ends
of cDNA fragments obtained from Tan digestion. The subsequent PCR amplification
uses a series of designed arbitrary primers selectively annealed to the adaptor junctions.
This approach avoids the problem of 3’-end bias and limits the number of primers
required to screen the complete genome of eukaryotic organisms to 64 pairs (Liang &

Pardee, 1992; Liang et al., 1994; Guimaraes et al., 1995; Liang, 1998).

Differential display works via the systematic amplification of selected sub-populations of
mRNA using arbitrary primers and resolution and visualization of those fragments on
denaturing polyacrylamide sequencing gel. It thus allows for side-by—side comparison of
potentially all expressed genes in a systematic and sequence-dependent manner among
related cells. Since most cellular processes and responses to toxicants are driven by the
temporal and spatial expression of mRN A, differential display is a very useful technique

in examining and comparing mRNA expression profiles under given conditions or

75

treatments. Due to this unique feature, differential display has been utilized in a wide
range of applications including developmental biology, cancer research, neuroscience,
endocrinology and many other fields, including toxicological studies (Higuchi et al.,

2003; Mong etal., 2002; Matsuba et al., 1998).

More recently, methods of screening gene expression based on DNA microarrays have
been developed. In these methods, target cDNA sequences or oligonucleotides are
spotted or synthesized in situ onto a glass slide or ‘array’. Fluorescently labeled cDNA
fragments generated from control and exposed tissues or organisms are hybridized to the
array and differences in gene expression are visualized as differences in the fluorescent
signal intensity of the specific target ‘spots’ or ‘features’. The automation of array
production, hybridization and analysis has lead to the development of arrays with in
excess of 100,000 discrete features per slide. Similarly, automation and analysis software
have greatly reduced the complexity of the data produced. With the use of specific
arrays, it is a relatively simple and fast procedure to assess the expression of over 10,000

genes with considerable statistical rigor.

While the speed and accuracy of microarray methods is generally unquestioned, these
methods are limited to the analysis of gene expression in the specific species for which
the array was designed. In contrast, the differential display technique can be used on any
species in which the investigator is interested. However, the process of analyzing all

possible gene transcripts on polyacrylamide gels is labor intensive. In addition, many of

76

the bands isolated from the gels contain multiple gene products making unequivocal

identification of the altered genes difficult and additionally time consuming.

Of considerable significance to interpretation of data derived from these two methods is
the question of data comparability. Specifically how does sequence specific gene array
data compare to data generated from the entire genome by differential display given that
the ‘pools’ of gene products analyzed are different and the analytical and statistical
methods applied generate distinctly different data. In this study we compared the two
gene expression techniques to determine the degree of ‘comparability’ or ‘agreement’

when the same mRN A samples were analyzed by the two methods.

MATERIALS AND METHODS
Chemicals
Perfluorooctane sulfonic acid (PFOS) used in the in vitro experiments was obtained from
3M (St. Paul, MN). The PFOS (potassium salt) used for in vivo experiments was
purchased from Fluka Chemicals (Switzerland), chemical analysis revealed that it was

essentially the same as the product obtained from 3M.

Cell Culture and Treatment

H4IIE rat hepatoma cells were cultured in 100 mm disposable tissue culture dishes at
37°C under sterile conditions (pH=7.4) in a humidified 5/95% C02/air incubator (Forma
Scientific, Model 8173). Cells were grown in Dulbecco’s Modified Eagle Medium

(DMEM, Sigma D-2902, Sigma, St. Louis MO), supplemented with 10% fetal bovine

77

serum (FBS, Hyclone, Logan, UT). At confluence, cells were removed from the dish with
trypsin/EDTA (Hyclone, Logan, UT), then split into four tissue culture plates. Twenty
four hours after splitting, cells were dosed with PFOS to achieve final concentrations of 2
mg/L and 50 mg/L, methanol was used as solvent control, and the blank control received

no dose. Cells were incubated for 72 hr after exposure.

RNA extraction and purification

Total RNA from cell culture samples was extracted using TriPure Isolation Reagent
(Boehringer Ingelheim, Germany) using the manufacturers recommended procedures.
The optical densities of RNA samples were measured at 260 nm and 280 nm on
spectrophotometer, and the 260/280 ratio was calculated. RNA concentration was
quantified using optical density at 260 nm. The quality of RNA was evaluated by the

appearance of distinct 185 and 28s ribosomal RNA bands on 1% agarose gels.

Differential Display Procedure

Differential display was conducted using reagents and procedures from the Display
System Profile Kit (Buena Vista, CA). Total RNA was used as a template for cDNA
synthesis using reverse Transcriptase and an oligo-dT primer. After synthesis from total
RNA, double-stranded cDNA was digested with Taq I restriction enzyme, which is a 4-
base-cutter that leaves a 5'-overhanging end. Following digestion, two different
specifically constructed DNA adaptors are ligated to the ends of the cDNA fragments.
One of the adaptors, the EP adaptor, has a 5'-overhang and an "extension-protection

group" on the 3'—end, which prevents 3' to 5' synthesis filling in the overhang. Each

78

reaction uses a labeled O-extension 5'-primer that anneals to the ligated EP adaptor, and a
specific displayPROBE 3'-primer that anneals to the junction between the standard
adaptor and the cDNA insert. The extension-protection group on the EP adaptor prevents
amplification of cDNA fragments that have EP adaptors on both ends. Three bases of the
displayPROBE primer extend into the cDNA sequence. It is these three bases that make
the displayPROBE specific for certain cDNA sequences, and also, prevents amplification
of cDNA with standard adaptors on both ends since both ends would need to have the
same three bases to be amplified. To amplify all variations, 43 or 64 different display
probe primers are required, each for a separate PCR reaction that amplifies a different set
of cDNA fragments. For less complex prokaryotic cDNA, the 64 display probe primers
were combined in pairs reducing the number of reactions to 32. Each PCR reaction
amplifies 400 or more fragments, which is referred to as an "expression window." The 64
reactions, used for a eukaryotic sample, produce approximately 25,000 distinct cDNA
fragments. Because of the design, each amplified fragment should be represented in two
different expression windows in the RFDD-PCR analysis (Display Systems manual 2.1,

1999).

RF DD-PCR was conducted using procedures recommended by the Manufacturer
(Display Systems Biotech Inc., Buena Vista CA). After amplification and labeling with

33F, fragments were separated on polyacrylamide sequencing gels. After gel drying,
amplicons were visualized by autoradiography at -80°C over night. Differentially

displayed bands were identified, excised from the gel by overlaying the film on the gel,

79

and DNA was extracted by placing the gel slice in buffer overnight. DNA was then

amplified by PCR with the same primers used for the initial amplification.

Subcloning of PA GE Bands

Direct sequencing of the PCR products extracted from gels was attempted but
demonstrated the presence of more than one amplicon sequence in each gel band.
Cloning of target genes was conducted using pGEM-T easy vector system (Promega,
Madison, WI) with the T-A cloning technique. PCR products were ligated into plasmids
and then transformed into E. coli JM109 competent cells. Positively transformed cells
were selected via blue-white screening, 6 colonies were picked from each plate.
Plasmids were purified using Wizard plus SV minipreps from Promega (Madison, WI).
DNA sequencing was conducted using dideoxynucleotide labeling technique at the
Michigan State University Macromolecular Structure Facility. The sequences obtained
were used to interrogate the Gene Bank database (http://www.ncbi.nlm.nih.gov/) using

the multiple sequence alignment BLOSUM 62 BLAST program (Altschul et al. 1997).

GeneArray Analysis

The RNA samples described above were also analyzed by GeneChip Array technology.
The Affymetrix rat genome U34A gene chip array was purchased from Affymetrix Inc.
(Santa Clara CA). The oligonucleotide probes on U34A array cover approximately 8800
known genes and Expressed Sequence Tags (ESTS) in the rat genome. Transcripts were

selected from Genebank Unigene build 34 and the dBEST database. The methods and

80

results of that study are reported extensively elsewhere (Hu et al 2003a) and will not be

discussed here.

RESULTS

Identify and Subclone Diflerentiallv Expressed Genes

To determine gene expression changes associated with PFOS exposure, H4IIE cells
treated with PFOS at 2 or 50 mg/L were compared with untreated cells and solvent
control treated cells. The differential display technique was able to identify distinct gel
bands, which represented differentially expressed genes (Figure 3.1). Gene products that
were either induced or inhibited by exposure to PFOS could be identified as increase or
decrease of gel band intensities. RT-PCR differential display was conducted, in
duplicate, using 32 of the possible 64 primer combinations from the Display Systems kit.
Theoretically, since each primer set covers approximately 400 amplicons in its expression
window, 32 primer sets should result in a sum of 12,500 mRNA fragments. Since less
than 11,000 mRNAs from the rat genome have been sequenced and functionally
annotated, the 32 primer sets should be able to give a fairly broad, though not complete,
coverage of the rat genome. The number of genes analyzed is comparable to the number

analyzed by Gene Chip analysis (Hu et al., 2003a).

After separation of the reaction products on sequencing gels and examining the gel band
intensities carefully, 55 amplicons were identified that were altered consistently in
duplicates based on comparison between the controls and the treated groups. Of these 55

amplicons, 34 appeared to be affected in a completely present/absent fashion (the bands

81

appeared only in control or treatment group, but not both). The remaining 21 amplicons
were partially increased or decreased (the bands were present in all treatment groups but

with different intensities).

The 55 bands were excised from the sequencing gels and re-amplified using the short
arbitrary primers used in the original amplification step. Direct sequencing of the PCR
products was attempted, but was not successful due to the short length of the amplicons
and the presence of multiple sequences in each band. Therefore, each amplicon was sub-
cloned into a plasmid vector which was then used to transform E. coli host cells. After
sub-cloning amplicon DNAs, 29 of the original gel bands were able to be prepared in

sufficient quality and quantity to permit direct sequencing.

Six bacterial colonies were selected from the transformed E. coli hosts containing
plasmids from each gel band, and the inserted DNAs were sequenced yielding 154
sequences suitable for sequence comparison analysis. There were generally two or more
different sequences present in each gel band (Table 3.1). Clones where 6 out of six
colonies contained the same DNA insert were considered to be unambiguously identified.
Clones for which 4 or 5 colonies contained the same amplicon were considered to be

identified with relatively great certainty (Table 3.1).
After sequencing the putatively differentially expressed amplicons, the sequences were

used to search the Gene Bank database with the BLAST algorithm (Altschul et al., 1997).

The searches were first conducted against all Rattus norvegicus sequences since the

82

exposures were conducted with a rat cell line. If the initial searches did not match known
rat sequences searches were conducted against all mammalian sequences in the data base.
Of the 154 sequences searched, 120 matched rat sequences, 13 matched mouse sequences
and 1 matched a human sequence. Thirteen sequences did not match any known
mammalian sequences in the database. As expected, sequence similarity ratings were
highest for comparison to rat sequences and were lower for comparisons to mouse and
human sequences. Of the six clones isolated from band 5_8, none showed homology to
any known mammalian sequence although all of the six clones contained the same 89 bp
amplicon. When compared to the entire Gene Bank database, the highest degree of
homology for the 5_8 band sequence was to a plant ferredoxin gene but that homology
only extended over a 23 bp region. Since the cells used in this study were of rat origin,
this weak homology probably represented a random coincidence. Of the original 29 gel
bands 12 contained a clear ‘majority sequence’ as indicated by greater than 50% of the

sequenced clones representing one gene product (Table 3.1).

Comparison with GeneChip Results

The results of a GeneChip analysis of the same samples used in the current analysis have
been previously reported (Hu et al., 2003a). Due to the different biases inherent in the
two gene expression profiling methods a comparison of results from the two methods was
conducted. To permit a numerical comparison with the fold induction values determined
in the GeneChip analysis, each gene altered in the differential display data set was
assigned a score ranging from -2 to 2 based on the degree of inhibition (-2 = strong

inhibition of the band, -1 = moderate decrease in band intensity) or induction (2 = strong

83

induction of band, 1 = moderate increase in band intensity) under the different treatment
conditions. There was little correspondence in the estimated degree of alteration of
expression determined by the two methods (Table 3.2). This is not surprising as the
differential display analysis and band selection are based on the subjective assessment of

the intensity of the specific gel bands.

The GeneChip analysis reported that the expression of over 400 genes, from the 8790
genes and ESTs on the array, were significantly altered by treatment with PFOS (Hu et
al., 2003a). Of the genes whose expression was altered 161 increased at 2 mg/L PFOS;
74 decreased at 2 mg/L PFOS; 76 increased at 50 mg/L PFOS and 99 decreased at 50
mg/L PF OS. Of the gene whose expression increased 32 were induced in both treatments
while only 5 were identified as being decreased at both treatment concentrations.
Concordance of results between the two treatment concentrations was higher for the
differential display results. This is not surprising as a common response for the two
adjacent samples would be more likely to be visually identified as a definite alteration in

gene expression when differentially expressed gel bands were being identified.

DISCUSSIONS

PF OS has been shown to alter a variety of properties of the biological membranes (Hu et
al., 2002a, 2002b). It is therefore of interest that the differential display method
demonstrated alterations in the expression of several genes related to membrane structure

and function. Specifically, differential display indicated a decrease in the expression of

84

desmoplakin rabin 3, profilin and guanine nucleotide exchange factor. Desmoplakin are
important proteins in desmosomes which serve as intercellular junctions and attachment
sites for intermediate filaments (Meng et al., 1997). They are thus important in the
maintenance of the intercellular ‘skeleton’ of tissues. Rabin3 is an inhibitor of Rab3A, a
small Ras-like GTPase expressed in neuroendocrine cells where it is associated with
secretory vesicle membranes and controls exocytosis (Brondyk et al., 1995). Although
Rab3A appears to be limited to neuroendocrine cells Rabin3 is expressed in a wide range
of tissues (Brondyk et al., 1995) where it presumably interacts with proteins whose
fimction is homologous to Rab3A. Guanine nucleotide exchange factors are also
involved in the formation of the cell skeleton and in membrane trafficking in various
ways particularly in regulating the activity of Rab and Ras type proteins (Ganesan et al.,

1999)

While PFOS does have many effects at the level of the cell membrane it has also been
demonstrated to alter the expression of a number of genes involved in lipid metabolism
(Hu et al., 2003a) and has been shown to alter the lipid status in rodents and primates in
vivo (Seacat et al., 2002a, 2003). Genechip analysis of the same samples used in the
current study indicated that PFOS specifically induced the peroxisomal fatty acid
oxidation pathway but not the same enzyme systems in the mitochondria (Hu et al.,
2003a). Interestingly, the differential display and gene chip analyses demonstrated a
decrease in the expression of complex I NADqubiquinone oxidoreductase. This
complex is the first membrane bound electron transport complex of the mitochondrial

respiratory chain and so accounts for up to 40% of the proton-translocating capacity of

85

the respiratory chain. Loss of activity in this proton-translocating complex could result in
a need to increase peroxisomal fatty acid oxidation as observed or, alternatively,
increased peroxisomal fatty acid oxidation could result in a lower energy demand on the
mitochondria and so result in a down regulation of the mitochondrial electron transport
chain. Complex I has been shown to be susceptible to hydrophobic inhibitors (Okun er
al., 1999) and the highly electronegative nature of the PFOS molecule suggests that it
would have an propensity to modulate electron transport and translocation. Interference
with complex I could also explain the observation that PFOS is a weak non-ionophoric

mitochondrial uncoupler (Starkov and Wallace 2002).

Both methods of expression analysis used have inherent biases. For example, the need to
visually identify gel bands whose intensity is different across treatments biases the
differential display data set towards those genes whose expression is altered the greatest.
In contrast the Genechip data analysis treats all gene expression levels equally. However,
the genes represented on the GeneChip are strictly limited to those which have been
isolated and identified and sequenced by other researchers. In contrast the differential
display method is capable of identifying any gene product whose expression is altered.
Therefore the data sets generated by these two methods tend to be diametrically opposed.
One data set (GeneChip) is biased towards sequences of toxicological interest while the
other (differential display) attempts to identify every possible gene but is mechanistically

biased towards the greatest alterations in expression level.

86

The differential display method relies on the ability to separate and visually identify gel
bands whose intensity varied with different treatments. As such this procedure is open to
interpretation by the individual investigator. Therefore the critical step in this process is
one of observation by the individual conducting the analysis. As such, this method is
limited by the resolving power of the PAGE system used and by ability of the human eye
to detect ‘significant’ changes in expression profiles. After determination and excision of
bands which are differentially expressed, a series of procedures are required to
unequivocally determine which of the possible multiple cDNAs present in the band is the
gene product which is differentially expressed. This is achieved by amplifying and
subcloning the sequences in the band such that only a single sequence is inserted into
each bacterial host. In the simplest case where all bacterial cells contain the same insert
we can assume that the original gel band contains only a single amplicon. In those cases
where the majority of the amplicons represent one gene product it can also be safely
assumed that the most abundant amplicon represents the differentially expressed gene. In
cases where several amplicons are present in similar proportions it is not possible to

identify the gene product which is differentially expressed.

In summary the differential display method is labor intensive and requires application of
considerable professional judgment. This makes the method less than ideal for rapid
screening and assessment of large numbers of stressors. However, the method is of use
in determining critical modes of action and is an ‘open platform’ method suitable for use

in species and tissues where significant DNA sequence information is not available.

87

The differential display method was able to detect alterations in expression of genes not
present on the microarray used. In some instances, for example desmoplakin, the
differential display method also allowed a tentative identification of the function of the
product as it was homologous to genes in other species. In the case of one gene identified
by differential display, the sequence was not homologous to any known mammalian
genes, this product would be an obvious candidate for further investigation to determine

the nature and function of the gene product.

In contrast to the differential display method, the gene array technique provides a
statistically robust identification of each amplicon and the degree to which it is altered
relative to the control conditions. Essentially all the amplicon identification procedures

are carried out during the preparation of the array.

Based on the quality of the data including the sensitivity, specificity and reproducibility
of the results, gene chip arrays appear to provide a greater wealth of information than
differential display (Yuen et al., 2002). Because of the current need for visual detection
of differences in gel band patterns differential display is limited to only abundantly
expressed messenger RNAs. Theoretically, separation of RNA fragments on a well-cast
polyacrylamide sequencing gel can identify one single nucleotide difference, in practice,
however, this is rarely the case. The separation of gene fragments is limited by several
factors, including the quality of the denatured polyacrylamide gel and the means of
detection. In this study we used 33P—labeled NTP to assist visualization, and differentially

expressed genes were identified by visual examination. Therefore, the results are not

88

quantitative. In contrast, the gene chip technique utilizes fluorescent dye labeling and
computer image analysis, which makes gene chip results specific and quantitative.
Another limitation of differential display method as discussed previously is the relatively
great incidence of false positives. The ambiguity in determining the gene identities using
similarity search of sequence databases add another level of complication. Whereas, the
gene chip array has well-defined gene identities, and using the specially designed
perfect/miss match strategy serving as internal control allows for unambiguous
identification of significant alterations. Even with these limitations, over-all there was
still a general agreement between the results obtained by use of the differential display
and gene chip methods. It is somewhat naive to assume that the two techniques would
provide identical information. Rather they provide complimentary information, this
explains the difficulty encountered in trying to obtain a direct comparison of the results.
The gene chip technology provides precise definition of relatively small changes in gene
expression in a biased sample of genes present in the genome. In contrast the visual
identification in the differential display analysis provides evidence of large changes in
specific gene products from the entire genome but gene identification is equivocal and
relies on probabilistic approaches to identification. It is probably most significant that
this comparison of the two techniques provides evidence of a general agreement in
mechanistic interpretation rather than a direct comparison. In general genes whose

expression was apparently altered in one technique was similarly altered in the other.

Both gene expression techniques clearly deliver a wealth of information. In the case of

the gene array the amplification process and data reduction software delivers higher

89

'quality’ information. In essence a great deal of the probabilistic interpretation of the data
is done before the data reaches the researcher. In contrast the differential display method
requires that the researcher filters data to remove false positives and interpret sequence
matches to identify genes of interest. In the gene array technique the gene identity is
known exactly for each probe set on the array. These basic differences in the two

methodologies result in data that should be seen as complimentary.

As to economic and practical consideration, differential display is less expensive than
gene chips. All the equipment and reagents needed for differential display technique can
be easily obtained in most molecular biology laboratories. On the other hand, gene chips
are expensive themselves and the technique requires a specially designed work station,
and scanner, furthermore the reagents used for labeling cRNA and the software used for
analyzing the results are also expensive. However, when we look at the cost efficiency
aspects, differential display is relatively more time and labor intensive. The choice of
which technique is more appropriate will be based on considerations of the research

question at hand.

90

Table 3.1. Genes identified to be differentially expressed due to PFOS exposure in
H4IIE cells in vitro. “Clone seqs” indicates the number of clones sequenced from each

original gel band. The “majority sequence”, if present, is the sequence that was present

in 50% or more of the sequenced clones. The “Proportion Majority” is the number of

clones out of the total that were the majority sequence.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Gel clone majority Proportion Name
Band seqs sequence Majority
l_l 6 None
l_10 3 None
l_ll 6 None
l_12 6 NM_017313 6/6 rabin 3 (RABIN3)
l_2 7 XM_225259 4/7 similar to (Desmoplakin I (DPI))
l_3 8 None
Mk1 protein (Mk1), mRNA
l_4 6 NM_134399 3/6 (homologous to profilin)
similar to RIKEN cDNA
l_5 5 XM_213242 4/5 0610011B04
Ribosomal DNA external
l_7 6 X16321 4/6 transcribed spacer 1 (ETSI)
4_2 6 None
4_3 3 None
5_1 3 None
mitogen-activated protein kinase-
5_lO 6 AY197741 3/6 activated protein kinase2
5_11 6 None
5_12 3 None
5_13 10 None
5_2 3 None
5_3 6 None
5_4 6 None
5_5 6 None
5_7 7 NM_080907 5/7 protein phosphatase 4
5_8 6 No mammalian equivalent
5_9 6 AF 476964 5/6 Transferring-like mRNA
6_2 5 None
6_3 6 NM_053556 6/6 maternal G10 transcript (G10
similar to mouse guanine nucleotide-
7_l 4 XM_225548 2/4 exchange factor (LOC307098)
7 ;2 2 NM_017245 2/2 eukaryotic elongation factor 2
7_4 6 V01270 3/6 Ribosomal 188, 5.88, 288

 

91

 

Table 3.2. Alterations in gene expression (as fold change from control) assessed by
differential display and Genechip analysis. “dd” and “go” refer to differential display or

GeneChip analysis at 2 or 50 mg/L as indicated.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Band majority sequence (5d (513 gzc g3 Gene

l_12 NM_017313 -1 -1 0 -1.2 Rabin3

l_2 XM_225259 0 -l desmoplakin

l_4 NM_134399 -l -l -1.1 -1.1 profilin

l_5 XM_213242 0 -1 NADqubiquinone oxidoreductase
l_7 X16321 O 1 ribosomal ETSl

5_10 AY197741 -l -1 1.1 1.3 protein protein kinase 2

5_7 NM_080907 -1 -1 protein phosphatase 4

5_9 AF47 6964 1 1 1.8 -l .9 transferrin like

6;3 NM_053556 1 2 -1.1 -l.l G10 protein

7_1 XM_225548 0 -1 -1.1 -1.2 guanine nucleotide exchange factor
7_2 NM_017245 0 -l 1.5 1.4 elongation factor

7_4 V01270 -1 -2 1 -l .2 ribosomal genes

 

92

Figure 3.1

 

 

 

 

 

BC SC P2 P50

 

 

 

BC SC P2 P50

 

 

93

 

 

 

CONCLUSION

The results from gene expression analysis and cell bioassays can be summarized as

follows:

Overall genome-wide gene expression analysis revealed the modulation of several
critical metabolic and regulatory pathways by PFOS exposure. Genes that were
significantly up regulated by PFOS were involved in fatty acid metabolism, drug and
xenobiotic metabolism, and hormone regulation. Down regulated genes were mostly
coding for key factors involved in signal transduction pathways and the regulation of
neural system function.

Among the pathways regulated by PFOS exposure in the GeneChip experiment, fatty
acid B-oxidation appeared to be the major pathway affected. Further interpretation of
the data revealed that most key enzymes involved this pathway were consistently and
significantly induced in the peroxisomes but not in the mitochondria.

GeneChip expression analysis also indicated that distinct patterns of gene regulation
were resulted from the in vitro vs. in vivo exposure. However, when gene expression
profiles derived from long-term and short-terrn in vivo exposure were compared, a
much greater level of consistency in gene regulation was observed.

Comparison of differential display and gene chip technique based on PFOS in vitro
treatment results indicated that the mechanistic interpretation derived from these two
methods was in general agreement, although the list of significantly modulated genes
generated from the two methods was more complimentary than directly comparative.
A contrast of the advantages and disadvantages with differential display and gene

chip was listed in table C. l.

94

0 Results from cell bioassays using model ligands in combination of PFOS indicated
that PF OS increased the membrane permeability selectively to those ligands tested.

0 Membrane related bioassays using fluorescent probes and flowcytometer revealed
that PFOS treatment significantly increased the membrane fluidity and decreased the
mitochondrial membrane potential in a dose dependent fashion. PFHS and PFBS,
with shorter carbon chain length, showed no effects on those membrane-associated

properties within the concentration range tested.

Results from the current study provide useful information on various aspects of the
molecular and biochemical toxicity of PFOS. A structure activity relationship can be
derived with the evidence provided in this study in combination with previous findings in
regarding to PFFAs. Fatty acids with different chain lengths have different cellular
functions and are metabolized by different systems in the cell. Fatty acids can be
incorporated into cellular membrane phopholipids or can be utilized to synthesize other
lipids or steroids. Fatty acids can also be broken down for energy generation. Some fatty
acids are known to affect protein kinase pathways, thus modulating cell signaling
cascades. Although PFOS has a similar structure with endogenous fatty acids, short chain
fatty acids are generally not present in normal adult rat tissues. PFOS is therefore
‘foreign’ to the cell and the sulfonic acid functional group is relatively non-reactive
making PFOS metabolically non-degradable. Therefore, PFOS should exhibit membrane-
related effects as well as alter fatty acid metabolism. Results fiom the current study
support this prediction. PF OS was proven to affect membrane-associated properties, such

as membrane permeability, plasma membrane fluidity, mitochondria membrane potential,

95

and previously reported gap junctional intercellular communication. PFOS treatment
also up regulated fatty acid B-oxidation, altered hormone and cholesterol synthesis, and
down regulated signal transduction pathways. Furthermore, studies conducted on PFFAs
with different chain lengths, including PFOS, PFHS and PFBS, indicated at least some of

the observed effects were associated with the molecules chain length.

Results from this study confirmed and elaborated on the gene expression modulation by
PFOS in the rat. The results also supported using toxicogenomic evaluation for the
detection of molecular mechanisms of toxicity. One issue of concern with genome wide
gene expression analysis is whether it is feasible to extrapolate these mRNA level effects
to protein expression and functional responses at the cellular or whole animal level. This
has been a point of discussion within the field of toxicogenomics. Predicting protein or
cellular level function and toxicity from alterations in gene expression profiles must be
done with caution. One should not assume a direct correlation between RNA and protein
concentration or enzyme activity without further testing. Gene expression profiles exhibit
strong temporal variation, and are complicated with numerous factors, such as genetic
background, age, gender, nutrient conditions, as well as social behavior. Furthermore,
modulation at the post-translational level, such as protein phosphorylation or changes in
protein 3-D conformations, would never be predicted based solely on genomic data.
Since changes at the transcriptional level occur rapidly after exposure, and respond
sensitively to relative low level treatment, gene expression analysis should be treated as
an early-stage screening tool, which can provide hypotheses and predictive models for

further investigation. Especially with the high throughput nature of this technology,

96

genome wide gene expression analysis provides a theoretically unlimited source of
information, which keeps growing with the increasing amount of genomic sequence
annotation. When coupled with other “omics” technology, such as proteomics and
metabonomics, genomics technology, including the two examples discussed in this study,
will become much more powerful in applications such as predictive toxicology, drug

discovery and development, biomarker identification, and risk assessment.

Table C 1. Comparison between differential display and genechip techniques

 

 

 

Differential display GeneChips
'Open Platform - Very sensitive and reproducible
Advantages 0No need for sequence infor. - Gene specific
-Not lrmrted by specres - Quantitative
-Relatively cheaper
- Labor intensive -Close Platform
Disadvantages - Time consuming -Less flexible . .
- Frequent “false- ~Need sequence rnforrnatron
positive” -Only available for several species
0 not quantitative ~More expensive

 

97

APPENDICES

METHODS AND PROTOCOLS

98

A. RNA Extraction and Purification

. Cell culture medium was aspirated and cells were washed with 5 ml PBS twice. 1.5
ml tri-reagent was added into each cell culture dish, and allowed to immerse the cells
for 3 min. A sterile policeman was used to scrape the cells off the bottom of the plate,
and cell suspension was then passed through a pipette tip several times before it was
transferred to 2 m1 RNase free Eppendorf tube.

Rat liver samples were rinsed with PBS solution, then cut into approximately 100 mg
per piece and transferred to a typhoon tube. 1.5 ml of tri-reagent was added into each
tube, and the liver samples were homogenized. The tissue homogenate was then
transferred to 2 ml RNAse free Eppendorf tube.

. The cell and tissue lysates were incubated for 5 min at room temperature to ensure
complete dissociation of nucleoprotein complexes. 150 pl of chloroform was added to
each tube. The tubes were shaken vigorously for 15 sec and allowed to stand at room
temperature for 15 min.

. Samples were centrifuged at 12,000 x g for 15 min at 4°C. Three layers were formed
in the tube: the upper colorless aqueous phase contains RNA, the interphase and
lower red organic phase contain protein and DNA. The upper phase was carefully
transferred to a new Eppendorf tube.

. 400 pl of isopropanol was added into each tube, and mixed thoroughly by inversion.
The tube was incubated for 10 min at —80°C to allow RNA precipitation.

. Samples were spun at 12,000 x g for 10 min at 4°C. The supernatant was discarded,

and 750 pl 75% Ethanol was added in the tube, and the pellet was washed by

vortexing in Ethanol. Samples were then centrifuged at 7500 x g for 5 min at 4°C,

99

and the supernatant was discarded. Excess ethanol was removed fi'om the RNA pellet
under vacuum.

. The RNA pellet was re-suspended with 20 pl DEPC treated water, and dissolved
completely by incubating in a 55°C-60°C water-bath for 15 min.

. The optical density of RNA samples were measured at 260 nm and 280 nm on
spectrophotometer, and the 260/280 ratio was calculated. This value for pure RNA
sample should be as close to 2 as possible. RNA concentration was estimated using
the OD260 value based on the observation that OD260 of l is equivalent to an RNA
concentration of 40 pg/ml.

. The quality of RNA samples was tested on a 1% agarose gel. Two distinct bands at
188 and 28s appeared after electrophoresis, verifying for pure and undegraded

samples.

100

B. Restriction Fragment Differential Display Polymerase Chain Reaction (RFDD-PCR)

Outline

Cell culture (H4IIE)

l
Exposure to PF OS (for 72 hrs)

Total RNA extraction

Jr

RT reaction and template preparation

l
PCR amplification

l
Separation on sequencing gel
Isolation of differentially expressed gene
l
Re-amplification of gene fragments of interest
Subcloning of isolated genes
l
Sequencing
Determination of the identity and bio-significance of the genes

l

Confirming results

Differential display experiment was conducted using kit from display systems, which is

part of Azign Bioscience (Copenhagen, Denmark).

RT reaction and template preparation

1. Based on the concentration of RNA samples determined using spectrophotometer,
dilute appropriate volume of RNA sample in DEPC-water to make up the final
concentration of 1 pg total RNA per tube.

2. Set up the first strand cDNA synthesis reaction as following:

Total RNA (1.0 pg) 10.0 pl

101

Anchored primer 1.5 pl

10 x cDNA buffer] 2.5 pl

dNTP mix 5.0 pl

display Therrno-RT 1.0 pl

sterile water 5.0 pl
Incubate at 42 °C for 2 hr.

3. Prepare the second strand synthesis reaction mixture as following:

First-strand mixture 25 pl
10 x cDNA buffer 2 7.5 pl
dNTP mix 2.5 pl
DNA polymerase I 1.2 pl
RNAse H 0.8 pl
Sterile water 38.0 pl

Incubate at 16 °C for 2 hr.

4. Perform a phenol/chloroform extraction by adding to each 75 pl reaction mixture 125
pl water and 200 pl phenolzchloroform, and vortexing, then centrifuge for 5 min at
12,000 x g at 4°C.

5. Transfer the aqueous phase to a 1.5 ml eppendorf tube containing 20 pl 3 M sodium
acetate (pH 5.2) and 400 pl 96% ethanol, and precipitate at -80°C overnight.

6. Centrifuge the precipitated samples at 12,000 x g for 15 min at 4°C. Discard
supernatant, and wash the pellet in 50 pl 70% ethanol. After centrifugation at 12,000
x g for 5 min at 4 °C, air-dry the pellet, and dissolve it in 20 pl sterile water.

7. Confirm the efficiency of cDNA synthesis by running 10 pl of the re-suspended
cDNA on a 0.8% agarose gel. A cDNA smear should be visible from approximately
100 to 200 base pair.

8. Prepare template by a endonuclease digestion reaction followed by adaptor ligation

reaction:

102

10 x display PROFILE buffer 2.0 pl

cDNA sample 10.0 pl
Taq I restriction enzyme 0.5 p]
Sterile water 7.5 pl

After incubation at 65 °C for 2 hr, add the following:

10 x displayPROFILE buffer 0.75 pl
adaptor mix 0.75 pl
ATP 1.25 pl
T4 DNA ligase 0.3 pl
Sterile water 4.45 pl

Incubated at 37 °C for 3 hr.

9. Confirm the quality of template by running control PCR reaction as following:

DisplayTAQ FL 10 x reaction buffer 2.0 pl
dNTP mix 0.8 pl
control primer 8.0 pl
display TAQ FL 0.3 pl
template 0.2 pl
sterile water 8.7 pl

Perform PCR reaction for 30 cycles using the following condition:
94 °C, 30 sec
55°C, 30 sec
72°C, 1 min
When running 10 pl reaction mix on a 1.5 % agarose gel, a DNA smear from

approximately 50 to 1000 base pair should be visible.

PCR amplification

1. End-labeling for radioactive reaction

Set up reaction mix as following:

10 x display PROFILE buffer 0.10 pl
O-extension primer 0.40 pl
[y33P]-ATP 0.20 pl
T4 polynucleotide kinase 0.02 pl

103

Sterile water 0.28 pl
Incubate at 37 °C for 30 min.
2. Amplification of template

Set up PCR reaction mix as following:

DisplayTAQ FL 10 x reaction buffer 2.0 pl
dNTP mix 0.8 pl
labeled primer 1.0 pl
display TAQ FL 0.3 pl
display PROBE 4.0 pl
template 0.2 pl
sterile water 11.7 pl

Amplify each template using the following PCR profile:

94°C, 1 min
the first 10 cycles
94°C, 30 sec

60°C, 30 sec (reduce by 05°C for each cycle until 55°C is reached)

72°C, 1 min
continue with another 30 cycles
94°C, 30 sec
55°C, 30 sec
72°C, 1 min

Expression on the sequencing gel
1. Casting 5 % poly acrylamide sequencing gel

1.) make up 10 x TBE buffer as following:

Tris base 108 g
Boric acid 55 g
EDTA 97g

Add distilled water up to 1 liter, autoclave, pH 8.3

2.) make up gel solution as following:

Urea 63 g
10 x TBE buffer 15 ml
30 % acrylamide/bis 25 ml

add distilled water up to 150 ml

104

dissolve with low heat, filter sterilized, de-gas for 15 min.

3.) Case gel using sequensing gel unit from Biorad (Hercules, CA). Wash sequencing
gel plates thoroughly using soap and hot water, rinse with distilled water followed
by spray with ethanol.

4.) Assemble gel cell and holders following the instruction on the manual.

5.) After de-gas, add 165 p1 TEMED and 165 pl 25 % ammonium persulfate, mix
gently. Fill up the big syringe with gel solution, carefully get rid of air bubbles.
Insert the tip of the tube connected to the syringe to the small hole at the bottom
of the gel cell, push syringe slowly. Take extreme care while casting the gel, no
air bubble should form between the plates. Insert the comb on top of the gel, with
the flat edge down. Hold the top of the gel with four clamps, let the gel stand for
about 1 hr to become solid.

. After gel is cast, take out the comb, revert it and insert back, this forms 49 wells on

the top of the gel. Pre-run the gel with 1 x TBE buffer at 75 W for 30 min. Loading

buffer can be loaded in pre-run to check the quality of the gel.

. Take out 4 p1 of PCR reaction, add 3 pl loading buffer to each tube, incubate at 85 °C

for 5 min. Load 5 pl of sample to each well on the gel. Run the sequencing gel at 80

W for 3 hr.

. After run-off, discard running buffer, carefully separate the plate, use filter paper to

pick up the gel, put pre-wet membrane on top of the gel. Dry in the gel dryer for 3 hr.

. When the gel is completely dry, in the dark room, put the gel in the exposure cassette.

Mark one comer of the gel with alignment marker. Fit film on top, expose for 48 hr.

105

Put exposure cassette in —80°C for best exposure result. Develop film after 2 days

exposure.

Isolating the differentially expressed gene

Compare the DNA profile among samples, especially the PFOS treated ones and the
solvent control. Identify differentially expressed gene, both the density of the band and
the presence of specific band. Overlay the film on top of the gel, pin at the band of
interest. Use steriled razor blade cut off the both the band on the film and the underlying
gel, and re-dissolve it in 50 pl sterile water. Heat up samples at 95°C for 15 min, and spin

down for 10min at 14,000 x g. Transfer the supernatant to a fresh microcentrifuge tube.

Re-amplification of gene fragments of interest

To re-amplify the gene fragments of interest, set up the PCR reactions as the following:

display 10X reaction buffer 4.0 pl
dNTP mix (SmM each) 1.6 pl
O-extension primer (10 pM) 0.8 pl
display probe (1 pM) 8.0 pl
gene fragment solution 5.0 pl
display TAQ 0.6 pl
sterile water 20 pl

Amplify the reaction with 30 cycles of PCR using the following profile:

94°C, 30 sec for strand separation;
55°C, 30 sec for anneal;
72°C, 1 min for amplification.

Separate PCR products on 1 % low melting agarose gel at 75 V for 45 min. Cut out gel
slices containing single band, and extract DNA using Ultrafree-DA devices from Amicon
(Charlotte, NC). Place gel slices into the pro-assembled filtering unit and spin at 5,000 xg

for 10 min at room temperature. Store DNA samples at —20°C.

106

Subclaning of isolated genes

Due to the short length of the differential display probes, subcloning of isolated DNA

fragments is necessary before they can be subjected to sequencing. Cloning of target

genes is conducted using pGEM-T easy vector system with T-A cloning technique from

Invitrogen (Carlsbad, CA). The pGEM-T easy vector is constructed by cutting normal

vector using EcoR-V restriction enzyme and adding a 3’ terminal thymidine to both ends.

These single 3’-T overhangs at the insertion site greatly improve the efficiency of ligation

of a PCR product into the plasmid, since PCR product generated by Taq DNA

polymerase often have a single 3’-A overhangs in a template-independent fashion.

Purify DNA fragments isolated from sequencing gel using Qiagen RNeasy kit (Valencia,

CA), and concentrate for subsequent ligation into plasmid. Set up the ligation reactions

along with appropriate controls in the following way:

 

 

 

 

 

 

 

 

Target reaction Positive control Background control
2X ligation buffer 5 pl 5 pl 5 pl
PGEM-T easy vector 1 pl 1 pl 1 pl
PCR product X pl 0 pl 0 p1
Control insert DNA 0 pl 2 pl 0 pl
T4 DNA ligase 1 pl 1 pl 1 pl
Distilled water (lO-x) pl 1 pl 3 pl

 

 

 

 

 

Mix the reaction components by pipetting and incubate for 1hr at room temperature.

107

*Positive control of insert DNA is used to determine ligation efficiency, typically 100
colonies should be observed. Background control with no insert DNA is used to
determine the number of background blue colonies resulting from undigested vectors.
Transformation control with uncut plasmid was used to check the transformation

efficiency with competent cells.

Transformation is conducted using JM109 high efficiency competent cells from Promega
(Madison, WI). Thaw one vial of competent cell on ice for 5 min, and label four 5 ml
round bottom culture tubes. Add 2 pl of each ligation product, 50 pl of competent cell
suspension to each tube, and mix gently. Incubate on ice for 20 min and heat shock on
42°C heat block for exact 45 sec. Transfer to ice and add 950 pl SOC medium, shake for
1.5 hr in 37°C water-bath at about 150 RPM. During the meanwhile, take out 8 LB plate,
on each plate add 100 pl IPTG and 20 pl X-Gal solution, spread over using metal
policeman (rinse with ethanol and burn on fire, cool down before each use), and warm up
the plate to room temperature. After incubation, add 100 pl of each transformation

solution to the plate, spread with metal policeman, incubate at 37°C over night.

Successful cloning of an insert into the pGEM-T vector interrupts the coding sequence of
B-galactosidase, so positive recombinant clones can be identified with color screening.
After overnight incubation, plates containing positive control should show a great amount
of white colonies, and very few blue colonies for background control. Pick 6 white
colonies from target plate using sterilized tooth pick and insert it into 5 ml culture tube
containing 2 ml LB medium. Incubate the tube in 37°C water-bath for 12-15 hr, while

shaking at 125 RPM.

108

After overnight incubation, competent cells grow to an intensity by which tooth-pick
could not be seen through the cell suspension. Transfer cell suspension to 2 ml
centrifugation tube and purify plasmid using Wizard plus SV minipreps from Promega
(Madison, WI). Harvest 2 ml of bacterial culture by centrifugation for 5 min at 10,000 x
g. Discard the supernatant and blot the inverted tube on a paper towel to remove excess
media. Add 250 pl of cell suspension solution and completely resuspend cell pellet by
vortexing. Add 250 pl of cell lysis solution and mix by inverting the tube four times, and
incubate for 5 min. Add 10 pl of alkaline protease solution and mix by inverting the tube
four times. Incubate for 5 min at room temperature. Add 350 pl of neutralization solution
and immediately mix by inverting the tube four times. Centrifuge the cell lysate at 14,000
x g for 10 min at room temperature. Transfer the clear lysate to the spin column, avoid
disturbing or transferring any precipitate. Centrifuge the supernatant at 14,000 x g for 1
min at room temperature. Remove the spin column and discard the flow through from the
collecting tube. Add 750 pl of wash solution, previously diluted with 95 % ethanol to
spin column. Spin at 14,000 x g for l min, and discard flow through. Repeat the wash
procedure using 250 pl wash solution. Centrifuge at 14,000 x g for 2 min. Transfer the
spin column to a fresh microcentrifuge tube and elute the plasmid DNA by adding 100 pl
of nuclease-free water to spin column. Let it sit for 10 min, then centrifuge at 14,000 x g

for 1 min. Store sample at -20°C.

Reagents used for cloning:

109

LB medium: in a 2 L flask add Bacto-trypton 10 g, Bacto-yeast extract 5g, NaCl 5 g, 800
ml distilled water, adjust pH to 7.0 with NaOH and add volume up to l L. Sterilize with
autoclaving.

LB plate: add 15 g agar to l L LB medium and autoclave. Allow the medium to
cooldown to 50°C then add 1 ml of ampicillin stock solution and mix well. Pour
approximately 25 ml of the medium into eath petri-dish, let it harden. Store at 4°C.
Ampicillin stock: add 1 g ampicillin in 10 ml distilled water to make 100 mg/ml stock,
split into 10 tubes with 1 ml in each, and store at -20°C.

IPTG stock: add 1.2 g IPTG in 50 ml distilled water, filter sterilize, and store at 4°C.
X-Gal stock: dissolve 100 mg X-Gal in 2 ml dimethyl-formamide, and cover the bottle
with aluminum foil, store at -20°C.

SOC medium: add 2 g bacto-tryptone, 0.5 g bacto-yeast extract, 1 m1 1 M NaCl, and
0.25 ml 1 M KCl in 90 ml distilled water, autoclave and cool down to room temperature.
Add 1 ml 2 M megnisium stock (10.17 g Mng and 12.33 g MgSO4 in 50 ml water)

andlml 2 M glucose, bring volume to 100 ml, and filter sterilize.

Sequencing and Sequence comparison

Transfer 10 pl of each cloning sample (concentrate to 50 ng/pl) to 96 round bottom well
plate, and submit the plate to Genomics facility on MSU campus for sequence analysis.
DNA sequencing is conducted through dideoxynucleotide labeling technique using the
ABI 3730 Genetic Analyzer from Applied Biosystems (Foster City, CA). Obtained
sequences can then be submitted to Genebank (th://ncbi.nlm.nih.gov) for sequence

comparison. Since different gene sequences with the same length could co-migrate to the

110

same position, typically 2-3 different sequences could be obtained from each of the 6
white colonies picked from original plate. Multiple sequence alignment BLAST program
using BLOSUM62 was used to make sequence comparison, the top 10 significant hits

were examined, and sequence identity determined.

lll

C. Affymetrix GeneChip® Array

Overview of GeneChips method
Animal Exposure

l

Total RNA extraction

l
cDNA synthesis

l
Biotin-labeled cRNA synthesis

Target-grobe hybridization
Probe array washing and staining
Probe array scan

l
Data analysis
A flfvmetrix GeneChip Array build-up
GeneChip probe arrays are manufactured using technology that combines
photolithographic methods and combinational chemistry. A glass substrate is coated with
linkers containing photo liable-protecting groups. Then, a mask is applied on top of glass
surface so that only selected portions of the probe array are exposed to ultraviolet light.
Illumination removes the protecting groups enabling selective nucleoside to be added on
the end of linker compound by chemical coupling reaction. Next a different mask is
applied and the cycle of illumination and chemical coupling is performed again. By
repeating this process, a specific set of oligonucleotide probes is synthesized at known

positions. This 1 cm2 glass substrate is then mounted in a plastic cartridge, which then

serves as hybridization chamber as well as protecting shield.

112

cDNA synthesis

Synthesis of cDNA from total RNA was performed using the SuperScript Choice System
obtained from Gibco BRL life Technologies (Cheshire, UK). The amount of 16 pg high
quality total RNA dissolved in 10 pl DEPC water was incubated at 70°C with 1 pl 100
pmol/pl T7-(dT)24 primer for 10 min. T7-(dT)24 primer (5’ — GGC CAG TGA ATT GTA
ATA CGA CTC ACT ATA GGG AGG CGG — (dT)24 —3’) was obtained from Genset
Corp. (Boulder CO). 4 pl of the 5 X first strand cDNA buffer, 2 pl of 0.1 M DTT and 10
mM dNTP mixture were added to the tube, and allowed to equilibrate to 42°C for 2 min,
before a 2 pl of S811 reverse transcriptase was added in the tube. The first strand cDNA
synthesis reaction took 1 hr at 42°C. At the end of reaction the samples were placed on

ice and centrifuged briefly. The following reagent were added to each tube:

DEPC-treated water 91 pl
5 X second strand reaction buffer 30 pl
10 mM dNTP mix 3 pl
10 U/pl DNA ligase 1 pl
10 U/ pl DNA polymerase I 4 pl
2 U/ pl Rnase H 1 pl

The tube was gently tapped to mix and incubated at 16°C for 2 hr. At the end of
incubation, 2 pl T4 DNA polymerase was added and incubation went on for another
5mins. A 10 pl 0.5 M EDTA was added to the reaction mixture. Double stranded cDNA
was cleaned up using phase-lock-gel (PLG) phenol/chloroform extraction tube from
Eppendorf (W estbury, NY). A volume of 162 pl phenolzchloroformzisoamyl alcohol
(saturated with 10 mM Tris-HCl and 1 mM EDTA, pH 8.0) was added to the cDNA

synthesis preparation, and vortexed for 2 min. Then the mixture was transferred to PLG

113

tube, which was centrifuged at 12,000 x g for 2 min. The aqueous upper phase was
transferred to a flesh tube and 0.5 volumes of 7.5 M ammonia acetate and 2.5 volumes of
absolute ethanol (prechilled) were added to each tube, and vortexed, followed by
immediate centrifugation at 12,000 x g for 20 min. The supernatant was removed and the
pellet was washed with 0.5 ml 80 % prechilled ethanol. After centrifugation at 12,000 x g
for 5 min, the washing step was repeated one additional time. The pellet was air dried and

resuspended in 12 pl DEPC water.

Synthesis ofbiotin-labeled cRNA

An in vitro transcription reaction was performed to produce biotin-labeled cRNA from
the cDNA obtained from previous step using the BioArray Higtheld RNA transcript
labeling kit from Affymetrix Inc. (Santa Clara, CA). To an Eppendorf tube the following

were added;

CDNA template 5 p1
DEPC water 17 pl
10 X reaction buffer 4 pl
10 X biotin labeled ribonucleotides 4 pl
10 X DTT 4 pl
10 X Rnase Inhibitor 4 pl
20 X T7 RNA polymerase 2 pl

The reagents were mixed and the tube was transferred to a 37°C water-bath incubating
for 4.5 hr. The contents of the tube were mixed every 30 min by gently tapping on the
tube. The cRNA samples were cleaned up using an Rneasy spin column from Qiagen
(Valencia, CA) to remove unincorporated NTPs and followed by ethanol precipitation.

Each sample was split into two tubes, and the volume was adjusted to 100 pl with Rnase-

114

free water. The 350 pl of the RLT lysis buffer was added to each tube. Samples were
mixed with vigorous shaking and a volume of 250 pl 100 % pure ethanol was added into
the lysate, and mixed well by pipetting. The mixtures were applied directly to the Rneasy
mini spin column sitting on a collection tube. The set was then centrifuged at 8000 x g for
15 sec. The collection tube was discarded and the column was transferred to a fresh tube.
An aliquot of 500 pl of salt buffer RPE was added in each column and the set was spun
for another 15 sec at 8000 x g. The washing step was repeated once more, and the column
was centrifuged for 2 min at maximum speed (14,000 x g) to dry the Rneasy membrane.
The column was then transferred to a new collection tube, and 40 pl DEPC water was
pipetted onto the membrane and allowed to sit at RT for 5 min. The RNA samples were
eluted by centrifugation for 1min at 10,000 x g, this step was repeated one more time to
allow for complete elution. A 0.5 volume of 7.5 M ammonia acetate and 2.5 volumes of
absolute ethanol (prechilled) were added to each tube, and vortexed. RNA samples were
allowed to precipitate at —20°C overnight. The samples were centrifuged at 12,000 x g for
30 min at 4°C. Supernatant was removed and pellet was washed with 0.5 ml 80 %
prechilled ethanol. After centrifugation at 12,000 x g for 5 min, the washing step was
repeated one additional time. The pellet was air dried and resuspended in 30 pl DEPC

water.

The yield of Biotin-labeled cRNA was determined using spectrophotometric analysis.
The optical density (OD) value at 260 nm and 280 nm was measured to determine sample
concentration and purity, applying the convention that lOD at 260nm equals 40 pg/ml

RNA. Only samples with OD 260/280 ratio between 1.9 and 2.1 are acceptable. Since the

115

cRNA was obtained using total RNA as starting material, an adjusted cRNA yield was
calculated using the following formula to reflect carryover of unlabeled total RNA:
Adjusted cRNA yield = RNAm — (RNAmml) (Y)
RN Am = amount of cRNA measured after the in vitro transcription reaction (pg)
RNAtom. = starting amount of total RNA (pg)
Y = fraction of cDNA reaction used in the in vitro transcription reaction
For efficient hybridization the cRNA was fragmented using 20 pg cRNA sample and 8 pl
of 5 X fragmentation buffer diluted with RNase-free water up to the volume of 40 pl. The
mixture was incubated at 94°C for 35 min, and transferred to ice following the
incubation.

5 X fragmentation buffer:

4.0 ml 1M Tris acetate pH 8.1

0.64 g MgOAc

0.98 g KOAc

DEPC-treated water to 20 ml
Reaction mixture was mixed thoroughly and filtered through a 0.2 pm vacuum filter unit.
Gel electrophoresis of the cRNA product was conducted to estimate the yield and size
distribution of labeled transcripts. Samples were loaded into the 1 % agarose gel, and
electrophoresize for 45 min at 75 V. Gel was stained using ethidium bromide for 10 min
and destained with destilled water for 30 min. Sample distribution was visualized using

GelDoc system from BioRad (Hercules, CA).

T he following steps were performed in the Gemonics facility on MS U campus.

116

Target Hybridization

The hybridization cocktail was prepared following the table, and the volume was scaled

 

 

 

 

 

 

 

 

 

up for multiple chips.
Component Amount Final Concentration Source
Fragmented cRNA 15 pg 0.05 pg / pl from previous steps
Control Oligonucleotide 5 pl 50 pM Affymetrix
20 X Eukaryotic 15 pl 100 pM Affymetrix
Hybridization Controls
Herring Sperm DNA 3 p] 0.1 mg/ml Promega
Acetylated BSA 3 p] 0.5 mg/ml Gibco BRL
2 X Hybridization 150 pl 1 X Affymetrix
Buffer
Rnase-free water To a final volume

of 300 pl

 

 

 

 

The hybridization cocktail was heated to 99°C for 5 min in a heated block and was then
transferred to 45°C heat block for 5 min for equilibration. The hybridization mixture was
centrifuged for 5 min to remove any insoluble material. Meanwhile, the chip array was
warmed up to room temperature, and wetted with 250 pl 1 X hybridization buffer, and
incubated at 45°C for 10 min with rotation. After the buffer solution was removed from
the probe array cartridge, it was filled with 250 pl volume of the clarified hybridization
cocktail solution. The probe array was then placed in rotating incubator at 45°C and

allowed to hybridize for 16hrs while rotating at 60 RPM.

Washing, Staining and Scanning Probe arrays
The GeneChip Fluidics Station 400 from Affymetrix was used to wash and stain the
probe arrays. It was operated using GeneChip software. Refer to the GeneChip Fluidics

Station 400 user’s manual for instructions on connecting and addressing multiple fluidics

117

 

stations. The experiment was first defined by entering experiment name and parameters
in GeneChip software. Priming the fluidics station was conducted by turning on
appropriate modules and filling in the buffer reservoir with non-stringent wash buffer.
After 16hrs of hybridization, the hybridization cocktail was removed from the probe
array, and the array was refilled with non-stringent wash buffer and allowed to equilibrate
to room temperature. Streptavidin phycoerythrin working solution obtained from
Molecular Probes (Eugene, OR) was prepared immediately before usage. The following

components was mixed in a microcentrifuge tube:

2X stain buffer 300 pl
DEPC water 270 pl
50 mg/ml acetylated BSA (Gibco BRL) 24 pl
1 mg/ml streptavidin phycoerythrin 6 pl

The antibody amplification solution was prepared by mixing the following components in

a microcentrifuge tube:

2 X stain buffer 300 p1
DEPC water 266.4 pl
50 mg/ml acetylated BSA 24 pl
10 mg/ml normal goat IgG (Sigma chemical) 6 pl
0.5 mg/ml biotinylated antibody (Vector Laboratories) 3.6 pl

The probe array was inserted into the designated module of the fluidics station, and
microcentrifuge tubes containing the staining buffer and antibody staining buffer were
place in appropriate sample holder. In the fluidics station protocol drop-down list, the
appropriate control protocol for washing and staining steps were selected and following

the instruction on LCD window display to start. Eject message would be displayed on

118

 

LCD window when the whole protocol was completed. The probe array was then

removed from the module and kept at 4°C in the dark until ready for scanning.

The scanner was also connected with the station and controlled by the GeneChip
software. Prior to scanning, make sure the scanner option was set to pixel value = 3 pm,
wavelength = 570 nm, and the laser was warmed up for at least 15 min. The probe array
was inserted into the sample holder. From the control drop-down manual, the experiment
name corresponded to the probe array was selected, and the image scanning was
performed by clicking the start button. Each probe array was scanned twice with each
time approximately 5 min. The computer work station automatically overlaied the two
scanned images and averaged the intensities of each probe cell for the greatest assay

sensitivity. Each complete image was then saved as data image file with .dat extension.

GeneChips Data Analysis

Each image file was analyzed and data was retrieved using Affymetrix data mining tool.
The outcome file was then stored in a database built using Microsoft Access. Initial data
analysis was also conducted using Microsoft Access query design. Further Cluster
analysis, Genetree construct and pathway analysis were conducted using GeneSpring

software obtained from SiliconGenetics (Redwood City, CA).

119

REFERENCES

120

REFERENCES

. Affymetrix® Microarray Suite 5.0 User’s Guide.

. Altschul, S.F., Madden, T.L., Schiiffer, A.A., Zhang, J., Zhang, Z., Miller, W.,
and Lipman, DJ. (1997) Gapped BLAST and PSI-BLAST: a new generation of
protein database search programs. Nucleic Acids Res. 25: 3389-3402.

. Beguinot, F., Tramontano, D., Duilio, C., Formisano, S., Beguinot, L., Mattioli,
P., Mancini, M., and Aloj, SM. (1985) Alteration of erythrocyte membrane lipid
fluidity in human obesity. J. Clin. Endocrinol. Metab 60: 1226-1230.

. Beguinot, F., Beguinot, L., Tramontano, D., Duilio, C., Formisano, S., Bifulco,
M., Ambesi-Impiombato, RS, and Aloj, SM. (1987) Thyrotropin regulation of
membrane lipid fluidity in the FRTL-S thyroid cell line. Its relationship to cell
grth and functional activity. J. Biol. Chem. 262:1575-1582.

. Berthiaume, J., and Wallace, KB. (2002) Perfluorooctanoate, perfluorooctane
sulfonate, and N-ethyl perfluorooctane sulfonamide ethanol; peroxisome
proliferation and mitochondrial biogenesis. T oxicol. Lett. 129: 23-32.

. Biegel, L.B., Liu, R.C. M., Hurtt, M.E., and Cook, J.C. (1995) Effects of
ammonium perfluorooctanoate on Leydig cell function: in vitro, in vivo, and ex
vivo studies. T oxicol. Appl. Pharmacol. 134: 18-25.

. Boujrad, N., Vidic, B., Gazouli, M., Culty, M., and Papadopoulos, V. (2000) The
peroxisome proliferator perfluorodecanoic acid inhibits the peripheral-type
benzodiazepine receptor (PBR) expression and hormone-stimulated mitochondrial
cholesterol transport and steroid formation in Leydig cells. Endocrinology 141(9):
3137-3148.

. Brondyk, W.H., McKiernan, C.J., Fortner, K.A., Stabila, P., Holz, R.W., and
Macara, LG. (1995) Interaction cloning of Rabin3, a novel protein that associates
with the Ras-like GTPase Rab3A. Mol. Cell Biol. 15(3): 1137-1143.

121

10.

11.

12.

13.

14.

15.

16.

17.

18.

Cook, J.C., Murray, S.M., Frame, SR, and Hurtt, ME. (1992) Induction of
Leydig cell adenomas by ammonium perfluorooctanoate: a possible endocrine-
related mechanism. T oxicol. Appl. Pharmacol. 113: 209-217.

Cossarizza, A., Baccarani-Contri, M., Kalashnikova, G., and Franceschi, C.
(1993). A new method for the cytofluorimetric analysis of mitochondrial
membrane potential using the J-aggregate forming lipophilic cation 5,5’,6,6’-
tetrachloro-1,l ’,3,3’-tetraethy1 benzimidazol carbocyanine iodide (JC-l).
Biochem. Biophy. Res. Comm. 197: 40-45.

Diaz, M.J., Chinje, E., Kentish, P., Jamot, B., George, M., and Gibson, G. (1994)
Induction of cytochrome P450 4A by the peroxisome proliferator perfluoro-n-
octanoic acid. Toxicology. 86: 109-122.

Display Systems Biotech Inc. (1999) Restriction Fragment Differential Display.
Kit manual version 2.1. Display Systems Biotech Inc., Buena Vista, CA. pp19.

Douglass et al., (1995) PCR differential display identifies a rat brain mRNA that
is transcriptionally regulated by cocaine and amphetamine. J. Neurosciences. 15:
2471-2481.

Duplus, E., Glorian, M., and Forest, C. (2000) Fatty acid regulation of gene
transcription. J. Biol. Chem. 275: 30749-30752.

Galbiati, F., Razani, B., and Lisanti, MP. (2001) Emerging themes in lipid rafts
and caveolae. Cell 106: 403-411.

Ganesan, A.K., Vincent, T.S., Olson, J .C., and Barbieri, J .T. (1999) Pseudomonas
aeruginosa exoenzyme S. disrupts Ras-mediated signal transduction by inhibiting

Guanine nucleotide exchange factor-catalyzed nucleotide exchange. J. Biol.
Chem. 274: 21823-21829.

Gao, L. et al., (1998) A novel response to dioxin. Induction of ecto-ATPase gene
expression. J. Biol. Chem. 273:15358-15365.

GeneChip® Expression Analysis Technical Manual.

122

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Gerhold, D.L., Liu, F., Jiang, G., Li, Z., Xu, J., Lu, M., Sachs, J.R., Bagchi, A.,
Fridman, A., Holder, D.J., Doebber, T.W., Berger, J ., Elbrecht, A., Moller, DE,
and Zhang, BB. (2002) Gene expression profile of adipocyte differentiation and
its regulation by peroxisome proliferator-activated receptor-gamma agonists.
Endocrinology 143: 2106-21 18.

Giesy, J. P., and Kannan, K. (2001) Global distribution of perfluorooctane
sulfonate in wildlife. Environ. Sci. T echnol. 35: 1339-1342.

Giesy, J. P., and Kannan, K. (2002) Perfluorochemical surfactants in the
environment. Environ. Sci. T echnol. 35: 1339-1342.

Goldenthal, E.I., Jessup, D.C., Geil, R.G., Jefferson, N.D., Arceo, R.J., and
Ruecker, FA. (1978) 90-day subacute rat study. Study No. 137-085, International
Research and Development Corp. Mattawan, MI. Available on USEPA public
docket AR-226.

Green, S. (1992) Commentary: receptor-mediated mechanisms of peroxisome
proliferators. Biochemical Pharmacology 43: 393-401.

Green, C.B. et al., (1996) Identification of a novel vertebrate circadian clock-
regulated gene encoding the protein noctumin. Proc. Natl. Acad. Sci. USA. 93:
14884-14888.

Guimaraes, M.J. et al., (1995) Differential Display by PCR: novel findings and
applications. Nucleic Acids Res. 23: 1832-1833.

Hahn, M.E., and Chandran, K. (1996) Uroporphyrin accumulation associated with
cytochrome P4501A induction in fish hepatoma cells exposed to aryl hydrocarbon
receptor agonists, including 2,3,7,8-tetrachlorodibenzo-p-dioxin and planar
chlorobiphenyls. Arch. Biochem. Biophys. 329: 163-174.

Hahn, M.E., Lamb, T.M., Schultz, M.E., Smolowitz, R.M., and Stegeman, JJ.
(1993) Cytochrome P4501A induction and inhibition by 3,3’,4,4’-
tetrachlorobiphenyl in an Ah receptor-containing fish hepatoma cell line (PLHC-
1). Aquat. T oxicol. 26: 185-208.

Hamadeh, H.K., Bushel, P.R., Jayadev, S., Martin, K., DiSorbo, 0., Siegber, S.,
Bennett, L., Tennant, R., Stoll, R., Barrett, C., Blanchard, K., Paules, RS,

123

29.

30.

31.

32.

33.

34.

35.

36.

37.

Afshari, C.A., (2002a) Gene expression analysis reveals chemical-specific
profiles. T oxico]. Sci. 67: 219-231.

Hamadeh, H.K., Bushel, P.R., Jayadev, S., Martin, K., DiSorbo, 0., Bennett, L.,
Lee, L., Tennant, R., Stoll, R., Barrett, C., Barrett, K., Paules, R.S., Blanchard, K.,
and Afshari, C.A., (2002b) Prediction of compound signature using high density
gene expression profiling. T oxico]. Sci. 67 : 232-240.

Hanhijarvi, H., Ophaug, RH, and Singer, L. (1982) The sex-related difference in
perfluorooctanoate excretion in the rat. Proc. Soc. Exp. Biol. Med. 171: 50-55.

Hansen, K.J., Clemen, L.A., Ellefson, M.E., and Johnson, H0. (2001)
Compound-specific, quantitative characterization of organic fluorochemicals in
biological matrices. Environ. Sci. T echno]. 35: 766-770.

Haughom, B., and Spydevold, O. (1992) The mechanism underlying the
hypolipemic effect of perfluorooctanoic acid (PFOA), perfluorooctane sulphonic
acid (PFOSA) and clofibric acid. Biochimica et Biophysica Acta. 1128: 65-72.

Hightower, LE. and Renfro, J .L. (1988) Recent applications of fish cell culture to
biomedical research. J. Exp. Zoo]. 248: 290-302.

Higuchi, E., Oridate, N., Furuta, Y., Suzuki, S., Hatakeyama, H., Sawa, H.,
Sunayashiki-Kusuzaki, K., Yamazaki, K., Inuyama, Y., and Fukuda, S. (2003).
Differentially expressed genes associated with CIS-diarnminedichloroplatinum
(II) resistance in head and neck cancer using differential display and CDNA
microarray. Head Neck 25: 187-193.

I-lu, W.Y. (2000) In vitro studies of the biochemical toxicity of perfluorooctane
sulfonic acid and its possible interaction with 2,3,7,8 — tetrachlorodibenzo-p-
dioxin. MSc Thesis, Michigan State University, East Lansing, MI 48824.

Hu, W.Y., Jones, P.D., DeCoen, W., King, L.B., Fraker, P., Newsted, J.L., and
Giesy, J .P. (2002a) Alterations in cell membrane properties caused by
perfluorinated compounds. Comp. Biochem. Physio]. 135: 77-88.

Hu, W.Y., Jones, P.D., Upham, B.L., Trosko, J.E., Lau, C., and Giesy, J.P.
(2002b) Inhibition of gap junctional intercellular communication by

124

38.

39.

40.

41.

42.

43.

45.

perfluorinated compounds in rat liver and dolphin kidney epithelial cell lines in
vitro and Sprague-Dawley rats in vivo. T oxicol. Sci. 68: 429-436.

Hu, W.Y., Jones, P.D., Celius, T., and Giesy, J .P. (2003a) Identification of genes
responsive to perfluorooctane sulfonic acid using gene expression profiling.
European J. T oxicol. Pharmacol. (In Press).

Hu, W.Y., Jones, P.D., DeCoen, W., and Giesy, J .P. (2003b) Comparison of gene
expression methods to identify genes responsive to perfluorooctane sulfonic acid.
Environmental Toxicology and Pharmacology (In Press).

Ikeda, T., Aiba, K., Fukuda, K., and Tanaka, M. (1985) The induction of
peroxisome proliferation in rat liver by perfluorinated fatty acids, metabolically
inert derivatives of fatty acids. J. Biochem. 98: 475-482.

Jefferson, J.R., Powell, D.M., Rymaszewski, Z., Kukowska-Latallo, J., Lowe,
J .B., and Schroeder, F. (1990) Altered membrane structure in transfected mouse

L-cell fibroblasts expressing rat liver fatty acid-binding protein. J. Biol. Chem.
265: 11062-11068.

Jones, P.D., Hu, W.Y., DeCoen, W., Newsted, J .L., and Giesy, JP. (2002) The
binding of perfluorinated chemicals to serum proteins. Environ. T oxico]. Chem.
(In Press).

Just, W.W., Gorgas, K., Hartl, F .U., Heinemann, P., Salzer, M., and Schimassek,
H. (1989) Biochemical and zonal heterogeneity of peroxisome proliferation
induced by perfluorocarboxylic acids in rat liver. Hepatology 9: 570-581.

. Kannan, K., Koistinen, J ., Beckmen, K., Evans, T., Gorzelany, J .F., Hansen, K.J.,

Jones, P.D., Helle, E., Nyrnan, M., and Giesy, J.P. (2001a). Accumulation of
perfluorooctane sulfonate in marine mammals. Environ. Sci. T echnol. 35: 1593-
1598.

Kannan, K., Franson, J.C., Bowerman, W.W., Hansen, K.J., Jones, P.D., and
Giesy, J .P. (2001b) Perfluorooctane sulfonate in fish-eating water birds including
bald eagles and albatrosses. Environ. Sci. T echnol. 35: 3065-3070.

125

46.

47.

48.

49.

50.

51.

52.

53.

54.

Kawashima, Y., Kobayashi, H., Miura, H., and Kozuka, H. (1995)
Characterization of hepatic response of rat to administration of perfluorooctanoic
acid and perfluorodecanoic acid at low level. Toxicology 99: 169-178.

Kawasaki, K., Yin, J.J., Subczynski, W.K., Hyde, J.S., and Kusumi, A. (2001)
Pulse EPR detection of lipid exchange between protein-rich raft and bulk domains
in the membrane: methodology development and its application to studies of
influenza viral membrane. Biophys. J. 80: 738-748.

Kramer, V.J., Helferich, W.G., Bergman, A., Klasson-Wehler, E., and Giesy, J .P.
(1997) Hydroxylated polychlorinated biphenyl metabolites are anti-estrogenic in a

stably transfected human breast adenocarcinoma (MCF7) cell line. T oxicol. Appl.
Pharmacol. 144: 363-376.

Kudo, N., Bandai, N., Suzuki, E., Katakura, M., and Kawashima, Y. (2000)
Induction by perfluorinated fatty acids with different carbon chain length of

peroxisomal B-oxidation in the liver of rats. Chemico-Biologica] Interaction 124:
119-132.

Kudo, N., Suzuki, E., Katakura, M., Ohmori, K., Noshiro, R., and Kawashima, Y.
(2001) Comparison of the elimination between perfluorinated fatty acids with

different carbon chain length in rats. Chemico-Biologica] Interactions 134: 203-
216.

Lau, C., Thibodeaux, J .R., Hanson, R.G., Rogers, J .M., Grey, B.E., Stanton, M.E.,
Butenhoff, J .L., and Stevenson, LA. (2003) Exposure to perfluorooctane
sulfonate during pregnancy in rat and mouse. 11: postnatal evaluation. T oxico].

Sci. 74: 382-392.

Liang, P., and Pardee, AB. (1992) Differential display of eukaryotic messenger
RNA by means of the polymerase chain reaction. Science 257: 967-971.

Liang, P. et al., (1992) Differential display and cloning of messenger RNAs from
human breast cancer versus mammary epithelial cells. Cancer Research. 52:
6966-6968.

Liang, P. et al., (1994) Differential display using one-base anchored oligo-dT
primers. Nucleic Acids Res. 22: 5763-5764.

126

 

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

Liang, P. (1998) Factors ensuring successful use of differential display. Methods
in Enzymology. 16: 361-364.

Liu, R.C.M., Hahn, C., and Hurtt, ME. (1996) The direct effect of hepatic
peroxisome proliferators on rat Leydig cell firnction in vitro. T oxico]. Appl.
Pharmacol. 30: 102-108.

Liu, R.C., Hurtt, M.E., Cook, BC, and Biegel, L.B. (1996b) Effect of
peroxisome proliferator, ammonium perfluorooctanoate (C8), on hepatic
Aromatase activity in adult male Cr1:CD BR rats. F undam. Appl. T oxico]. 30(2):
220-228.

Lockhart, D.J., and Winzeler, EA. (2000) Genomics, gene expression and DNA
arrays. Nature 405: 827-836.

Luebker, D.J., Hansen, K.J., Bass, N.M., Butenhoff, J .L., and Seacat, A.M. (2002)
Interactions of fluorochemicals with rat liver fatty acid-binding protein.
Toxicology 176: 175-185.

Masuda, M., Kuriki, H., Komiyama, Y. et al. (1987) Measurement of membrane
fluidity of polyrnorphonuclear leukocytes by flow cytometry. J. Immunol. Meth.
96: 225-231.

Maniatis, T., and Reed, R. (2002) An extensive network of coupling among gene
expression machines. Nature 416: 499-505.

Matsuba, T., Keicho, N., Higashimoto, Y., Granleese, S., Hogg, J .C. Hayashi, S.,
and Bondy, GP. (1998) Identification of glucocorticoid- and adenovirus ElA-

regulated genes in lung epithelial cells by differential display. Am. J. Respir. Cell
Mo] Bio]. 18: 243-254.

Matz, M.V., and Lukyanov, SA. (1998) Different strategies of differential
display: areas of application. Nucleic Acids Research 26: 5537-5543.

Meng, J.J., Bornslaeger, E.A., Green, K.J., Steinert, P.M., and 1p, W. (1997) Two-
hybrid analysis reveals fundamental differences in direct interactions between
desmoplakin and cell type-specific intermediate filaments. J. Biol. Chem. 272
(34): 21495-21503.

127

65.

66.

67.

68.

69.

70.

71.

72.

73.

Mong, J.A., Krebs, C., and Pfaff, D.W. (2002) Perspective: micoarrays and
differential display PCR-tools for studying transcript levels of genes in
neuroendocrine systems. Endocrinology 143: 2002-2006.

Oboum, J.D., Frame, S.R., Bell, R.H., Longnecker, D.S., Elliott, G.S., and Cook,
J .C. (1997) Mechanisms for the pancreatic oncogenic effects of the peroxisome
proliferator wyeth-14,643. T oxico]. Appl. Pharmacol. 145: 425-436.

Okun, J .G., Lummen, P., and Brandt, U. (1999) Three classes of inhibitors share a
common binding domain in mitochondrial complex I (N ADqubiquinone
oxidoreductase). J. Biol. Chem. 274: 2625-2630.

Olson, C. T., and Andersen, M. E. (1983) The acute toxicity of perfluorooctanoic

and perfluorodecanoic acids in male rats and effects on tissue fatty acids. T oxicol.
Appl. Pharmacol. 70: 362-372.

Permadi, H., Lundgren, B., Anderson, K., and Depierre, J. W. (1992) Effects of
perfluoro fatty acids on xenobiotic-metabolizing enzymes, enzymes which
detoxify reactive forms of oxygen and lipid peroxidation in mouse liver.
Biochemical Pharmacology. 44: 1 183-1 191.

Pons, M., Gagne, D., Nicolas, J .C., Mehtali, M. (1990) A new cellular model of
response to estrogens: a bioluminescent test to characterize anti-estrogen
molecules. Biotechniques 9: 450-459.

Pownall, H.J., and Smith, LC. (1989) Pyrene-labeled lipids: versatile probes of
membrane dynamics in vitro and in living cells. Chemistry and Physics of Lipids.
50: 191-211.

Ravis, W.R., Hoke, J .F ., and Parsons, D.L. (1991) Perfluorochemical erythrocyte
substitutes: disposition and effects on drug distribution and elimination. Drug
Metabolism Reviews. 23: 375-411.

Reo, N.V., Narayanan, L., Kling, KB, and Adinehzadeh, M. (1996)
Perfluorodecanoic acid, a peroxisome proliferator, activates phospholipase C,
inhibits CTP: phosphocholine cytidylyltransferase, and elevates diacylglycerol in
rat liver. T oxicologv Letters. 86: 1-11.

128

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

Richter, C.A., Tieber, V.L., Denison, MS, and Giesy, J.P. (1997) An in vitro
rainbow trout cell bioassay for aryl hydrocarbon receptor-mediated toxins.
Environ. T oxico]. Chem. 3: 543-550.

Sanderson, J.T., Aarts, J.M., Brouwer, A., Froese, K.L., Denison, MS, and
Giesy, J .P. (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.
T oxico]. Appl. Pharmacol. 137: 316-325.

Sargent, J., and Seffl, R. (1970) Properties of perfluorinated liquids. Fed. Proc.
29: 1699-1703.

Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T., and
Butenhoff, J .L. (2002a) Subchronic toxicity studies on perfluorooctane sulfonate
potassium salt in cynomolgus monkeys. T oxicol. Sci. 68: 249-264.

Seacat, A.M., Thomford, P.J., and Butenhoff, J .L. (2002b) Terminal observations
in Sprague-Dawley rats after lifetime dietary exposure to potassium
perfluorooctane sulfonate. T oxicologist 66: 185 (Abstract ID: 906).

Seacat, A.M., Thomford, P.J., Hansen, K.J., Clemen, L.A., Eldridge, S.R.,
Elcombe, CR, and Butenhoff, J.L. (2003) Sub-chronic dietary toxicity of
potassium perfluorooctane sulfonate in rats. T oxicol. Sci. 183: 117-131.

Shinoda, K., and Nomura, T. (1980) Miscibility of fluorocarbon and hydrocarbon
surfactants in micelles and liquid mixtures. Basic studies of oil repellent and fire
extinguishing agents. J. Phys. Chem. 84: 365-369.

Simpson, E.R., Mahendroo, M.S., Means, G.D., Kilgore, M.W., et a]. (1994)
Aromatase cytochrome P450, the enzyme responsible for estrogen biosynthesis.
Endocrine Rev. 15: 342-355.

Sohlenius, A.K., et a]. (1994) Effects of perfluorooctanoic acid, a potent
peroxisome proliferator in rat, on morris hepatoma 7800Cl cells, a rat cell line.
Biochimica et Biophysica Acta. 1213: 63-74.

Sohlenius, A. K., Reinfeldt, M., Backstrom, K., Bergstrand, A., and Depierre, J.
W. (1996). Hepatic peroxisome proliferation in vitamin A deficient mice without

129

84.

85.

86.

87.

88.

89.

90.

91.

a simultaneous increase in peroxisomal Acyl-CoA oxidase activity. Biochemical
Pharmacology. 51: 821-827.

Sokol, R.J., Straka, M.S., Dahl, R., Devereaux, M.W., Yerushalmi, B.,
Gumpricht, E., Elkins, N., and Everson, G. (2001) Role of oxidant stress in the
permeability transition induced in rat hepatic mitochondria by hydrophobic bile

acids. Pediatr Res 49: 519-531.

Starkov, A.A., and Wallace, KB. (2002) Structural determinants of
fluorochemical-induced mitochondrial dysfunction. T oxico]. Sci. 66: 244-252.

Takagi, A., Sai, K., Umemura, T., Hasegawa, R., and Kurokawa, Y. (1991) Short-
terrn exposure to the peroxisome proliferators, perfluorooctanoic acid and
perfluorodecanoic acid, causes significant increase of 8-hydroxydeoxyguanosine
in liver DNA of rats. Cancer Lett. 57(1): 55-60.

Thibodeaux, J.R., Hanson, R.G., Rogers, J.M., Grey, B.E., Barbee, B.D.,
Richards, J.H., Butenhoff, J.L., and Stevenson, LA. (2003) Exposure to
perfluorooctane sulfonate during pregnancy in rat and mouse. 1: maternal and
prenatal evaluations. T oxicol. Sci. 74: 369-381.

Tobin, K.A.R., Ulven, S.M., Schuster, G.U., Hermansen Steineger, H., Andresen,
S.M., Gustafsson, J .A., Nebb, H1 (2002) Liver X receptors as insulin-mediating
factors in fatty acid and cholesterol biosynthesis. J. Biol. Chem. 277: 10691-
10697.

Upham, B.L., Deocampo, N.D., Wurl, B., and Trosko, J.E. (1998) Inhibition of
gap junctional intercellular communication by perfluorinated fatty acids is
dependent on the chain length of the fluorinated tail. Int. J. Cancer. 78: 491-495.

Vanden Heuvel, J .P., Davis, J .W., Sommers, R., and Peterson, R.E. (1992a) Renal
excretion of perfluorooctanoic acid in male rats: inhibitory effect of testosterone.
J. Biochem T oxico]. 7(1): 31-36.

Vanden Heuvel, J .P., Kuslikis, BL, and Peterson, R.E. (1992b) Covalent binding
of perfluorinated fatty acids to proteins in the plasma, liver and testes of rats.
Chem. Biol. Interact. 82(3): 317-328.

130

92.

93.

94.

95.

96.

97.

98.

99.

100.

Vanden Heuvel, J.P., Sterchele, P.F., Nesbit, D.J., and Peterson, RE. (1993)
Coordinate induction of acyl-CoA binding protein, fatty acid binding protein and

peroxisomal B-oxidation by peroxisome proliferators. Biochim. Biophys. Acta.
1177(2): 183-190.

Wallace, K.B., Luebker, D.J., Butenhoff, J.L., and Seacat, A.M. (2001)
Perfluorooctane sulfonate and 2-(N-ethylperfluorooctanesulfonamido)-ethy1

alcohol are peroxisome proliferators in rats, but not guinea pigs. T oxicologist 60:
348 (Abstract ID: 1657).

Wan, Y., and Nordeen, SK, (2002) Overlapping but distinct gene regulation
profiles by glucocorticoids and progestins in human breast cancer cells. Mo].
Endocrinol. 16: 1204-1214.

Ylinen, M., Hanhijarvi, H., Jaakonado, J., and Peura, P. (1989) Stimulation by
oestradiol of the urinary excretion of perfluorooctanoic acid in the male rat.
Pharmacol. T oxico]. 65: 274-277.

Ylinen, M., and Auriola, S. (1990) Tissue distribution and elimination of

perfluorooctanoic acid in the rat after single intraperitoneal administration.
Pharmacol. T oxico]. 66: 45-48.

Youssef, J., and Badr, M. (1998) Extraperoxisomal targets of peroxisome
proliferators: mitochondrial, microsomal, and cytosolic effects. Implications for
health and disease. Crit. Rev. T oxicol. 28(1): 1-33.

Yuen, T., Wurmbach, E., Pfeffer, R.L., Ebersole, B.J. and Sealfon, SC. (2002)
Accuracy and calibration of commercial oligonucleotide and custom cDNA
microarrays. Nucl. Acids. Res. 30: e48.

Zhang, L., and Medina, D. (1993) Gene expression screening for specific genes
associated with mouse mammary tumor development. Mo]. Carcinog. 8: 123-26.

Zoeteweij, J .P., Water, B.V., DeBont, H.J., Mulder, G.J., and Nagelkerke,
J .F . (1992) Involvement of intracellular Ca2+ and K+ in dissipation of the
mitochondrial membrane potential and cell death induced by extracellular ATP in
hepatocytes. Biochem. J. 288: 207-213.

131

        

     
 

        

Iiiilinllin llilil “

W
3 1293

ll