Androgen-dependent sexual dimorphism in pituitary tryptophan hydroxylase expression: relevance to sex differences in pituitary hormones

Abstract

Serotonin is a biogenic monoamine conserved across phyla that is implicated in diverse physiological and behavioural functions. On examining the expression of the rate-limiting enzymes in serotonin synthesis, tryptophan hydroxylases (TPHs), in the teleost medaka (Oryzias latipes), we found that males have much higher levels of tph1 expression as compared with females. This robust sexual dimorphism was found to probably result from the direct stimulation of tph1 transcription by androgen/androgen receptor binding to canonical bipartite androgen-responsive elements in its proximal promoter region. Our results further revealed that tph1 expression occurs exclusively in pro-opiomelanocortin (pomc)-expressing cells and that the resulting serotonin and its derivative melatonin inhibit the expression of the pituitary hormone genes, fshb, sl and tshb. This suggests that serotonin and/or melatonin synthesized in pomc-expressing cells act in a paracrine manner to suppress pituitary hormone levels. Consistent with these findings and the male-biased expression of tph1, the expression levels of fshb, sl and tshb were all higher in females than in males. Taken together, the male bias in tph1 expression and consequent serotonin/melatonin production presumably contribute to sex differences in the expression of pituitary hormones and ultimately in the physiological functions mediated by them.

1. Introduction

Serotonin, a monoamine neurotransmitter that is highly conserved across animal phyla, affects a wide variety of physiological and behavioural functions including stress response, aggression, feeding, cognition and mating behaviour [1]. It is also a precursor of melatonin, a pleiotropic indoleamine hormone best known for its regulatory role in circadian and seasonal rhythms [2]. Because males and females show profound differences in many serotonin-dependent functions mentioned above, it seems reasonable to assume that serotonin acts in a sex-dependent manner in the brain. Indeed, sex variations in brain serotonin levels that probably underlie the differences in these functions have been repeatedly demonstrated in rodents, where females have moderately, but significantly, higher levels as compared with males [3–7].

Tryptophan hydroxylase (TPH) is the rate-limiting enzyme in serotonin synthesis and a specific enzymatic marker for serotonergic neurons. In vertebrates, two paralogous forms of TPH, referred to as TPH1 and TPH2, have been identified [8]. Consistent with the female-biased sex difference in brain serotonin levels, studies in rodents and primates have revealed that TPH1 and TPH2 expression is dependent on oestrogen, the predominant sex steroid in females [9–12]. Subsequent studies have further shown that the effect of oestrogen on TPH1 and TPH2 is mediated by oestrogen receptor 2 (ESR2, also known as ERβ), one of the nuclear oestrogen receptors that act as ligand-gated transcription factors [13–15].

In addition to the brain, a substantial amount of serotonin is synthesized in peripheral organs, where it acts in an autocrine or paracrine manner on local cells [16]. For example, expression of TPH has been observed in the pituitary gland in several vertebrate species, including rodents and humans [17–19]. Considering that the pituitary plays a central role in the endocrine system by secreting multiple tropic hormones that stimulate other endocrine organs, it seems likely that serotonin synthesized locally in the pituitary may have a specific endocrine function. To our knowledge, however, this hypothesis has not been specifically explored in the literature. Moreover, there remain several other important questions, including the identity of the pituitary cells responsible for TPH expression, the regulatory mechanisms that control TPH expression in the pituitary and the possible existence of sex differences therein.

In recent years, small teleost fish, such as zebrafish (Danio rerio) and medaka (Oryzias latipes), have emerged as potent model organisms for studying the physiology and pathophysiology of serotonergic signalling [20–23]. In the present study, we have examined the expression of TPH in medaka, identifying a substantial sex difference in the expression of tph1 in the pituitary. Unexpectedly and contrary to the findings in the rodent brain described above, this sex difference was male-biased in medaka and attributable to a direct stimulatory effect of androgen, the predominant sex steroid in males, on tph1 transcription. In addition, our findings led us to assess the potential involvement of sex-dependent tph1 expression in sex differences in pituitary hormone expression.

2. Material and methods

(a). Animals and drug treatment

Medaka of the d-rR strain were bred and maintained at 28°C under a 14/10 h light/dark photoperiod. They were fed three to four times per day with live brine shrimp and commercial pellet food. Sexually mature, spawning adult fish (aged three to six months) were used for all analyses unless otherwise stated. All sampling was conducted at 1–2.5 h after onset of the light period.

For in vivo drug treatment studies, male fish were maintained in water containing 250 ng ml⁻¹ of the androgen receptor (AR) antagonist cyproterone acetate or vehicle alone (ethanol; final concentration 0.002%) for 9 days. Female fish were maintained in water containing 100 ng ml⁻¹ of the prominent, non-aromatizable teleost androgen 11-ketotestosterone (11KT) or vehicle alone (ethanol; final concentration 0.001%) for 9 days.

(b). Real-time PCR

For each sample, the pituitaries from 15 fish were pooled (except for organ culture studies, where the pituitaries from three fish were pooled) and total RNA was isolated by using the RNeasy Lipid Tissue Mini Kit (Qiagen, Hilden, Germany) with DNase treatment (Qiagen). Complementary DNA was synthesized by using the SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA) with random primers provided in the kit. Real-time PCR was run on a LightCycler 480 System II (Roche Diagnostics, Basel, Switzerland) using the LightCycler 480 SYBR Green I Master (Roche Diagnostics). For every reaction, a standard curve was generated to determine the efficiency and linearity of the amplification, and melting curve analysis was conducted to ensure that a single amplicon was produced in each sample. Normalized values were obtained by dividing the relative amount of target transcripts by that of the β-actin transcript (actb; NM_001104808) in the same sample. The amount of the actb transcript was not significantly different between male and female pituitaries and among serotonin-, melatonin- and vehicle-treated pituitaries. The primers used for real-time PCR are listed in electronic supplementary material, table S1.

(c). In situ hybridization

Complementary DNA fragments of medaka tph1, ARs (ara and arb), the pituitary hormones (follicle-stimulating hormone β subunit (fshb), luteinizing hormone β subunit (lhb), thyroid-stimulating hormone β subunit (tshb), growth hormone (gh), somatolactin (sl), prolactin (prl) and pro-opiomelanocortin (pomc)) and aralkylamine N-acetyltransferases (AANATs; aanat1a, aanat1b and aanat2) (for details, see electronic supplementary material, table S2) were PCR-amplified and transcribed in vitro to generate digoxigenin (DIG)- or fluorescein-labelled antisense probes using the DIG RNA Labelling Mix and Fluorescein RNA Labelling Mix (Roche Diagnostics), respectively. Single and double in situ hybridization were performed according to Hiraki et al. [24] and Takeuchi & Okubo [25], respectively. Detailed procedures are available in electronic supplementary material.

(d). Transcriptional activity assay

A medaka bacterial artificial chromosome (BAC) clone (ola1-104G05) containing the tph1 locus was obtained from National BioResource Project (NBRP) Medaka. The proximal promoter region of tph1 was sequenced and analysed for the presence of putative androgen-responsive elements (AREs) by using Match (public v.1.0) and Jaspar (v.5.0_alpha) with default settings.

Genomic DNA fragments upstream of the first methionine codon of tph1, containing 3808 or 1523 bp of the proximal promoter region plus the first 90 bp of exon 1, were PCR-amplified from the BAC clone and ligated into the HindIII site of the pGL4.10 luciferase reporter vector (Promega, Madison, WI, USA). The full open reading frames of medaka ara (NM_001104681) and arb (NM_001170833) were PCR-amplified and ligated into the pcDNA3.1/V5-His-TOPO expression vector (Thermo Fisher Scientific). COS-7 cells were co-transfected transiently with either the 3808 or 1523 bp luciferase reporter construct, either the Ara or Arb expression construct and an internal control vector pGL4.74 by using Lipofectamine LTX (Thermo Fisher Scientific). Six hours after transfection, the cells were treated with 11KT at doses of 0, 10⁻¹⁰, 10⁻⁸ and 10⁻⁶ M in phenol red-free Dulbecco's modified Eagle's medium supplemented with 5% charcoal-stripped fetal bovine serum for 18 h. The cell extracts were assayed for luciferase activity by a GloMax 20/20n Luminometer (Promega) using the Dual-Luciferase Reporter Assay System (Promega). Assays were performed in duplicate and repeated three times independently.

Additional assays were performed with 5′-truncated and point-mutated luciferase reporter constructs to determine the AREs responsible for androgen induction of tph1 transcription. Constructs 5′-truncated to 910 and 410 bp upstream of the tph1 transcription start site were generated as described above. Constructs carrying point-mutated putative AREs were generated by using the PrimeSTAR Mutagenesis Basal Kit (Takara Bio, Shiga, Japan). Given that an ARE half-site is sufficient for transcriptional activation [26], mutations were introduced into either one or both half-sites of each of the two putative AREs (at positions −1282 and −1019 relative to the transcription start site) in the 1523 bp luciferase reporter construct. Each half-site was mutated into a HindIII recognition site sequence (AAGCTT). Assays with these mutated 1523 bp constructs and the wild-type 1523 bp construct were performed as described above, except that a single dose of 11KT (10⁻⁶ M) was used.

(e). Pituitary organ culture and drug treatment

Because male pituitaries are expected to contain large amounts of endogenous serotonin and melatonin, female pituitaries were used to evaluate the effects of serotonin and melatonin on the expression of pituitary hormones. For each sample, the pituitaries from three female fish were pooled and cultured ex vivo in Leibovitz's L-15 medium supplemented with 5% charcoal-stripped fetal bovine serum as described in [27]. The medium was further supplemented with 10⁻⁵ M serotonin or melatonin or vehicle alone. After 24 h of culture at 25°C, the pituitaries were processed for total RNA extraction, cDNA synthesis and real-time PCR as described above.

(f). Statistical analysis

Quantitative data were expressed as means with error bars representing standard error of the mean. When comparing data between sexes, individual values were also reported to determine whether sexual dimorphism exists (i.e. little or no overlap between female and male values). For comparison of relative values, the values for control samples (grey column(s) in each graph) were arbitrarily set to 1, and the other values were adjusted accordingly. Statistical analyses were performed by using GraphPad Prism (GraphPad Software, San Diego, CA, USA). See electronic supplementary material for a detailed description of the analysis pipeline. The results of statistical analyses are summarized in electronic supplementary material, table S3.

3. Results

(a). The pituitary shows androgen-dependent sexual dimorphism in tph1 expression

Analysis of the expression of TPH genes in the adult pituitary by real-time PCR showed that tph1 was much more abundantly expressed in males than in females (figure 1a). All male individuals had clearly higher expression levels as compared with female individuals, indicating that this difference between males and females represents a true ‘sexual dimorphism’. By contrast, levels of tph2 expression did not differ between the sexes (figure 1b).

Figure 1.

The pituitary shows androgen-dependent sexual dimorphism in tph1 expression. (a,b) Levels of tph1 (a) and tph2 (b) expression in the adult female and male pituitary (n = 8 per sex). ***p < 0.001. (c) Total area of tph1 expression in the female (grey columns) and male (blue columns) pituitary at different ages (n = 5 per sex and age). *p < 0.05. (d) Representative micrographs of tph1 expression in the female and male pituitary at different ages. Scale bars represent 100 µm. (e,f) Total area (e) and representative micrographs (f) of tph1 expression in the pituitary of adult males treated with vehicle alone or androgen receptor antagonist (AR-ant) (n = 5 per treatment). ***p < 0.001. Scale bars represent 100 µm. (g,h) Total area (g) and representative micrographs (h) of tph1 expression in the pituitary of adult females treated with vehicle alone or 11-ketotestosterone (11KT) (n = 5 per treatment). **p < 0.01. Scale bars represent 100 µm. (Online version in colour.)

To further investigate the sex differences in tph1 expression, we performed in situ hybridization in pituitaries from fish at different ages, including one month (when secondary sexual characteristics begin to appear), two months (when fish are juvenile and have not yet spawned), three months (when fish are sexually mature and have spawned) and seven months (when fish have regressed and the frequency of spawning has declined). tph1 was expressed in the intermediate lobe of the pituitary in both sexes at all ages, and the total area of expression gradually increased as the pituitary increased in size with age, particularly in males (figure 1c,d). Significant male-biased sexual dimorphism was evident as early as two months of age and persisted thereafter. A male bias was also noted at one month of age but was not statistically significant.

These results led us to assume that a factor associated with male sexual development, perhaps androgen, plays a key role in mediating the sexual dimorphism in tph1 expression. We tested this idea by first treating adult males with the AR antagonist cyproterone acetate and adult females with the primary teleost androgen 11KT, and then examining tph1 expression in their pituitaries by in situ hybridization. As expected, treatment of males with cyproterone acetate caused a marked reduction in the total area of tph1 expression (figure 1e,f), whereas treatment of females with 11KT increased it (figure 1g,h).

Collectively, these results reveal that there is male-biased sexual dimorphism in tph1 expression in the intermediate lobe of the pituitary that is largely, if not entirely, mediated by androgen/AR signalling.

(b). Androgen directly activates transcription of tph1 through AREs in its proximal promoter region

To test the possibility that androgen directly acts on tph1-expressing cells in the pituitary, we first determined whether these cells coexpress ARs. Double in situ hybridization for tph1 and each of the two AR subtypes (ara and arb) revealed that most tph1-expressing cells also expressed both ara and arb (figure 2a).

Figure 2.

Androgen directly activates transcription of tph1 through AREs in its proximal promoter region. (a) Coexpression of tph1 and androgen receptor genes (ara and arb) in the same pituitary cells. Left and middle panels show representative images of, respectively, tph1 (green) and ara/arb (magenta) expression in the same sections; right panels show the merged images with nuclear counterstaining (blue). Scale bars represent 10 µm. (b) Schematic diagram of the tph1 locus depicting the relative location of the seven canonical bipartite ARE-like sequences in the tph1 proximal promoter region. Bent arrow indicates the transcription start site. The nucleotide sequences of the ARE-like sequences at positions −1282 and −1019, which were found to be functional, are shown. (c–f) Stimulation of tph1 transcriptional activity by 11-ketotestosterone (11KT). COS-7 cells were transfected with a luciferase reporter construct containing 3808 bp (c,d) or 1523 bp (e,f) of the tph1 proximal promoter region, together with an expression construct for Ara (c,e) or Arb (d,f). The cells were then treated with different concentrations of 11KT, and the relative luciferase activity was determined. *p < 0.05; **p < 0.01; ***p < 0.001 (versus untreated control). (g,h) Effects of 5′-truncation of the tph1 proximal fragments on the stimulation of tph1 transcriptional activity by 11KT. The cells were transfected with a luciferase reporter construct containing 1523, 910 or 410 bp of the tph1 proximal promoter region, together with an Ara (g) or Arb (h) expression construct. The cells were treated with (blue columns) or without (grey columns) 11KT, and luciferase activity was assayed. *p < 0.05. (i,j) Effects of mutations in the ARE-like sequences at positions −1282 and −1019 on the stimulation of tph1 transcriptional activity by 11KT. The cells were transfected with a wild-type 1523 bp luciferase reporter construct (wild-type) or a construct carrying a mutation in the ARE-like sequence at position −1282 (mut-1282) or −1019 (mut-1019), together with the Ara (i) or Arb (j) expression construct. Mutations were introduced into the 5′ half-site (−5′), 3′ half-site (−3′) or both half-sites (-both) of the ARE-like sequences. The cells were treated with (blue columns) or without (grey columns) 11KT, and luciferase activity was assayed. *p < 0.05; **p < 0.01. (Online version in colour.)

We then examined the ability of androgen to directly stimulate tph1 transcription. The search for possible AREs within the 5′ proximal promoter region of tph1 identified seven canonical bipartite ARE-like sequences at positions −3721, −3537, −1282, −1019, −790, −646 and −555 relative to the transcription start site (figure 2b). A luciferase-based transcriptional activity assay using the 3808 bp proximal fragment of tph1 revealed that 11KT induced luciferase activity in a dose-dependent manner in the presence of either Ara or Arb, with much higher induction levels observed with Arb (figure 2c,d). An additional assay using the 5′-truncated 1523 bp proximal fragment of tph1 yielded similar results, although significant differences were not observed with Ara (figure 2e,f). Further 5′-truncation of the tph1 proximal fragments to 910 and 410 bp resulted in complete loss of 11KT induction of luciferase activity (figure 2g,h), indicating that the cis-elements responsible for induction by 11KT are located in the region between −1523 and −910 bp relative to the transcription start site.

This region contains two ARE-like sequences (at positions −1282 and −1019); therefore, we introduced point mutations into both sequences in the 1523 bp luciferase reporter construct and determined the effects on luciferase activity. In the presence of Ara, all of the introduced mutations resulted in no significant induction by 11KT (figure 2i). In the presence of Arb, the mutations introduced into either or both half-sites of the sequence at position −1282 led to reduced but still significant induction of luciferase, whereas those introduced at position −1019 more effectively attenuated the induction, with no significant induction of luciferase observed for the mutations of both half-sites (figure 2j). In addition, when both sequences at positions −1282 and −1019 were mutated, 11KT seemed to suppress rather than induce luciferase activity (figure 2j).

Taken together, these results suggest that 11KT acts directly on pituitary tph1-expressing cells to stimulate the transcription of tph1, and that this stimulatory action of 11KT is mediated mainly by binding of Arb to canonical bipartite AREs at positions −1282 and −1019 of the tph1 proximal promoter, with the ARE at position −1019 assuming a major role. In addition, the results suggest that both the 5′ and 3′ half-sites of these AREs contribute to the stimulatory action of 11KT.

(c). tph1 is expressed exclusively in pomc-expressing cells in the pituitary

Next, we examined the identity of the pituitary cells showing sexually dimorphic expression of tph1. Double in situ hybridization for tph1 and various pituitary hormone genes showed that expression of tph1 almost completely overlapped with that of pomc and was immediately adjacent to, but did not overlap with, expression of sl in the intermediate lobe (figure 3a,b). There was no overlapping or immediately adjacent expression of tph1 with any of the other pituitary hormone genes including fshb, lhb, tshb, gh and prl, which were primarily expressed in the anterior lobe (figure 3a). These results showed that sexually dimorphic expression of tph1 in the pituitary occurs exclusively in pomc-expressing cells in the intermediate lobe.

Figure 3.

tph1 is expressed exclusively in pomc-expressing cells in the pituitary. (a) The spatial expression pattern of tph1 was compared with that of pituitary hormone genes. Left and middle panels show representative images of the expression of, respectively, tph1 (magenta) and the indicated pituitary hormone gene (fshb, lhb, tshb, gh, sl, prl or pomc; green) in the same sections; right panels show the merged images with nuclear counterstaining (blue). Scale bars represent 50 µm. (b) Higher magnification images of tph1 and sl or pomc expression in the same sections. Scale bars represent 20 µm. (Online version in colour.)

(d). Relevance of sexually dimorphic expression of tph1 in the pituitary to sex differences in pituitary hormones

The expression of tph1 in pomc-expressing cells suggested that serotonin is locally synthesized and released by these cells. To test the possibility that serotonin synthesized by these cells is further metabolized to melatonin, we examined the pituitary for expression of AANATs, the rate-limiting enzyme in the synthesis of melatonin from serotonin. Teleosts, including medaka, have three paralogous genes encoding AANAT: aanat1a, aanat1b and aanat2 [28]. Real-time PCR demonstrated that all of these genes were expressed in the pituitary (electronic supplementary material, figure S1a). A small, but significant, sex difference favouring males was observed for the expression of aanat1a and aanat1b. Only aanat1b was expressed at a detectable level in the pituitary as measured by in situ hybridization (electronic supplementary material, figure S1b). Subsequent double in situ hybridization revealed that aanat1b was coexpressed with tph1 in the pituitary (figure 4a). These results suggest that serotonin synthesized by tph1-expressing cells is further metabolized to melatonin locally in the pituitary.

Figure 4.

Relevance of sexually dimorphic expression of tph1 in the pituitary to sex differences in pituitary hormones. (a) Coexpression of tph1 and aanat1b in the same cells in the pituitary (n = 8 per sex). Left and middle panels show representative images of, respectively, tph1 (green) and aanat1b (magenta) expression in the same sections; right panel shows the merged image with nuclear counterstaining (blue). Scale bars represent 20 µm. (b) Effects of serotonin and melatonin on the expression of the indicated pituitary hormone genes (fshb, lhb, tshb, gh, sl, prl and pomc) in a pituitary organ culture system (n = 8 per treatment). *p < 0.05; **p < 0.01. (c) Levels of pituitary hormone gene (fshb, lhb, tshb, gh, sl, prl and pomc) expression in the adult female (grey columns) and male (blue columns) pituitary (n = 8 per sex). The data are split into two graphs for clarity. **p < 0.01; ***p < 0.001. (Online version in colour.)

On the basis of these results, we hypothesized that serotonin and melatonin produced locally in the pituitary might have autocrine and/or paracrine effects on the expression of pituitary hormone genes. To test this idea, we investigated the effects of serotonin and melatonin on pituitary hormone gene expression by using an ex vivo pituitary organ culture system. Real-time PCR demonstrated that the expression levels of fshb and sl in cultured pituitaries were significantly reduced by both serotonin and melatonin supplementation (figure 4b). The level of tshb expression was also reduced by melatonin supplementation, whereas serotonin and melatonin had no significant effects on lhb, gh, prl and pomc expression (figure 4b).

These results led us to assume that the sexual dimorphism in tph1 expression and consequent differences in serotonin and/or melatonin production may ultimately contribute to differences in the expression of pituitary hormone genes between the two sexes. Based on this framework, we determined whether there are sex differences in the expression of pituitary hormone genes in the adult medaka pituitary. Real-time PCR analysis showed female-biased sexual dimorphism in the expression of five genes: all male specimens showed lower expression levels of fshb, lhb, tshb, gh and sl (figure 4c). By contrast, male-biased sexual dimorphism was observed in pomc expression, whereas no sex difference was found in prl expression (figure 4c).

4. Discussion

In the present study, we identified a robust sex difference in tph1 expression in the intermediate lobe of the medaka pituitary. To our knowledge, this is the first report of a difference in TPH expression in the pituitary between males and females. Moreover, the observed difference is more pronounced than any data reported previously for TPH, representing a true sexual dimorphism. Also noteworthy is that the sex difference in tph1 expression in the medaka pituitary is male-biased, in contrast to the rodent brain, where females typically have higher serotonin levels than males [29,30]. A few previous studies have shown a male predominance in TPH expression or serotonin synthesis, similar to the present study. For example, in adult humans, the mean rate of serotonin synthesis is 1.5-fold higher in the male brain than in the female brain [31]; and in juvenile mice, Tph1 expression in the trigeminal ganglion is twofold higher in males than in females [32]. In the adult medaka brain, males show slightly higher expression of tph1 and tph2 in the telencephalic and thalamic nucleus, respectively [33]. Together with these observations, the present findings suggest that the extent and direction of sex differences in TPH expression and serotonin synthesis vary with species, tissue and developmental stage more greatly than has been previously appreciated.

We further investigated the underlying mechanism and physiological relevance of the sexually dimorphic expression of tph1. Our results enable us to propose a likely model for these issues, as shown in figure 5. Our examination of males and females treated with an AR antagonist and non-aromatizable androgen, respectively, revealed that pituitary tph1 expression is largely dependent on androgen/AR signalling. Our subsequent transcriptional activity assay, together with the detection of AR expression in pituitary tph1-expressing cells, demonstrated that androgen can act directly on these cells mainly through Arb, which activates tph1 transcription by binding to the proximal promoter region of tph1 in response to androgen. These findings are interesting given that it is generally assumed in mammals that ESR2, but not AR, is expressed on serotonergic neurons, and that any effects of androgen on these neurons, including the modulation of TPH expression, are mediated by either indirect mechanisms involving other cells or by activation of ESR2 after aromatization of androgen to oestrogen [34–37]. Thus, the molecular mechanisms underlying sex steroidal regulation of serotonin synthesis may be fundamentally different between the pituitary and the brain or between medaka and mammals.

Figure 5.

Model of the potential mechanism underlying sexually dimorphic expression of tph1 in the pituitary and its potential relevance to sex differences in pituitary hormones. Androgen, the predominant sex steroid in males, acts directly on pomc-expressing cells in the pituitary to stimulate the transcription of tph1, resulting in male-biased sexual dimorphism in tph1 expression in these cells. This stimulatory action of androgen is mediated mainly by androgen-dependent binding of Arb to two AREs located in the proximal promoter region of tph1. The male-biased tph1 expression leads to more efficient synthesis in males of serotonin, which can be further converted to melatonin by Aanat1b. The resulting serotonin and melatonin act in a paracrine manner to repress the expression of fshb, sl and tshb, probably contributing to female-biased sex differences in the expression of the encoded pituitary hormones. (Online version in colour.)

We also identified two canonical bipartite AREs in the tph1 proximal promoter as the cis-elements responsible for androgen induction of tph1 transcription. To our knowledge, this is the first identification of endogenous cis-elements involved in androgen/AR signalling in basal vertebrates. Notably, the two AREs identified can act in a synergistic fashion to activate tph1 transcription, as seen by a comparison of data between the construct with wild-type sequence and that with mutations introduced into both half-sites of the AREs in figure 2j. Considering that the canonical AREs identified are only 263 bp apart and that multiple non-canonical ARE half-sites in close proximity can act synergistically [38,39], the half-sites of these canonical AREs may interact to some extent with one another.

It should be also noted that the sexually dimorphic pattern of pituitary tph1 expression can be almost completely reversed between males and females by altering the androgen milieu in adulthood. In rodents, by contrast, the effects of the adult sex steroid milieu on brain TPH expression and serotonin levels tend to be small or even non-significant [15,40]. This difference may reflect the fact that teleosts are highly sexually labile across their lifespan. In teleosts, including medaka, altering the sex steroid milieu effectively lead to the reversal of sex-typical behavioural, neural and gonadal phenotypes, even in adulthood [41,42]. The pituitary, as well as the brain and gonad, may possess lifelong sexual lability in teleosts.

Our subsequent investigations revealed for the first time that pomc-expressing cells are the source of tph1 expression in the pituitary, and that serotonin synthesized in these cells can be further converted to melatonin because the cells also express aanat1b. Expression of aanat1b was also observed in cells adjacent to pomc-expressing cells (probably sl-expressing cells), suggesting that serotonin may be converted to melatonin in these adjacent cells. It is well established that serotonin synthesized in the brain activates both POMC neurons in the arcuate nucleus and POMC cells in the pituitary [29]. We thus speculated that serotonin synthesized in the pituitary might act in an autocrine manner to stimulate pomc expression. Contrary to our expectation, however, studies using an organ culture system revealed that serotonin and melatonin do not affect pomc expression, but instead exert inhibitory effects on the expression of fshb, sl and tshb. These results suggest that serotonin/melatonin synthesized in pomc-expressing cells acts in a paracrine manner to suppress FSH, SL and TSH. This idea is strengthened by the observation that fshb- and tshb-expressing cells are in close proximity to, and sl-expressing cells are intermingled with, pomc-expressing cells. In further support of this idea, studies in rats have shown that serotonin and melatonin have an inhibitory effect on the release of FSH and that melatonin has an inhibitory effect on plasma TSH levels [43,44]. It thus seems probable that the male-biased tph1 expression leads to more pronounced suppression of these pituitary hormones in males than in females. Consistent with this presumption, we found that the expression levels of fshb, sl and tshb are all sexually dimorphic, with males having lower levels than females. Taken together, the male predominance in tph1 expression and its effects on serotonin/melatonin production supposedly contribute to sex differences in the expression of these pituitary hormones in favour of females.

The present findings can also be interpreted in the context of the mechanism of action of androgen on pituitary hormones. Considering the induction of pituitary tph1 expression by androgen, it seems reasonable to assume that the effects of androgen on FSH, SL and TSH are mediated, at least in part, by expression of tph1 and the consequent production of serotonin/melatonin. Although no definite evidence is available for SL and TSH, evidence exists showing that androgen regulates the expression of FSH by acting both directly on the pituitary and indirectly through gonadally derived inhibin and hypothalamic pathways involving, for example, gonadotropin-releasing hormone [45,46]. Although the effects of androgen on FSH seem to vary depending on species, physiological state and experimental design, a direct inhibitory effect at the pituitary level has been demonstrated in primates and several teleost species, including medaka [45,47]. However, the mechanism for this direct inhibitory effect remains largely unknown. Our results suggest that this mechanism may involve the AR-mediated transcriptional activation of tph1 in pomc-expressing cells, followed by the paracrine actions of the resulting serotonin and melatonin.

In summary, our study has identified the sexually dimorphic expression of tph1 in the medaka pituitary and provided conclusive evidence for its underlying mechanism and probable physiological relevance. As a master endocrine organ that controls a myriad of important physiological functions, the pituitary merits special attention. Our results suggest that tph1 expressed in the pituitary may affect the synthesis of pituitary hormones that primarily regulate gonadal steroidogenesis and gametogenesis (FSH), skin coloration (SL), and thyroid cell growth and function (TSH). The androgen-dependent sexual dimorphism in the signalling pathway identified in this study supposedly constitutes part of the mechanism underlying the differences in these functions between males and females. It would be worth exploring in future studies whether and to what extent the present findings in medaka are applicable to other species, including humans.

Ethics

The care and use of animals were done in accordance with the guidelines of the Institutional Animal Care and Use Committee of the University of Tokyo. The committee requests the submission of an animal-use protocol only for use of mammals, birds and reptiles, in accordance with the Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology of Japan (Ministry of Education, Culture, Sports, Science and Technology, Notice No. 71; 1 June 2006). Accordingly, we did not submit an animal-use protocol for this study, which used only teleost fish and thus did not require approval by the committee.

Data accessibility

All data generated or analysed during this study are included in this published article and its electronic supplementary material.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) in Japan and the Japan Society for the Promotion of Science (JSPS) (MEXT/JSPS grant nos 13J04816 (to Y.K.-S.), 17H06429 (to K.O.) and 19H03044 (to K.O.)).