The phenotypic and transcriptomic effects of developmental exposure to nanomolar levels of estrone and bisphenol A in zebrafish

Abstract

Estrone and BPA are two endocrine disrupting chemicals (EDCs) that are predicted to be less potent than estrogens such as 17β-estradiol and 17α-ethinylestradiol. Human exposure concentrations to estrone and BPA can be as low as nanomolar levels. However, very few toxicological studies have focused on the nanomolar-dose effects. Low level of EDCs can potentially cause non-monotonic responses. In addition, exposures at different developmental stages can lead to different health outcomes. To identify the nanomolar-dose effects of estrone and BPA, we used zebrafish modeling to study the phenotypic and transcriptomic responses after extended duration exposure from 0 – 5 days post-fertilization (dpf) and short-term exposure at days 4 – 5 post fertilization. We found that non-monotonic transcriptomic responses occurred after extended duration exposures at 1 nM of estrone or BPA. At this level, estrone also caused hypoactivity locomotive behavior in zebrafish. After both extended duration and short-term exposures, BPA led to more apparent phenotypic responses, i.e. skeletal abnormalities and locomotion changes, and more significant transcriptomic responses than estrone exposure. After short-term exposure, BPA at concentrations equal or above 100 nM affected locomotive behavior and changed the expression of both estrogenic and non-estrogenic genes that are linked to neurological diseases. These data provide gaps of mechanisms between neurological genes expression and associated phenotypic response due to estrone or BPA exposures. This study also provides insights for assessing the acceptable concentration of BPA and estrone in aquatic environments.

Graphical Abstract

1. Introduction

Endocrine disrupting chemicals (EDCs) impact aquatic life and human health via acute or chronic exposures throughout life. EDCs interact with the normal activity of the body by activating or blocking hormone receptors, disrupting synthesis and degradation of hormones. Anthropogenic sources of EDCs in the environment include human liquid and solid wastes, as well as leaching from manufactured products with EDCs, such as phthalates and Bisphenol A (BPA). Contaminated surface water is a major EDC exposure route to aquatic life and may also contribute to human exposure via food chain pathways.¹ EDC levels in surface waters range from nano- to micro-molar with higher concentrations where surface waters receive the treated effluent of wastewater treatment plants and hospital sewage.^1,2 However, there is lack of rigorous scientific research to determine the health effects and risks of exposure to these low-dose EDCs. Except for long-studied chemicals, e.g. BPA, regulatory authorities around the world have not started to initiate regulatory actions to mitigate any environmental sources of EDC, due to the uncertainties related to long-term adverse human health effects under low-dose exposure, particularly during sensitive developmental windows.³

This study focuses on estrone and BPA, two estrogenic compounds that are frequently detected in surface water mainly because of human usage or wastes.¹ Estrone, an estrogen responsible for sexual development, is naturally produced by ovaries, adipose tissue, and adrenal glands. Postmenopausal women undergoing hormone therapy are typically prescribed estrone or estradiol,⁴ the latter of which can be converted to estrone, thus also resulting in estrone excretion in urine and faces.⁵ Estrone removal efficiency during the wastewater treatment process is highly variable, and estrone level in the final treated effluent can even exceed the influent level.⁶ Therefore, estrone can be detected in surface waters in a range of 0.01 to 1 nM.^7–9 Very few studies are available for the toxicity of estrone on aquatic organisms. The more potent estrogens, 17β-estradiol and 17α-ethinylestradiol, have received greater attention and are known to impair fish reproduction functions, including sex differentiation and gamete development,^10,11 and cause skeletal related abnormalities.¹² Estrone is about 3 to 54 times less potent than 17β-estradiol and 17α-ethinylestradiol, respectively.¹³ Recently, Ankely et al. found that waterborne estrone can be converted to 17β-estradiol metabolically in fish; therefore, the predicted potency of environmental estrone may be underestimated.⁹

BPA is a synthetic chemical that is largely produced for manufacturing polycarbonate plastics and epoxy resins in a variety of industrial and consumer products. Studies have shown that BPA can leach from a variety of plastic products, including food packages, water bottles,¹⁴ and PVC pipes in distribution systems and household water pipelines.¹⁵ Human potential exposures from these sources range up to 1 nM,¹⁶ much lower than what most studies to-date have focused on. Similar to other EDCs, BPA cannot be fully removed by the wastewater treatment process¹⁷, and thus has been detected in surface waters in a range of 0.1 to 100 nM.^1,18,19 Evidence suggests that BPA has possible health risks for aquatic wildlife and humans. Exposure to micromolar levels of BPA is known to change the sexual and neural development in fish species through both estrogenic and non-estrogenic pathways.^1,20,21 Toxicity of BPA in humans is based on epidemiological evidence, and includes disruption of fertility, pregnancy outcomes, neurodevelopment and behavior of newborns, as well as increased risk of metabolic disorder, liver and kidney diseases, and various cancers.²²

Since the endocrine system is sensitive and responsive to even very low concentrations of hormones, the health effects of estrone and BPA at nanomolar levels are in need of investigation. Hormone doses and biological effects can potentially have a non-linear relationship, e.g. a non-monotonic response, in which the strongest observable biological effect does not occur at higher concentrations but rather at lower concentrations. Additionally, specific exposure windows at different developmental stages can lead to different health outcomes. Non-monotonic dose responses were found in mice models after exposure to micromolar levels of BPA, including neurotoxic effects, such as a decrease in neuron numbers at the adult phase²³, and metabolic disruption, such as reduced glucose tolerance and fat mass at the fetal phase²⁴. Similar health effects due to BPA exposure were also shown in humans; specifically a non-monotonic relationship was found between human urinary BPA concentration and declines in sperm concentration, motility, and morphology²⁵, as well as between maternal urinary BPA and birth weight of newborns.²⁶ Only a few studies have investigated the potential health effects of exposure to estrogenic EDCs at concentrations as low as nanomolar levels. Chen et al. discovered a non-monotonic relationship between BPA and changes in sperm function and reproduction in adult zebrafish (Danio rerio) after a chronic exposure at early developmental (embryonic) stages.²⁷ A short-term developmental BPA exposure induced non-monotonic sex-specific changes on social behavior in adult zebrafish.²⁸ The more potent estrogen, 17α-ethinylestradiol, also induced non-monotonic vertebral malformations in Fathead minnows (Pimephales promelas) after developmental exposure.¹² More research is needed to evaluate if estrone, with relatively weaker estrogenic activity, could produce non-monotonic dose-responses at nanomolar levels.

The purpose of this study is to use the zebrafish model to assess the health effects of estrone and BPA at nanomolar exposure levels during critical early developmental windows. We designed our studies with two strategies: 1) to differentiate responses to exposure throughout embryogenesis versus during early larval development by evaluating a 5-day exposure from 0 – 5 days post-fertilization (dpf) and a 24-hour exposure at days 4 – 5 post fertilization); and 2) to identify the associated diseases and pathways implicated in human exposure, as well as the mechanisms potentially causing the observable effects by evaluating both phenotypic and transcriptomic changes. The results of this study will provide valuable insights into risk assessment to estimate the acceptable concentration of BPA and estrone in aquatic environments.

2. Materials and Methods

2.1. Fish husbandry

Zebrafish wild-type AB strain were housed in a recirculating system (Aquaneering. California, USA) on a 14 h: 10 h light/dark cycle with reverse-osmosis (RO) water buffered with Instant Ocean salts (Spectrum Brands, Wisconsin, USA). Water temperature was maintained at 27 – 30°C. Adult fish were fed Aquatox Fish Diet flakes (Zeigler, PA, United States) twice daily. During spawning, adult zebrafish were bred in spawning tanks with a sex ratio of 2 females: 1 male. A total of 360 embryos were collected 4 hours post fertilization (hpf), approximately at sphere stage of embryonic development. The embryos were cleaned with 58 ppm bleach (diluted from Chlorax bleach, the Clorox Co., Oakland, CA) for 5 minutes, rinsed, and sorted into groups for their respective estrone and BPA exposures at 28°C for 5 days. Zebrafish use protocols were approved by the Institutional Animal Care and Use Committee at Wayne State University, according to the National Institutes of Health Guide to the Care and Use of Laboratory Animals.

2.2. Chemical exposure

The stock solutions of estrone (CAS 53-16-7, Spectrum Laboratory Products, USA) and BPA (CAS 80-05-7, Sigma Aldrich, USA) were prepared in 100% acetone at an initial concentration of 10 μM and 1 mM, respectively. Embryos were dosed with 0.01, 0.1, 1, 10, or 100 nM estrone or 1, 10, 100, 1,000, or 10,000 nM BPA. Control fish were placed in vehicle (0.01% acetone (v/v) in buffered RO water (66 mg/L Instant Ocean Salt and 54 mg/L sodium biocarbonate, Aquarium Systems Inc, Mentor, OH). The diluted chemical solutions for exposure were prepared daily from aliquots of the stock solution. For each chemical, all of the collected embryos were exposed at two different periods: 0 – 5 dpf, defined as “extended duration” exposure, or 4 – 5 dpf, defined as “short-term exposure”. Exposures were performed in 6 well trays, and each well was plated with 7 mL water and at most 30 embryos per exposure concentration. Each day, water changes were performed, new exposure solution was added, and mortality was recorded. After the 5-day incubation period, larval fish were rinsed 3 times with saline RO water (600 mg/L Instant Ocean Salt (Aquarium Systems Inc, Mentor, OH) to end the chemical exposure.

2.3. Abnormality screening

Zebrafish embryos were visualized and screened at 24, 48, 72, 96, and 120 hpf for mortality and abnormalities by stereomicroscope (M165C, Leica Microsystem, Germany). The endpoints assessed were number of unhatched embryos, skeletal deformities, improperly inflated swim bladder, yolk sac edema, heart edema, and total abnormalities. Embryos were screened using 6.7X magnification with detailed evaluation occurring at a magnification of 50X. The abnormality rates across all exposure concentrations for each chemical were compared by chi-square test and pairwise comparison with Bonferroni corrections.

2.4. Behavioral test

After abnormality screening, the control and exposed larval fish with swim bladder and without abnormality were chosen for behavioral testing. A total of 24 larval zebrafish were tested for each exposure concentration with one larva per well (in 2 mL fish water) in a 24 well plate. Fish were allowed to acclimate for at least an hour at 27 °C before being placed into the DanioVision Observation Chamber (Noldus Information Technology, Wageningen, Netherlands) with a constant temperature at 28.5 ± 0.5 °C. The behavioral assay consisted of 3-minute light and dark alternating periods with a total of four light-dark cycles (24 min). Larval movements, including distance traveled, velocity, turn angle, and angular velocity were integrated every 30 sec. All behavioral tests were performed between 14:00 and 22:00.

The raw data was exported from EthoVisionXT14 into a spreadsheet in order to perform quality control. The data series were not normally distributed, therefore interquartile range (IQR) method was used to remove outliers from light and dark cycles. In the light cycle, data series were excluded if two serial data points were larger than 75^th percentile plus 1.5 of IQR after the 1:00 minute mark. In the dark cycle, data series were excluded if two serial data points were smaller than the median of the light data series. Lastly, data series that had a mean ratio of light:dark series equal or larger than 0.9 were removed. The behavioral data were then analyzed using ANOVA and Tukey’s HSD tests. Significance was considered at p value smaller than 0.05. The quality control and statistics were conducted using R (http://www.r-project.org).

2.5. RNA isolation

Larval fish were euthanized with tricaine methanesulfonate solution (150 mL fish water, 0.06 g tricaine methanesulfonate, and 0.1 g sodium bicarbonate) individually for 10 minutes and then pooled in RNALater™ (Thermo Fisher, Waltham, MA) at 4 °C for 24 hours. Five fish were pooled as one individual sample and three pools were collected per concentration of each chemical and submitted for transcriptome sequencing. The RNALater™ solution was removed after 24 hours, and the larval fish pool was stored in −80°C until RNA isolation. RNA was isolated using the RNeasy Lipid Tissue Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s specifications. RNA purity was measured with Qubit® 3.0 Fluorometer (Invitrogen, Carlsbad, CA).

2.6. Transcriptomic analysis

3’ mRNA-seq libraries were prepared from isolated RNA using QuantSeq 3’ mRNA-Seq Library Prep Kit FWD for Illumina (Lexogen, Vienna, Austria). Samples were normalized to 40 ng/μL (total input of 200 ng in 5 μL) and amplified at 17 cycles. Libraries were quantified using a Qubit® 2.0 Fluorometer and Qubit® dsDNA Broad Range Assay Kit (Invitrogen, Carlsbad, CA), and run on an Agilent TapeStation 2200 (Agilent Technologies, Santa Clara, CA) for quality control. The samples were sequenced on a HiSeq 2500 (Illumina, San Diego, CA) in rapid mode (single-end 50 bp reads). Reads were aligned to D. rerio (Build danRer10) using the BlueBee Genomics Platform (BlueBee, Rijswijk, The Netherlands). Quality control of sequencing is shown in Appendix A: Table S1. Differential gene expression between the control and exposure treatment was evaluated using DEseq2 (available through GenePattern; Broad Institute, Cambridge, Massachusetts). Genes with significant changes in expression, as defined by absolute log2 fold change value ≥ 0.75 and p-value < 0.05, were uploaded into Ingenuity Pathway Analysis software (IPA; QIAGEN Bioinformatics, Redwood City, CA) for analysis using RefSeq IDs as identifiers. Z-score was generated by IPA software to predict the activation (>0) or inhibition (<0) of a pathway.

3. Results

3.1. Extended duration estrone and BPA exposures at embryogenesis showed more apparent phenotypic responses than short-term exposures at the early larval development stage

3.1.1. Estrone

Fig. 1 shows the rate of abnormalities for control fish and fish exposed to estrone for 5-days at 0 – 5 dpf or 24-hours at 4 – 5 dpf. The extended duration exposure to estrone during embryonic development led to significant skeletal abnormalities, uninflated swim bladder, and total abnormalities p < 0.05). However, these abnormality rates were not significantly different between exposed and control groups (p > 0.05). Nevertheless, estrone exposure concentrations of 1 to 100 nM had higher skeletal abnormalities (27% ± 3%) than 0 to 0.1 nM treatments (7% ± 2%). At 1 nM treatment in particular, both the rate of total abnormalities (61%) and uninflated swim bladder (53%) were the highest compared to 0.01, 0.1, 10, and 100 nM exposures with average rates of total abnormalities and uninflated swim bladder 38% ± 11% and 32% ± 12%, respectively. If larval fish were only exposed to estrone for a short term at the early larval development stage, the abnormality rates of all exposed larval zebrafish were similar to the control condition (p > 0.05).

Fig. 1.

Abnormality rate of control fish and fish exposed to estrone or Bisphenol A (BPA) for 5-days at 0 – 5 days post-fertilization (dpf) or 24-hours at 4 – 5 dpf.

*Condition is significantly different from control (0 nM; p < 0.05).

Fig. 2 shows the behavior analysis tracking the locomotor activity of larval fish at 5 dpf in the dark and light. As expected, larval fish moved more in the dark than in the light at all conditions. For extended duration estrone-exposed fish, the distance traveled during dark at 1 nM decreased significantly compared to control group (p < 0.05) (Fig. 2A). This result corresponded to our observation that 1 nM exposure resulted in the highest rate of total abnormalities after the 5-day exposure. During the short-term exposure, the larval fish exposed to the least (0.01 nM) and highest concentrations (100 nM) were both hypoactive compared to the control (p < 0.05).

Fig. 2.

Behavior analysis) after 5-day exposure at 0 – 5 days post-fertilization (dpf) or 24-hour exposure at 4 – 5 dpf to estrone (A) or Bisphenol A (BPA; B).

*Condition is significantly different from control (0 nM; p < 0.05).

3.1.2. BPA

After the 5-day exposure encompassing embryonic development, skeletal abnormalities and total abnormalities were significantly affected by BPA treatment (p < 0.05) (Fig. 1). Specifically, rates of skeletal abnormalities (31%) at the highest BPA exposure (10000 nM) were significant compared to the control group (11%, p < 0.05). The total abnormalities rates of the 100 (71%) and 10000 nM exposures (81%) were significantly higher than the control group (59%, p > 0.05). For larval fish exposed at 4 – 5 dpf, we found a trend of increasing skeletal abnormalities, uninflated swim bladders, and total abnormalities as the exposure concentration increased starting from 10 nM. However, none of these phenotypic changes were statistically different from the control group (p > 0.05).

For extended duration BPA exposed fish, the distances traveled during dark at 100 and 10000 nM were significantly higher than the control group (p < 0.05) (Fig. 2B). These two BPA exposure concentrations also caused more total abnormalities compared to 1, 10, and 1000 nM exposures. The short-term exposure, on the other hand, showed a different behavior profile, specifically fish exposed to all concentrations (except 1000 nM) traveled significantly less than the control group in the dark (p < 0.05).

3.2. Non-monotonic transcriptomic response occurred at 1 nM BPA and estrone after extended duration exposure

We compared gene expression levels of exposed fish at low (0.1 nM estrone and 1 nM BPA), middle (1 nM estrone and 100 nM BPA), and high concentrations (100 nM estrone and 10000 nM BPA) with the control group (Fig. 3). Overall, BPA exposures resulted in > 2.5 times the number of differentially expressed genes (1375) compared to estrone exposures (523; Fig. 4). All differentially expressed transcripts are shown in Appendix B.

Fig. 3.

Venn diagram of the number of genes differentially expressed after exposure to estrone or Bisphenol A (BPA) after 5-day exposure at 0 – 5 days post-fertilization (dpf; extended duration) or 24-hour exposure at 4 – 5 dpf (short term).

Fig. 4.

Venn diagram of differentially expressed genes after exposure to estrone 5-day exposure at 0 – 5 days post-fertilization (dpf, A), 24-hour exposure at 4 – 5 dpf (B), or Bisphenol A (BPA) 5-day exposure at 0 – 5 dpf (C) or 24-hour exposure at 4 – 5 dpf (D). The genes included in this diagram have adjusted p value < 0.05 and log2 fold changes > 0.75. The number inside the circle and intersection indicate the number of genes differentially expressed per condition. The number outside the circle indicates the exposure concentration (nM).

3.2.1. Estrone

A total of 445 and 83 genes were differentially expressed after long- and short-term estrone exposure, respectively. Among them, 5 genes (pck1, pnp4b, psmb5, dnajc5gb, and zgc:92590) were differentially expressed after both exposure periods. After extended duration embryonic exposure, 342 genes were exclusively expressed at 1 nM estrone, which was the most transcriptomic changes among all conditions. Much lower numbers of genes were differentially expressed at 0.1 (52 genes) and 100 nM (60 genes) estrone. The gene profiles across the 3 estrone exposure concentrations were distinct. Only a limited number of genes were shared by either 2 or all 3 conditions. Among them, apoea was differentially expressed at 0.1, 1, and 100 nM estrone concentrations. The expression levels of genes that were changed in more than one exposure condition is shown in Appendix A: Fig. S1. Different exposure periods also affected the transcriptomic changes. After short-term exposure at 4 – 5 dpf, only 1 and 100 nM estrone concentration had differentially expressed genes, and 98% (82) were differentially expressed at 100 nM. In addition, the transcriptomic response to estrone during both exposure periods showed a similar trend as changes in behavioral responses, i.e. the two conditions with the highest number of transcriptomic changes, extended duration 1 nM and short-term 100 nM, also showed significant changes in locomotive behavior compared to the control.

3.2.2. BPA

A total of 168 and 1296 genes were differentially expressed after long- and short-term exposure, respectively, with 90 genes shared between both exposure periods. After extended duration embryonic exposure, 166 and 2 genes were differentially expressed at the lowest (1 nM) and highest (10000 nM) BPA treatment, respectively. These two concentrations did not have differentially expressed genes in common. No genes had significant changes at 100 nM BPA exposure. Similar with estrone, short-term exposure showed distinct transcriptomic responses. Only the 2 higher BPA exposure conditions had differentially expressed genes: 309 at 100 nM and 502 at 10000 nM with 21 genes differentially expressed at both concentrations. The expression levels of genes that were changed in more than one condition is shown in Appendix A: Fig. S2. BPA transcriptomic responses did not fully correspond to the abnormality rates and locomotor activity after extended duration exposure. The larval fish exposed to 1 nM BPA had the highest number of differentially expressed genes among all conditions but did not have abnormality or locomotive changes. At 100 and 10000 nM conditions, BPA exposure showed significant skeletal abnormalities and locomotor changes, but no genetic response occurred simultaneously.

3.3. The gene expression profiles of estrone and BPA were implicated with neurological and endocrine related pathways

We used IPA software to assess the canonical pathways and diseases implicated by the differentially expressed genes. We found that BPA exposures altered several canonical pathways (Appendix A: Table S2). Among the top six pathways, oxidative phosphorylation and Eukaryotic Initiation Factor 2 (EIF2) signaling were activated only at 1 nM BPA after extended duration exposure, whereas these pathways were increasingly induced as concentrations increased from 100 to 10000 nM BPA after short-term exposure. The rest of the pathways were affected only after short-term exposure and mostly at 100 nM BPA. The genes relevant to the canonical pathways are shown in Appendix A: Fig. S3 – S5. Diseases of organismal injury and abnormalities, neurological diseases, and endocrine/reproductive system diseases were among the top 5 that affected the exposed fish for both estrone and BPA exposure (Appendix A: Table S3). Both the extended duration estrone and BPA exposures at 1 nM played essential roles in transcriptomic responses as more diseases were predicted at these two conditions than at higher concentrations. In terms of short-term exposure, 100 and 10000 nM BPA were implicated with the most diseases and pathways. Thus, these 4 conditions were determined to be critical in terms of concentration, as well as timing and duration of exposure.

Table 1 shows the differentially expressed genes in these critical conditions. We selected these key genes because their expression levels were larger than 0.75 log2 fold change, significantly different from control (adjusted p value < 0.05), and implicated in more than 50 diseases. The implicated pathways are primarily related to the neurological or endocrine systems (Table 1). Among the 33 neurological-related genes, only 1 gene was differentially expressed in both estrone and BPA exposures, only 6 genes were differentially expressed after estrone exposure, and 28 genes were differentially expressed after BPA exposure. A majority of the neurological-related genes are involved with nervous system development, including apoea²⁹, enox1³⁰, tgif1³¹, tubg1³², crybb1³³, rack 1³⁴, ndufv2³⁵, fat2³⁵, id1³⁶, atp6ap2³⁷, tshz2³⁸, rho³⁹, dock1³⁹, and ubb⁴⁰. Genes involved with motor or cognitive behaviors are ddc⁴¹, wwc1⁴², tln2a⁴², fat2³⁵, and arhgef10⁴³. A few genes were implicated with malfunctions or illness, including neurological abnormalities (slc7a1⁴⁴), neuropsychiatric illness (dbi⁴⁵, idh2⁴⁵, plexna2⁴⁶, rap1aa⁴⁶), and neurological diseases (rps19⁴⁷, dync1h1⁴⁸). The rest of the genes are involved in neuronal differentiation (fbl⁴⁹), neuroregeneration (ttr⁵⁰), and neuroprotection (sigmar1⁵¹).

Table 1.

Fold changes in gene expression after estrone or Bisphenol A (BPA) exposure at 0 – 5 days postfertilization (dpf) or 4 – 5 dpf.

Gene Symbol	Gene Name	Estrone	BPA
		0 – 5 dpf	0 – 5 dpf	4 – 5 dpf
		1 nM	1 nM	100 nM	10000 nM
Neurological
ttr	transthyretin	−1.4	0.6	0.8	1.6
apoea	apolipoprotein Ea	1	0.5	0.1	0.9
sigmar1	sigma non-opioid intracellular receptor 1	0.8	−0.3	1.0	−0.03
fbl	fibrillarin	0.8	−0.2	−0.3	−0.6
enox1	ecto-NOX disulfide-thiol exchanger 1	−0.9	0.2	−0.3	−0.2
ddc	dopa decarboxylase	−1.3	0.5	0.3	0.6
tgif1	TGFB-induced factor homeobox 1	−0.3	0.9	0.3	0.3
tubg1	tubulin, gamma 1	−0.3	0.9	0.3	0.3
cat	catalase	−0.6	1.0	0.7	0.9
cdk4	cyclin-dependent kinase 4	0.4	0.9	0.1	0.4
ndufb6	NADH:ubiquinone oxidoreductase subunit B	0.2	0.9	1.0	1.1
rps13	ribosomal protein S13	−0.05	1.0	1.2	1.4
rpl35a	ribosomal protein L35a	−0.03	1.0	1.0	1.3
pparab	peroxisome proliferator-activated receptor α b (Orthologous to human PPARA)	0.02	−0.3	−1.3	−1.2
slc7a1a	solute carrier family 7 member 1a	0.1	−1.2	−0.4	−0.6
crybb1	crystallin, beta B1	−0.7	1	1.4	1.5
rack1	receptor for activated C kinase 1	0	0.9	1.2	1.4
ndufV2	NADH:ubiquinone oxidoreductase core subunit V2	0.1	1.2	1.3	1.3
fat2	FAT atypical cadherin 2	0.3	−1.1	−1.1	−1.1
ndufs4	NADH:ubiquinone oxidoreductase subunit S4	0.2	0.6	1.2	1.5
id1	inhibitor of DNA binding 1, HLH protein	0.1	0.6	1.0	1.1
atp6ap2	ATPase H+ transporting accessory protein 2	−0.1	0.7	0.9	1.1
tshz2	teashirt zinc finger homeobox 2	−0.4	−0.4	−1.5	−1.6
dbi	diazepam binding inhibitor (GABA receptor modulator, acyl-CoA binding protein)	0.3	0.3	−0.2	1.1
rho	rhodopsin	−0.6	1	0.9	1.3
arhgef10	Rho guanine nucleotide exchange factor (GEF) 10	0.01	−0.5	−1	−1.2
dock3	dedicator of cytokinesis 3	−0.4	−0.7	−1.2	−1.3
wwc1	WW and C2 domain containing 1	0.2	−0.4	−1.0	−1.3
tln2a	talin 2a	0.04	−0.1	−1.2	−1.3
rps19	ribosomal protein S19	0.2	0.4	0.8	1.2
dync1h1	dynein, cytoplasmic 1, heavy chain 1	−0.2	−0.5	−1.1	−1.4
ubb	ubiquitin B	0.05	−0.3	−1.3	−0.7
plxna2	plexin A2	−0.05	−0.5	−1.1	−0.9
Reproduction
fabp6	fatty acid binding protein 2, ileal (gastrotropin)	−1.2	0.9	1	1.7
gemin5	gem (nuclear organelle) associated protein 5	0.9	0.04	−0.2	−0.5
atf3	activating transcription factor 3	0.9	0.1	0.1	0.3
txn	thioredoxin	0.8	−0.6	−0.6	−0.3
txn2	thioredoxin 2	0.5	0.8	1.1	1.4
txnipa	thioredoxin interacting protein a	−0.7	−0.08	1.0	1.3
txnipb	thioredoxin interacting protein b	−0.5	−0.6	−0.8	−1.2
gstp1	glutathione S-transferase pi 1	0.2	−0.03	1.2	1.7
prdx1	peroxiredoxin 1	0.3	0.3	0.5	1.3
prdx2	peroxiredoxin 2	−0.3	0.9	1.0	1.0
prdx3	peroxiredoxin 3	0.3	0.8	0.5	0.9
prdx4	peroxiredoxin 4	0.04	0.3	0.8	0.8
gpx1a	glutathione peroxidase la	−0.4	0.8	0.8	1.2
gpx1b	glutathione peroxidase lb	0.2	0.8	0.9	1.1
Lipid metabolism
fads2	fatty acid desaturase 2	−2	1	1	1.8
elovl5	ELOVL fatty acid elongase 5	0.9	−0.2	−1	−0.4
elovl2	ELOVL fatty acid elongase 2	−1.1	−0.06	−0.1	0.3
lpin1	lipin 1	−0.9	0.3	0.3	0.4
apoa1a	apolipoprotein A-1a	−0.1	0.7	0.8	1.3
apoa2	apolipoprotein A-II	−0.1	0.5	0.8	1.5
apoa4b.1	apolipoprotein A-IV.b, tandom duplicate 1	0.1	0.1	1.3	1.8
apoc2	apolipoprotein C-II	0.01	0.9	0.7	0.9
abcg2a	ATP-binding cassette, sub-family G (White), member 2a	0.3	−0.2	−0.8	−1.1
dnmt1	DNA (cytosin-5-)-methyltransferase 1	0.3	−0.7	−0.9	−1.4

The genes involved in endocrine pathways can be classified as associated with reproduction, lipid metabolism, cholesterol, or glucose metabolism. Among the 31 genes of interest, 19% genes (6) were only differentially expressed after estrone exposure, 65% (20) after BPA exposure, and 16% (5) after either estrone or BPA exposure. The types of estrone- and BPA- expressed genes were distinct from each other. For example, the shared genes, including fabp6 (involved with ovulation)⁵², fads2 (lipid metabolism)⁵³, cyp7a1 (cholesterol homeostasis in liver)⁵⁴, and ndufs4 (neurodegeneration due to mitochondrial dysfunction)⁵⁵, were down-regulated after estrone exposure, but upregulated after BPA exposure. Another shared gene, elovl5 (lipid metabolism)⁵³, was upregulated after estrone exposure, but down-regulated after BPA exposure.

The other category we explored were genes that were only differentially expressed after BPA exposures and associated with lipid, cholesterol, and glucose metabolism. Regardless of the exposure concentrations, all genes except apoa were down-regulated after short-term BPA exposure. The expression levels of apoa (involved with insulin resistance)⁵⁶, osbp (adipocyte differentiation)⁵⁷, dhc24 (cholesterol biosynthesis)⁵⁸, and agpat5 (glucose tolerance)⁵⁹ increased when the exposure concentration increased from 100 to 10000 nM. The other gene, atp8b1 (cholestasis)⁶⁰ was only differentially expressed at 100 nM BPA. Lastly, the genes involved with reproduction or lipid metabolism were mostly differentially expressed after estrone exposure. They are involved in reproduction and associated with pregnancy (txn⁶¹), development of follicle (gemin5⁶²), urethral abnormalities, as well as sexual differentiation during embryogenesis (atf3⁶³). The genes associated with lipid metabolism are involved in polyunsaturated fatty acid synthesis (elovl2, evlo5, and fads2⁵³) and regulation of gluconeogenesis and lipid metabolism (pck1⁶⁴).

Based on the observed phenotypic skeletal abnormalities, we also identified 4 differentially expressed genes associated with skeletal system development, of which only tnfrsf1a, a gene involved in signaling of skeletal muscle wasting,⁶⁵ was upregulated by estrone after extended duration exposure at 1 nM. The remaining 3 genes (eif3f, utrn, and xylt1) were solely responsive to BPA after extended duration exposure at 1 nM or short-term exposures at 100 and 10000 nM. Eif3f was upregulated by both long- and short-term BPA exposures. Overexpression of eif3f is associated with muscle hypertrophy,⁶⁶ while downregulation of utrn and xylt1 impacts skeletal cell differentiation⁶⁷ and growth of skeletal elements. ⁶⁸

4. Discussion

Our study shows that BPA exposure leads to distinctive and more significant effects than estrone exposure in larval zebrafish. Overall, BPA affected more than twice the number of differentially expressed genes than estrone when accounting for both long- and short-term transcriptomic responses. Among the BPA-specific responsive genes, the majority were implicated with neurological pathways. The extended duration BPA exposure induced more significant skeletal abnormalities than estrone exposures, as well as hyperactive locomotion, in contrast to the hypoactive behavior induced by estrone. BPA has been viewed as a less potent estrogenic compound because of relatively lower binding affinity to estrogen receptors compared to 17β-estradiol.⁶⁹ However, a higher proportion of free BPA is available in serum to target estrogen receptors and promote estrogen receptor-mediated effects.⁷⁰ In addition, BPA can bind to a variety of steroid and non-steroid receptors, such as androgen, thyroxine, and pregnane X receptors,⁷¹ and thus it induces more diverse genes and pathways than common estrogens like 17β-estradiol and 17α-ethinylestradiol.

Both estrone and BPA had distinctive phenotypic and transcriptomics responses between long- and short-term exposures. In terms of phenotypic responses, both chemicals had more abnormalities after extended duration than short-term exposure. BPA extended duration exposures at 100 and 10000 nM led to hyperactive behavior. The hyperactivity at 100 nM BPA may be induced by abnormal neurogenesis⁷² and can occur as early as the first 48 hour of embryo development²⁷. In our study, 10000 nM BPA also led to skeletal abnormalities, which could be associated with the suppression of neuron branching from the spinal cord and abnormal development of neuromast cells²¹. Our study did not observe locomotive responses and abnormalities at 1 nM BPA. Nevertheless, previous studies found that a similar concentration of BPA can cause hyperactivity after 0–5 dpf exposure^73,74 and abnormal heart morphology after 48-hour exposure at early embryogenesis⁷³. In terms of the number of differentially expressed genes, we found that extended duration estrone exposure had 362 more responsive genes than short-term exposure. Only 5 genes were shared between the two estrone exposure periods. In contrast, the transcriptomic response of BPA short-term exposure induced 1128 more differentially expressed genes than extended duration exposure, and 91 genes were shared between the two BPA exposure periods. With the highest number of differentially expressed genes, extended duration estrone and short-term BPA exposures were also implicated in the greatest number of pathways and diseases. Short-term BPA exposures were specifically associated with more pathways related to neurological and endocrine system diseases compared to the extended duration estrone exposures. The difference in the number of differentially expressed genes between long- and short-term exposures corresponded with Hao et al’s study that estrogenic genes can be regulated in a stage-dependent manner.⁷⁵ In addition, negative feedback loops may occur as the exposure duration is increased, leading to opposite responses between short- and extended duration exposures.⁷⁶

The discrepancy of phenotypic and transcriptomic responses was observed after extended duration BPA exposure. BPA-associated phenotypic responses after extended duration 100 and 10000 nM exposures did not have significant corresponding transcriptomic responses. Similarly, at the range of 0.01 nM to 10 uM BPA levels, Wang et al. found the transcriptome potencies were not fully correlated with the embryo phenotypic toxicities.⁷⁷ The phenotypic responses at these two concentrations may stem from non-genomic signaling pathways that lead to physiological changes, including neurologic disorders.^78,79 Larval fish exposed to extended duration 1 nM BPA had the highest number of differentially expressed genes, but no associated phenotypic or locomotive changes. Omics responses are usually more sensitive than phenotypic responses because they show potential adverse or non-adverse physiological sequelae at the molecular level.⁸⁰ It may be possible that these transcriptomic responses could lead to changes in later developmental stages or adulthood.

Not only did BPA affect expression of different genes than estrone, but even differentially expressed genes shared between BPA and estrone had opposite expression. We focused on the 90 shared genes that were differentially expressed after extended duration estrone exposure at 1 nM and short-term BPA exposures at 100 and 10000 nM (Appendix A: Fig. S6). Among these shared genes, 74% of the estrone-responsive genes were downregulated, while 90% of BPA-responsive genes were upregulated. The shared genes include lipid/cholesterol metabolism-related genes (e.g. cyp7a1 and fads2), reproduction-related genes (e.g. fabp6, txnipa, txnipb), and a neurological-related gene, ttr. Previous studies show that while cyp7a1, fabp6, and the human txnip ortholog can be downregulated by either 17α-ethinylestradiol or 17β-estradiol,^54,61,75 fads2 and ttr are upregulated.^81,82 In our study, BPA regulation of fads2, ttr, and the other txnip homologue gene, txnipb, induced the opposite expression direction as estrone, but was the same as estrogens.⁸² It is unclear why BPA, estrone, and common estrogens impose differential regulation of these shared genes. In aquatic environments, BPA and estrone may present as a mixture with other EDCs. Compared to single compound exposure, BPA and estrone mixtures may exert antagonist effects over these shared genes, causing completely different health outcomes in exposed humans or wildlife. We note that fads2 was implicated in more than 50 diseases and differentially expressed after all 4 critical exposure conditions, and thus may be a potential gene marker for lipid-related EDC-induced abnormalities. PPARα and sterol regulatory element binding protein can upregulate fads2,⁸³ but the specific mechanisms influencing fads2 expression needs more investigation.

Both estrone and BPA exposures showed non-monotonic transcriptomic responses after extended duration exposure and peaked at 1 nM. 77% and 99% of estrone- and BPA-responsive genes, respectively, were differentially expressed solely at 1 nM after extended duration exposure. Similar non-monotonic response at low nanomolar BPA levels were found in zebrafish larvae^74,84,85, ovary⁸⁶ and brain.⁷² Among the estrone-responsive genes, some affected embryogenesis, including neurological genes (e.g. sigmar1, fbl, enox1, ddc, ttr), a reproductive-related gene, atf3, and a lipogenesis gene, lpin1. Lpin1 was previously found to be non-monotonically induced by millimolar levels of BPA,⁹¹ but we did not observe the same result after BPA nanomolar exposures. The induction of non-monotonic responses at low doses is a common feature of hormone treatment and EDC exposure because physiologically relevant effects are innately sensitive to low levels of endogenous hormones. The non-monotonic responses can also be due to negative feedback loops caused by long-term administration/exposure or inactivation of receptors after saturation with ligands. In addition, some genes are only sensitive to low doses of estrogens and can be suppressed by higher exposures.⁹² Receptor selectivity for EDCs at different concentrations can also explain non-monotonic responses. For example, BPA mainly targets nuclear or membrane estrogen receptors at low concentrations, but binds to other receptors such as androgen⁹³ and thyroid receptors⁹⁴ at micromolar levels.

Several canonical pathways, notably oxidative phosphorylation, NF-κB signaling, and EIF2 signaling, were altered specifically following BPA exposures. Oxidative phosphorylation and EIF2 signaling were altered only at 1 nM after extended duration exposure, but increasingly affected at higher concentrations after short-term exposure. BPA affected the regulation of these pathways non-monotonically after extended duration exposure and monotonically after short-term exposure at embryogenesis. Several studies that focused on mice at postnatal phase found that micromolar to nanomolar levels of BPA cause mitochondrial dysfunction and decrease oxidative phosphorylation in digestive and neurological systems, thus increasing oxidative stress in the cells and inducing NF-κB signaling.^95–97 In contrast, we found that nanomolar BPA exposures increased activation of oxidative phosphorylation, while several genes in the NADH dehydrogenase complex (Ndufa and Ndufb), ubiquinol-cytochrome c reductases (Uqcr), and cytochrome c oxidase subunits (Coxc) were upregulated at 1 nM extended duration, as well as 100 and 10000 nM short-term exposures. Associated with the enrichment of oxidative phosphorylation, NF-κB signaling was inactivated after BPA exposures at 100 and 10000 nM. Lee et al. found that NF-κB activity of neurological cells in immature mice can be inhibited, but by much higher levels of BPA (20–50 μM).⁹⁸ We also observed activation of EIF2 signaling after BPA exposures, which occurs in response to environmental stress.⁹⁹ While all three pathways are related to neurotoxic effects, further studies should focus on how nanomolar BPA alters oxidative phosphorylation at the specific cell or tissue level in the nervous system at different developmental stages.

As seen in the current study, hypoactive behaviors observed after short-term BPA exposures can be linked to the transcriptomic changes based on pathway analysis results and previous literature. These genes include atp6ap2, dync1h1, wwc1, tln2a, ubb, id1, dock3, and plxna2. The upregulation of atp6ap2 and downregulation of dync1h1 are both implicated in increased neurological signs, movement disorders, motor dysfunction, seizure disorder, and cognitive disorders.^37,48 WWC1 and tln2 were found to be co-expressed in the brain tissues of Alzheimer’s patients, so these 2 genes may be involved in the same signaling pathways.⁴² The downregulation of wwc1 has been implicated with synaptic depression and neurological signs. The upregulation of id1 and downregulation of dock3, plxna2, pparab, and ubb are involved with neurological cell development.^39,40,46,100 Dock3 was implicated with decreased proliferation of neuronal cells and outgrowth of neurites based on IPA. Downregulation of plxna2 can impair granule cell distribution in developing hippocampal cells and disrupt neurogenesis.⁴⁶ Inhibiting the function of the orthologues of ppara, pparaa and pparab reduces proliferation of neuronal and glial precursor cells,¹⁰⁰ and pparab was downregulated in our study. Ubb has been implicated with decreased neurite development. In fact, disruption of Ubb causes dysregulation of neural stem cell differentiation and impairs neuronal maturation in mouse embryonic brain.⁴⁰ While expression levels of the short-term BPA-responsive genes were primarily increased as exposure concentration increased (Appendix A: Fig. S2 and S3), Ubb was the only gene that exhibited a non-monotonic response at 100 nM.

Some estrone-responsive genes at 1 nM were related to neurological development during embryogenesis, yet their relationship with locomotive behavior is less clear than for BPA. Downregulation of ddc may be linked to hypoactive behavior as it is responsible for the synthesis of several neuroactive biogenic amines in the brain after 3 dpf in zebrafish. Reduced ddc expression impairs brain morphology and swimming activity in larval zerbafish,¹⁰¹ and motor and cognitive functions in humans⁴¹. The other 3 genes that were induced by 1 nM estrone were fbl, sigmar1, and apoea. Knock-out of fbl impairs vision and decreases brain volume of zebrafish.⁴⁹ Sigmar1 and apoea are related to the motor neuron¹⁰² and vision¹⁰³ functions in zebrafish, respectively. In addition, apoe and sigmar1 are identified as interactive risk factors of Alzheimer’s disease¹⁰⁴, as researchers found those who inherited both genes confer higher risk for Alzheimer’s disease as compared to those who carry only 1 of these genes.¹⁰⁵ However, we are uncertain about the relationship of hypoactive behaviors and upregulation of these genes after estrone exposure. Interestingly, sigmar1 was highly expressed in the brain of Alzheimer’s mouse models, but significantly decreased in Alzheimer’s patients.¹⁰⁶ As activation of sigmar1 is known to suppress the progression of Alzheimer’s disease pathogenicity in human⁵¹, the researchers hypothesized that sigmar1 may serve as different roles in animal models than human.¹⁰⁶ Further investigation is needed to determine the relationship of sigmar1 and neurological responses following estrone exposure in zebrafish. On the other hand, apoe can be induced by 17β-estradiol.²⁹ Given that estrone is a relatively weak estrogen, we found estrone induced apoea non-monotonically at 1 nM after extended duration exposure. Therefore, it is important to consider low dose and non-monotonic conditions while studying the neurological effects associated with apoea expression after estrone exposure.

We compared our gene profiles to the common biomarkers that are used to assess the potency of EDCs. Hao et al. suggested that vtg1, vtg3, esr1, f13a1a, epd, and cpn1 are potential gene markers to evaluate estrogenic potencies because they can be induced by 17β-estradiol at both early larval development stage and adulthood.⁷⁵ Jarque et al. proposed eight gene markers to evaluate if EDCs disrupted estrogen (cyp19a1b and vtg1), androgen (sult2st3 and cyp2k22), or thyroid (tpo, ttr, tra, and dio2) pathways in zebrafish.¹⁰⁷ Our study found that two of the proposed gene markers were differentially expressed for some of the BPA and estrone exposures (Appendix A: Table S4). The estrogenic marker, cyp19a1b, was differentially expressed after extended duration exposure at 100 and 10000 nM estrone, as well as 10000 nM BPA. Cyp19a1b is a common gene marker for estrogenic activity because of its essential role in the conversion of androgens to estrogen in neurological and reproductive systems. However, low nanomolar estrone or BPA did not induce upregulation of cyp19a1b expression. We detected that the thyroid gene marker, ttr, was upregulated by BPA at 100 and 10000 nM after short-term exposure, but downregulated by estrone at 1 nM after extended duration exposure. As mentioned earlier, ttr can be upregulated by 17β-estradiol via estrogen receptor pathways.⁸² Since BPA also upregulated ttr, the mechanistic pathway could be the same as 17β-estradiol. The sensitivity of gene markers is also related to their activity at specific cells and tissue at different developmental stages.¹⁰⁷ It is not too surprising that none of the proposed gene markers were sensitive enough to respond to 1 nM of either estrone or BPA. As previously mentioned, BPA and estrone affect completely different gene profiles throughout the developmental stage. In fact, only one gene, fads2, was differentially expressed for all critical conditions, and EDC-implicated pathways are likely triggered via distinct transcriptomic responses. Thus, the neurotoxicity of EDCs as a group may not be easily predicted by certain sets of gene markers.

In conclusion, our findings show that outcomes related to BPA exposure are more significant than for estrone exposure. During embryonic development, BPA alters gene expression at 1 nM non-monotonically. At larval developmental stage in zebrafish, BPA at concentrations equal or above 100 nM can alter locomotive behavior and gene expression patterns that are linked to neurological diseases. The neurotoxic effects that are induced by BPA are acutely observable and involve a variety of signaling pathways that include both estrogenic and non-estrogenic genes. Estrone is relatively less potent than other estrogens, as well as BPA based on the current study. Very few studies have focused on estrone toxicity. Our study is the first to show that estrone can induce a non-monotonic transcriptomic response and alter locomotive behavior after exposure during embryogenesis in zebrafish. We found that 1 nM is the lowest level for estrone to induce the most phenotypic and genetic responses, which is the same order of magnitude as the conclusions of Caldwell et al, that the predicted-no-effect concentration of estrone is 6 ng/L.¹³ More studies should take the non-monotonic response into account to evaluate if BPA induces more adverse effects at concentrations lower than 1 nM.

Highlights.

• Estrone and BPA exert phenotypic or transcriptomic responses as low as 1 nM.

• Estrone or BPA 0–5 dpf exposure causes non-monotonic transcriptomic changes.

• BPA 4–5 dpf exposure causes hypoactivity and expression of neurological genes.