A transgenic zebrafish model for monitoring glucocorticoid receptor activity

Abstract

Gene regulation resulting from glucocorticoid receptor and glucocorticoid response element interactions is a hallmark feature of stress response signaling. Imbalanced glucocorticoid production and glucocorticoid receptor activity have been linked to socio-economically crippling neuropsychiatric disorders, and accordingly there is a need to develop in vivo models to help understand disease progression and management. Therefore, we developed the transgenic SR4G zebrafish reporter line with six glucocorticoid response elements used to promote expression of a short half-life green fluorescent protein following glucocorticoid receptor activation. Herein, we document the ability of this reporter line to respond to both chronic and acute exogenous glucocorticoid treatment. The green fluorescent protein expression in response to transgene activation was high in a variety of tissues including the brain, and provided single cell resolution in the effected regions. The specificity of these responses is demonstrated using the partial agonist mifepristone and mutation of the glucocorticoid receptor. Importantly, the reporter line also modeled the temporal dynamics of endogenous stress response signaling, including the increased production of the glucocorticoid cortisol following hyperosmotic stress and the fluctuations of basal cortisol concentrations with the circadian rhythm. Taken together, these results characterize our newly developed reporter line for elucidating environmental or genetic modifiers of stress response signaling, which may provide insights to the neuronal mechanisms underlying neuropsychiatric disorders such as major depressive disorder.

Introduction

The vertebrate stress response system integrates sensory inputs from actual or perceived threats to homeostasis and uses this information to drive a neuroendocrine response through the hypothalamic-pituitary-adrenal (HPA) axis, providing an adaptive mechanism to overcome the stressor (Chrousos, 1998). Upon activation, the HPA axis initiates a signaling response that culminates with the secretion of cortisol into the bloodstream, where it functions as an agonist to the glucocorticoid receptor (GR). The activated GRs in turn regulate gene transcription through interactions with cognate glucocorticoid response elements (GREs) or crosstalk with other transcription factors (Heitzer et al., 2007). The affected genes regulate an array of physiological parameters, and thus perturbed HPA axis functioning often underpins the pathological phenotypes characteristic of neuropsychiatric, immune, metabolic, and cardiovascular disorders (Hill et al., 2010). Accordingly there is a burgeoning interest in characterizing genetic and environmental modifiers of the HPA axis, and identifying GR ligands that may be used to mitigate symptoms of disease. Until recently, in vitro models have been largely responsible for facilitating such drug discoveries (Michelini et al., 2010). Although in vitro models are cost-effective and high-throughput, their usefulness is constrained since the effects of GR ligands are contingent on the profile of their local signaling environment. Specifically, factors such as the tissue-specific GR densities and the available repertoire of regulatory complexes hamper the interpretation of in vitro data, and therefore necessitate the development of in vivo models (Chang et al., 2013, Rogatsky et al., 2003).

Given its advantages as a model organism, the zebrafish is becoming more frequently leveraged in the field of stress science (Lohr & Hammerschmidt, 2011, Steenbergen et al., 2011, Stewart et al., 2012). Indeed, the zebrafish stress response system is functionally homologous to that of mammals, although the teleost GR largely mediates both glucocorticoid and osmoregulatory actions—the later of which is typically ascribed to mineralocorticoid receptor signaling in mammals (Takahashi & Sakamoto, 2013). Like humans, the zebrafish stress response culminates with cortisol release following exposure to a stressor (Alsop & Vijayan, 2009). This response is functional in 4 day post-fertilization (dpf) larvae and incorporates complex features like glucocorticoid-mediated negative-feedback by 5 dpf (Alsop & Vijayan, 2008, De Marco et al., 2013). Consequently physiological and behavioral techniques are rapidly being developed and used to study stress response system signaling and biology within the zebrafish (Clark et al., 2011b). Recently, a zebrafish reporter of GR signaling has been developed using GRE consensus sequences to drive luciferase expression in the presence of active GRs (Weger et al., 2012). Despite the advantages this line offers over conventional in vitro models, the spatiotemporal resolution of luciferase expression is limited in its ability to capture the cell-type specific affects of GR signaling and the rapid dynamics of HPA axis signaling. Therefore, we sought to develop and characterize a zebrafish model of GR activation that instead uses 6 pairs of endogenous GRE half sites to drive the expression of a 4 h half-life enhanced green fluorescent protein.

Materials & Methods

Vector Construction

pSR4G_BH was constructed by inserting the 1.5 kb XmaI to NheI restriction endonuclease fragment of pKTol2cmlc-BFP that corresponds to the cardiac myosin light chain 2 (cmlc2) promoter, tagBFP coding region, and poly(A) signal into the 4.7 kb vector fragment of pKTol2_GRE-d4G from AgeI to NheI. pKTol2_GRE-d4G was made by cloning an 860 bp XhoI to BglII fragment of the d4EGFP coding region amplified from PL450-IRES-d4EGFP with CDS-d2EGFP-F1 [AACTCGAGAACCTAGGATGGTGAGCAAGGGCGA] and CDS-d2EGFP-R1 [AAAGATCTACACATTGATCCTAGCAGAAGCACAG] into a 3.8 kb vector fragment of pKTol2-GRE from BglII to XhoI (Grinevich et al., 2009). pKTol2-GRE was made by cloning a 470 bp XmaI to ClaI fragment of a synthetic DNA, SAM_GRE, and a 1.0 kb ClaI to AvrII fragment of pGBT-R15 (Clark et al., 2011a) containing the zebrafish beta-actin polyadenylation signal into pKTol2-SE (Clark et al., 2011c) between the XmaI and AvrII sites of the polylinker. pKTol2cmlc-BFP was made by moving the tagBFP coding region, a 720 bp XhoI to BglII fragment, from pKTol2gC-TagBFP into pKTol2cmlc-GFP, a 3.0 kb BglII to XhoI vector fragment that removes the GFP. pKTol2cmlc-GFP was made by cloning the XmaI to XhoI digested amplification of the zebrafish cmlc2 promoter (Huang et al., 2003) amplified from genomic DNA with MISC-cmlc2-F1 [TTCCCGGGCATTCATCCATCCTTTTCATCC] and MISC-cmlc2-R1 [TTCTCGAGTTCACTGTCTGCTTTGCTGTTGGT] into a 3.5 kb XhoI to XmaI vector fragment of pKTol2C-GFP (Hoeppner et al., 2012) that removed the mini-CAGs promoter. pKTol2gC-BFP was made by cloning a 720 bp XhoI to BglII fragment with the coding region of tagBFP amplified from pTagBFP-N (Evrogen) using CDS-tagBFP-F1 [TTCTCGAGACCATGAGCGAGCTGATTAAGGAGAAC] and CDS-tagBFP-R1 [TTAGATCTCAGCTTTAATTGAGCTTGTGCCCCAGTT] into pKTol2gC-GFP (Hoeppner et al., 2012).

The SAM_GRE DNA (Supplementary Figure 1) was synthesized by Integrated DNA Technologies. SAM_GRE contains 6 composite GREs, the mouse cFos minimal promoter, a non-coding exon that is a fusion of the mouse c-Fos exon 1 and carp beta-actin exon 1 (Fisher et al., 2006, Liu et al., 1990), and a mini-intron utilizing the splice-donor and splice-acceptor from the carp beta-actin intron 1 (Liu et al., 1990). We identified 12 zebrafish GREs ranging from a 75.0 % to 83.3 % match to the human consensus GRE within 20 kb of the elastin a (elna) gene (Ledo et al., 1994). Using a similar search strategy we identified an additional 37 zebrafish GREs from additional genomic sequences within various genes of interest, including crhr1, nr3c1, nr3c2, and pomc (Supplementary Table 1). Individual GREs matched between 8 and 11 of the 12 nucleotides of the human consensus GRE, with a median match of 10 of 12 nucleotides. However, the zebrafish consensus for these 49 putative GREs matched the human consensus perfectly. Judging the slight mismatch as biologically important, we decided to use composite GREs in our synthetic enhancer. 3 of the 6 GREs used the most common 5′ half site [AGAACA] and the other 3 used the most common 3′ half site [TGTTCT]. The other half of each full GRE was composed of other common half sites and is shown in Figure 1.

Figure 1. The pSR4G_BH Construct.

(a) A schematic of the pSR4G_BH construct, which consists of a stress response GFP reporter and blue heart marker. ITR, Tol2 inverted terminal repeat; GRE, glucocorticoid response element; Prom^min, mouse cFos minimal promoter; intron, carp beta-actin intron 1; d4EGFP, 4 h half-life enhanced green fluorescent protein; Poly(A), zebrafish beta-actin polyadenlyation signal; cmlc, cardiac myosin light chain 2 promoter; tagBFP, blue fluorescent protein. (b) An illustration of the 6 composite GRE sequences used in the pSR4G_BH construct.

Production and Care of the SR4G Transgenic Fish

Standard practices were used to produce transgenic zebrafish by co-injecting the pSR4G_BH plasmid (Figure 1a) with Tol2 transposase (Clark et al., 2011c). Injected fish were raised to adulthood and crossed to a dominant leopard (Cx41.8^+/tq270) line (Watanabe et al., 2005). F1 embryos were screened for tagBFP expression in their heart. The tagBFP⁺ families were screened based on their d4EGFP expression baseline and response to exogenous glucocorticoids before selection for further experiments. All zebrafish lines were housed in a recirculating system (Aquaneering, Inc.) at 28 °C with a 14-10 h light/dark cycle. The fish were fed brine shrimp twice a day, and supplemented with Adult Zebrafish Diet (Zeigler) pellets 1–2 times per day. All experiments were carried out with prior approval from Mayo Clinic’s Institutional Animal Care and Use Committee.

Experimental Protocol

Embryos were obtained from adult, heterozygous SR4G fish crossed to a dominant leopard (Cx41.8^+/tq270) line (Watanabe et al., 2005). The embryos from individual pair crosses were combined for each experiment, and thinned to 60 embryos per 100 × 15 mm petri dish (BD Falcon) with 25 mL of 0.5 X E2 embryo medium (Nüsslein-Volhard & Dahm, 2002). The embryos were raised at 28.5 °C with a 14-10 h light/dark cycle, and all nonviable embryos were removed daily. Transgenic larvae, identified by the presence of tagBFP expression in the heart, were sorted from their non-transgenic siblings at 2 dpf. The tagBFP⁺ SR4G siblings were then redistributed into petri dishes in groups of 60 with 25 mL of embryo medium. At 3 dpf the SR4G larvae were divided into groups of 10, 20, or 25 and transferred to 6-well tissue culture plates (BD Falcon) with each well containing 8 mL of embryo medium. The administration of chronic, overnight glucocorticoid treatment began in the evening at 4 dpf, and the corresponding experiments were performed at 5 dpf. All other experiments were started half an hour after the onset of the light cycle at 5 dpf. The samples collected for subsequent cortisol or d4EGFP transcript analysis were flash frozen in liquid nitrogen, and stored at −80 °C for post-treatment processing.

Chemical Treatments

For each compound used, a fresh working stock was prepared prior to its administration on the day of the experiment. A 20 mM stock of fluticasone propionate (Sigma-Aldrich) was made in dimethyl sulfoxide (DMSO), and then diluted to a final working concentration of 50 μM in water. Likewise, a 50 mM solution of hydrocortisone (Sigma-Aldrich) was prepared in DMSO, prior to being diluted to 50 μM in water. The nicotine (Acros Organics) and sodium chloride (Sigma-Aldrich) solutions were both made in water at working concentrations of 125 μM, and 500 mM respectively. 2 mL of each working solution was mixed with the larvae that were in 8 mL of embryo medium, resulting in a 5-fold dilution yielding treatment conditions of 10 μM fluticasone propionate (Sigma-Aldrich), 10 μM hydrocortisone (Sigma-Aldrich), 100 mM sodium chloride (Sigma-Aldrich), or 25 μM nicotine (Acros Organics). In order to account for any physical stress resulting from the administration of the treatments, 2 mL of vehicle (water or 0.25 % DMSO) was applied to the control groups. In contrast, mifepristone (Tocris) was diluted to 40 mM in DMSO and stored in aliquots at −20 °C, before being further diluted in DMSO to working concentrations of 100 μM or 300 μM on the day of the experiment. The mifepristone pretreatments were applied 2 h before the addition of any acute treatment. The pretreatment was administered by substituting 80 μL of embryo medium in each well with 80 μL of either the 100 μM or 300 μM working solutions, resulting in a further 100-fold dilution.

Imaging

The d4EGFP expression patterns resulting from chronic glucocorticoid treatment were captured in sagittal-, dorsal-, and ventral- oriented z-stacks at 50 X magnifications using an ApoTome microscope (Zeiss) with a 5 X/0.25 NA dry objective (Zeiss). Expression of tagBFP was imaged using a 31037 - Pacific Blue optical filter (Chroma Technology Corp), while d4EGFP expression was imaged using a MF101 - Spectrum Green optical filter (Chroma Technology Corp). Each image is a composite of both rostral and caudal maximal image projections of z-dimension stacks obtained from the same larva. Likewise, composite images of the d4EGFP expression were created after acute glucocorticoid treatment using rostral and caudal sagittal-oriented maximal image projections of z-dimension stacks obtained from the same larva at 50 X magnifications. The BFP channel was opened in the rostral images of sagittal planes to document any tagBFP expression driven by the cmlc2 promoter. Additionally, rostral d4EGFP expression patterns resulting from acute and chronic hydrocortisone treatment were captured in dorsal-oriented z-stacks at 100 X magnifications using a Lightsheet Z.1 microscope (Zeiss) equipped with a 20 X/1.0 NA water-dipping objective (Zeiss) with 0.5 X zoom and a DAPI-GFP (beam splitter SBS LP 490; emission filters BP 420–470 and BP 505–545) optical filter (Zeiss). Each corresponding 100 X rostral image is a maximal image projection generated from z-dimension stacks of the respective plane. All 100 X rostral maximal image projections and movies are depicted using a custom 256 pixel lookup table (Figure 4 and Supplementary Movies 1–3). All larvae were treated with 0.2 mM phenylthiocarbamide (Sigma-Aldrich) at 1 dpf to inhibit pigment formation. The images were taken using the previously described SCORE technique (Petzold et al., 2010).

Figure 4. Glucocorticoid treatment induces d4EGFP throughout the brain.

Changes in d4EGFP levels in SR4G larvae following no treatment (Control), 4 h 10 μM hydrocortisone (Acute Cortisol) treatment, or overnight 10 μM hydrocortisone (Chronic Cortisol) treatment. (a) A schematic of the planes (P 1–4) imaged and used to generate the representative maximal image projections. (b) Representative rostral maximal image projections of z-dimension stacks of d4EGFP expression acquired at 100 X magnifications. The effects of each treatment on d4EGFP expression are represented in planes 1–4 (P 1–4) using a custom 256 pixel lookup table.

Cortisol Extraction, and ELISA

Samples frozen in 800 μL of phosphate-buffered saline (PBS) were thawed, and homogenized on ice for 90 s with a Vibra-Cell VCX 130 (Sonics and Materials Inc.) ultrasonic processor. The cortisol was extracted from 640 μL of the resulting homogenate three times using 3 mL of ethyl ether (Fisher Scientific), which was then evaporated off by placing the test tubes in a 42 °C water bath for approximately 12–18 h. The samples were then reconstituted with 500 μL of PBS at 4 °C for approximately 30 h, vortexing occasionally. The cortisol in each sample was quantified using a Cortisol EIA Kit (Cayman Chemical). The cortisol levels were normalized to the amount of protein in each of the samples, which was determined by following the microplate procedure of a Pierce BCA Protein Assay Kit (Thermo Scientific) using 25 μL aliquots of either bovine serum albumin standards or sample homogenate for the unknowns. The cortisol extraction efficiency was calculated to be approximately ~70 %, which is similar to previously published results (Alsop & Vijayan, 2008). The calculated extraction efficiency was used to correct all recorded cortisol values to the actual amount of cortisol per sample.

mRNA Isolation, cDNA Synthesis, and qRT-PCR

Samples frozen in 350 μL RLT Buffer (Qiagen) with 1 % 2-mercaptoethanol (Sigma-Aldrich) were thawed, and homogenized with a 21-gauge needle (Kendall). The mRNA was isolated using a phenol-chloroform extraction with MaXtract High Density Tubes (Qiagen). The subsequent mRNA purifications were carried out with a RNeasy Mini Kit (Qiagen) and RNase-Free DNase Set (Qiagen). The first-strand cDNA was then synthesized with 1 μg of mRNA from each sample using 100 ng/μL of random hexamers (Integrated DNA Technologies), Superscript II Reverse Transcriptase (Invitrogen), and RNaseOut Recombinant Ribonuclease Inhibitor (Invitrogen). The qRT-PCR reaction mixture was prepared with a SensiFAST SYBR Lo-ROX kit (Bioline), and primers from Integrated DNA Technologies that amplified either eef1a1l1 (Forward 5′-CCGTCTGCCACTTCAGGATGTGT-3′, Reverse 5′-TTGAGGACACCAGTCTCCACACGA-3′) or d4EGFP (Forward 5′-CGAGCAACTGAGGATCCCATTCTCT-3′, Reverse 5′-CACCCCGGTGAACAGCTCCT-3′). All qRT-PCR reactions were carried out on a Bio-Rad C1000 Touch Thermal Cycler CFX 96 Real-Time System with a polymerase activation step at 95 °C for 2 min, followed by 40 3-step cycles of 95 °C for 5 s, 58 °C for 10s, and 72 °C for 20 s, and finally a melting curve analysis cycle of 95 °C for 10 s, 65 °C for 5 s, and 95 °C for 50 s.

Targeted Mutagenesis via TALENs

Transcription activator-like effector nucleases (TALENs) were designed using the free, open access software available online at http://www.talendesign.org (Neff et al., 2013). A TALEN targeting exon 2 of nr3c1 (left TALEN recognition sequence 5′-TTGGGAACAGCTCGC-3′, right TALEN recognition sequence 5′-GATCTTTCTGCAGAC-3′) was assembled following the previously described Golden Gate method, using pT3TS-GoldyTALEN as the final destination vector (Bedell et al., 2012, Cermak et al., 2011). The resulting expression plasmids were linearized using Sac1 endonuclease, and mRNA was transcribed using an mMESSAGE mMachine T3 Kit (Life Technologies). One-cell SR4G embryos were microinjected with 50 pg of mRNA encoding the TALEN pair. The TALEN recognition sites flank a 15 bp spacer sequence that contains a PvuII restriction enzyme site (5′-CCACAGCTGTCGTCG-3′, PvuII underlined) (Supplementary Figure 3), which was used to detect indels in somatic tissue with restriction fragment length polymorphism analysis as previously described (Bedell et al., 2012). In short, the region around exon 2 of nr3c1 was amplified (Forward 5′-CTCTCCTTTCAGAGCTGCCGACAA-3′, Reverse 5′-GGTGGTCTTGATGGCTTACCTGGAAT-3′), and digested with PvuII. In cases where the PvuII site was modified through non-homologous end joining following the formation of a double-strand break by the TALEN pair, the amplicon was resistant to digestion. Once TALEN activity was confirmed, the remaining larvae were raised and treated with chemicals according to the Materials & Methods Experimental Protocol and Chemical Treatments subsections. Following chemical treatment, the larvae were mounted in a 2.5 % methylcellulose (Sigma-Aldrich) solution containing 1: 20,000 eugenol (Sigma-Aldrich), and imaged on an Axio Scope.A1 microscope (Zeiss) equipped with a 5 X/0.25 NA dry objective (Zeiss) and a MF101 - Spectrum Green optical filter (Chroma Technology Corp). Images were captured with a D3 digital camera (Nikon), using a 1/1000 s shutter speed to take bright-field photographs and a 2 s shutter speed to capture d4EGFP photographs. All larvae were treated with 0.2 mM phenylthiocarbamide (Sigma-Aldrich) at 1 dpf to inhibit pigment formation.

Statistical Analysis

Each time point collected for d4EGFP transcript analysis was performed in triplicate, providing biological triplicates (10 or 20 larvae/n, n = 3). Each time point collected for cortisol analysis was performed in triplicate and assayed with the Cortisol EIA Kit (Cayman Chemical) in triplicate, providing both biological and technical triplicates (25 larvae/n, n = 3). All corresponding data points are expressed as means ± S.E.M., and two-tailed independent-sample t-tests were used to make comparisons between groups.

Results

Selection of Zebrafish GREs and Vector Design

To identify zebrafish GREs, we examined the elastin gene and 20 kb of flanking sequence on either side for regions that matched the GRE consensus sequence [AGAACA NNN TGTTCT] because the human elastin gene is known to contain GRE sites (Ledo et al., 1994). We identified 12 potential zebrafish GREs that loosely matched the consensus sequence, matching from 8–10 of the 12 consensus nucleotides. Using other zebrafish genomic sequences, we identified an additional 37 potential GRE sites. In no case did we observe a perfect match to the GRE consensus sequence, rather we saw a range from 8–11 nucleotides matching with a median match of 10 nucleotides (Supplementary Table 1). However, building a consensus sequence of these 49 zebrafish GREs did result in an exact match to the human GRE consensus sequence. Therefore in recreating a synthetic GRE enhancer, we thought that imperfect consensus sites might better replicate biological responses to glucocorticoids. The final pSR4G_BH vector uses 6 zebrafish composite GREs separated by 12 bp spacers (Figure 1a and 1b). For each of the 6 GREs either the most common left half site [AGAACA] or the most common right half site [TGTTCT] was used. The other half site for each GRE is unique within the synthetic 6 X GRE of pSR4G_BH, and they represent the next 3 most common left and right half GRE sites from our collection (Supplementary Table 1). The final plasmid connects the 6 X GRE enhancer to a cFos minimal promoter, mini-intron, and d4EGFP sequence. This expression cassette should produce a 4 h half-life GFP in response to activated GRs. In addition there is a second cassette that uses the cmlc2 promoter to produce tagBFP in the heart. The blue heart can be used to select transgenic fish without a need to activate the d4EGFP expression through stress or pharmacogenomics. The whole cassette is placed within a Tol2 transposon to facilitate transgenesis.

Production and Selection of the SR4G Transgenic Line

Fluticasone propionate is a potent glucocorticoid commonly used in drug preparations prescribed to treat inflammatory disorders of the respiratory tract, including asthma and allergic rhinitis (Bernstein et al., 2004, Verona et al., 2003). This compound exerts its effects by binding to GRs with a high affinity, and inducing the corresponding transactivation or transrepression of gene expression (Smith & Kreutner, 1998). In order to establish the SR4G reporter line as model for identifying alterations in GR signaling, we treated some tagBFP⁺ SR4G larvae with 10 μM fluticasone propionate overnight and examined d4EGFP expression in untreated and treated larvae. In many lines, the chronic drug administration induced a visually striking increase in d4EGFP production through numerous tissue types in each of the planes that were imaged (Figure 2). A line with low background fluorescence and a strong signal that behaved similarly to other lines was selected for all further experiments.

Figure 2. Overnight synthetic glucocorticoid treatment induces d4EGFP expression.

Representative maximal image projections of z-dimension stacks of SR4G larvae treated overnight with either 0.25 % DMSO (DMSO) or 10 μM fluticasone propionate (Fluticasone). The effects of each treatment on d4EGFP expression are illustrated in the sagittal, dorsal, and ventral planes at 50 X magnifications. tagBFP expression driven by the cmlc2 promoter was documented in the rostral sagittal plane of the larvae.

Acute Response to Exogenous Glucocorticoids

In order to facilitate high-throughput drug screening it is necessary for the SR4G transgenic line to generate a robust response to not only chronic, but also acute GR agonist exposure. Therefore we examined the ability of 5 dpf SR4G larvae to generate a response to GR signaling following a 4 h exposure to either synthetic or natural glucocorticoids. Fish treated with a 10 μM dose of fluticasone propionate or hydrocortisone demonstrated a discernible elevation of d4EGFP fluorescence relative to the DMSO treated control (Figure 3a). Similarly, an analysis of the d4EGFP transcript levels revealed a significant increase relative to time 0 following treatment with either fluticasone propionate [t(4) = 20.59, p < 0.001] or hydrocortisone [t(4) = 4.62, p = 0.010] (Figure 3b). In contrast, the 4 h DMSO treated control exhibited a slight decrease in d4EGFP transcript relative to time 0, suggesting that fluctuations occur in basal GR signaling (Figure 3b). We subsequently tested the specificity of these responses by pretreating the fish with the GR antagonist mifepristone prior to glucocorticoid exposure. A 2 h pretreatment with 10 μM mifepristone significantly attenuated the robust ~35-fold increase in d4EGFP transcript production resulting from fluticasone propionate treatment [t(4) = 17.38, p < 0.001] (Figure 3b). Likewise, a 2 h mifepristone pretreatment significantly attenuated the robust ~20-fold increase in transcript production following hydrocortisone treatment [t(4) = 3.87, p = 0.018] (Figure 3b). The attenuated responses observed following a mifepristone pretreatment were significantly increased relative to time 0 for both the mifepristone-pretreated fluticasone group [t(4) = 5.97, p = 0.004], as well as the mifepristone-pretreated hydrocortisone group [t(4) = 6.06, p = 0.004] (Figure 3b). Additionally, the administration of mifepristone alone resulted in statistically significant ~4-fold increase in d4EGFP transcript relative to time 0 [t(4) = 9.48, p < 0.001] (Figure 3b). The increase in transcript levels after mifepristone treatment corresponded to an increase in d4EGFP production (Supplementary Figure 2). As an alternative means of demonstrating the specificity of transgene activation, the GR was disrupted using highly efficient transcription activator-like effector nucleases (TALENs) engineered to target exon 2 of the nr3c1 locus (Supplementary Figure 3a and 3b). TALEN injection prevented the visually discernable increase of d4EGFP expression in SR4G larvae following 2 h fluticasone treatment (Supplementary Figure 3c). This result was variable between individual larvae, reflecting the mosaic nature of somatic TALEN activity. Taken together these data demonstrate the ability of the transgene to specifically model the tissue-dependent effects of acute GR signaling.

Figure 3. Acute glucocorticoid treatment induces transgene activation.

Changes in d4EGFP levels in SR4G larvae 4 h after the administration of either 0.25 % DMSO (DMSO), 10 μM hydrocortisone (Cortisol) or 10 μM fluticasone propionate (Fluticasone). (a) Representative sagittal maximal image projections of z-dimension stacks of d4EGFP expression acquired at 50 X magnifications. tagBFP expression driven by the cmlc2 promoter was documented in the rostral sagittal plane of the DMSO treated larva. (b) The fold-change in d4EGFP transcript levels relative to time 0. In some groups, larval fish were pretreated for 2 h with 10 μM mifepristone (+M). Data points are means ± S.E.M. (10 larvae/n, n = 3). ** indicates a data point significantly different from the 0 h DMSO group (t-test; p < 0.01). † indicates a data point significantly different from the 4 h Cortisol group (t-test; p < 0.05). ‡ indicates a data point significantly different from the 4 h Fluticasone group (t-test; p < 0.001).

Transgene Activation in the Brain

Disruptions in GR signaling are tightly linked to numerous neuropsychiatric disorders such a major depressive disorder, mandating the development of in vivo models to clarify the etiologies of such disorders and the specific role that neuronal GR signaling plays in their onset and potentiation (De Kloet et al., 2007). Accordingly, we subsequently characterized the effects of hydrocortisone treatment on d4EGFP expression in the central nervous system. Both acute 4 h and chronic overnight treatment with 10 μM hydrocortisone resulted in elevated d4EGFP expression relative to the control in each of the rostral planes imaged at 100 X magnifications (Figure 4 and Supplementary Movies 1–3). Widespread d4EGFP expression was observed throughout the brain, with the expression intensifying with increasing lengths of glucocorticoid exposure. The expression of d4EGFP in the brain of SR4G larvae was heterogeneous, and permitted for a single-cell resolution of the effected tissues.

Transgene Activation via the Endogenous Stress Response

After establishing the capability of the SR4G line to rapidly respond to glucocorticoid treatment in both a time- and tissue- specific manner, we examined if it could reflect the more transient dynamics of cortisol signaling during the endogenous stress response. Zebrafish are a freshwater cyprinid, and their larvae are known to synthesize cortisol when stressed with hyperosmotic conditions (Alderman & Bernier, 2009). The whole-body cortisol levels of 5 dpf larvae were significantly elevated 15 min [t(4) = 6.50, p = 0.003], 20 min [t(4) = 6.26, p = 0.003], and 30 min [t(4) = 3.37, p = 0.028] after 100 mM NaCl treatment (Figure 5a). A peak in whole-body cortisol levels was observed 20 min after exposure to hyperosmotic conditions, after which the levels returned to baseline (Figure 5a). The secretion of cortisol preluded the elevation of d4EGFP transcript in SR4G larvae, which reached a significant increase 1 h [t(4) = 4.97, p = 0.008] after exposure to hyperosmotic conditions (Figure 5b). A maximum fold-change was observed at 2 h post-treatment, after which the transcript levels returned to baseline (Figure 5b). Nicotine was used as an alternative activator of the HPA axis, since this compound is known to elicit a cortisol response in zebrafish (Cachat et al., 2011). The administration of 25 μM nicotine also resulted in a significant increase in d4EGFP transcript production relative to time 0 [t(4) = 4.53, p = 0.012], thereby demonstrating the sensitivity of the transgene to a variety of physiological stressors known to induce glucocorticoid production (Supplementary Figure 4).

Figure 5. Cortisol and d4EGFP transcript levels are elevated after hyperosmotic stress.

(a) Whole-body cortisol levels after the administration of 100 mM NaCl. Data points are means ± S.E.M. (25 larvae/n, n= 3, performed in technical triplicates). * indicates a data point significantly different from time 0 (t-test; p < 0.05). ** indicates a data point significantly different from time 0 (t-test; p < 0.01). (b) The fold-change in d4EGFP transcript levels relative to time 0 in SR4G larvae after the administration of 100 mM NaCl. Data points are means ± S.E.M. (20 larvae/n, n = 3). ** indicates a data point significantly different from time 0 (t-test; p < 0.01).

Basal Glucocorticoid Receptor Signaling Cycles with the Circadian Rhythm

Given the high spatiotemporal resolution offered by the pSR4G_BH transgene in response to exogenous or endogenous glucocorticoid exposure, we investigated the endogenous signaling dynamics during homeostatic conditions. Zebrafish larvae are diurnal, and are increasingly used in circadian rhythm research (Hurd & Cahill, 2002, Ziv et al., 2005). Since basal cortisol levels are known to cycle with this rhythm, we examined the d4EGFP transcript levels in unstressed SR4G larvae every 2 h over a 26 h timeline that started half an hour after the onset of the light cycle (Debono et al., 2009, Dickmeis et al., 2007). The transcript levels were highest around the beginning of the 5 dpf light cycle, significantly decreased throughout the course of the day [4 h: t(4) = 7.32, p = 0.002; 6 h: t(4) = 3.93, p = 0.017; 8 h: t(4) = 6.12, p = 0.004; 10 h: t(4) = 5.31, p = 0.006; 12 h: t(4) = 4.40, p = 0.012; 14 h: t(4) = 5.76, p = 0.005; 18 h: t(4) = 4.36, p = 0.012; 20 h: t(4) = 3.98, p = 0.016], and then gradually returned to their previous peak level by the start of the 6 dpf light cycle (Figure 6). The notable exception to this trend occurred at 16 h where a smaller peak in transcript was observed 2 h after the beginning of the dark cycle.

Figure 6. The d4EGFP transcript levels fluctuate with the circadian rhythm.

The fold-change in basal d4EGFP transcript levels relative to time 0 in SR4G larvae during the 14-10 h light/dark cycle. Data points are means ± S.E.M. (10 larvae/n, n = 3). * indicates a data point significantly different from time 0 (t-test; p < 0.05). ** indicates a data point significantly different from time 0 (t-test; p < 0.01).

Discussion

Zebrafish are easily subjected to a plethora of genetic manipulations, have a remarkably well-conserved stress response system, produce large clutches of embryos, and feature ex-utero and transparent embryonic and larval development that allows for the imaging of live animals and in vivo analysis. This suite of characteristics makes the zebrafish particularly well suited for facilitating drug development and neurodevelopmental research—two understudied fields of stress science (Steenbergen et al., 2011). The SR4G reporter line presented herein is intended to exploit these advantages in order to provide an in vivo tool that expedites discoveries in these fields. For instance, natural and synthetic glucocorticoids are used to treat numerous diseases and therefore it is of high medical significance to identify GR ligands with differential binding specificities, improved clearance properties, and minimal side effects (Austin et al., 2002). Consistent with the design of our construct the administration of both natural and synthetic glucocorticoids elicited an increase in d4EGFP transcript production. It is worth noting that an acute 4 h ligand treatment is sufficient to drive up to a ~35-fold change in transcript from the pSR4G_BH cassette. This rapid, robust response makes this model ideal for high-throughput drug screening, and provides a large working range for dose curve analysis. These responses may be attenuated by pretreatment with mifepristone or with an injection of transcription activator-like effector nuclease mRNA directed against the nr3c1 locus, thereby demonstrating the specificity of transgene activation. However, mifepristone pretreatment did not block the entire response and exposure to mifepristone alone led to a ~4-fold change in d4EGFP transcript. In vitro studies have previously demonstrated that mifepristone has partial agonist activity that is dependent on both the cell-type and the density of GRs (Peeters et al., 2008, Zhang et al., 2007). Our data suggests that the SR4G reporter line is able to corroborate mifepristone’s in vitro partial agonist activity in vivo for the first time. To this end it is advantageous that the pSR4G_BH construct employs the d4EGFP reporter, which makes it possible to not only quantify the net change in d4EGFP levels resulting from GR ligand exposure but to also visualize the tissue-specific changes GR signaling dynamics. After observing the ability of the reporter line to model these tissue-dependent effects, we then sought to specifically investigate GR signaling within the central nervous system. Both chronic overnight and acute 4 h hydrocortisone treatment induced widespread heterogeneous activation of the transgene through the brain. The intensity of d4EGFP expression was dependent on the length of treatment time and provided a single cell resolution, enabling the visualization of distinct neural populations involved with GR signaling. This level of resolution was also available in the control group, where transgene activation would be solely dependent on basal levels of endogenous glucocorticoids. Collectively, these properties should enable the SR4G reporter line to accelerate the identification of novel modifiers of GR activity, clarify the tissue-specific pharmacology of GR signaling, and enable the continued optimization of glucocorticoid treatments.

Given the sensitivity of the pSR4G_BH cassette to exogenous GR agonist exposure, we interrogated the changes in transcript expression resulting from the initiation of the endogenous stress response. As anticipated, exposure to hyperosmotic stress led to a sharp increase in the whole-body cortisol levels before returning to baseline after 1 h, a trend that was reflected by the d4EGFP transcript levels over the course of a day. A similar response was observed from the reporter following an acute treatment with nicotine, demonstrating the ability of the SR4G line to respond to glucocorticoid production triggered by different types of stressors. Interestingly, a consistent decrease in d4EGFP transcript was observed between the samples collected at time zero and the unstressed negative control groups collected later in the experiments, suggesting that the transgene may be responding to endogenous fluctuations in basal cortisol levels. Therefore, we examined the transcript levels in unstressed SR4G larvae over the course of an entire day. In contrast to the early rising trend observed following hyperosmotic stress, the d4EGFP transcript levels in unstressed fish exhibited a peak around the beginning of each light cycle and then decreased throughout the day. This diurnal pattern reflects the cycle of circulating cortisol levels resulting from the circadian rhythm. Therefore the changes in transcript levels observed following hyperosmotic stress or nicotine treatment are actually larger than what is depicted, since the post-treatment values are expressed as a fold-change relative to time zero rather than their corresponding unstressed time point in the light-dark cycle. We do expect that d4EGFP protein synthesis would similarly capture these temporal dynamics of GR activation due to the incorporation of a destabilizing PEST sequence in the pSR4G_BH construct.

Taken together, our results demonstrate the ability of SR4G larvae to provide a high-resolution model of GR signaling, under both unstressed and stressed conditions. These properties allow the SR4G reporter line to exhibit a broad translational potential that extends beyond drug development applications. As the use of zebrafish to study the stress response system has increased, so too has the adaptation of techniques to investigate stress-aggravated neuropsychiatric disorders with this model organism. To date, behavioral paradigms have been developed to model a number of neuropsychiatric disorders in the zebrafish including aspects of major depressive disorder, generalized anxiety disorder, and substance use disorder (Braida et al., 2007, Clark et al., 2011b, Petzold et al., 2009, Stewart et al., 2012, Stewart et al., 2011, Ziv et al., 2013). Although stress and disrupted GR signaling are intimately associated with the precipitation and potentiation of these neuropsychiatric disorders, their specific etiologies remain unclear (De Kloet et al., 2007). The use of the SR4G reporter line with existing models of neuropsychiatric disease should clarify these relationships by enabling a detailed assessment of when and where GR activity imbalances occur, establishing how acute and chronic stress disrupts this signaling, and ascertaining the specific roles of GR signaling in the presentation of disease. As the neuroanatomical atlas of the zebrafish brain becomes increasingly defined, the high-resolution provided by the SR4G reporter line will allow description of specific cellular substrates that are involved in glucocorticoid receptor signaling. Additionally, neuropsychiatric disorders are known to be highly heritable disease states, a characteristic that furthers the potential applications of the SR4G line (Foley et al., 1998, Kendler et al., 1999). The high fecundity of zebrafish coupled with advances in genome-engineering methodologies makes this model organism ideal for identifying quantitative trait loci associated with these disorders. Transcription activator-like effector nuclease and clustered regularly interspaced short palindromic repeat technologies are becoming increasingly accessible and have enabled the targeted mutagenesis of loci, allowing for investigations of how genetic variability translates into disease (Bedell et al., 2012, Hwang et al., 2013, Jao et al., 2013). Establishing mutant lines and crossing them with SR4G fish will permit the investigation of specific genetic contributions to the development and functioning of the stress response system. Although the SR4G line has clear potential for improving the understanding and treatment of neuropsychiatric disorders, the field of stress science is broad and so are the potential applications of this novel tool. These applications may encompass topics ranging from investigating environmental modifiers of HPA axis activity including drugs of abuse like nicotine or environmental stressors like pollutants, to clarifying the role of GR signaling following an injury to the central nervous system. By using this high-resolution reporter line in conjunction with existing models, it should facilitate the characterization of GR ligands and investigation of how environmental and genetic factors contribute to endogenous HPA axis functions, which in turn should provide novel insights for the management of stress-aggravated pathologies.