Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Identification and characterisation of androgen– responsive genes in zebrafish embryos Eva Fetter, Sona Smetanova, Lisa Baldauf, Annegret Lidzba, Rolf Altenburger, Andreas Schüttler, and Stefan Scholz Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b01034 • Publication Date (Web): 26 Aug 2015 Downloaded from http://pubs.acs.org on August 27, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 28
Environmental Science & Technology
1
Identification and characterisation of androgen–responsive genes in zebrafish embryos
2
Eva Fetter1*, Soňa Smetanová1,2, Lisa Baldauf1, Annegret Lidzba1, Rolf Altenburger1, Andreas
3
Schüttler1, Stefan Scholz1
4
1
5
Permoserstraße 15, 04318 Leipzig, Germany
6
2
7
Faculty of Science, Kamenice 753/5, 625 00 Brno, Czech Republic
UFZ – Helmholtz Centre for Environmental Research, Department Bioanalytical Ecotoxicology,
RECETOX – Research Centre for Toxic Compounds in the Environment, Masaryk University,
1 Environment ACS Paragon Plus
Environmental Science & Technology
Page 2 of 28
8
Abstract
9
Responsive genes for fish embryos have been identified so far for some endocrine pathways but not
10
for androgens. Using transcriptome analysis and a multiple concentration–response modelling, we
11
identified putative androgen–responsive genes in zebrafish embryos exposed to 0.05–5000 nM 11–
12
ketotestosterone for 24 hours. Four selected genes with sigmoidal concentration–dependent
13
expression profiles (EC50=6.5–30.0nM) were characterised in detail. The expression of cyp2k22 and
14
slco1f4 was demonstrated in the pronephros, lipca was detected in the liver and sult2st3 was found
15
in the olfactory organs and choroid plexus. Their expression domains, the function of human
16
orthologs and a pathway analysis suggested a role of these genes in the metabolism of hormones.
17
Hence, it was hypothesised that they were induced to compensate for elevated hormone levels. The
18
induction of sult2st3 and cyp2k22 by 11–ketotestosterone was repressed by co–exposure to the
19
androgen receptor antagonist nilutamide supporting a potential androgen receptor mediated
20
regulation. Sensitivity (expressed as EC50 values) of sult2st3 and cyp2k22 gene expression
21
induction
22
ketotestosterone~testosterone>progesterone>cortisol>ethinylestradiol) correlated with their known
23
binding affinities to zebrafish androgen receptor. Hence, these genes might represent potential
24
markers for screening of androgenic compounds in the zebrafish embryo.
after
exposure
to
other
steroidal
hormones
(11–
25 26
Keywords: endocrine disruptors, microarray, animal alternatives, Danio rerio, gene expression,
27
transcriptome, chemical hazard assessment, toxicogenomics
28
2 Environment ACS Paragon Plus
Page 3 of 28
Environmental Science & Technology
29
1. Introduction
30
Chemicals that have the potential to disrupt the endocrine system of humans and wildlife are of high
31
concern given the possible long–term impact on health and population development. The focus has
32
been so far on estrogen–, androgen– and thyroid hormone regulation and an array of different in
33
vitro and in vivo test systems has been developed, targeting e.g. the binding to hormone receptors,
34
receptor transcriptional activation, enzymatic activities and protein– or hormone production [1-3].
35
Particularly for the screening of many compounds or environmental samples in vitro cellular
36
systems provide advantages such as specificity, simplicity and ability to automate the analysis.
37
Alternatively, fish embryos have been suggested as a promising approach to screen for endocrine
38
disruption [4]. In contrast to cellular assays, zebrafish embryos represent a complex system closely
39
related to adult animals [5]. Further advantages of fish embryos in contrast to cellular assays are
40
that they allow testing of hormonal effects in the context of other relevant endpoints such as
41
neurotoxicity [6] or teratogenicity [7].
42
Given the small dimension, transparency, and rapid ex utero development, the ease of production
43
and the fact that many hormone signalling pathways are at least partially established already in
44
embryonic stages [8], the zebrafish embryo as model organism is attractive and increasingly used in
45
toxicological studies. This includes endocrine disruption using endpoints such as gene expression of
46
target genes [9, 10], fluorescence of reporter genes/proteins in transgenic lines [11-14] or in in vivo
47
immunoassays [15]. According to animal welfare regulations embryonic stages of fish are not
48
protected and are therefore considered as alternatives to the testing of adult animals. [16].
49
To date several assays using fish embryos are available for the screening of (anti)estrogens or
50
goitrogens [1]. An appropriate assay for androgens is available only using a construct of the
51
stickleback spiggin gene promoter with a reporter gene in medaka embryos [17]. Other androgen
52
signalling–related sensitive genes have not yet been discovered for embryonic stages.
53
Transcriptome analyses revealed responsive genes for androgen receptor agonists [18] and
54
antagonists [19] in adult zebrafish liver and gonads, such as hydroxysteroid dehydrogenases 3 Environment ACS Paragon Plus
Environmental Science & Technology
55
(hsd17β3, hsd3β7, and hsd11β2) and cytochrome P450s (cyp2k7 and cyp17a1). These genes are
56
involved in androgen biosynthesis and metabolism. For fish embryos potential anti–androgen
57
responsive genes were identified for early stages (2 days post fertilisation) [20], but a detailed
58
concentration–response and ligand specificity analysis has not been conducted. Therefore, in this
59
study we aimed at identifying native androgen responsive genes with a concentration–dependent
60
expression in post–hatched embryonic stages of zebrafish using microarray analysis. We selected
61
the late embryonic period since previous studies with zebrafish have shown a higher sensitivity
62
against exposure to endocrine disruptors for late embryonic stages [11, 15] presumably due to the
63
onset of hormone synthesis [21] or the expression of nuclear receptors [22, 23]. Transcriptome data
64
were obtained from exposure experiments with a series of concentrations of 11–ketotestosterone
65
(0.05 – 5000 nM). We aimed at identifying genes directly regulated through the androgen receptor
66
(AR). Therefore, the time course of the expression profile and the effect of a co–exposure to the
67
androgen antagonist nilutamide and agonist 11–ketotestosterone on the gene expression were
68
investigated in more detail for the two most responsive candidate genes (sult2st3 and cyp2k22).
69
Furthermore, we assessed their specificity by comparing the response to various hormones. Since
70
little was known about the identified target genes, we also performed in situ hybridisations to
71
localise the expression domain and to obtain further information on the potential function of the
72
genes.
73
2. Experimental section
74
2.1
75
2.1.1 Chemicals
76
For the exposure experiments, the following chemicals were used (CAS registry number, purity and
77
manufacturer are given in parentheses): 11–ketotestosterone (53187–98–7, ≥98% Sigma Aldrich);
78
17α–ethinylestradiol (57–63–6, ≥98% Sigma Aldrich), cortisol (50–23–7, ≥98% Sigma Aldrich),
79
dimethylsulfoxide–DMSO (67–68–5, ROTIDRY ≥99.5%; ≤200 ppm H2O), nilutamide (63612–50–
Exposure experiments
4 Environment ACS Paragon Plus
Page 4 of 28
Page 5 of 28
Environmental Science & Technology
80
0, ≥98% Sigma Aldrich), progesterone (57–83–0, ≥99% Sigma Aldrich), testosterone (58–22–0,
81
≥98% Sigma Aldrich).
82
2.1.2 Zebrafish maintenance and exposure of embryos
83
We used the UFZ–OBI wild type strain, which has been established from a stock of a local breeder
84
and kept for several generations at the UFZ. Fish were cultured at 26±1°C at a 14:10 h light: dark
85
cycle in a recirculating tank as described by Westerfield [24]. Husbandry and experimental
86
procedures were performed in accordance with the German animal protection standards and were
87
approved by the Government of Saxony, Landesdirektion Leipzig, Germany (Aktenzeichen 75–
88
9185.64).
89
Zebrafish embryos were exposed under static conditions for 24 h (96–120 hpf in the qPCR
90
experiments, 72–96 and 96–120 hpf in the in situ hybridisation experiments) with the exception of
91
the gene expression studies on different exposure durations where the exposure length was varied
92
between 1-96 h (see details in chapter 3.5). The short exposure period was chosen in order to
93
primarily detect genes directly regulated by hormones and to minimize potential secondary effects
94
(e.g. autoregulatory feedback mechanisms). For the microarray analysis exposure to the non–
95
aromatizable fish androgen 11–ketotestosterone was conducted in the range of 0.05–5000 nM in a
96
single replicate but with three groups of control embryos (incubated in embryonic medium). For
97
each sample 40–50 embryos were exposed in a volume of 100 ml. At the end of the exposure
98
embryos were pooled in order to obtain enough material for microarray analysis and to avoid the
99
necessity of further amplification steps and to obtain candidate genes with robust expression
100
patterns. The 11–ketotestosterone concentration in the exposure media were analysed using an 11–
101
ketotestosterone ELISA kit (Biomol GmbH, Hamburg, DE). About 80% of the nominal
102
concentrations were detected after 24 h of exposure (Table S1). Validation experiments using qPCR
103
were performed in three independent experiments pooling 40–50 embryos for each sample. For the
104
combined exposure with 11–ketotestosterone and nilutamide two independent experiments were
105
performed pooling 40–50 embryos for each sample. 5 Environment ACS Paragon Plus
Environmental Science & Technology
Page 6 of 28
106
Penetration of RNA probes may be limited in later embryonic stages. Therefore, the exposure
107
window of 72–96 hpf was used in addition to the 96–120 hpf exposure to 500 nM 11–
108
ketotestosterone for part of the in situ hybridisation experiments. At least 25 individuals were
109
analysed for each hybridisation probe and treatment, and a hybridisation has been repeated at least
110
twice with the same probe in independent experiment.
111
To prepare 11–ketotestosterone solutions no solvents were used. For all other compounds, embryos
112
were exposed to dilutions of DMSO stock solutions (1 mg/mL for AR agonists and 7 mg/mL for
113
nilutamide) in embryonic medium prepared according to the OECD guideline 236 [25]. An
114
exposure of embryos to 0.1% DMSO served as solvent control. Dilutions of the stocks were
115
prepared freshly before starting the exposure. Embryos were kept in an incubator at 26°C and a
116
light cycle of 14:10 hours light: dark. Only embryos showing no malformations were subjected to
117
gene expression analysis.
118
2.2
119
2.2.1 Microarray experiment
120
Total RNA was isolated from homogenates of embryos using phase–lock tubes (5Prime GmbH,
121
Hamburg) and Trizol (Invitrogen, Darmstadt) according to the manufacturer’s instructions. RNA
122
was further purified using RNeasy Mini kit (Qiagen, Hilden, Germany). RNA concentrations were
123
measured
124
Spectrophotometer (Peqlab, Erlangen, Germany) and the RNA quality was checked using an
125
Agilent 2100 Bioanalyzer (Agilent Technologies, Böblingen, Germany). RNA integrity values
126
ranged from 9.6 to 10. RNA was labelled with the one–colour low input labelling and hybridization
127
kit (Agilent Technologies) according to the manufacturer instructions. Scanning of the arrays was
128
performed at Genovia GmbH (Zwenkau, Germany) using an Agilent DNA Microarray scanner.
129
Microarray studies were performed using a commercial zebrafish oligonucleotide array
130
(representing the whole zebrafish genome) based on the Ensembl zebrafish genome version 3 but
Transcriptome analysis using microarrays
by
spectrophotometric
absorption
at
260
nm
6 Environment ACS Paragon Plus
on
a
NanoDrop
ND–1000
Page 7 of 28
Environmental Science & Technology
131
adapted to a custom 8x60k design (Amadid G4102A, Agilent). The annotation of the probes has
132
been updated to the version 9 mapping the oligonucleotide probes to the zebrafish genome
133
according to the method described by Arnold, et al. [26]. In brief, Agilent probe sequences were
134
each aligned to the Zebrafish Genome (Zv9/danRer7, Jul. 2010) using BLAT [27]. Only hits with a
135
minimum overlap of 95% were considered. All probes mapped more than once in the genome were
136
disregarded. After mapping, the sequences were annotated using R version 3.0.1 and the biomaRt
137
package [28, 29].
138
2.2.2 Statistical analysis of microarray data
139
Fluorescent intensities of individual microarray spots were extracted using the Agilent Feature
140
Extraction software (version 10.7.3.1). Raw microarray data were converted to log2 values and
141
were quantile–normalised. For further statistical analysis, fold changes in relation to the mean of the
142
controls were calculated for each treatment. In order to identify genes potentially regulated via the
143
AR, a multiple concentration–response modelling using a logistic regression was performed. The
144
four parametric log–logistic Hill concentration–response model commonly used in toxicology for
145
concentration–response analysis was applied for the fitting of fold change data.
146
y = E0 + (Emax - E0) × (C^ δ / (EC50 ^ δ +C^ δ))
147
E0 represents the minimum effect level, Emax the maximum effect level, EC50 the half maximum
148
effect concentration, δ the slope of the curve and C the exposure concentration.
149
Probes showing a concentration–dependent response were identified and ranked by calculating the
150
Akaike's Information Criterion (AIC) for each fit. The AIC was used as an indicator for the
151
goodness of fit and is based on a maximized value of likelihood function and the number of
152
parameters of the fitted model. Given the complexity and computational requirements of such a
153
modelling it was restricted to genes showing at least a ratio of four between the maximum and
154
minimum gene expression values across the treatments. Modelling was conducted for individual
155
probes (all oligos were handled separately) after rescaling of the fold change data between 0 and 1
156
(minimum–maximum normalisation). The rescaling of data prior to the concentration response 7 Environment ACS Paragon Plus
(1)
Environmental Science & Technology
Page 8 of 28
157
modelling was necessary to be able to use the AIC as an indicator for the general goodness of fit
158
among all tested responses of probes, since this criterion is dependent on the scale of the data. The
159
curve fitting and the estimation of AIC were conducted using an R script based on the ‘drc’and
160
‘epicalc’ R packages [30, 31].
161
For further analyses, we selected probes based on the EC50 values and the AIC. The selected probes
162
were clustered using the TMEV software package (version 4.9) and HCL clustering function with
163
Pearson correlation [32]. In contrast to the concentration response–modelling, the cluster analysis
164
was conducted with original (not rescaled) data.
165
For the identification of biological functions and pathways associated with the changes of gene
166
expression,
167
[https://analysis.ingenuity.com]. As input data, we used here the probes selected for the cluster
168
analysis as quantile normalised log2 ratios with respect to the mean of the controls. We used the
169
“core analysis” function with a cut–off of 0.5 log2 ratio to identify over represented
170
canonical pathways. As in the core analysis treatments are handled separately, a subsequent
171
comparative analysis of the single core analyses was done.
172
2.3
173
2.3.1 Gene expression measurements with real–time quantitative PCR (qPCR)
174
For the qPCR experiments total RNA was extracted from control or exposed zebrafish embryos at
175
120 hpf (extraction was performed as specified at the microarray experiment). 2 µg of total RNA
176
was reversely transcribed with RevertAidTM H Minus Reverse Transcriptase (Fermentas, Leon–
177
Roth, Germany) in 20 µl reaction mix according to the manufacturer instructions. qPCR was carried
178
out using a Step–One–Plus PCR System (Applied Biosystems, Darmstadt, Germany) and
179
SensiMixTM SYBR with ROX as passive reference dye (Bioline, Luckenwalde, Germany) in 12.5 µl
180
reaction volume in three technical replicates for each experimental replicate. gapdh was used as
181
housekeeping gene. Full name of target genes and of the housekeeping gene, their NCBI reference
Ingenuity
Pathway
Analysis
(version
18030641)
was
performed
Validation of microarray experiment and characterisation of the candidate genes
8 Environment ACS Paragon Plus
Page 9 of 28
Environmental Science & Technology
182
sequence identification code and literature references are listed in Table S3. Primer sequences,
183
amplicon lengths and PCR protocol are listed in Table S4. The validation of gapdh as housekeeping
184
gene is described in the supporting information (Figure S3 and Table S5). Melting curve analysis
185
was conducted to ensure the gene specificity of the primers. Relative expression levels were
186
determined by using the ∆∆Ct method [33]. In order to minimize variation across experiments, data
187
were mean–centred. Modelling of concentration response curves was performed with the Hill model
188
(see equation 1) using GraphPad Prism v 5.01 (GraphPad Software Inc., La Jolla, CA, USA) by
189
fixing E0 to 1.
190
2.3.2 Toxicokinetics–toxicodynamic modelling of the time course of gene expression changes
191
The time course of changes in gene expression in embryos exposed to 500 nM 11–ketotestosterone
192
was described by a toxicokinetics–toxicodynamic model using an indirect effect model. The
193
description of the indirect effect model and details on the modelling are given in the supporting
194
information. In brief, for the description of time course of internal concentration we used a TK
195
model for zebrafish embryos proposed by Kühnert, et al. [34]. The TD model was built under the
196
assumption that gene expression is regulated by the binding of 11–ketotestosterone (ligand) to the
197
AR resulting in an activation of the receptor via conformational changes of the AR. These activated
198
hormone receptor complexes are transferred into the nucleus where they bind to the androgen
199
response elements of the target gene promoter and trigger transcription (Figure S6).
200
Here we applied two types of indirect effect models (direct stimulation and direct stimulation &
201
indirect inhibition) used commonly in pharmacological studies [35] to describe the time course of
202
gene expression profiles. We solved the differential equations numerically using a fourth–order
203
Runge–Kutta method in Berkeley Madonna (version 8.3.18). Using Madonna's “curve fitter”
204
option, we estimated kinetic parameters and took these to analyse the plausibility for describing the
205
observations.
9 Environment ACS Paragon Plus
Environmental Science & Technology
206
2.3.3 Localisation of gene expression with in situ hybridisation
207
The expression domain of the target genes was investigated by in situ hybridisation in treated (500
208
nM 11–ketotestosterone) and untreated zebrafish embryos in order to reveal first evidence on the
209
possible function of the genes. However, the design of the in situ hybridisations did not aim at
210
demonstrating an increase in tissue–specific expression in 11–ketotestosterone exposed embryos but
211
to obtain information on the potential function of the genes. To localise the expression domain of
212
sult2st3, a control experiment was conducted with the clu gene, a marker of the zebrafish choroid
213
plexus (Figure S4).
214
A fragment of 600–900b of the cyp2k22, sult2st3, clu, slco1f4, and lipca genes were amplified
215
using RT–PCR. The PCR reaction was conducted with GoTaq® Polymerase (Promega GmbH,
216
Mannheim, Germany) according to the manufacturer instruction. Fragments of cyp2k22, slco1f4,
217
clu and lipca were cloned in a PCRII vector (Invitrogen, Darmstadt, Germany), linearised and
218
purified. Fragments of sult2st3 were amplified with primers including a T7 promoter added to the 5’
219
end of the reverse primer (see PCR protocol in the supporting information). Primers and amplicon
220
length are given in Table S4. Synthesis of digoxigenin labelled RNA probes and whole mount in
221
situ hybridizations were performed according to Thisse and Thisse [36] using linearised plasmids
222
(cyp2k22, clu, slco1f4 and lipca) or the PCR fragment (sult2st3). For cyp2k22, clu, slco1f4 and lipca
223
a control in situ hybridisation with sense RNA probes was performed.
224
A Leica stereo microscope (MZ16F) equipped with a Leica colour digital camera (DFC 350FX)
225
(Leica, Wetzlar, Germany) was used to image embryos from in situ hybridisation mounted in
226
glycerol.
227
3. Results
228
3.1
229
The multiple concentration–response modelling of transcriptome data was restricted to 6877 probes
230
of the microarray that showed at least a ratio of four between the maximum and minimum gene
Microarray experiment
10 Environment ACS Paragon Plus
Page 10 of 28
Page 11 of 28
Environmental Science & Technology
231
expression values across treatments. Since 11–ketotestosterone is known to transactivate AR
232
already in low concentrations (EC50=~1 nM) [37], we selected probes showing an EC50
