Fluorescent Reporter Zebrafish Line for Estrogenic Compound Screening Generated Using a CRISPR/Cas9-Mediated Knock-in System

Abstract

An increasing number of compounds in our diet and environment are being identified as estrogenic, causing serious and detrimental effects on human, animal, and ecosystem health. Time- and cost-effective biological tools to detect and screen these compounds with potential high-throughput capabilities are in ever-growing demand. We generated a knock-in zebrafish transgenic line with enhanced green fluorescent protein (EGFP) driven by the regulatory region upstream of vitellogenin 1 (vtg1), a well-studied biomarker for estrogenic exposure, using CRISPR/Cas9 technology. Exposure to 17β-estradiol (E2: 0–625 nM) starting at 4-h post-fertilization in dechorionated embryos resulted in the significant induction of hepatic EGFP with $\ge$5 nM E2 as early as 3-days post-fertilization. Concentration- and time-dependent increase in the percentage of hepatic EGFP-positive larvae and extent of fluorescence expression, categorized into 3 expression levels, were observed with E2 exposure. A strong correlation between the levels of EGFP mRNA, vtg1 mRNA, and EGFP fluorescence levels were detected. Image analysis of the area and intensity of hepatic EGFP fluorescence resulted in high-fidelity quantitative measures that could be used in automated screening applications. In addition, exposure to bisphenol A (0–30 μM) resulted in quantitative responses showing promise for the use of this transgenic line to assess estrogenic activity of endocrine-disrupting chemicals. These results demonstrate that this novel knock-in zebrafish reporter allows for distinct screening of in vivo estrogenic effects, endpoints of which can be used for laboratory testing of samples for estimation of possible human and environmental risks.

Chemicals with estrogenic potential from both natural and anthropogenic sources are prevalent in our food and environment (Atkinson et al., 2003; Baronti et al., 2000). Exposure to exogenous estrogens have been reported to adversely impact a variety of physiological functions including reproduction in humans and wildlife (Adeel et al., 2017; Andersen et al., 2003; Iguchi et al., 2002; Kidd et al., 2007). Nevertheless, only a handful of existing chemicals have been tested for estrogenic potential. One limitation has been the paucity of sensitive, cost-effective, and high-throughput whole vertebrate screening methods. Conventional testing using rodents or other mammals are time-consuming and expensive. Invertebrate screening systems such as Daphnia magna and Caenorhabditis elegans as well as various in vitro systems have also been evaluated and/or used successfully but lack the complex systemic response and toxicodynamics of a vertebrate animal to estrogenic compounds, and encounter core challenges to extrapolation of results.

Zebrafish, Danio rerio, is becoming widely used in studying the effects of endocrine-disrupting chemicals (EDC). Its short generation time, transparent embryos/larvae, well-studied genome, and sensitivity to exposure to EDCs, extra-uterine development allowing direct exposure to chemicals, are among the advantages encouraging its use (Caballero-Gallardo et al., 2016; Miscevic et al., 2012; Segner, 2009). Although biotransformation capabilities between fish and mammals could be slightly different and thus may require caution in extrapolation of results to humans, there remains clear advantages; zebrafish has been proposed as a close to ideal in vivo model with high-throughput screening capabilities (Padilla et al., 2012; Sipes et al., 2011; Truong et al., 2014).

Expression of vitellogenin (VTG), the precursor of egg yolk proteins in oviparous species, is a commonly used biomarker for assessing exposure to estrogens and estrogen-mimicking compounds in teleosts (Denslow et al., 1999; Kausch et al., 2008; Muncke and Eggen, 2006; Sumpter and Jobling, 1995). Vitellogenin is synthesized in the liver of mature female zebrafish at reproductive age and transported to developing oocytes through blood circulation and stored. Juveniles and males of reproductive age do not normally express VTG, but it can be induced upon exposure to exogenous estrogenic compounds (Jin et al., 2009; Muncke and Eggen, 2006; Sumpter and Jobling, 1995). Vitellogenin protein expression (Maradonna et al., 2013; Van den Belt et al., 2004) as well as gene expression (Jin et al., 2009; Maradonna et al., 2013; Muncke and Eggen, 2006) have been used to detect estrogenic exposure using homogenate of whole zebrafish and primary hepatocytes derived from zebrafish. However, these methods have limitations and challenges, requiring labor-intensive and time-consuming sample processing and readout. In addition, the use of an in vivo system is important in the assessment of estrogenic compounds because their actions are far more complex than an effect on a single cell type (eg, hepatocytes), involving feedback of endocrine system.

To address some of these pitfalls, several biosensing transgenic lines have been made to monitor vtg1 expression in situ. Medaka and zebrafish line that expresses green fluorescence protein (gfp) reporter gene under the control of the vtg1 gene promoter have been generated (Chen et al., 2010; Zeng et al., 2005). In the zebrafish model, however, co-treatment with 1-phenyl-2-thio-urea was necessary to prevent pigment formation to allow GFP visualization (Chen et al., 2010). 1-phenyl-2-thio-urea is a goitrogen that blocks thyroid hormone synthesis and disrupts associated metabolic functions (Elsalini and Rohr, 2003), further confounding toxicological interpretations. Most recently, another zebrafish line that expresses mCherry under the control of the vtg1 promoter was reported (Bakos et al., 2019). Similarly, transgenic fish lines capturing other targets of estrogenic actions such as estrogen response element (ERE) (Gorelick and Halpern, 2011; Green et al., 2018; Lee et al., 2012; Legler et al., 2000), Cyp19a1b (Kim et al., 2009; Tong et al., 2009), and choriogenin H (Kurauchi et al., 2005) have also been developed using fluorescence and/or luciferase reporters.

All of the previous reports, however, have used conventional methods to generate transgenic lines by introducing DNA constructs that harbor specific promoters/enhancers into the genome by random insertions. These processes involve time-consuming steps, such as characterization of the promoter/enhancer regions of the target genes, lower success rates, and in many instances, would not fully recapitulate the expression of the endogenous genes simply because some enhancers of gene expression are located hundreds of kilobases away. Thus, we sought to develop a novel and efficient biosensor for detecting exposure to estrogenic compounds by trapping endogenous gene regulation using precise CRISPR/Cas9-targeted homology-independent insertions which is a newly emerging technique (Auer et al., 2014; Kimura et al., 2014). This technique allows for a simple, flexible, and highly efficient targeted insertion of marker fluorophores into the promoter region of the target gene, and thus, fully recapitulating its temporal and spatial expression.

We successfully developed a knock-in zebrafish transgenic line with an enhancer trapped enhanced green fluorescent protein (EGFP) driven at the vitellogenin 1 (vtg1) locus and tested the fidelity of EGFP expression to endogenous vtg1 expression, and sensitivity to both steroidal and non-steroidal estrogens. Our results suggest that this novel approach allows for high-fidelity detection of in vivo estrogenic effects, endpoints of which can be used to estimate human and environmental risk. This is also the first report to suggest the usefulness of knock-in CRISPR/Cas9 technology in the generation of zebrafish lines for toxicological screening purposes with high-throughput capabilities.

MATERIALS AND METHODS

Zebrafish

Optically transparent Casper-mutant zebrafish (White et al., 2008) were bred and raised in the fish facility at Cornell University. Broodstock fish were held in a controlled recirculating system (Z-Hab System, Aquatic Habitats, Apopka, Florida), with a 14:10-h light:dark photoperiod and at a temperature of 28 ± 1°C. Water conductivity was maintained at 500 ± 50 µS, pH was 7.4 ± 0.2, and total ammonia, nitrate/nitrite levels were below recommended thresholds. Breeding was performed in 2-l breeding tanks with dividers. Breeders were placed in the tanks the night before at a 1:2 male:female ratio and dividers were removed the next morning to allow fish breeding in a timely manner. Growing embryos and larvae were maintained in nursery cups within the controlled recirculating system until the age of 10-days post-fertilization (dpf), then moved to 3-l tanks. Adult fish were fed twice daily with a combination of hatched Great lake Artemia nauplii and ground fish flakes (TetraMin, Tetra USA, Blacksburg, Virginia), whereas larvae were fed powdered larval food (AP100; Zeigler, Gardners, Pennsylvania) and A. nauplii are added to their diet at 10-dpf. All fish maintenance and experimental protocols were approved by the Institutional Animal Care and Use Committee at Cornell University.

CRISPR/Cas9-mediated targeted EGFP knock-in system

A CRISPR/Cas9-mediated targeted knock-in via in site-specific non-homologous end joining was performed in zebrafish as previously reported (Auer et al., 2014; Kimura et al., 2014) (Figure 1) with some modifications. This method allowed for the development of a transgenic line with a vtg1 regulatory region driving the expression of EGFP. In brief, the donor plasmid was generated by introducing an hsp70 promoter sequence (Halloran et al., 2000) and an EGFP sequence encapsulated with 2 Mbait sequences (Auer et al., 2014; Kimura et al., 2014) into a pBluescriptSK plasmid. Two single-guide RNAs targeting sites within 1000 bp upstream of vtg1 were selected (vtg1-sgRNA1 and vtg1-sgRNA2; see Supplementary Table 1 for sequences) using a sgRNA Designer website (Broad Institute, Cambridge, Massachusetts). To describe the mechanism in brief, vtg1-sgRNA directs CRISPR/Cas9-mediated cleavage at the target regulatory region of the vtg1; the Mbait-sgRNA induces cleavage of the donor plasmid. These cleavages trigger homology-independent DNA repair that results in an integration of the cleaved donor plasmid at the cleaved vtg1 regulatory region with high frequency. This integration allows for enhancer trapping, in which cis-regulatory elements of vtg1 drive the hsp70 promoter resulting in EGFP expression. Integration may occur in either the forward or reverse direction, but both remain functional at different efficacies (Auer et al., 2014; Kimura et al., 2014).

Figure 1.

A diagram showing the CRISPR/Cas9-mediated homology-independent knock-in strategy adopted in generating tg[vtg1:EGFP] zebrafish for the current study. Three solutions of Cas9 nuclease pre-mixed with (1) vtg1-sgRNA (for genome digestion) or (2) Mbait-sgRNA (for plasmid digestion) and (3) donor plasmid are co-injected into one-cell stage Casper-mutant embryos. Concurrent digestion of the donor plasmid and the genome (600–900 bp upstream of vtg1 transcriptional start site) using sgRNA-guided Cas9 nuclease occurs causing a double-strand DNA break. The cis-regulatory sequences of vtg1 acts on the hsp70 promoter, which induce EGFP expression in cells that express vtg1 (enhancer trapping). Abbreviations: EGFP, enhanced green fluorescent protein; hsp, heat shock protein; vtg1, vitellogenin1.

Synthetic oligos of Alt-R CRISPR-Cas9 CRISPR RNA (crRNA of target sequences), Alt-R CRISPR-Cas9 universal transactivating crRNA (tracrRNA, cat#1072532), and Alt-R S.p. CRISPR-Cas9 nuclease (cat#1081058) were ordered from Integrated DNA Technologies (Skokie, Illinois). The crRNAs and tracrRNA were resuspended in nuclease-free duplex buffer (30 mM HEPES, pH 7.5; 100 mM potassium acetate) to final concentrations of 100 µM. Each sgRNA was then mixed with the universal tracrRNA and the duplex buffer at a ratio of 1:1:2 then heated at 95°C for 5 min to create the crRNA/tracrRNA complex. The CRISPR-Cas9 nuclease was diluted with Cas9 working buffer (20 mM HEPES, 150 mM KCl, pH 7.5) to a final concentration of 1 µg/µl. Aliquots of each sgRNA (crRNA/tracrRNA complex) and Cas9 nuclease were made and stored at −80°C until use.

Generation of tg[vtg1:EGFP] zebrafish transgenic line

On the day of the injection, the injection mixture was prepared by combining the vtg1-sgRNA (sgRNA1 or sgRNA2) and Mbait-sgRNA each with Cas9 nuclease in 1:1 ratio in a 2 µl mixture and incubating at 37°C for 10 min to allow Cas9/sgRNA complex to form. After letting the mixture cool down to room temperature, the donor plasmid was added and this final injection mixture consisting of sgRNAs (∼60 ng/µl each of crRNA), Cas9 nuclease (200 ng/µl), and the donor plasmid (50 ng/µl), was injected into 1-cell stage Casper embryos (Figure 1). Total of >2500 1-cell stage embryos were injected (∼1.4 nl intracellular injection) for each sgRNA (vtg1-sgRNA1 and vtg1-sgRNA2) across ∼14 sessions. Larvae with positive eye fluorescence at 3-dpf were raised into adulthood. Founders were identified for vtg1-sgRNA1 and vtg1-sgRNA2 by outcrossing with regular Caspers and testing for germline transmission by examining F1 larvae for EGFP fluorescence in the eyes and further confirming positives via EGFP induction in liver after exposure to 100–125 nM of 17β-estradiol (E2) from 4-h post-fertilization (hpf) to 5-dpf. Once founders were identified, outcrossing with regular Caspers were repeated using the original F0 founders to select F1 larvae with positive eye expression. These positive F1 larvae were raised to adulthood, then in-bred or out-crossed to create F2 and subsequent generations. Heterozygotes resulting from a cross of F1 × Casper were used for experiments for further evaluation of this zebrafish line.

Test exposures

Estradiol (E2) and bisphenol A (BPA) (Sigma-Aldrich, Saint Louis, Missouri) were dissolved into DMSO and highest treatment media was made by diluting with 10% Hank’s balanced salt solution (HBSS). Subsequent treatment media were prepared by serially diluting this highest treatment media with 10% HBSS containing 0.0034% DMSO to yield 1, 2.5, 5, 25, 125, and 625 nM E2, and 5, 10, 20, and 30 μM BPA (final concentration of 0.0034% DMSO in all treatment media). All treatment media, including control (10% HBSS, 0.0034% DMSO) were prepared fresh daily and maintained at 28 ± 1°C until use. At 2-hpf, embryos were collected and dechorionated using enzymatic digestion. Subsequently, at 5-hpf, 40 healthy embryos were randomly assigned for each treatment in 60 ml glass petri dishes containing 40 ml of the different treatment media (n = 5 replicate dishes/treatment). At 3-dpf, embryos were checked for eye expression and those without expression were removed, resulting in 15–22 embryos per replicate. Embryos were reared in a temperature-controlled chamber (28 ± 1°C) at 14:10-h light:dark photoperiod. Treatment media were changed daily, and mortalities were recorded. Observations were made through 3, 4, 5, and 6-dpf (to allow 72, 96, 120, 144-h of exposure from 5-hpf), and samples were collected at 6-dpf.

Evaluation of fluorescence and ranking

At 3, 4, 5, and 6-dpf, embryos/larvae were screened under an epifluorescence stereomicroscope (M205 FA, Leica Microsystems Inc, Buffalo Grove, Illinois) to confirm their transgenic status and identify the prevalence and ranks of EGFP expression in the liver after E2 exposure. Enhanced green fluorescent protein expression in the liver was ranked as 3 intervals: 1–10 foci, 10–100 foci, and >100 foci (Figure 2) subjectively by a human eye through an epifluorescent microscope. Defined spots of visible EGFP fluorescence was quantified as foci. At 6-dpf/termination, larvae were sub-categorized based on the number of liver EGFP foci and sorted into 4 different plates according to EGFP expression ranks. After the sorting was complete during 1 trial, 10 larvae from each ranking group were selected randomly in a blinded fashion for further imaging. Images were acquired using a CMOS camera (Hamamatsu, Japan) for further quantitative analysis.

Figure 2.

Hepatic EGFP expression in 6-dpf tg[vtg1:EGFP] zebrafish following E2 exposures. A, Representative epifluorescence images of 6-dpf tg[vtg1:EGFP] larvae showing no liver EGFP foci (0 foci), less than 10 foci (1–10 foci), between 10 and 100 foci (1–100 foci), and more than 100 foci (>100 foci) with exposures to E2 (125 nM). Higher magnification at the site of the liver is showed using different filters: GFP (bandpass Ex 470/40, Em 525/50; Ch1), mCherry (bandpass Ex 560/40, Em 630/75; Ch2), and merge (Ch1+Ch2). Lower magnification image showing the entire zebrafish with additional dark field transillumination overlay is shown to the right (Ch1+Ch2+DF). Additional mCherry filter was used to show the non-specific autofluorescence of the zebrafish gut to distinguish liver EGFP expression. Arrow (↑) indicates positive liver EGFP expression and asterisks (*) indicates positive eye expression. All scale bars represent 200 μm. B, Casper mutant (6-dpf) showing autofluorescence (Ch1+Ch2+DF) in the gut region. No green eye expression is seen. C, Percentage of larvae showing liver EGFP expression categorized by liver EGFP expression ranks (1–10, 10–100, and >100 foci) at 6-dpf after 0, 2.5, and 125 nM E2 exposure in 2 different vtg1-sgRNA target sites (vtg1-sgRNA1 and vtg1-sgRNA2). The results are expressed as a mean percentage of each ranks shown in stacks (n = 5). Significant differences are marked by lower case letters when at least one rank shows a significance (a and b). Abbreviations: EGFP, enhanced green fluorescent protein.

Quantitative evaluation of fluorescence by imaging software

In order to assess if subjective ranking of expression levels by human can be replaced with more objective methods, quantification of liver EGFP expression was performed using ImageJ (NIH) for 10 larvae for each expression rank. After eliminating background autofluorescence using the threshold tool, changes to (1) fluorescence area, (2) mean fluorescence intensity, (3) maximum fluorescence intensity, and (4) integrated fluorescence density of the targeted liver EGFP were analyzed and compared between the different ranks.

Gene expression analysis

To determine if the vtg1 and EGFP mRNA expression, in the liver as well as in the eye, correlate with the visual liver fluorescence expression, abdominal and head regions of the larvae were collected separately (on ice) for RNA extraction. Larvae were selected in the blinded fashion as described in the larval selection for imaging, and pooled samples (5 heads or abdomens; n = 5/tissue/expression ranks) were lysed using TissueLyser II (Qiagen, Valencia, California) and stored at −80°C until further processing. Total RNA was extracted with Trizol reagent (Thermo Fisher Scientific, Waltham, Massachusetts) followed by a cleanup with a RNeasy MinElute Cleanup Kit (Qiagen) and quantified using the Nanodrop 2000c spectrophotometer (Thermo). The purity of total RNA was evaluated with 260/280 and 260/230 ratio, with both being >1.8. All samples were treated with Deoxyribonuclease (DNase) I (Thermo) and reverse transcribed using a high-capacity reverse transcription kit (Applied Biosystems, Foster City, California) with Oligo(dT)18 primers (Thermo). Primer sequences used for target genes (EGFP and vtg1) and housekeeping genes (rpl13a and ß-actin) are shown in Supplementary Table 2. For quantitative PCR (qPCR), the reaction mixture per well comprised of 10 ng total RNA equivalent of cDNA template, 10 µM each of forward/reverse primers, SYBR Select Master Mix (Applied Biosystems), and autoclaved MilliQ water. Reactions were performed in 10 µl volumes in 384-well plates using QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems). The C_t (threshold cycle) value was obtained and the relative amount of amplicon was calculated using the relative standard curve method. Both housekeeping genes tested (rpl13a and ß-actin) did not show any statistically significant difference between groups from the same tissue type. Comparatively, as rpl13a showed better uniformity across different groups, it was selected as a reference gene for this study.

Statistical analysis

All data were analyzed using SPSS 25.0 (IBM-SPSS Inc, Chicago, Illinois) and presented as mean + standard error (SEM). Assumptions of normality and homogeneity of variance were tested, and in some cases, log-transformed data or non-parametric Kruskal-Wallis test were used. Multilevel linear regression followed by a Bonferroni correction was used to compare changes between concentrations and lengths of exposure (with and without expression ranks as a dependent variable) for E2 and BPA exposures. One-way ANOVA followed by post hoc Tukey’s test was used for the rest of the comparisons. Differences were determined significant at p ≤ .05.

RESULTS

Generation of Knock-in tg[vtg1:EGFP] Zebrafish

Successful expression of the knock-in transgene was observed with restricted EGFP fluorescence to the eyes, starting at 48∼72-hpf in tg[vtg1:EGFP] zebrafish. Total of >2500 eggs were injected for each of the sgRNA across ∼14 injection sessions. The prevalence of positive eye EGFP expression of the surviving larvae at 3-dpf varied per injection session (0%–10% positive rate). One female F0 founder was identified out of 21 vtg1-sgRNA1 injected embryos with positive eye expression raised to adulthood (4.8%), whereas 2 male founders were identified out of 26 vtg1-sgRNA2 injected embryos raised to adulthood (7.7%). Each of the F0 founders generated showed germline transmission in their offspring at a rate of 2%–20% when bred to regular Casper. In subsequent generation, it was confirmed that all fish carried the transgene expressed EGFP in the eye from 3-dpf, with the earliest expression starting from 48-hpf.

Hepatic EGFP Expression Following E2 Exposure

Exposure of tg[vtg1:EGFP] zebrafish to E2 (125 nM) resulted in increased fluorescence restricted to the liver in both vtg1-sgRNA1 and vtg1-sgRNA2 derived lines (Figure 2A). In both control Casper (Figure 2B) and tg[vtg1:EGFP] zebrafish, we consistently observed background autofluorescence associated with the gut. Test exposure of 2.5 and 125 nM E2 showed that vtg1-sgRNA1 had higher sensitivity in inducing liver EGFP expression than the vtg1-sgRNA2 derived lines (Figure 2C). Therefore vtg1-sgRNA1 lineage was selected for further experiments herein.

To test the estrogenic sensitivity of the tg[vtg1:EGFP] zebrafish transgenic line developed, embryos were exposed to different concentrations of E2. There was a general dose-dependent increase (Figure 3A;p < .001) and a time-dependent increase (Figure 3B;p < .001) in percentage of EGFP-positive larvae. At all time-points (3, 4, 5, and 6-dpf; Figure 3A), hepatic EGFP foci were detected in the tg[vtg1-EGFP] zebrafish larvae at concentrations ≥2.5 nM E2. Significant differences from the control group were, however, only observed at E2 exposure concentrations ≥5 nM at all time-points (23.7 ± 7.30%, 47.91 ± 6.72%, 68.84 ± 4.98%, and 80.47 ± 3.68%, for 3, 4, 5, and 6-dpf, respectively, with 5 nM). Exposure to 125 nM E2 resulted in the highest percentage of larvae with EGFP foci in their livers (80.82 ± 7.21%) at 3-dpf compared with other treatment groups. E2 exposure at or above 5 nM showed an 80%–100% response at 6-dpf. The only treatment group that reached a consistent 100% response, however, was 125 nM group, at 5 and 6-dpf.

Figure 3.

Presence of liver EGFP expression with E2 exposures in tg[vtg1:EGFP] larvae. A, Percentage of larvae showing liver EGFP expression grouped by data collected at different larval age (3, 4, 5, and 6-dpf) during different concentrations of E2 exposures (0, 1, 2.5, 5, 25, 125, and 625 nM). B, Same data are shown in different format—grouped by different concentration comparing different larval stage (3, 4, 5, and 6-dpf). All exposures started at 5-hpf. The results are expressed as a mean percentage and error bars represent SEM (n = 5). Significant differences (p < .05) in responses to exposure at different concentrations (A) or lengths of exposure (B) are marked by lower case letters (a–d). Abbreviations: EGFP, enhanced green fluorescent protein; E2, estradiol.

In performing these E2 exposures, it was noticed that individual tg[vtg1-EGFP] hepatic EGFP expression levels within each concentration group could be ranked at 3 intervals/levels: (1) 1–10 foci, (2) 10–100 foci, and (3) >100 foci (Figure 2). Therefore, changes in the percentages of larvae with different EGFP expression ranks were evaluated using a multilevel linear regression model (Figure 4). Significant differences in percentages of larvae with different EGFP expression ranks were observed between the different exposure concentrations (Concentration * Rank, p < .001; Figure 4A) and the different exposure periods (Time * Rank, p < .001; Figure 4B) tested. There were also significant interactions between concentration, exposure duration, and expression rank (Concentration * Time * Rank, p < .001). Integrating information on these intervals/ranks indicated that the number of larvae with 1–10 hepatic EGFP foci were the first to be observed in the 2.5 nM E2 group at 3-dpf (Figure 4B). Larvae with 10–100 hepatic EGFP loci started to emerge in this same treatment group only at 5-dpf. Larvae with >100 hepatic EGFP loci could be observed only at concentrations ≥5 nM E2 (Figure 4B) but was observed as early as 3-dpf with 25 and 125 nM E2 (Figure 4A). Exposure to 125 nM E2 at 6-dpf induced the greatest percentage of larvae showing >100 hepatic foci among all experimental groups (82.11 ± 6.21%). Significant differences among groups were affected when ranking information was incorporated into the statistical analysis. For example, there were no significant differences in total percentage of positive larvae in 125 nM E2 groups between exposure periods (Figure 3B) but when semi-quantitative ranking information was included, significances were detected in at least one of the rankings between some exposure periods (Figure 4B).

Figure 4.

Level of liver EGFP expression with E2 exposures in tg[vtg1:EGFP] larvae. A, Percentage of larvae categorized by liver EGFP expression ranks (1–10, 10–100, and >100 foci) grouped by data collected at different larval age (3, 4, 5, and 6-dpf) during different concentrations of E2 exposures (0, 1, 2.5, 5, 25, 125, and 625 nM). B, Same data are shown in a different format—grouped by different concentration comparing different larval stage (3, 4, 5, and 6-dpf). All exposures started at 5-hpf. The results are expressed as a mean percentage of each ranks shown in stacks (n = 5). Significant differences (p < .05) between any of the ranks in responses to exposure at different concentrations (A) or lengths of exposure (B) are marked by lower case letters (a–d). Abbreviations: EGFP, enhanced green fluorescent protein; E2, estradiol.

Correlation between vtg1 and EGFP mRNA Levels in Response to E2 Exposure

tg[vtg1:EGFP] larvae expressed EGFP in their eyes under normal rearing conditions but liver EGFP foci were only induced in response to E2 at different expression levels at 3∼6-dpf (Figure 2). To assess whether subjectively selected liver EGFP expression ranks of tg[vtg1:EGFP] larvae correlate with vtg1 and egfp mRNA expression levels, gene expression analysis was performed on pooled samples with different levels of EGFP expression. To determine if gene expression in the eyes are also affected by E2 exposure, zebrafish larvae was divided into 2 sections (head and abdomen). No significant differences in EGFP and vtg1 expression levels were recorded between the head samples collected from the different groups (Figs. 5A and 5B, respectively), whereas statistically significant differences in EGFP and vtg1 expression levels were recorded between the abdomen samples collected from the different groups (p < .001). A group of larvae that did not express any EGFP foci in their livers at 6-dpf despite E2 exposures (group 0) did not show any significant difference in vtg1 and EGFP expression levels when compared with the control group that were not exposed to E2 (Figs. 5A and 5B, respectively). Significantly higher levels of EGFP and vtg1 expression were recorded in all groups in abdominal samples of larvae showing liver EGFP foci categorized under any of the 3 positive expression ranks. Moreover, the fold increase in expression levels of EGFP and vtg1 in groups of larvae showing 10–100 liver EGFP foci (125.54 ± 55.17 and 6245.32 ± 1889.30, respectively) was significantly higher than the fold increase in expression levels of EGFP and vtg1 observed in groups of larvae showing 1–10 liver EGFP foci (8.65 ± 1.40 and 701.69 ± 200.44, respectively). Also, the fold increase in expression levels of EGFP and vtg1 in groups of larvae showing >100 liver EGFP foci (1991.70 ± 578.53 and 101 040.48 ± 32 582.87, respectively) was significantly higher than the 2 lower severity ranks (Figs. 5A and 5B).

Figure 5.

Expression of mRNA in 6-dpf tg[vtg1:EGFP] larvae exposed to E2 (1–625 nM) and grouped by liver EGFP expression ranks (0, 1–10, 10–100, and >100 foci). Zebrafish larvae were dissected into the head and abdominal region to assess the mRNA expression of (A) enhanced green fluorescence protein (EGFP) and (B) vitellogenin 1 (vtg1) in the eyes versus the liver. Expression of each target gene was normalized to ribosomal protein L13a (rp13a) as housekeeping gene and are shown as the mean of fold change to the control (exposed to 10% Hank’s with 0.0034% DMSO) of the same body region. Error bars represent SEM (n = 5). Significant differences (p < .05) between groups of the same body regions are denoted with lower case letters (a–d). Abbreviations: EGFP, enhanced green fluorescent protein; E2, estradiol.

Quantitative Analysis of Fluorescence between Different EGFP Expression Ranks

To assess if quantification of changes in EGFP expression in the livers of tg[vtg1:EGFP] zebrafish larvae is possible with the use of imaging software, and whether its findings correlate with those of the ranks selected subjectively, images of larvae showing different expression ranks were analyzed using ImageJ. Changes in the area, mean intensity, maximum intensity, and the integrated density of EGFP fluorescence were recorded. The area of EGFP fluorescence in the liver after E2 treatments significantly increased in all 3 positive expression ranks compared with the control group (no foci) and also showed significant differences between the different ranks (Figure 6A; 0.53 ± 0.13, 7.69 ± 1.03, and 54.73 ± 4.38 × 10³ µm² for 1–10, 10–100, and >100 foci, respectively, vs control group). Changes in mean EGFP fluorescence intensity were significantly different from the control group in all expression ranks (1.75 ± 0.22, 1.54 ± 0.34, and 13.48 ± 1.99 × 10³ for 1–10, 10–100, and >100 foci, respectively). The highest expression rank group was significantly higher from lower rank groups, but the difference could not be observed between 1–10 and 10–100 foci rank groups (Figure 6B). Maximum EGFP intensities were significantly higher in all expression rank groups (2.96 ± 0.43, 5.75 ± 1.02, and 42.29 ± 4.67 × 10³ for 1–10, 10–100, and >100 foci, respectively) compared with the control, and the ranks were significantly different from each other (Figure 6C). Finally, the integrated density, a product of area and mean fluorescence intensity, showed significant increase in all 3 expression ranks (0.1 ± 0.03, 1.1 ± 0.22, and 75.6 ± 11.8 × 10⁷ for 1–10, 10–100, and >100 foci, respectively) compared with the control group and in increasing order of ranks (Figure 6D).

Figure 6.

Quantitative analysis of EGFP expression by imaging software in 6-dpf tg[vtg1:EGFP] larvae exposed to E2 (1–625 nM) and grouped by liver EGFP expression ranks. The area (A), mean intensity (B), maximum intensity (C), and the integrated density (D) of liver EGFP fluorescence was analyzed using Image J after background subtraction of autofluorescence. Results expressed as mean + SEM (n = 10). Significant differences (p < .05) between groups are denoted with lower case letters (a–d). Abbreviations: EGFP, enhanced green fluorescent protein; E2, estradiol.

Prevalence and Extent of EGFP Expression following BPA Exposure

To test the sensitivity of the tg[vtg1:EGFP] zebrafish transgenic line with exposure to non-steroidal estrogens, embryos were exposed to different concentrations of BPA from 5-hpf until 6-dpf and percentage of larvae expressing EGFP foci in the liver was monitored (via subjective sorting/counting) at 6-dpf. There were significant differences (p < .001; Figure 7A) in the percentage of larvae with positive liver EGFP expression between different exposure concentrations. Enhanced green fluorescent protein liver expression was not observed in any larvae of control and 5 μM BPA exposure group at 6-dpf, but was first detected in 10 μM exposure group (3.33 ± 3.33%; Figure 7A). Significant increases from the control groups were observed in 20 and 30 μM (23.37 ± 5.20 and 48.14 ± 3.98%, respectively, p < .05). Similar to E₂, the induction of liver EGFP in tg[vtg1:EGFP] larvae occurred at different expression ranks after BPA exposure (Figure 7B). Changes in exposure concentrations, expression ranks, and the interaction of both variables were found to have significant effects (p < .001, p < .001, and p < .01, respectively). Exposure to 10 μM BPA resulted in percentages of larvae with EGFP foci in their livers categorized under 1–10 foci only, whereas larvae exposed to 20 and 30 μM BPA concentration have shown percentages of larvae with EGFP foci in their livers categorized under 1–10 or 10–100 foci. None of larvae had >100 foci in any of the groups. In addition, deformities, in the form of an unabsorbed yolk sac, pericardial edema, and microphthalmia were significantly increased in larvae exposed to 30 μM BPA compared with larvae exposed to 20 μM BPA (23% and 17%, respectively, p < .05). Deformities were not detected in larvae exposed to lower 2 concentrations (5 and 10 μM BPA) or in the vehicle control.

Figure 7.

Liver EGFP expression in 6-dpf tg[vtg1:EGFP] larvae with exposures to bisphenol A (BPA). Percentage of larvae showing positive liver EGFP expression (A) and same larvae categorized by liver EGFP expression ranks, 1–10 and 10–100 foci, (B) after exposure to BPA from 5-hpf to 6-dpf. The results are expressed as mean percentage + SEM or as mean percentage of each rank shown in stacks (n = 5). Significant differences (p < .05) between concentrations are marked by lower case letters (a–c). For (B), significance is noted when at least one of the ranks show a difference between concentrations. Abbreviations: EGFP, enhanced green fluorescent protein; BPA, bisphenol A.

DISCUSSION

With a large proportion of chemicals used in the consumer market being untested for potential endocrine-disrupting capabilities, there is a dire and increasing need for efficient cost-effective screening methods using whole animal organisms. Zebrafish is an ideal model system for a variety of reasons including ease of exposures and abilities to generate transgenic lines with fluorescent markers that can be monitored in vivo. In this study, we have successfully generated a transgenic zebrafish line, Tg[vtg1:EGFP], using a CRISPR/Cas9-mediated knock-in approach (Kimura et al., 2014), in which vtg1 transcriptional regulators drive the expression of EGFP.

Exposure to E2 concentrations as low as 5 nM induced EGFP expression in the livers of tg[vtg1:EGFP] larvae as early as 3-dpf and up to 6-dpf. Exposure to higher concentrations of E2 or longer exposure periods resulted in higher percentages of larvae expressing EGFP in their livers. This lowest observed effect concentration (LOEC) of 5 nM E2 seem to be consistent with previous report by Muncke et al. (2007), where vtg1 mRNA expression assessed by real-time qPCR was induced at 10 nM E2 but not at 1 nM with a 5-day embryonic exposure. The reported LOEC in a tg[ere-zvtg1:gfp] was 100 ng/l (0.37 nM) E2 (Chen et al., 2010) but with much longer exposure than the current study (13 days total, 7–20-dpf). The earliest induction of GFP expression in larvae of tg[mvtg1:gfp] medaka was ∼10-dpf after being exposed to 5 µg/l (18.36 nM) E2 starting 64-hpf (Zeng et al., 2005). In the same study, exposure of 3-month old males to E2 for 30 days resulted in LOEC of 0.5 µg/l (1.84 nM) E2 in a live fish, although liver dissection did result in an even lower LOEC of 0.05–0.1 µg/l (0.18–0.37 nM) E2. Bakos et al. (2019) more recently reported that fluorescence was detected at 100 ng/l (0.37 nM) of E2 with 2 day exposure (3–5-dpf) with calculated integrated density generated from images. Adopting this imaging analysis would likely increase the sensitivity of our line as well, although it remains to be tested. In addition, Bakos et al. (2019) also evaluated sensitivity in homozygous fish whereas our study only used heterozygous fish. Our preliminary study showed a greater response using tg[vtg1:EGFP] homozygous fish (unpublished), thus use of homozygous fish may be a consideration for increased sensitivity in future studies.

Studies have shown that exposure to supraphysiologic concentrations of estrogens is required to attain physiologically relevant concentrations in vivo and that E2 uptake increases with increasing concentrations, duration of exposure, and developmental stage (Souder and Gorelick, 2017). Although, the E2 concentrations used in the current as well as in other in vivo studies, can be considered magnitudes higher than physiologically relevant, 5 nM E2 concentration in water can be calculated to correspond to 190 pM E2 in the zebrafish, assuming 3.8% uptake (Souder and Gorelick, 2017), making the dosage physiologically relevant. Extending exposure duration beyond 6-dpf may result in higher sensitivity. It has also been reported that juvenile fish during gonadal differentiation (Legler et al., 2000) and adult fish, particularly males (Bakos et al., 2019) are more sensitive to effects of estrogens than larvae. It is likely using older and male fish will increase the sensitivity substantially in tg[vtg1:EGFP] as well, however not ideal for high-throughput application as it would require feeding of the fish beyond 6-dpf.

Liver EGFP induction in response to E2 exposure occurred at different expression ranks in tg[vtg1:EGFP]. Although Zeng et al. (2005) also classified the extent of EGFP expression into different levels in their medaka studies, statistical analysis was not performed to evaluate the value of this categorization. In our study, additional information on the extent of liver EGFP expression was evaluated by statistical analysis providing added insight into effects of E2 concentrations on vtg1 expression, revealing an inverted U-dose response curve, typical of steroid hormones. Such a semi-quantitative ranking/scoring approach has been successfully used in the histopathological evaluation of EDC effects such as the intersex condition in fish (Abdel-Moneim et al., 2017; Bateman et al., 2004). We confirmed that our ranking system corresponded positively with changes in liver vtg1 and egfp expression by real-time qPCR. This shows promise for semi-automation for the screening of estrogenic compounds using imaging software. Our findings suggest that the transgenic line developed in this study provides a sensitivity to estrogenic exposure that is in alignment with endogenous vtg1 mRNA induction, earlier onset of vtg1 expression in a larval fish, with relatively short exposure time, thus showing promise for high-throughput pipelines.

Bisphenol A is known for its estrogenic potency several orders of magnitude lower than E2. The tg[vtg1:EGFP] was able to sense the estrogenicity of BPA in a dose-dependent manner. As with E2 exposure, previous report of vtg1 mRNA expression analyzed by qPCR was consistent with our findings—significant induction with 10 µM BPA, but not with 5 µM after 5 days of exposure (Muncke et al., 2007). Sensitivity of our fish line to BPA also seems to be similar to other in vivo transgenic lines developed for screening of estrogenic compounds. Chen et al. (2010) reported GFP liver induction in about 5% of fish with BPA exposure at 1 and 10 mg/l (4.38 and 43.8 µM). Zeng et al. (2005) reported GFP fluorescence in a live fish at 5 mg/l (21.9 µM) and in a dissected liver of a fish at 1 mg/l (4.38 µM). Although it is difficult to compare the sensitivity of the 3 lines because both previous studies used longer exposure duration (13 and 21 days) in an older fish than the current study, our line generated greater percentages of positive fish (maximum of 48%) with only 6 days of exposure within the similar ranges of dosage tested. Gorelick and Halpern (2011) reported 90% of larvae positive for GFP in the liver and heart of tg[5xERE:GFP] line at 2.3 mg/l (10 µM) BPA with 5-day exposure. They also reported that 0.23 mg/l (1 µM) BPA caused GFP induction only in the heart and not in the liver, so the response in the liver seems very comparable to the current study, although biomarker used is different (estrogen receptor binding vs vtg1 expression). More recently, Bakos et al. (2019) reported detectable mCherry expression using tg[vtg1:mCherry] embryos at 1 mg/l (4.38 µM) BPA with 2 days of exposure (3–5-dpf).

The dose of BPA tested in this study, 5–30 µM is certainly higher than concentrations considered environmentally relevant. Highest level of BPA reported in river waters in the USA is 12 µg/l (52.56 nM) with median detected concentration at 0.14 µg/l (Kolpin et al., 2002). Reported human serum BPA concentrations are 0.2–20 ng/ml (0.88–87.61 nM) (Vandenberg et al., 2007). This raises a limitation to the in vivo early embryonic stage model to detect estrogenic compounds at environmentally relevant levels, and sample concentration may be necessary before detection/screening. In comparison, use of in vitro approaches such as fish primary hepatocyte, MCF-7 (E-screen), yeast estrogen system (YES), have BPA EC50 values in ranges of 140 nM∼84 mM, can be considered more sensitive. Nevertheless, inconsistencies with in vivo results due to lack of metabolic and endocrine feedback persist as confounding factors [reviewed in (Navas and Segner, 2006)]. In most cases, purpose of developing high-throughput screening is not necessarily to determine environmental safety levels, or to “detect” estrogenic potential in an environmental sample, but to find chemicals that have the potential for certain effects (eg, estrogenic response) when safer chemical alternatives or novel compounds for certain applications are being evaluated. Therefore, chemicals can be tested at higher doses to determine its estrogenic capability and their EC50 or LOEC used to calculate relative estrogenicity to E2. Regardless, testing compounds using multiple approaches would be necessary to ensure consistency and care in interpretation.

The EGFP eye expression was not initially anticipated when designing CRISPR/Cas9 system, it became apparent that this was driven by the hsp70 promoter. We adopted a design reported by Kimura et al. (2014) which included the hsp70 promoter for enhancer trapping based on their previous experience of generating BAC-transgenic zebrafish (Satou et al., 2012, 2013), because it not only yielded increased levels of transgene expression, but also allowed forward and reverse integrations to be functional. Halloran et al. (2000) also reported zebrafish eye fluorescence driven by the hsp70 promoter in transgenic offspring by 48-hpf. Although Kimura et al. did not specifically mention the expression of EGFP in the eyes, we confirmed that they also made this observation (personal communication). In the current study, we confirmed that EGFP mRNA remains unchanged in the head region, even when individual fish are grouped according to the liver EGFP expression ranks with E2 exposure. The presence of eye EGFP in the early embryonic stage was useful in identifying offspring with transgene incorporation, because physiological liver vtg1 induction occurs only later in development and only in females. Therefore, when the targeted gene is not normally expressed in early embryonic stages, use of hsp70 promoter for enhancer trapping can be very useful for screening positive embryos.

Albeit the utility, the use of hsp70 promoter is not without potential for negative consequences in uncontrolled environments. Exposure to heat shock (42°C) for 1-h induced diffuse whole body EGFP expression in 100% of tg[vtg1:EGFP] zebrafish larvae (Clark et al., unpublished data). Although this induction is unlikely to be problematic under standard conditions in a fish facility (usually ∼28.5°C), a possibility exists that hsp70-inducing compounds could enhance global EGFP background expression independent of vtg1 enhancer trapping. We, however, did not detect any differences in hsp70 mRNA expression with increasing liver EGFP expression (data not shown).

Changes in vtg1 expression, however, was several magnitudes higher than EGFP. The differences in qPCR primer efficiency and/or use of heterozygous tg[vtg1:EGFP] zebrafish in the current experiment may at least partially explain this. It is also possible that the enhancer trapping is not working as efficiently as expected. Experimenting with alternative enhancer trap loci on the vtg1 upstream regulatory region could potentially result in transgenic lines with even higher sensitivity.

In summary, we first report the generation of a tg[vtg1:EGFP] zebrafish transgenic line using CRISPR/Cas9-mediated knock-in method to serve as a biosensing screening for exposure to estrogenic compounds. This line is sensitive to exposure to steroidal (E2) and non-steroidal (BPA) estrogenic EDC with responses observed as early as 3-dpf and has the potential to be used for high-throughput screening applications.

DECLARATION OF CONFLICTING INTERESTS

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.