Aryl hydrocarbon receptor-independent toxicity of weathered crude oil during fish development.

Abstract

Polycyclic aromatic hydrocarbons (PAHs), derived largely from fossil fuels and their combustion, are pervasive contaminants in rivers, lakes, and nearshore marine habitats. Studies after the Exxon Valdez oil spill demonstrated that fish embryos exposed to low levels of PAHs in weathered crude oil develop a syndrome of edema and craniofacial and body axis defects. Although mechanisms leading to these defects are poorly understood, it is widely held that PAH toxicity is linked to aryl hydrocarbon receptor (AhR) binding and cytochrome P450 1A (CYP1A) induction. Using zebrafish embryos, we show that the weathered crude oil syndrome is distinct from the well-characterized AhR-dependent effects of dioxin toxicity. Blockade of AhR pathway components with antisense morpholino oligonucleotides demonstrated that the key developmental defects induced by weathered crude oil exposure are mediated by low-molecular-weight tricyclic PAHs through AhR-independent disruption of cardiovascular function and morphogenesis. These findings have multiple implications for the assessment of PAH impacts on coastal habitats.

Full text

Page 1

Environmental Health Perspectives • VOLUME 113 | NUMBER 12 | December 2005
1755
Research
Every recent assessment of coastal habitats
worldwide, from tropical reefs to temperate
estuaries, has cited land-based pollution
or runoff as a major threat to aquatic eco-
system health (Fabricius 2005; Li and Daler
2004; Pew Oceans Commission 2003; U.S.
Commission on Ocean Policy 2004). As per-
vasive components of runoff from impervious
surfaces, polycyclic aromatic hydrocarbons
(PAHs) are a part of this problem, and there is
very little understanding of their biologic
impacts on aquatic resources. Because of
urbanization and increased heavy vehicle use,
storm water runoff and atmospheric depo-
sition are now the largest sources of aquatic
PAH contamination (Li and Daler 2004;
Lima et al. 2002; National Research Council
2003; Van Metre and Mahler 2003; Van
Metre et al. 2000). An understanding of the
effects of PAHs on aquatic organisms is essen-
tial to understanding fully the impacts of
urbanization and non-point source pollution
on coastal habitats.
On a smaller scale, oil spills have provided
a more conspicuous view of the impacts
of PAH pollution on aquatic resources.
Hydrocarbons from oil spills can persist in
nearshore sediments for decades or longer and
have long-term effects on aquatic ecosystems
(Peterson et al. 2003; Reddy et al. 2002; Short
et al. 2004). The deleterious effects of PAHs
on fish early-life stages were investigated
extensively after the 1989 Exxon Valdez oil
spill in Prince William Sound, Alaska, which
contaminated nearshore and intertidal spawn-
ing grounds for Pacific herring (Clupea pallasi)
and pink salmon (Oncorhynchus gorbuscha)
with Alaska North Slope (ANS) crude oil.
Field and laboratory studies in these species
and others demonstrated a common syn-
drome of oil-induced embryolarval toxicity
that occurs in a range of teleosts, including
marine, freshwater, temperate, and tropical
species (Carls et al. 1999; Couillard 2002;
Heintz et al. 1999; Marty et al. 1997; Pollino
and Holdway 2002). This was characterized
by pericardial and yolk sac edema, jaw reduc-
tions, and curvature of the body axis.
Increased weathering of crude oil enriches the
fraction of tricyclic PAHs and their alkylated
homologs and increases the frequency of mal-
formations (Carls et al. 1999; Heintz et al.
1999). Additionally, delayed mortality also
occurred in the absence of external malforma-
tions, as indicated by the reduced oceanic sur-
vival of pink salmon exposed to weathered
crude oil as embryos and released as smolts
(Heintz et al. 2000).
The mechanisms leading to PAH-associ-
ated malformations and sublethal effects dur-
ing fish development are unknown. Most
PAHs bind the aryl hydrocarbon receptor
(AhR), a ligand-activated basic-helix-loop-
helix-Per-Arnt-Sim family transcription factor
that controls the expression of a battery of
genes encoding enzymes that convert PAHs to
water-soluble derivates that are excreted,
including mixed-function oxygenases such as
cytochrome P450 1A (CYP1A) family mem-
bers (Nebert et al. 2004). Although CYP1A
for decades has been the most widely used
biomarker for PAH exposure (Whyte et al.
2000), its role as a bioindicator of PAH toxic-
ity has been debated. Genetic analysis in the
mouse has led to a dual model in which the
AhR pathway mediates both an adaptive
response by which xenobiotic compounds are
metabolized and detoxified, and a toxic
response whereby receptor activation results in
negative impacts in the exposed animal
(Nebert et al. 2004; Schmidt and Bradfield
1996). Generally, the toxic response occurs
with AhR ligands that are poor substrates for
CYP enzymes, in particular, halogenated aro-
matic hydrocarbons such as dioxins and poly-
chlorinated biphenyls. Because these ligands
are resistant to metabolism, they accumulate
in tissues and persistently activate the AhR.
On the other hand, PAHs are classically asso-
ciated with the adaptive response, by which
they are eliminated from tissues. Nevertheless,
some high-molecular-weight PAHs such as
benzo(a)pyrene are converted to carcinogenic
reactive intermediates by CYP1A (Phillips
1983), and it is widely held that much of the
acute toxicity of PAHs is due to oxidative
stress and cellular damage arising from
CYP1A catalytic activity.
In fish embryos, PAH and dioxin toxicities
are usually equated because exposure to potent
Address correspondence to J.P. Incardona,
Environmental Conservation Division, Northwest
Fisheries Science Center, National Oceanic and
Atmospheric Administration, 2725 Montlake Blvd. E,
Seattle, WA 98112 USA. Telephone: (206) 860-
3347. Fax: (206) 860-3335. E-mail: john.incardona@
noaa.gov
Supplemental Material is available online at http://
ehp.niehs.nih.gov/docs/2005/8230/supplement.pdf
We thank J. Stegeman and B. Woodin for provid-
ing monoclonal antibody 1-12-3; T. Linbo for fish
husbandry; and P. Swanson and K. Peck-Miller for
manuscript reviews.
This work was supported by the National Oceanic
and Atmospheric Administration Coastal Storms
Program; by grants to J.P.I. and N.L.S. from the
California Department of Fish and Game’s Oil Spill
Response Trust Fund through the Oiled Wildlife
Care Network, University of California, Davis; and
by a grant to H.T. from the Japanese Ministry of
Education, Culture, Sports, Science and Technology,
and Cooperative Research from Active Research in
Rakuno Gakuen University. J.P.I. was supported by
the National Academies/National Research Council
Research Associateships Program.
The authors declare they have no competing
financial interests.
Received 19 April 2005; accepted 10 August 2005.
Aryl Hydrocarbon Receptor–Independent Toxicity of Weathered Crude Oil
during Fish Development
John P. Incardona,1Mark G. Carls,2Hiroki Teraoka,3Catherine A. Sloan,4Tracy K. Collier,1and Nathaniel L. Scholz1
1Ecotoxicology and Environmental Fish Health Program, Environmental Conservation Division, Northwest Fisheries Science Center,
National Oceanic and Atmospheric Administration, Seattle, Washington, USA; 2Auke Bay Laboratory, Alaska Fisheries Science Center,
National Oceanic and Atmospheric Administration, Juneau, Alaska, USA; 3Department of Toxicology, School of Veterinary Medicine,
Rakuno Gakuen University, Ebetsu, Japan; 4Environmental Assessment Program, Environmental Conservation Division, Northwest
Fisheries Science Center, National Oceanic and Atmospheric Administration, Seattle, Washington, USA
Polycyclic aromatic hydrocarbons (PAHs), derived largely from fossil fuels and their combustion, are
pervasive contaminants in rivers, lakes, and nearshore marine habitats. Studies after the Exxon Valdez
oil spill demonstrated that fish embryos exposed to low levels of PAHs in weathered crude oil develop
a syndrome of edema and craniofacial and body axis defects. Although mechanisms leading to these
defects are poorly understood, it is widely held that PAH toxicity is linked to aryl hydrocarbon recep-
tor (AhR) binding and cytochrome P450 1A (CYP1A) induction. Using zebrafish embryos, we show
that the weathered crude oil syndrome is distinct from the well-characterized AhR-dependent effects
of dioxin toxicity. Blockade of AhR pathway components with antisense morpholino oligonucleotides
demonstrated that the key developmental defects induced by weathered crude oil exposure are medi-
ated by low-molecular-weight tricyclic PAHs through AhR-independent disruption of cardiovascular
function and morphogenesis. These findings have multiple implications for the assessment of PAH
impacts on coastal habitats. Key words: cardiovascular function, fish development, non-point source
pollution, oil spill. Environ Health Perspect 113:1755–1762 (2005). doi:10.1289/ehp.8230 available
via http://dx.doi.org/ [Online 10 August 2005]

Page 2

AhR ligands such as 2,3,7,8-tetrachloro-
dibenzo-p-dioxin (TCDD) induces a super-
ficially similar syndrome (Peterson et al. 1993).
In zebrafish (Danio rerio), a brief exposure to
TCDD shortly after fertilization results in the
appearance of vascular dysfunction, pericardial
and yolk sac edema, and anemia in hatching-
stage larvae at 72–96 hr postfertilization (hpf)
(Belair et al. 2001; Henry et al. 1997). These
effects of TCDD exposure require a func-
tional AhR. Because of genome duplication,
many teleosts have two AhR genes, AhR1 and
AhR2 (Hahn 2002). AhR1 protein is most
similar in structure to the single mammalian
AhRs, whereas AhR2 is divergent. Although
TCDD is a ligand for both receptors from
several fish species, AhR2 transcripts are more
abundant and widely distributed (Hahn
2002; Karchner et al. 1999), and in zebrafish
only AhR2 was found to be a functional
receptor for TCDD and other common halo-
genated AhR ligands (Andreasen et al. 2002).
Consistent with these findings, targeted
knockdown of AhR2 in zebrafish embryos
with antisense morpholino oligonucleotides
(MOs) prevented all of the toxic effects of
TCDD that occurred within the time frame
of morpholino efficacy (Prasch et al. 2003;
Teraoka et al. 2003). However, AhR-depen-
dent developmental defects were CYP1A
independent. A CYP1A morpholino did not
alter dioxin toxicity in zebrafish embryos,
implicating other AhR target genes in dioxin
pathophysiology (Carney et al. 2004).
A previous analysis showed that micro-
molar concentrations of individual tricyclic
PAHs representing the homologous series
most abundant in weathered crude oil (fluo-
rene, dibenzothiophene, and phenanthrene)
caused a syndrome of edema and craniofacial
and body axis defects after dose-dependent
cardiac dysfunction that was first observed at
about 36 hpf (Incardona et al. 2004). These
compounds caused cardiac arrhythmias that
are characteristic of drugs known to block car-
diac K+channels of the human ether-a-go-
go–related gene (HERG) family (Langheinrich
et al. 2003). Among four-ring compounds,
chrysene (9 µM), which is enriched in highly
weathered crude oil, was nontoxic, whereas
pyrene (1–5 µM), which is generally absent
from weathered ANS crude oil, induced a syn-
drome with features similar to TCDD expo-
sure that occurred between 80 and 96 hpf.
Here, we show that a) both toxic and non-
toxic PAHs induce CYP1A in zebrafish
embryos, acting tissue specifically through
AhR1 and AhR2; b) pyrene toxicity is AhR
dependent, whereas tricyclic PAH toxicity is
not; c) weathered ANS crude oil causes a syn-
drome in zebrafish embryos that is both clearly
distinct from TCDD toxicity and consistent
with cardiac dysfunction expected from the
most abundant tricyclic compounds; and
d) the cardiovascular toxicity of weathered
crude oil is independent of both AhR1 and
AhR2. Rather than mediating the embryo-
larval toxicity of weathered crude oil, the AhR
pathway confers a measure of protection
against the pathophysiologic effects of tricyclic
PAHs on the developing fish heart.
Materials and Methods
Chemicals. Dibenzothiophene (> 99%),
phenanthrene (> 99.5%), pyrene (> 99%),
and chrysene (98%) were obtained from
Sigma-Aldrich (St. Louis, MO, USA). Stock
PAH solutions were made in dimethyl sulfox-
ide (DMSO; tissue culture grade; Sigma-
Aldrich) at 10 mg/mL, except chrysene
(1 mg/mL). DMSO was ≤ 0.1% in exposure
solutions.
Zebrafish exposures. Wild-type AB strain
zebrafish were maintained and fertilized
eggs processed as previously described
(Incardona et al. 2004). Fish were treated
humanely and anesthetized when necessary.
Exposures to individual model PAH com-
pounds were carried out in plastic six-well
plates (15–25 embryos in 3 mL) with a static
renewal protocol at 28.5°C as described previ-
ously (Incardona et al. 2004). All exposures
used doses that were above the solubility of
the compounds and that, for toxic PAHs,
produced effects in 100% of the embryos
(Incardona et al. 2004). Each experiment was
replicated at least three times. For crude oil
exposures, embryos were incubated statically in
a water-accommodated fraction (WAF) of
crude oil, or in the continuously flowing efflu-
ent from a gravel column. The WAF was pre-
pared by an overnight high-energy spin of
50 mL ANS crude [partially weathered by
heating to 70°C until 20% reduction in mass
(Marty et al. 1997)] in 30 L zebrafish system
water using a large fiberglass tank and a paint
mixer fixed to a fan motor. The WAF con-
tained an estimated 2.8 mg/L total PAH (ini-
tial) but has the disadvantage of a high
proportion of other oil components such as
alkanes because a large fraction is present in
particulate or colloidal form. WAF was diluted
with system water to 1:2, 1:5, 1:10, and 1:100,
and static exposures were performed in 30 mm
glass Petri dishes (25 embryos in 4 mL) in an
incubator at 28.5°C.
We used oiled gravel columns to achieve a
more environmentally relevant exposure.
System water was passed by gravity through
gravel (4–6 mm grain diameter) coated with
partially weathered ANS crude oil (6.0 g
oil/kg gravel) to model conditions in oiled
intertidal substrate as described previously
(Marty et al. 1997; Short and Heintz 1997).
Controls were similarly incubated in water
passed through clean gravel. Dosing columns
were 2-L glass beakers filled with 1.3 kg rock,
with water flow directed to the bottom
through 6 mm glass tubes. Columns were
placed in glass baking dishes set at a slight
angle (~ 6°); effluent overflowed from the tops
of the columns, filling the baking dishes as a
reservoir for exposing embryos in replicate
open glass 30 mm Petri dishes (n = 4–5;
~ 25–50 embryos/dish). Temperature was
maintained with submersible aquarium
heaters. Column flow was initiated 1 day
before embryo exposure to further weather the
oil and remove any particulates. Exposures
were started at 4–8 hpf. Temperature (nearest
0.5°C) and flow rate (nearest milliliter) were
recorded as often as hourly during the day and
at one or two time points during the night
[Supplemental Table 1; Supplemental Material
available online (http://ehp.niehs.nih.gov/docs/
2005/8230/supplement.pdf)]. Although
embryos developed more slowly at the experi-
mental temperatures, they were staged accord-
ing to the published standard series (Kimmel
et al. 1995), and all developmental times are
reported as hpf at standard temperature
(28.5°C). We performed six oiled gravel exper-
iments; two included uninjected embryos only
(n > 300), three included AhR2 morphants
(n > 200), two included AhR1/AhR2 double
morphants (n = 145), and one included
CYP1A morphants (n = 87). Standard statisti-
cal analyses were carried out with Microsoft
Excel 2004 for Mac (Microsoft Corporation,
Redmond, WA, USA).
Morpholino injections. AhR1 cDNA and
genomic sequences are available as GenBank
AF258854 (GenBank 2005) and Ensembl
ENSDARG00000020046 (Ensembl 2005),
respectively, and AhR2 cDNA sequence as
GenBank AF063446 (GenBank 2005). All
morpholinos were synthesized by GeneTools
(Philomath, OR, USA) and are listed below
(mismatch nucleotides in control morpholinos
are indicated by lowercase letters). The trans-
lation-blocking MOs targeting zfAHR2
(5´-TGTACCGATACCCGCCGACA-
TGGTT-3´ for zfAHR2-MO, 5´-TGaAC-
CcATACCCGCCGtCATcGTT-3´ for the
negative control 4Mis-AHR2-MO) and
zfCYP1A (5´-TGGATACTTTCCAGT-
TCTCAGCTCT-3´) have been described pre-
viously (Teraoka et al. 2003). Splice-blocking
morpholinos for zfAhR1 were designed to tar-
get the exon 2/intron 2 splice donor site
(5´-CTTTTGAAGTGACTTTTGGCC-
CGCA-3´ for E2I2-MO, 5´-CTTTTcAAcTG-
AgTTTTGcCCCcCA-3´ for 5Mis-E2I2-MO)
and the intron 2/exon 3 splice acceptor site
(5´-GTTCAGGGTTACTGCAAAAGAAAT-
3´ for I2E3-MO). Morpholinos were injected
at the 1–4 cell stage (0.25–1 hpf) as previously
described (Incardona et al. 2004; Teraoka et al.
2003), and embryos were allowed to recover in
system water at 28.5°C to 50% epiboly
(5–6 hr) before use in exposure studies. AhR1
morpholinos were labeled with fluorescein,
Incardona et al.
1756
VOLUME 113 | NUMBER 12 | December 2005 • Environmental Health Perspectives

Page 3

whereas AhR2 and CYP1A morpholinos were
not. For injections involving AhR1, embryos
were selected on an epifluorescent stereoscope
based on fluorescence intensity and an even
distribution in blastomeres.
Imaging of live embryos/larvae, immuno-
fluorescence, and confocal microscopy. Digital
still micrographs were obtained and video-
microscopy of live embryos and larvae per-
formed as described previously (Incardona
et al. 2004). Antibodies used were monoclonal
1-12-3 against fish CYP1A (Park et al. 1986),
anti-myosin heavy chain monoclonal MF20
(Developmental Studies Hybridoma Bank,
University of Iowa, Iowa City, IA, USA)
(Bader et al. 1982), and anti-atrial myosin
heavy chain monoclonal S46 (Berdougo et al.
2003). For CYP1A immunofluorescence,
embryos were fixed overnight in 4% phos-
phate-buffered paraformaldehyde, and
MF20/S46 immunofluorescence was assessed
in embryos fixed in either paraformaldehyde or
methanol plus 10% DMSO. Processing for
immunofluorescence was carried out as
described previously (Incardona et al. 2004).
Secondary antibodies were AlexaFluor488-con-
jugated goat anti-mouse IgG1 (S46) and
AlexaFluor568-conjugated goat anti-mouse
IgG2b (MF20), both from Molecular Probes
(Eugene, OR, USA). Immunolabeled embryos
were mounted in glycerol or 3% methylcellu-
lose and imaged using a Zeiss LSM 5 Pascal
confocal system with Ar and HeNe lasers (Carl
Zeiss Advanced Imaging Microscopy, Jena,
Germany). For semiquantitative comparisons,
treated or control embryos were marked by tail
clipping and mixed for antibody labeling,
mounted together, and imaged with identical
settings.
PAH analysis. Water samples (200 mL)
were collected and stored in brown glass bottles
with 20 mL dichloromethane at 4°C for up to
7 days before extraction. After addition of
deuterated internal standards, samples were
extracted twice with dichloromethane (25 mL
each time) for 2 min each with 2 min separa-
tion times using 1-L separatory funnels. For
quality assurance, a known mixture of PAHs
was added to 200 mL zebrafish system water
(spiked blank) and extracted. The extracts were
processed and hydrocarbons analyzed by gas
chromatography–mass spectrometry using
selected ion monitoring as previously described
(Sloan et al. 2004). The accuracy of the hydro-
carbon analyses was estimated by recoveries
from the spiked blank, which ranged from 69
to 111% for 17 different PAHs (naphthalene,
102%; fluorene, 85%; dibenzothiophene, not
determined; phenanthrene, 98%; chrysene,
83%). Total PAH concentrations were calcu-
lated by summing concentrations of individual
PAHs. Relative PAH concentrations were cal-
culated as the ratio of PAH concentration to
the total PAH concentration.
Results
Differential activation of AhR1 and AhR2 by
model PAHs. We first examined the relation-
ship between CYP1A induction via the AhR
and the developmental toxicity of individual
PAH compounds [structures shown in
Supplemental Figure 1; Supplemental Material
available online (http://ehp.niehs.nih.gov/docs/
2005/8230/supplement.pdf)]. To determine
the roles of the two zebrafish AhRs, we used
the previously described (Prasch et al. 2003;
Teraoka et al. 2003) translation-blocking
AhR2 morpholino (AhR2-MO), and designed
AhR1 splice-blocking morpholinos targeting
either the exon 2/intron 2 junction or the
intron 2/exon 3 junction (Supplemental
Figure 1; Supplemental Material available
online (http://ehp.niehs.nih.gov/docs/2005/
8230/supplement.pdf)]. Morpholinos designed
to block splicing or translation have similar
efficacy (Draper et al. 2001). Coinjection of
AhR2-MO and AhR1 exon 2/intron 2 mor-
pholino failed to produce normal embryos, so
all studies used intron 2/exon 3 morpholino
(AhR1-MO). Nonfunctional control mor-
pholinos included the AhR2 sequence with a
four-base mismatch (AhR2-MIS) and the
AhR1 exon 2/intron 2 sequence with a five-
base mismatch (AhR1-MIS).
At nominal concentrations ranging from
10 to 60 µM, phenanthrene or dibenzo-
thiophene causes dose-dependent changes in
cardiac rhythm ranging from bradycardia
through partial (2:1) to complete atrio-
ventricular (AV) conduction block (Incardona
et al. 2004). In embryos exposed to approxi-
mately 30–60 µM phenanthrene or dibenzo-
thiophene with AV block and edema
[Figure 1C, Supplemental Movie 1; Supple-
mental Material available online (http://ehp.
niehs.nih.gov/docs/2005/8230/supplement.
pdf)], relatively weak CYP1A immuno-
fluorescence was observed predominantly in
vessels most proximal to the heart, including
the first aortic arch and the carotid artery
(Figure 1D,G). However, CYP1A immuno-
fluorescence was not observed in endothelial
cells lining the heart in embryos exposed to
either phenanthrene or dibenzothiophene
(Figure 1D,G). Injection of AhR2-MO largely
blocked CYP1A induction by dibenzo-
thiophene and phenanthrene (Figure 1F,H;
Table 1). However, despite this effect on
CYP1A induction, AhR2 morphants (i.e., MO
injected) were not protected from phenan-
threne- or dibenzothiophene-induced cardiac
dysfunction (Figure 1E, Table 1). The same
types of cardiac arrhythmia were observed in
AhR2 morphants exposed to phenanthrene or
dibenzothiophene at the same developmental
stage as controls [~ 36 hpf; data not shown,
Supplemental Movie 1; Supplemental Material
available online (http://ehp.niehs.nih.gov/docs/
2005/8230/supplement.pdf)]. Although we
did not quantify changes in the dose response,
AhR2 morphants generally had a higher degree
AhR-independent toxicity of petrogenic PAHs
Environmental Health Perspectives • VOLUME 113 | NUMBER 12 | December 2005
1757
Figure 1. AhR2 knockdown prevents CYP1A induction by phenanthrene and dibenzothiophene, but not car-
diac dysfunction. (A–F) Lateral light microscopic views of live embryos at 48 hpf (anterior at left) are paired
with corresponding ventral confocal images (anterior at top) of CYP1A (green) and myosin heavy chain (red)
immunofluorescence. (A, B) Embryo exposed to solvent (DMSO). (C, D) AhR2-MIS–injected embryo exposed
to 28 µM phenanthrene. (E, F) AhR2 morphant exposed to 28 µM phenanthrene. Black arrowheads and
arrows (C, E) indicate pericardial and yolk sac edema, respectively. CYP1A immunofluorescence induced
by phenanthrene (D) in the cranial division of the internal carotid artery (CrDI) and optic artery (OA) was
blocked by AhR2-MO injection (F). Solid white arrowheads indicate cross-reactive immunofluorescence in
the jaw cartilage, and unfilled white arrowheads indicate the ventricular myocardium. (G, H) Higher magni-
fication confocal images showing CYP1A (green) and myocardial myosin heavy chain (red) immunofluores-
cence at 48 hpf in embryos with cardiac dysfunction after exposure to 28 µM dibenzothiophene (lateral
views with anterior at left). In an uninjected embryo (G), the proximal portion of the mandibular arch (AA1,
arrow) is CYP1A+, whereas the ventricular endothelium is CYP1A–(asterisk). Only cross-reactive signal is
seen in the jaw cartilage (arrowhead) in an AhR2 morphant (H). Bars = 100 µm (A–F) and 50 µm (G, H).

Page 4

of AV block than did controls at a given
phenanthrene or dibenzothiophene concentra-
tion, consistent with a protective effect of
CYP1A induction. Injection of AhR1-MO
alone did not prevent CYP1A induction by
phenanthrene (data not shown) and, in combi-
nation with AhR2-MO, did not prevent tri-
cyclic PAH toxicity (Supplemental Movie 1;
Supplemental Material available online (http://
ehp.niehs.nih.gov/docs/2005/8230/supple-
ment.pdf)]. Therefore, although the tricyclic
PAHs phenanthrene and dibenzothiophene
induce CYP1A weakly in some blood vessels
through AhR2, the primary toxicity of these
PAHs in fish embryos is AhR independent,
and their cardiac effects are not associated with
AhR activation or CYP1A induction in the
endocardium.
In contrast to the tricyclic PAHs, the effects
of pyrene during zebrafish development over-
lap considerably with those previously reported
for TCDD exposure. Despite widespread
induction of CYP1A throughout the vascular
endothelium observed as early as 36 hpf
(Table 1 and data not shown), the overt signs
of pyrene toxicity do not appear until after
80 hpf (Figure 2). By 100 hpf, uninjected or
AhR2-MIS–injected embryos exposed to
pyrene showed edema (Figure 2A,C and data
not shown), cell death in the neural tube
(Figure 2E), and anemia (Figure 2G), as well as
CYP1A immunofluorescence in the liver
(Figure 2K) and throughout the vascular
endothelium of the trunk and head (Figure
2I,K). Most pyrene-exposed larvae die later in
the fifth day of development (Table 1).
Injection of AhR2-MO largely prevented the
morphologic and lethal effects of pyrene expo-
sure (Figure 2B,D,F,H; Table 1) and markedly
reduced the levels of CYP1A immuno-
fluorescence assayed at both 48 hpf (Table 1)
and 100 hpf (Figure 2J,L). At comparable time
points, CYP1A morphants exposed to pyrene
generally had defects that were less severe than
those of controls. For example, pericardial
edema was more severe at 96 hpf in AhR-
MIS–injected larvae than in CYP1A mor-
phants, indicated by pericardial cross-sectional
areas of 0.034 ± 0.008 mm2and 0.021 ±
002 mm2, respectively (n = 10 for each, t-test
p < 0.001). However, CYP1A-MO injection
did not ultimately protect from pyrene toxicity
because all of the defects associated with pyrene
exposure (including lethality) appeared in
CYP1A morphants 12–18 hr later than in con-
trols, possibly due to loss of MO efficacy at
later stages (Table 1). These findings indicate
that, like phenanthrene and dibenzothiophene,
pyrene selectively activates AhR2, but unlike
these tricyclic PAHs, pyrene toxicity is AhR
dependent. Moreover, delay of pyrene toxicity
in CYP1A morphants suggests the involvement
of a toxic CYP1A-derived pyrene metabolite.
A trivial explanation for the absence of
embryonic toxicity for some compounds [e.g.
chrysene (Incardona et al. 2004)] could be a
lack of tissue uptake during the course of the
exposure. For example, because of differences in
water solubility of PAHs with different ring
arrangements but similar molecular weight
(e.g., 0.67 µM pyrene vs. 0.009 µM chrysene),
the effective exposure levels would be very dif-
ferent. However, exposure to 9 µM chrysene,
which produces no apparent toxic effects,
resulted in robust CYP1A immunofluorescence
throughout the epidermis and the vascular
endothelium (Figure 3A,G). Most cranial
(Figure 3G and data not shown) and trunk ves-
sels (data not shown), as well as the endothelial
cells lining the cardiac ventricle (Figure 3H),
expressed CYP1A after chrysene exposure.
Remarkably, injection of AhR2-MO blocked
induction of CYP1A in the epidermis by chry-
sene while leaving the vascular induction intact
and generally increased (Figure 3B). In con-
trast, injection of AhR1-MIS (Figure 3C) or
AhR1-MO (Figure 3D) had no effect on
CYP1A induction by chrysene, but coinjection
of both AhR1-MO and AhR2-MO markedly
reduced both epidermal and endothelial
CYP1A immunofluorescence (Figure 3E).
Injection of CYP1A-MO eliminated virtually
all CYP1A immunofluorescence associated with
chrysene exposure (Figure 3F). These findings
indicate that chrysene can activate both AhR1
and AhR2 in a tissue-specific manner.
Although either AhR1 or AhR2 can mediate
the vascular induction of CYP1A by chrysene,
the epidermal induction of CYP1A by chrysene
is AhR2 specific. However, AhR activation and
CYP1A induction by chrysene are not associ-
ated with any overt developmental toxicity or
cardiac dysfunction (despite CYP1A induction
in cardiac endothelial cells).
Embryonic cardiac dysfunction and the
weathered crude oil syndrome. We exposed
zebrafish embryos to ANS crude oil weathered
using two methods: generation of a WAF by
Incardona et al.
1758
VOLUME 113 | NUMBER 12 | December 2005 • Environmental Health Perspectives
Figure 2. AhR2 morphants are resistant to pyrene toxicity. Control (uninjected or AhR2-MIS injected) and
AhR2 morphant embryos were exposed to 5 µM pyrene through 100 hpf. Lateral (A, B) and dorsal (C, D)
views showing edema (arrows) in uninjected larvae. Higher magnification light micrographs of the trunk
region of uninjected (E, G) and AhR2 morphant (F, H) larvae showing cell death (E, granular appearance) in
the neural tube (nt, neural tube; nc, notochord) and a reduction of erythrocytes (G, arrows) in the ventral
aorta (VA) and caudal vein (CV) of uninjected larvae. (I–L) CYP1A immunofluorescence in the trunk (I, J) and
head regions (K, L) of pyrene-exposed larvae. In AhR2-MIS–injected larvae (I, K) the vasculature (DLAV,
dorsal longitudinal anastomotic vessel; Se, intersegmental vessels; AA, branchial arches) and liver are
CYP1A+, whereas only weak signal is seen in the liver of the AhR morphant (J, L). In (E–L,) anterior is to the
left and dorsal at top. Bars = 200 µm (A–D) and 50 µm (E–L).
Table 1. AhR2 morpholino prevents pyrene toxicity but not tricyclic PAH toxicity (%).
Treatment
AhR2-MIS + 28 µM phenanthrene
AhR2-MO + 28 µM phenanthrene
Uninjected + 28 µM dibenzothiophene
AhR2-MO + 28 µM dibenzothiophene
Uninjected + DMSO
Uninjected + 5 µM pyrene
AhR2-MIS + 5 µM pyrene
AhR2-MO + 5 µM pyrene
CYP1A-MO + 5 µM pyrene
CYP1A+at 48 hpf
100 (n = 17)
7 (n = 27)
100 (n = 20)
0 (n = 17)
0 (n = 43)
100 (n = 39)
94 (n = 16)
3 (n = 31)
7 (n = 59)
Cardiac dysfunction at 48 hpf
100 (n = 22)
100 (n = 72)
100 (n = 20)
100 (n = 17)
—
—
—
—
—
Viability at 106 hpf
—
—
—
—
100 (n = 90)
9 (n = 93)
0 (n = 70)
86 (n = 56)
92 (n = 39)a
—, not applicable.
aMortality delayed 18 hr relative to control; 0% viable at 124 hpf.

Page 5

an overnight high-energy spin of an oil–water
mixture and incubation in the effluent from a
column loaded with oiled gravel (OGE) or
control (clean) gravel (CGE). Although chem-
ical analysis demonstrated that these two
methods produced different degrees of weath-
ering [Table 2, Supplemental Figures 2 and 3;
Supplemental Material available online
(http://ehp.niehs.nih.gov/docs/2005/8230/
supplement.pdf)], exposure of embryos to
both preparations (WAF exposed and OGE
exposed) produced identical types of biologic
effects (Figure 4 and data not shown).
In general, the defects in WAF-exposed
embryos were more severe, consistent with its
higher degree of weathering (i.e., larger tricyclic
PAH fraction). By 63 hpf, WAF- or OGE-
exposed embryos showed a suite of defects that
overlapped considerably with exposure to indi-
vidual model tricyclic PAHs, but with addi-
tional features not observed with the model
compounds. Grossly, all embryos exposed to
weathered crude oil showed dorsal curvature of
the body axis; mild to moderate pericardial
edema was seen in OGE-exposed embryos
(Figure 4B,D), and yolk sac edema was seen in
WAF-exposed embryos (data not shown). In
both types of exposures, changes in cardiac
function were the earliest observed defects.
Mild pericardial edema and reduced blood
flow associated with poor cardiac contractility
and bradycardia were apparent in WAF-
exposed embryos at 33 hpf (data not shown)
and by 36–39 hpf in OGE-exposed embryos
[Supplemental Movie 2; Supplemental
Material available online (http://ehp.niehs.nih.
gov/docs/2005/8230/supplement.pdf)].
Usually, the first and mildest sign of cardiac
dysfunction was regurgitation of erythrocytes
from the atrium into the yolk sac [Supple-
mental Movie 3; Supplemental Material avail-
able online (http://ehp.niehs.nih.gov/docs/
2005/8230/supplement.pdf)]. Staining with
the chamber-specific antibodies MF20 and
S46 demonstrated that subtle delay or disrup-
tion of cardiac looping was associated with the
earliest observed defects in cardiac function
(Figure 4E,F). Cardiac looping was assessed
quantitatively by measuring the angle between
the cardiac chambers relative to the left–right
body axis (Figure 4E,F). CGE-exposed
embryos fixed at 39 hpf had a mean inter-
chamber angle of 28 ± 5° (n = 6), whereas
OGE-exposed embryos with weak contractility
had a mean interchamber angle of 50 ± 2° (n =
7, t-test p < 0.01). All WAF- or OGE-exposed
embryos showed severely abnormal cardiac
looping at subsequent stages (typically by
54–64 hpf), and exposures were terminated at
72–80 hpf. The late cardiac morphology typi-
cally showed chambers that were stretched
along the anterior–posterior axis, with the ven-
tricle stiff and reduced in diameter and the
atrium relatively dilated (Figure 4D). Atypical
movement was often observed in the wall of
the cardiac outflow tract or bulbus arteriosus
[Supplemental Movie 4; Supplemental
Material available online (http://ehp.niehs.nih.
gov/docs/2005/8230/supplement.pdf)]. In
most cases, and in particular in WAF-exposed
embryos, both cardiac chambers ultimately
collapsed into a stringlike structure when
embryos were exposed for longer durations
(data not shown).
Intracranial hemorrhage and doming of
the head due to ventricular edema was very
common among embryos exposed to either
preparation (85% of OGE-exposed embryos
in one experiment, n = 66) and most often
involved the mesencephalon/third ventricle
and hindbrain/fourth ventricle (Figure 4H).
Hemorrhage occasionally involved the
branchial arches or eye but was never observed
in tissues outside the head (data not shown).
The sensitive period for intracranial hemor-
rhage was late during the second day of devel-
opment to early in the third day; although not
observed at 30 hpf, hemorrhage first occurred
between 30 and 39 hpf, with the number of
affected embryos maximal by 58–63 hpf.
Finfold defects involving all fins and con-
sisting of irregular edges or blisters were also
common by 63–72 hpf (94% of OGE-exposed
embryos, n = 66; Figure 4J,L). No abnormali-
ties were observed in CGE-exposed embryos.
Although we did not observe arrhythmias
typical of AV conduction block in OGE-
exposed embryos, the total PAH levels were
well below the levels at which either phenan-
threne or dibenzothiophene alone blocks AV
conduction. The key finding is that the com-
plex PAH mixture that comprises weathered
crude oil caused early cardiac dysfunction simi-
lar to effects of model tricyclic PAHs, rather
than a syndrome that arises later during the lar-
val period, such as that associated with pyrene
or TCDD exposure. Weathered crude oil also
produced additional defects in zebrafish
embryos (intracranial hemorrhage and finfold
defects) that either are not observed with expo-
sure to individual model PAHs or dioxins, or
arise during an earlier developmental stage
than AhR-mediated TCDD toxicity.
A protective role for the AhR pathway.
Embryos exposed to WAF or OGE showed
identical robust patterns of CYP1A induction
by 36 hpf. In OGE-exposed embryos, CYP1A
immunofluorescence (Figure 5, green) was
strong in the epidermis (Figure 5B), cranial
vasculature (Figure 5C), trunk vasculature
(data not shown), and the endothelium lining
both the atrium (Figure 5D) and ventricle
AhR-independent toxicity of petrogenic PAHs
Environmental Health Perspectives • VOLUME 113 | NUMBER 12 | December 2005
1759
Table 2. Summary of PAH levels in weathered crude
oil exposures.
Weathered oil preparation
(experiment)
WAF day 1
WAF day 4
OGE day 1 (column 1)
OGE day 4 (column 1)
OGE day 0 (column 6)
OGE day 2 (column 6)
OGE day 7 (column 6)
Total PAH
(µg/L)
1,549
264
78.0
53.5
111.1
53.2
52.7
Tricyclic
PAHs (%)
43.7
55.8
16.9
24.1
17.6
25.8
28.9
Figure 3. Chrysene induces CYP1A through both AhR1 and AhR2. All images show CYP1A immunofluores-
cence at 72 hpf (A, B, G, H) or 48 hpf (C–F) after exposure to 9 µM chrysene from 6 hpf. (A–F) Lateral epiflu-
orescent images with anterior to the left in uninjected (A), AhR2 morphant (B), AhR1-MIS–injected (C), AhR1
morphant (D), AhR1/AhR2 double morphant (E), and CYP1A morphant (F) embryos. Epidermal CYP1A is seen
as punctate fluorescence on the surface of the embryos. Immunofluorescent signal in the otic capsule and
jaw cartilage was often observed in unexposed embryos. This signal was resistant to CYP1A morpholino (F)
and is therefore likely to represent a cross-reactive protein. (G, H) Confocal images of uninjected chrysene-
exposed embryos. (G) Three-dimensional confocal projection (180 µm series of optical sections) of CYP1A
immunofluorescence, ventral view with anterior at top. Arrows indicate CYP1A+blood vessels; AA1,
mandibular arch; CrDI, cranial division of the internal carotid artery; OA, optic artery; ORA, opercular artery.
(H) Confocal optical section through the cardiac chambers (anterior at top) with CYP1A (green) and myosin
heavy chain (red) immunofluorescence. The asterisk (*) indicates CYP1A+endothelial cells lining the ventri-
cle. Bars = 250 µm (A–F) and 50 µm (G, H).