Gene expression profiles revealing the mechanisms of anti-androgen- and estrogen-induced feminization in fish

Abstract

Environmental anti-androgens are increasingly being recognized as potential contributing factors in the chemically induced feminization of wild fish because, by blocking androgen action, they can produce phenotypic effects similar to environmental estrogens. The molecular mechanisms by which anti-androgens and estrogens exert feminizing effects, however, have not been systematically compared. Using a targeted approach, we profiled the expression responses of a suite of 22 genes involved in reproduction, growth and development (processes controlled by androgens and estrogens) in the liver and gonad in adult male and female fathead minnow (Pimephales promelas) exposed to the model anti-androgen flutamide and the model synthetic estrogen 17α-ethinylestradiol (EE₂). Both flutamide (320 μg/L) and EE₂ (10 ng/L) produced phenotypic effects indicative of feminization (induction of plasma vitellogenin, reduced gonadosomatic index, and reduced secondary sex characters), although for the chosen test concentrations EE₂ was the more potent. For the genes studied, flutamide and EE₂ produced distinct expression profiles, suggesting that they largely operate via distinct molecular mechanisms. As examples, in liver EE₂ (but not flutamide) exposure up-regulated estrogen receptor (ER) α mRNA, whereas flutamide exposure increased ERβ and ERγ mRNAs in males and resulted in decreased androgen receptor (AR) mRNA in females. In the testis, flutamide up-regulated genes coding for enzymes involved in androgen biosynthesis (cytochrome P450 17 [CYP17] and 11β-hydroxysteroid dehydrogenase [11β-HSD]) implying an inhibitory action on androgen negative feedback pathways. EE₂, in contrast, inhibited the expression of enzymes involved in androgen biosynthesis (CYP17, 11β-HSD and 17β-hydroxysteroid dehydrogenase [17β-HSD]). There were also some commonalities in the molecular mechanisms of flutamide and EE₂ action, including the down-regulation of gonadal sex steroid receptor expression (gonadal AR and ovarian ERα), increased expression of genes coding for estrogen-producing enzymes (cytochrome P450 19A and B [CYP19A and CYP19B]), decreased expression of genes involved in testis differentiation (anti-Mullerian hormone [AMH] and doublesex and mab-3 related transcription factor 1 [DMRT1]), and decreased expression of hepatic genes which mediate wider physiological processes such as somatic growth (growth hormone [GH], GH receptor [GHR], insulin-like growth factor-I [IGF-I], IGF-I receptor [IGF-IR], thyroid hormone receptor α [TRα] and β [TRβ]).

Introduction

It is the ratio of androgens to estrogens that produces a ‘male’ versus a ‘female’ hormonal milieu in vertebrates, with elevated levels of androgens compared with estrogens in males and vice versa in females. It is now firmly established that a wide range of natural and anthropogenic chemicals present in the aquatic environment have the capacity to disrupt this hormonal balance in fish and, in turn, alter physiological function, most notably sexual development and function (endocrine disrupting chemicals [EDCs]; Tyler et al., 1998). To date, the focus of endocrine disruption research has been on chemicals that act as agonists of the estrogen receptor (ER) and have the ability to ‘feminize’ male fish (so-called environmental estrogens). Comparatively less attention has been directed towards EDCs with other modes of action, such as chemicals which bind to the androgen receptor (AR). However, it is now firmly established that some environmental contaminants, including some chemicals already classified as environmental estrogens, are (sometimes additionally) AR antagonists (environmental anti-androgens; Gray et al., 1994, Kelce et al., 1995, Sohoni and Sumpter, 1998). Furthermore, there is increasing recognition that an anti-androgenic mode of action may also play a part in eliciting the chemically induced feminization of the male reproductive system (Kelce and Wilson, 1997).

By competitively binding to the AR and blocking the action of endogenous androgens, anti-androgens can create an ‘estrogenic environment’, thus producing symptoms indicative of estrogen exposure (Sohoni and Sumpter, 1998). Both estrogen and anti-androgen exposure can therefore lead to similar phenotypic effects, even though they may act via different modes of action. Many of these effects are feminizing ones, including increased plasma estrogen levels (Makynen et al., 2000, MacLatchy et al., 2003, Jensen et al., 2004), decreased testis size and retarded spermatogenesis in males (Baatrup and Junge, 2001, Bayley et al., 2002, Kinnberg and Toft, 2003, Jensen et al., 2004, Pawlowski et al., 2004), reduced ovarian growth, altered oocyte development and decreased fecundity in females (Makynen et al., 2000, Brion et al., 2004, Jensen et al., 2004, Nash et al., 2004, Pawlowski et al., 2004, Mandiki et al., 2005), the presence of a female-like reproductive duct (ovarian cavity) and/or oocytes in the testes of male fish (Kiparissis et al., 2003, Brion et al., 2004, Nash et al., 2004), decreased prominence of male secondary sex characters and/or increased prominence of female secondary sex characters (Brion et al., 2004, Panter et al., 2004, Pawlowski et al., 2004), and female-skewed sex ratios (Bayley et al., 2002, Brion et al., 2004).

Research into endocrine disruption is increasingly being conducted at the level of gene expression as this approach provides a mechanistic understanding of the chemical effect and enables the comparison of mechanisms of action of different EDCs. Studies on gene expression have shown that chemicals which act through distinct modes of action induce unique and diagnostic gene expression ‘fingerprints/signatures’ (where the genes are not identified) or ‘profiles’ (where the genes are identified), even when they share some phenotypic effects (e.g. Hamadeh et al., 2002a). In contrast, chemicals which act through the same mode of action (e.g. estrogens: Watanabe et al., 2003, Moggs et al., 2004) share extensive commonalities in their gene expression fingerprints/profiles.

Research into the effects of anti-androgens in fish has, to date, been characterized by a heavy reliance on biochemical and histopathological endpoints and, in contrast to estrogens, the molecular mechanisms of action of anti-androgens are largely unknown. Some recent work on anti-androgens employing gene expression endpoints has shown that anti-androgens have a molecular signature distinct from that of estrogens (Larkin et al., 2002, Moens et al., 2006), suggesting that anti-androgens and estrogens exert their feminizing effects in fish via different molecular mechanisms. However, the majority of the genes studied in that work were unannotated and, therefore, the gene (or gene pathway) responses underlying the molecular signatures were largely unaddressed. Moreover, no phenotypic effect measures were recorded in those studies, so the links between altered gene expression patterns and physiological implications were not addressed.

In this work, we compared the effects of a model anti-androgen, the non-steroidal pharmaceutical flutamide (2-methyl-N-(4-nitro-3-[trifluoromethyl]phenyl)propanamide), and a model estrogen, the steroidal pharmaceutical estrogen 17α-ethinylestradiol (EE₂), on the expression of 22 genes (via real-time PCR) centrally involved in reproduction, growth and development (processes controlled by androgens and estrogens) in the fathead minnow (Pimephales promelas), in conjuction with phenotypic level effect measures of somatic growth, gonadal growth, the female yolk protein precursor plasma vitellogenin (VTG) and secondary sex character development. EE₂, the active ingredient of the human contraceptive pill, is known to be a widespread contaminant of the aquatic environment (Desbrow et al., 1998). Flutamide, however, has no known environmental relevance but is a pure AR antagonist, unlike anti-androgens with known environmental relevance (such as the fungicide vinclozolin and the insecticide DDT metabolite p,p′-DDE) which appear to have multiple modes of action [e.g. vinclozolin is also a mineralocorticoid and progesterone receptor antagonist and a ER agonist (Molina-Molina et al., 2006) and p,p′-DDE is also an ER agonist (Sohoni and Sumpter, 1998)]. Furthermore, on the basis of mammalian studies, flutamide has been proposed as a model AR antagonist for EDC research (U.S. EPA, 1998) and flutamide, and its metabolite 2-hydroxyflutamide, bind competitively to the fathead minnow AR (Ankley et al., 2004, Makynen et al., 2000). The data presented show clearly that it is possible to differentiate between the effects of a model anti-androgen and model estrogen through studies on gene expression in fish, and our data further demonstrate that the feminizing effects exerted by these two classes of EDCs occur via both distinct and common gene pathways.