Creating Females? Developmental Effects of 17α-Ethynylestradiol on the Mangrove Rivulus’ Ovotestis

Abstract

Interest in the occurrence and fate of trace organic contaminants in the aquatic environment and their potential effects on all organisms has increased over the past two decades. Researches on contaminants have included both natural and synthetic estrogenic contaminants, neuroactive pharmaceuticals, and other endocrine disrupting chemicals that are mediated by the androgen and progesterone receptors. Exposure to very low concentrations (ng/L or parts per trillion) of compounds such as 17α-ethynylestradiol (EE₂), a synthetic estrogen, can affect gonadal development, viability and production of eggs, fertilization rate, and sexual differentiation in fishes. Researchers and aquaculturists have used exposures to relatively higher concentrations of androgens and estrogens, for example 17α-methyltestosterone and EE₂, respectively, to direct sexual differentiation in a number of fishes. Rivulus is an androdioecious teleost that in nature exists mostly as selfing, simultaneous hermaphrodites as well as a small number of males that outcross with hermaphrodites. No one has either collected females in the wild or created functional females in the laboratory. This study had two goals: (1) to develop a reliable protocol to produce female rivulus to enable downstream technologies such as embryo injections and (2) to investigate developmental effects of EE₂ on the sexual outcome, reproductive health, and relevant gene expression in rivulus. With these goals in mind, we exposed newly hatched rivulus to nominal concentrations of 0.1, 0.5, or 1.0 parts per million (ppm) EE₂ for 4 weeks, grew them to maturity in control water, and then compared egg production; production and viability of embryos; age of reproductive maturity; and gene expression in the brain, gonad, and liver. Expression levels of seven genes with known relevance to gonadal development and function (cyp19a1b, cyp19a1a, dmrt1, figα, ERα, ERβ, and vtg) were measured using quantitative polymerase chain reaction (PCR). There was a significant decrease in cyp19a1a gene expression in the brain, corresponding to increased exposure to EE₂. Gonadal gene expression for cyp19a1a, ERα, and dmrt1 also decreased in response to EE₂. Vtg expression in the liver was unaffected. Our hypothesis that exposure to EE₂ during gonadal differentiation would direct female development was not supported by the data. However, treated fish exhibited impaired reproductive health that included reduced expression of relevant genes and, importantly, decreased fertility, increased sterility, and delay of age of reproductive maturity. The results of this study suggest that the development and maintenance of a simultaneous hermphrodite ovotestis may be particularly sensitive to its hormonal milieu.

Introduction

Interest in the occurrence and fate of trace organic contaminants in the aquatic environment and their subsequent effects on aquatic organisms has steadily increased over the past few decades. Of particular concern are endocrine disrupting chemicals (EDCs), which have the potential to disrupt internal homeostasis and hormone-regulated physiological processes such as development and reproduction. It is known that fish exposed to EDCs exhibit abnormalities in sexual development and the reversibility of those effects can vary with EDC concentration and with duration of exposure (Schãfers et al. 2007); however, few studies have addressed whether exposure in early life to EDCs, particularly estrogenic compounds, can permanently alter long-term reproductive function in fish (Maunder et al. 2007; Coe et al. 2010). The long-term effect of exposure during early developmental to 17α-ethynylestradiol (EE₂), a potent synthetic estrogen found in human birth control pills, is of particular interest.

The mangrove rivulus, Kryptelobias marmaratus, is a functional simultaneous hermaphrodite that reproduces by internal self-fertilization (Harrington 1961; Tatarenkov et al. 2009). In the wild, they exist as an androdioecious species, with mostly hermaphrodites and some males. While males are rare, functional female rivulus have never been observed in the wild or in the laboratory. Rivulus can be classified by the isogenic strain from which they are descended, but genetic differences among strains support the occurrence of natural outcrossing (Mackiewicz et al. 2006). Outcrossing is hypothesized to occur when a male fertilizes an egg from a hermaphrodite, which escaped internal fertilization, or from hermaphrodites that can shut down spermiation and ovulate only under certain environmental conditions. Male rivulus are rare in nature, but can be produced in the laboratory either through exposure to low temperatures prior to hatching (which is believed to cause development into males in the wild) or by exposure to an exogenous androgen: for example, 17α-methyltestosterone (Kanamori et al. 2006). Hermaphroditic rivulus develop a bilobed gonad termed an ovotestis, which contains functional ovarian and testicular tissue (Harrington 1975). During gonadogenesis, rivulus possess histologically undifferentiated gonadal tissue from 0 days posthatching (dph) to approximately 28 dph when ovarian, then testicular tissue at 68 dph, are discernible; by approximately 105 dph, the mature rivulus is capable of internal self-fertilization (Harrington 1975; Soto et al. 1992; Cole and Noakes 1997). Research has narrowed the developmental window in which there is a sensitivity to hormones that affect gonadal differentiation and expression of relevant genes, to between late embryogenesis and 8 dph for 17α-methyltestosterone (Kanamori et al. 2006) and between the peri-hatching period and 28 dph for EE₂ (unpublished report, Schaughency 2005).

Estrogens, as well as estrogenic EDCs such as EE₂, function through two different estrogen receptors (ERα and ERβ). A third estrogen receptor (ERγ) has been found in some fishes; however, the existence of ERγ is unknown in rivulus (Lee et al. 2006; Orlando et al. 2006). Estrogen promotes an upregulation of its own receptor and, once activated, ERs can function as regulators of gene expression (Menuet et al. 2002). Activated ERs can function directly through estrogen response elements (EREs) present in gene-promoter regions, or they can act indirectly to promote changes in gene expression through various pathways (Marchand et al. 2000; von Hofsten et al. 2002; Matthews and Gustafsson 2003; Kitano et al. 2007). In vertebrates, estrogens are synthesized from androgens through the catalytic action of cytochrome P450 aromatase, encoded by the cyp19 gene (Simpson et al. 1994). In fish, two distinct cyp19 gene loci, cyp19a1a (ovarian type) and cyp19a1b (brain type), encode structurally and functionally different aromatase isoforms (Kwon et al. 2001; Liu et al. 2007). Cyp19a1a functions in ovarian development for several teleost species, whereas cyp19a1b is involved with neuronal tissue development (Pellegrini et al. 2005) and may play a role in sexual differentiation of zebrafish (Danio rerio) and other species (Chang et al. 1997; Guiguen et al. 1999; Kitano et al. 1999; Kwon et al. 2001; Trant et al. 2001; Sawyer et al. 2006; Wang et al. 2007; Yamaguchi et al. 2007). Sexual differentiation of fish is particularly sensitive to the ratio of androgens to estrogens; thus, this ratio is critical anddependent on the availability of cyp19 enzyme. EREs have been identified in the cyp19a1b promoter, while ERE-halves were found in both cyp19a1a and cyp19a1b promoters in medaka (Oryzias latipes), goldfish (Carassius auratus), zebrafish, rainbow trout (Oncorhynchus mykiss), and gilthead seabream (Sparus aurata) (Callard et al. 2001; Kazeto et al. 2001; Tchoudakova et al. 2001; Tong and Chung 2003). The exact function of ERE-halves is not fully understood; however, the presence of an ERE indicates the potential regulation by estrogen and their activated estrogen receptors.

Vitellogenin (vtg) is a female-specific glycoprotein produced in the liver as a precursor of yolk during the vitellogenesis stage of oogenesis. Vtg synthesis is controlled by 17β-estradiol released from the gonads (Wahli et al. 1981). Several studies have linked exposure of male fish to low levels of EE₂ to the intersex phenotype in which oocyte development, accompanied by increased vtg expression, has been found in the testicular tissue (Van den Belt et al. 2002). However, the long-term effect of exposure to EE₂ during early development on vtg expression is not known in rivulus. For other species, such as zebrafish, the increase in vtg expression due to exposure to EE₂ is reversible in fish removed from treatment for at least 24 days (Van den Belt et al. 2002).

Dmrt1 (doublesex and mab-3-related transcription factor 1), which is found in both males and females but with expression significantly higher in males, is highly conserved and is expressed in the gonads of several species of teleosts (Guan et al. 2000; Kobayashi et al. 2003; Guo et al. 2005). Dmrt1 has also been shown to repress transcription of cyp19a1a, in Nile tilapia (Oreochromis niloticus) (Wang et al. 2010). Investigations of the effect of exposure in early life to EE₂ in zebrafish, a species with no known sex chromosomes, have shown a reduction in dmrt1 gene expression which also has been shown to be associated with significantly reduced gonadal development in males (Schulz et al. 2007). In addition, sex reversal brought on by treatment with estrogenic compounds has been shown to downregulate dmrt1 gene expression (Guan et al. 2000; Kobayashi et al. 2003).

Figα (gene factor in the germline, α), a female-specific gene marker, has also been implicated as playing important roles in early sex determination (Scholz et al. 2003). Like many other genes in the sex determination and differentiation pathways, figα appears to be conserved as it encodes a protein similar to that produced by figα in mice (Kanamori et al. 2008). This gene is also expressed in many vertebrate species, including mice, human, zebrafish, and medaka (Liang et al. 1997; Kanamori 2000; Huntriss et al. 2002; Onichtchouk et al. 2003). Figα is an oocyte-specific gene marker, making it a useful tool for determining the development of ovarian tissue.

Sex steroids are known to affect sexual differentiation in fish and that the balance between estrogens (17β-estradiol) and androgens (testosterone and 11-ketotestosterone) is of great importance. Rougeot et al. (2007) observed that for several teleost species a shift in the hormonal balance in favor of 17β-estradiol resulted in feminization, whereas a shift toward 11-ketotestosterone resulted in masculinization. In zebrafish, exposure to EE₂ can affect gonadal development, viability and production of eggs, fertilization rate, and sexual differentiation (Nash et al. 2004).

To further our elucidation of genes that regulate gonadogenesis and their product pathway in rivulus, it would be advantageous to control the exact time of fertilization so as to use injections to alter gene expression or to create transgenic rivulus. Given the variability of the timing of release of embryos from the parent rivulus hermaphrodite, currently this is not possible. We already know we can direct the development of male rivulus via exogenous 17α-methyltestosterone (Kanamori et al. 2006), and data from a student’s project in our laboratory suggest that we can direct females’ development similarly, but with EE₂ (unpublished report, Schaughency 2005).

Thus, this study had two goals: (1) to develop a reliable protocol for producing female rivulus and (2) to investigate the potential effects of EE₂ during early development on the sexual development and relevant gene expression in adult rivulus.

Materials and methods

Fish

Rivulus embryos were collected from a single isogenic strain (03-RhlC, (Tatarenkov et al. 2010) and incubated in groups of 20 embryos per 5″-diameter bowl at 24 ± 1°C, 16 h light:8 h darkness photoperiod, in approximately 200 ml of 15 parts per thousand (ppt) seawater (reverse osmosis water plus Instant Ocean sea salts; Aquarium Systems, Inc., Menton, OH) and allowed to hatch within approximately 15 days. Newly hatched fry were collected and randomly placed in different treatment groups: 15 fry per treatment, 15 ppt seawater treatment negative control, 1.0 parts per billion (ppb) EtOH solvent control, and nominal concentrations of 0.1, 0.5, or 1.0 ppm EE₂ (Sigma-Aldrich, St. Louis, MO) in 15 ppt seawater at one fry per 2.5″-diameter bowl, containing 20 ml of treatment water and maintained under the conditions stated above. Fry remained in treatment for 28 days. A 50% water change, containing the appropriate amount of ethanol or EE₂, was carried out twice a week for each bowl and treatment. Fry were fed freshly hatched brine shrimp (Artemia salina) daily ad libitum. After 28 days, fry were removed from treatment and placed in 5″-diameter bowls with approximately 200 ml of 15 ppt seawater until they reached sexual maturity at 105 dph under the conditions stated above. Fish were maintained for an additional 3 months to monitor the production of embryos. Fish were anesthetized with 150 ppm of buffered MS-222 (Western Chemical Inc., Ferndale, WA), and then euthanized by cervical dislocation. The brain (including pituitary), gonad, and liver were collected and stored in RNAlater (Ambion, Foster City, CA) at −20°C until gene-expression assays via quantitative, real-time PCR (QPCR, see below). This research was carried out under the guidance of the Institutional Animal Care and Use Committee (protocol R-09-56), University of Maryland, College Park.

Analysis of reproductive health

After the 28-day treatment period, all rivulus were placed individually in 5″-diameter bowls with a Teflon mesh resting approximately 15 mm off the bottom of the bowl. All other conditions remained the same as stated above. The mesh was large enough to allow fertilized embryos and presumed unfertilized eggs to fall through to the bottom of the bowl, but it was too small for the mature rivulus to pass through. This was done to prevent infanticide. Embryo production was monitored twice a week. At the time of each observation, all embryos/eggs were removed and placed in a 2.5″-diameter bowl with 15 ppt seawater for analysis under a microscope. Fertilized embryos, as well as unfertilized eggs, were counted and documented. Fertility rate and posthatching age at the start of embryo/egg production were also documented.

RNA isolation

Total RNA was isolated from tissues preserved in RNAlater using Trizol reagent (Gibco-BRL, Bethesda, MD) according to the manufacturer’s instructions. Total RNA samples were digested with RNase-free DNase I and purified using the RNeasy Mini Kit (Qiagen, Valencia, CA). The quantity and the quality of the RNA were determined by spectrophotometry and by denaturing agarose gel electrophoresis, respectively.

Cloning of genes

Partial complementary DNA (cDNAs) for the experimental rivulus genes ERα, ERβ, cyp19a1a, cyp19a1b, figα, dmrt1, and vtg were amplified by reverse transcriptase–PCR using primers designed from regions of high homology across teleost species (Table 1). Similarly, the ribosomal protein L8, rpl8, was used as a reference housekeeping gene (Table 1). As a template for PCR, first-strand cDNA was synthesized from the same total quantity of RNA for each tissue type (brain, 0.5 µg/µL; liver 1 µg/µL; and gonad, 0.8 µg/µL). cDNA was synthesized using the SuperScript^TM III First-Strand Synthesis System (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. For each tissue type, total RNA concentration was normalized across treatments so that the same amount of RNA went into each cDNA reaction. PCR products of appropriate size were cloned into the pPrime vector using the 5PRIME Perfect PCR Cloning Kit (Fisher Scientific). Sequences for all PCR products were verified using a BigDye terminator Cycle Sequencing kit (PE Biosystems, Foster City, CA) with the M13(-20) primer and analyzed on the Applied Biosystems 3730 Genetic Analyzer. Sequences were compared to known sequences using NCBI BLAST.

Table 1.

Gene	Sequence (5' → 3'): forward and reverse	Annealing Temperature (°C)	Primer Concentration (nM)	Product Size (bp)	GenBank Accession Number
ERα	TGTACTATTGACAGGAATCG	57.5	250	75	AB251458.1
	TTACGAAGTGGGTATGATG
ERβ	ACATATTGGAGTTGAGGA	57.5	250	106	AB251457.1
	TATCAGCATGTCGAAGAT
cypl9a	TGAAGCCGTCGATGATGTCA	60.6	350	102	AB251460.1
	TGAAGCTCTCCAGAACCTGCA
cypl9b	GTGGTTGACTTCACGATGCGT	54.4	350	101	AB251459.1
	CAACCTTGGCCTCATGCA
Figα	CCACAGAGGACAGCGATA	58.5	350	78	DQ683743.1
	GGTATCAGTTCATCATTCAAGTTG
dmrt1	CAACTTCTACCAGCCGTCACGCTAC	59	250	133	DQ683742.1
	GTGTCCTCTCAGTACCGAATGCATTCG
VTG	CCTGAGCAAATCGCAACT	59.1	250	97	AY279214.1
	CCCATAAAGTCTACGATCTGC
rpl8	TGACAAGCCCATCCTGAAGGC	59	350	102	NM_001104909.1
	GGCTATGAATCCTGTTGAGCA

Real-time QPCR

Gene expression was quantified using QPCR performed on the CFX96™ Real-Time System (BioRad, Hercules, CA). A standard curve with six points was made from a 5-fold serial dilution of plasmid DNA containing the amplicon of interest. To run QPCR, 2 µL of standards or samples were combined with a 48-µL mix of SYBR Green PCR Master Mix (BioRad), appropriate verified forward and reverse primers (Table 1), and water. Standards and samples were run in triplicate on a MicroAmp 96-well reaction plate (Applied Biosystems, Foster City, CA). QPCR cycles were as follows: 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 1 min at the optimal temperature (Table 1). The primers were designed using BioRad Beacon Designer (v7.8) and were synthesized by Integrated DNA Technologies (http://www.idtdna.com). Melt curve analysis was conducted to confirm that a single product was produced. The resulting amplicon was run on a gel to confirm size and was sequenced and BLASTED to confirm accuracy of the product.

Statistical analyses

For all tests in which the H₂O and EtOH solvent controls were not statistically different, data were combined into a single control group for analysis. Gene-expression data are expressed as mean ± 1 SEM. Differences in gene expression between controls and groups exposed to EE₂ were analyzed by one-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) Multiple Comparisons test. Data were checked for normality and homogeneity of variance, and when necessary were transformed. Fertility-rate percentages were arcsin transformed before analysis by one-way ANOVA, followed by Tukey’s HSD Multiple Comparisons test. Chi-square and Fisher’s exact test were used to analyze the age at which adult rivulus became reproductively active, as well as to assess the number of fish in each treatment that were presumed to be sterile. Statistical analyses used SYSTAT (Systat Software, v. 11, Chicago, IL) and significance was defined as P < 0.05.

Results

Gene expression in the brain

The effect of EE₂ during early development on expression of cyp19a1a and cyp19a1b messenger RNA (mRNA) in the adult rivulus brain plus pituitary was measured using QPCR. As shown in Fig. 1A, cyp19a1a was significantly reduced by approximately 20% in all EE₂ treatments (P < 0.05). Cyp19a1b was not affected by treatment (Fig. 1B). Basal gene expression levels, as determined by the expression level seen in the control group, for cyp19a1b were approximately 130-fold higher than in the cyp19a1a control group. The reference housekeeping gene, rpl8, was unaffected by EE₂ at all concentrations tested (Fig. 1C).

Fig. 1.

Effect of EE₂ (0.1, 0.5, and 1.0 ppm) on the expression of cyp19a1a, P = 0.005 (A); cyp19a1b, P = 0.136 (B); and rpl8, P = 0.283 (C) in the brains of adult hermaphroditic rivulus. Data are mean cDNA copy number ± SE of three technical replicates per fish with biological replicates of five fish per group. Significance is shown as groups with different letters.

Gene expression in the liver

The effect of exposure to EE₂ during early development on expression of vtg mRNA in the adult rivulus liver was measured using QPCR. There was no significant effect of EE₂ on vtg gene expression across any of the treatments tested (data not shown).

Gene expression in the gonad

The effect of exposure to EE₂ during early development on expression of cyp19a1a, cyp19a1b, ERα, ERβ, dmrt1, and figα mRNA in the gonad of adult rivulus was measured using QPCR. Expression levels in the gonad for cyp19a1b, ERβ, and figα were not affected by EE₂ during early development for any of the concentrations tested (Fig. 2B, D, and F). Gene expression for cyp19a1a significantly decreased with EtOH treatment and all EE₂ treatments (Fig. 2A). Gonadal expression of cyp19a1a was the only gene to be effected by the solvent control and, interestingly, this effect was not seen in cyp19a1a gene expression in the brain. At 1.0 ppm EE₂, gene expression for ERα was significantly reduced in relation to the control group, but 0.1 and 0.5 ppm of EE₂ had no effect (Fig. 2C). Dmrt1 was significantly reduced at exposures of 0.5 and 1.0 ppm EE₂, but the 0.1 ppm treatment did not have an effect (Fig. 2E). Basal gene expression levels, as determined by the expression in the H₂O group, for cyp19a1a were approximately 11-fold higher than at the levels seen in the cyp19a1b control group. Also, gene expression for ERβ was determined to be approximately 1.5-fold above ERα expression, as determined by expression levels in the control groups. The reference housekeeping gene, rpl8, was unaffected by EE₂ at all concentrations tested (Fig. 2G).

Fig. 2.

Effect of EE₂ (0.1, 0.5, and 1.0 ppm) on the expression of cyp19a1a, P < 0.001 (A); cyp19a1b, P = 0.591 (B); ERα, P = 0.008 (C); ERβ, P = 0.551 (D); dmrt1, P ≤ 0.001 (E); figα, P = 0.903 (F); and rpl8, P = 0.332 (G) in the gonads of adult hermaphroditic rivulus. Data are mean cDNA copy number ± SE of three technical replicates per fish with biological replicates of five fish per group. Significance is shown as groups with different letters.

Effect of EE₂ on sterility and fertility

Sterility rate was defined as the number of fish per treatment that were not able to produce a viable fertilized embryo, nonviable embryo, or a presumably unfertilized egg within 3 months of reaching adulthood. In this study 100% of the fish in both the H₂O and EtOH control groups were reproductively active within the time frame specified. In the 0.1 ppm EE₂ treatment group, 93% of the fish were reproductively active within 3 months of reaching adulthood, giving them a 7% sterility rate. This was not significantly different from the controls. However, both the 0.5 and the 1.0 ppm EE₂ groups showed a significant increase in the number of sterile fish per treatment compared to controls and to fish exposed to 0.1 ppm EE₂. The 0.5 and 1.0 ppm EE₂ treatment groups both had sterility rates of approximately 40% (Fig. 3A).

Fig. 3.

Effect of EE₂ (0.1, 0.5, and 1.0 ppm) on the sterility and fertility rates for adult hermaphroditic rivulus. (A) For each treatment group sterility rate was determined by dividing the number of fish per treatment that produced either an unfertilized egg or a nonviable embryo, by the total number of fish per treatment. Groups treated with 0.5 and 1.0 ppm EE₂ had a significant increase in the number of sterile fish, P < 0.001. (B) For each treatment group, fertility rate was determined by dividing the number of viable embryos by the total propagules produced, P < 0.001. Both for sterility and fertility, significance is shown as groups with different letters.

The fertility rate was defined as the number of viable and nonviable fertilized embryos divided by the total number of eggs/embryos produced within 3 months of reaching adulthood. For the purpose of this study only fish that were able to produce eggs/embryos were analyzed; all presumed sterile fish were removed from the analysis. Of the 100% of fish in the control group that were reproductively active, the average fertility rate was 71%. The 0.1 ppm EE₂ treatment group had a fertility rate of 61%, which was not significantly different from that of the control group. The 0.5 and 1.0 ppm EE₂ treatment groups’ fertility rates were 32% and 12%, respectively. The 0.5 ppm EE₂ treatment group’s fertility rate was significantly reduced compared to the control group but not to 0.1 ppm EE₂ group. The 1.0 ppm group’s fertility rate was significantly reduced compared both to the control and the 0.1 ppm EE₂ groups, but not different from that of the 0.5 ppm group (Fig. 3B).

Effect of EE₂ on age of reproductive maturity

Adult rivulus reach sexual maturity at approximately 105 dph when raised at a nominal 25°C. In order to analyze the effect of EE₂ during early development on sexual maturity, fish were monitored for embryo/egg production starting on 84 dph to ensure that the first instance of reproductive activity was documented. By age 105 dph, in the H₂O and EtOH treatment groups, respectively, 55% and 86% of the fish were reproductively active. All H₂O-treated fish were reproductively active by Day 140 with 82% being reproductively active by 112 dph. All fish treated with the EtOH solvent control were reproductively active by Day 125 with 86% being reproductively active by 112 dph. By comparison, the fish treated with EE₂ were much older before reproductive activity was observed. In the group treated with 0.1 ppm EE₂, only 92% ever produced an egg/embryo, and on 105 dph, only 15% were reproductively active with a majority of 85% being reproductively active by 154 dph. Similarly, the group treated with 0.5 ppm EE₂ had a maximum of 60% of the fish becoming reproductively active, and on 105 dph, none of the fish had begun to reproduce. Fifty-three percent of this group was reproductive by 147 dph. In the fish treated with 1.0 ppm EE₂, only 63% of the fish became reproductively active, and on 105 dph, only 13% were producing eggs/embryos. The maximum number of reproductively active fish was seen on 125 dph (Fig. 4).

Fig. 4.

Effect of EE₂ (0.1, 0.5, and 1.0 ppm) on the age at which adult hermaphroditic rivulus were first observed producing a viable embryo, a nonviable embryo, or an unfertilized egg. Data represent the proportion of fish that were reproductively active at each time point. Treatments on 84 dph and 91 dph were not significant. Treatments on 105 dph through to 203 dph were significantly different (shaded area and P < 0.001).

Discussion

Sex determination and differentiation in teleost fishes are extremely plastic, especially in comparison to eutherian mammals (Orlando and Guillette 2007; Baroiller et al. 2009; DeFalco and Capel 2009). We may not understand all of the differences in mechanisms regulating the apparent mammalian gonadal “fortress” versus the pliable fish gonad (DeFalco and Capel 2009). However, we know that sex determination in mammals is genetically driven and that in fishes genetics are involved, but the environment is also an important driver in many species. Most fish lack sex chromosomes and are highly sensitive to environmental factors, including temperature, social structure of the population, pH, and environmental contaminants, particularly EDCs, such as EE₂ (Devlin and Nagahama 2002).

Because primary and secondary sex characteristics in fishes are sensitive to hormones, researchers and aquaculturists have used treatments with androgens and estrogens, for example 17α-methyltestosterone and EE₂, respectively, to direct sexual outcome in a number of fishes (Donaldson 1996; Mylonas et al. 2010). In this study, we were interested in developing a protocol to reliably direct female development using EE₂. In a pilot study in our laboratory, we created all females in the treatment groups; however, our controls were insufficient, so these results left us unable to conclude that we had a protocol (unpublished report, Schaughency 2005).

In this study, fish were exposed aqueously to EE₂ at 0.1, 0.5, or 1.0 ppm from hatching to 28 dph. After treatment, fish were placed in fresh saltwater only and grown to adult stage. Expression levels of genes relevant to gonadogenesis and reproduction, and fertility rate, sterility rate, and age at reproductive maturity were measured. In the brain, there was no long-term effect on cyp19a1b expression. Our results suggest that in rivulus cyp19a1b expression may be resistant to EE₂ due to the hormonal requirements of the simultaneous, hermaphroditic ovotestis. An excess of circulating estrogens may negatively affect spermatogenesis; thus, in rivulus the cyp19a1b may be tightly controlled and resistant to upregulation. Neuroestrogens from brain aromatase are important for neurogenesis and thus vital for normal development and growth of the brain (Chang et al. 2005; Menuet et al. 2005). Thus, maintenance of cyp19a1b expression and function would be critical for survival of the organism and may be part of a mechanism that is tightly regulated and resistant to alterations. Also, at the time of tissue collection the fish had been removed from exposure to EE₂ for approximately 175 days, and it is possible that any effect caused by EE₂ on cyp19a1b was corrected over time. Interestingly, cyp19a1a, which is already relatively lowly expressed in the brain, was downregulated with EE₂ treatment, and this effect was not dose dependent. Given that cyp19a1a is lacking a full ERE (only contains a half ERE), and that the exact function of ERE-halves is not fully understood, suggests the possibility of a different regulatory mechanism (Chang et al. 2005). It is possible that EE₂ has a lasting, perhaps epigenetic, effect on an upstream regulator of cyp19a1a (Navarro-Martin et al. 2011).

The liver did not show an increase in vtg gene expression most likely because induction of vtg by estrogens and estrogen mimics is transitory in nature; thus, the expected upregulation vtg was likely returned to baseline levels through compensatory mechanisms after fish were removed from treatment (Sumpter et al. 1995; Lubzens et al. 2010).

In the adult rivulus’ gonadal tissue, cyp19a1b, ERβ, and figα were not affected by EE₂ treatment during early development. The lack of cyp19a1b regulation can be explained similarly as for the brain (see above). ERβ was unaffected by exposure to EE₂ suggesting that it may not be autoregulated by estrogen and/or estrogenic compounds. The other possibility is that any effect caused by EE₂ was reversed over time when fish were removed from treatment. The other gene investigated and that was unaffected by EE₂ was figα. Figα is a molecular marker for production of primary oocytes. In this study, EE₂ was given at high doses in part to try and induce a female phenotype, without the testicular tissue normally found in the ovotestis of the hermaphrodite. It makes sense that figα may not show an upregulation because gene expression may have been functioning at maximum capacity, given the ovarian tissue that was already present.

Cyp19a1a was downregulated in the solvent control (1.0 ppb EtOH) as well as in all EE₂ treatments. The effect of ethanol on cyp19a1a gene expression in the gonad was not seen in the brain. However, across all fish analyzed, as well as across triplicate measurements, the effect of ethanol was consistent. Studies of cancer and of alcoholism have shown that ethanol consumption can increase aromatase activity as well as ERα expression (Gordon et al. 1979; Etique et al. 2004; Menuet et al. 2005). However, those studies were conducted in mammals or with mammalian cell lines, usually focused on breast and liver tissue, and showed an upregulation, whereas in the present case we saw a downregulation of cyp19a1a. Studies of zebrafish showed that EE₂ causes an upregulation of ERα (Etique et al. 2004) similar to the findings in studies of mammals. Again, this is in contrast to the results of this study in which ERα was downregulated at the highest dose of EE₂. It is possible that ERα regulation, which in this species is not well understood, may be under additional regulatory pathways that may be negatively impacted by EE₂.

Studies of zebrafish have shown expression of dmrt1 in testicular somatic cells of adults (Martyniuk et al. 2007). In some species, such as medaka, dmrt1 has been associated with early testicular differentiation (Kanamori et al. 2006). Many studies confirm that dmrt1 is not expressed in females, but is a male-specific marker of spermatogenesis. In our study, there was a dramatic reduction in dmrt1 gene expression in response to EE₂. Its downregulation, but not ablation, seen here, verifies that EE₂ has a negative effect on male development in rivulus, but the doses or timing and duration of exposure used in this study were not enough to completely inhibit testicular formation. This confirms data on reproductive efficiency in which the ability to go through internal fertilization was severely hindered, but not destroyed.

Perhaps the most interesting result from our study is that rivulus exposed to EE₂ showed a severe overall decrease in reproductive health. There appears to be a dose-dependent effect of EE₂ on sterility rates with 100% of the fish in the control groups capable of reproductive activity while in the group treated with 0.1 ppm EE₂ about 7% were sterile and in the groups treated with 0.5 and 1.0 ppm EE₂ approximately 40% of the fish were sterile. Of the fish that were able to produce a viable embryo, nonviable embryo, or an unfertilized egg, the rate at which viable embryos were produced also had a dose-dependent response to EE₂, with the highest concentrations of EE₂ resulting in lower fertility rates. These results are not surprising because EE₂ is known to cause abnormal development of the male gonad including, but not limited to, the development of oocytes with the testicular tissue. EE₂ also has an adverse effect on female reproductive rate (Papoulias 2000; Balch et al. 2004). In that case one would hypothesize that even if the testicular portion of the ovotestis did develop, it would not have developed normally, resulting in reduced function, and this is supported by the data on gene expression for dmrt1 as discussed above. There is evidence in other species that the adverse effects of EE₂ can be reversed after exposure ceases (Hill and Janz 2003; Hahlbeck et al. 2004a, 2004b); however, rivulus continued to have severe decline in reproductive efficiencies, suggesting a higher sensitivity to this hormone. Although it is possible that the sterile fish in this study were females, upon dissection for assays of gene expression, we did not see evidence of large, mature follicles in these fish, which would indicate that they were egg-bound females. If we were able to examine the gonads histologically, we would most likely see immature oocytes and/or dysfunctional spermatogenesis and both would support the observed sterility. Another interesting result was the effect on the age at which rivulus became reproductively active. The control groups followed a similar timetable to that seen in several other studies (Harrington 1975; Maack and Segner 2004; Kanamori et al. 2006). However, in the EE₂-treated groups it took longer for the fish to reach sexual maturity. Again, this is most likely due to abnormal development of gonadal tissue. Delay in age of maturity could potentially have substantial effects on the health of wild rivulus populations if exposed to mixtures of EDCs that could alter the internal hormonal milieu in the fish.

In conclusion, we sought to develop a reliable protocol to produce female rivulus and to investigate the potential effects of EE₂ on the sexual development, reproductive health, and relevant gene expression in adult rivulus. Overall, our findings show that at concentrations not normally measured in the environment (ppm), but used in aquaculture to direct the sexual development of fishes, treatment of rivulus with EE₂ results in a gonad with reduced testicular function and overall reduced fertility. Clearly, more research needs to be done in order to develop a reliable protocol to produce females and to obtain a better understanding of the endocrine regulation of reproduction in this species. That said, the effect of the concentrations used for 1 month posthatching resulted in adults with greatly impaired reproductive health that includes reduced expression of relevant genes and, importantly, decreased fertility, increased sterility, and delay of age of maturity in the mangrove rivulus.

Funding

Conference support was provided by an NIH Conference Grant R13HD070622 from the Eunice Kennedy Shriver National Institute of Child Health & Human Development (BC Ring, Valdosta State University); SICB through the DCE, DCPB, DAB (E.F.O.) and the C. Ladd Prosser Fund (E.F.O.); and the College of Agriculture and Natural Resources, University of Maryland (UMD) (E.F.O.). We are grateful to the UMD, Department of Animal and Avian Sciences, for the funding to E.F.O. in support of this research and the faculty research assistantship to J.L.F.