To investigate the role of estrogen in the serial-sex changing fish Trimma okinawae, we isolated complementary DNAs encoding two distinct cytochrome P450 aromatase isoforms from adult ovary and brain (termed P450aromA and P450aromB, respectively). Sequence and phylogenic analyses showed that the goby P450arom forms belong to two separate CYP19 subfamilies. Transient expression of these cDNAs in HEK293 cells caused conversion of exogenous testosterone to estradiol-17β. RT-PCR showed that P450aromA was expressed in the brain, spleen, testis and ovary. P450aromB was expressed in the brain, liver, testis and ovary. In situ hybridization studies showed that P450aromA mRNA, but not P450aromB mRNA, was present in both ovary and testis. Positive signals were restricted to granulosa cells of vitellogenic follicles and interstitial cells of mature testis. Ovarian expression of both P450arom genes during the spawning cycle was examined by quantitative real-time RT-PCR. P450aromA transcripts increased during vitellogenesis and decreased prior to spawning. In contrast, P450aromB transcripts were barely detectable and did not correlate with ovarian development. These findings suggest that P450aromA, but not P450aromB, is involved in regulating ovarian vitellogenesis in goby.
Despite numerous endocrine studies on sex change in teleosts, no general mechanisms mediating sex change have emerged. The gobiid fish, Trimma okinawae, possesses ovarian and testicular tissues simultaneously in its gonad. Furthermore, T. okinawae is able to change sex repeatedly in both directions depending on its social surroundings (Sunobe and Nakazono, 1993; Kuwamura et al., 1993). A dominant male controls a harem of several females. Removal of this dominant male results in sex change of the largest female in the harem. If placed with a larger male, the once dominant male reverts to a female. Almost all sequential sex-changing fish either protogynous or protandrous reversibly change sex in one direction. As sex change in both directions can be socially manipulated, T. okinawae provides an excellent animal model to elucidate the mechanisms of male to female and female to male sex change.
Steroid hormones produced by gonads play crucial roles in ovarian function, sex differentiation and sexual maturation and behavior in vertebrates in general (Fostier et al., 1983; Maclusky and Naftolin, 1981; Wilson et al., 1981). In hermaphrodite fishes, estrogen plays a particularly important role in natural and experimentally-induced sex change (Chang et al., 1998; Fostier et al., 1993; Hunter et al., 1983). Our concern is the physiological role of estrogen in serial-sex changing gobiid fish, T. okinawae. Estrogens are synthesized in the gonads and brain by an enzymatic step using testosterone as a substrate. The terminal enzyme in the steroidogenic pathway is cytochrome P450 aromatase (P450arom).
To investigate the role of estrogen in a serial-sex changing gonad, we focused on P450arom. As a first step, we isolated and characterized P450arom cDNAs from goby. Subsequently, localization of P450arom in the gonad was observed by in situ hybridization. Additionally, changes in P450arom expression were examined during the natural ovarian cycle using real-time quantitative RT-PCR.
MATERIALS AND METHODS
Animals were collected from their natural habitat at Makurazaki, Kagoshima Prefecture, Japan and then kept in the laboratory. Fish were maintained under constant conditions (seawater at 27°C, 14L: 10D) in indoor tanks. Fish were fed commercial pellet and brine shrimp. Sex was determined by genital papillae (Sunobe and Nakazono, 1993; Grober and Sunobe 1996). Mature female and male fish were transferred to glass aquaria and reared under the above condition. The normal spawning cycle (∼4 days) was seen with no evidence of sex change. In this study, samples were obtained after spawning to prevent contamination with sex-changing samples.
Sample collection and RNA extraction
Adult fish were sedated in an ice bath and sacrificed by decapitation. Following removal, majority of the gonad was frozen in liquid nitrogen and stored at −80°C. The remaining gonadal fragment was fixed in 4% paraformaldehyde in 0.1M phosphate buffer for approximately 14 hours. After dehydration and paraffin embedding, gonads were sectioned at 5 μm. Total RNA was extracted from frozen tissues using the RNeasy mini kit with the RNase-free DNase set (Qiagen) according to manufacturer's protocols.
Oligonucleotides used as PCR primers are shown in Table 1. The degenerate primers of A-dn-Fw and -Rev were against highly conserved regions of the aromatase I-helix and heme-binding domains. Another degenerate primer pair (B-dn-Fw and -Rev) was designed on against highly conserved regions of P450aromB cDNA from channel catfish (AF417239), goldfish (U18974), zebrafish (NM_131642), Nile tilapia (AF295761) and rainbow trout (AJ311937). Gene-specific primers were synthesized based on our sequences. Primers actin-Fw and actin-Rev were selected on the goby β-actin cDNA sequence.
Primer sets used for RT-PCR, 5′ and 3′-RACE and real-time quantified RT-PCR for sequence and expression analysis of goby P450 aromtase A and B
Cloning of goby aromatase cDNAs
Goby aromatase was cloned first by generating a partial cDNA by reverse transcriptional-polymerase chain reaction (RT-PCR) using degenerate primers. Following the amplifications of the 5′-and 3′-cDNA ends by rapid amplification of cDNA ends (RACE), a cDNA encoding the complete coding region was obtained by a single RT-PCR reaction.
Described briefly as above, one microgram of total RNA from ovary was reverse transcribed using an oligo(dT) primer and Superscript II reverse transcriptase (Gibco-BRL) according to manufacturer's instructions. PCR was performed using degenerate primers (A-dn-Fw and -Rev, B-dn-Fw and -Rev) designed from highly conserved regions of known P450arom sequences. RT-PCR cycle conditions were 25 cycles, with 94°C 1 min, 55°C 1 min, 72°C 1.5 min. The amplicons were T-A ligated to pGEM-T-easy vector (Promega, Madison, WI, USA) and sequenced.
Based on the sequence information of the presumptive aromatase cDNA amplicon, rapid amplification of cDNA ends (RACE) procedures were performed to isolate the 5′ and 3′ ends of the cDNA (SMART cDNA library construction Kit; Clontech). In each RACE procedure, the initial PCR amplification was followed by a nested amplification. Gene-specific primer sets (GSP series) in combination with adaptor primers were used, respectively, for 5′-and 3′-RACE. RACE PCR cycling conditions were 35 cycles with 94°C 30 sec, 64°C or 56°C 40 sec, 70°C 120 sec.
The complete open reading frames (ORF) of the goby P450aromA and B were generated from ovarian RNA by RT-PCR with primers targeting the untranslated portions immediately upstream (AORF-Fw and BORF-Fw) and downstream of the ORF (AORF-Rev and BORF-Rev). The resulting PCR products were subcloned into pBluescript II KS-(Stratagene, LaJolla, CA, USA) designated gfAA/pBlue and gfAB/pBlue and sequenced in both directions.
DNA sequencing analysis
The resulting plasmid DNA was purified by the alkaline lysis method (Sambrook and Russell, 2001). Sequencing was performed for both strands with an Applied Biosystem model 377 sequencer after labeling with dye terminator cycle sequencing kit (Applied Bio-systems, Foster city, CA, USA) using vector and gene-specific primers. Sequence analysis was performed using DNASIS software (Hitachi Co, Ltd., Tokyo. Japan). Alignment of multiple protein sequence was performed using ClustalW multiple sequence alignment program and homology value (percent of amino acid sequence identity) was calculated by pair wise alignment. The phylogenetic tree was constructed using the Neighbor-joining method (Saitou and Nei, 1987) and viewed with TREE-view (Ver. 1.8). Program settings are detailed in the figure legend.
Transient expression in HEK293 cells
The cDNA inserts from goby gfAA/pBlue and gfAB/pBlue were subcloned into a cytomegalovirus expression vector (pcDNA3.1+, Invitrogen) at the EcoRI-XhoI restriction site and designated gfAA/pc and gfAB/pc. These clones were sequenced to confirm that the cDNA was inserted in the sense orientation. HEK293 cells were plated onto 6 cm tissue culture dish (Corning, Japan) in 4 ml Dulbecco's modified Eagle's medium (Nissui Pharmaceutical Co. Ltd., Tokyo, Japan) containing 10%(W/V) fetal calf serum (JRK, Japan) under 5% CO2 until 80% confluent. One microgram of purified recombinant expression vector was transfected into HEK293 cells using TfxTM-50 reagent (Promega) according to the suppliers protocol, followed by incubation at 37°C for 2 hours in 4 ml culture medium. The cells were then cultured for an additional 12 hours. After adding testosterone at a final concentration of 10 ng/ml (Sigma Chemical Co., St Louis, MO, USA), the plates were incubated for 48 hours. Culture medium and cells were separated by centrifugation at 1000 r.p.m. for 10 min and supernatant was stored at −80°C until assay. To measure levels of estrogen produced by HEK293 cells, the culture medium was extracted twice with diethyl ether and evaporated by vacuum centrifugation. The amount of estradiol-17β was measured using an estradiol enzyme immuno assay kit (Cayman, USA). According to the supplier's information, the cross-reaction for testosterone is 0.1%.
Tissue expression of the goby aromatase genes
Tissue distribution of goby aromatases was examined by RTPCR as follows: single-stranded cDNA was synthesized using Omni-script reverse transcriptase (Qiagen) from 0.25 μg of oligo(dT)-primed total RNA isolated from various tissues. PCR analysis of both P450aroms (primers ART-Fw and -Rev, BRT-Fw and -Rev, respectively) and β-actin (primers actin-Fw and -Rev) was carried out with the following cycle conditions: 30 cycles at 92°C for 30 sec, 55°C for 30 sec, and 72°C for 1 min. Ten-microliter aliquots of the PCR reactions were analyzed electrophoretically and the amplicons were visualized on UV transilluminator.
In situ hybridization analysis
Gonadal sections were hybridized with digoxygenin (DIG)-labelled goby P450aroms cRNA probe overnight at 58°C in hybridization buffer. Sense and antisense probes were labeled using standard transcription reactions and a DIG-labeling nucleotide mix (Boehringer Mannheim, Tokyo, Japan). Hybridization signals were detected with alkaline phosphatase-conjugate antibody according to manufacturer's protocol (Boehringer). Sections were incubated for 2 hours at room temperature with DIG antibody. After incubation sections were incubated with NBT/BCIP substrate (Boehringer) for detection of alkaline phosphatase activity.
Real-time quantitative RT-PCR
Abundance of goby aromatase transcript was determined as a measure of gene expression in ovary during the spawning cycle by real-time quantitative RT-PCR analysis (rtq-PCR) method using TaqMan technology published elsewhere (Perkin Elmer Applied Biosystems, 1997; Heid et al., 1996). Samples were reverse transcribed from 500 ng of total RNA in a 20-μl reaction volume using random hexamer primers and Omniscript reverse transcriptase (Qiagen). Standard copy number of both P450arom genes was estimated based on molecular weight and absorbance of gfAA/pBlue and gfAB/pBlue plasmid. Two sets of primer RP-A-Fw and -Rev, RP-B-Fw and -Rev (P450aromA, B respectively) were designed according to Primer Express software (version 1.5, PE Applied Bio-systems). The dilution templates for each of triplicate were tested in 25-μl of PCR mixture containing 2X SYBER Green master mix (PE Applied Biosystems) and 0.3 μM (in final concentration) of each primer. The PCR profile was as follows: one initial denaturation at 95°C for 15 min, 45 cycles of denaturation at 94°C for 15 sec, annealing at 52°C (both genes) for 15 sec, and extension at 72°C for 30 sec and final extension at 60°C for 5 min, followed by the dissociation curve analysis. The PCR amplifications and fluorescent detection were performed with ABI Prism Sequence Detector 7000. DNA contamination was checked by performing a control reaction in which there was no reverse transcribed RNA.
Isolation of two goby P450arom cDNAs
The degenerate RT-PCR generated an amplicon of the expected size (401 bp) from ovary. Ten clones of this ampl-icon were sequenced and all of the nucleotide sequences were P450aromA. A single set of nested 3′ RACE reactions using sense primers 3GP1 and 3GP2 in combination with adapter primers amplified the entire 3′ terminus of cDNA (including the poly-A tail). On the other hand, a nested 5′ RACE reaction using the antisense primers 5GP1 and 5GP2 stopped short of the translation start codon. Therefore, a second 5′RACE using primers 5GP3 and 5GP4 was conducted to amplify the remainder of the 5′ end and a portion of the untranslated region. Goby P450aromA cDNA contained 2126 nucleotides with a putative 1545 bp open reading frame (ORF) and polyadenylation signal, ATTAAA, located 30 bp upstream of the poly (A)+ tail. This cDNA encoded a 515-amino acid protein with a calculated molecular mass of 58 KDa.
The initial RT-PCR using the degenerate primers amplified a partial cDNA of the expected size (1038 bp) from ovary. Seventeen clones of this amplicon were sequenced and the nucleotide sequence of all clones was P450aromB. RT-PCR of brain mRNA also generated P450aromB amplicons. The RACE procedures led to the isolation of the 5′ and 3′ ends of the cDNA covering the ORF and untranslated region. Specific RT-PCR to generate a full-length clone using primer BRT-Fw and -Rev amplified a single amplicon. The nucleotide sequence of this clone was identical to that of RACE clones. The goby P450aromB cDNA contained 1790 nucleotides with a putative 1494 bp ORF. A putative polyadenylation signal is located 28 bp upstream of the polyA tail. This cDNA encodes 498-amino acid protein with a calculated molecular mass of 57 KDa.
Fig. 1 shows the deduced amino acid sequence of the two goby aromatase aligned with other reported P450arom forms of zebrafishA (AF226620); zebrafishB (NM_131642); Nile tilapiaA (U72071); Nile tilapiaB (AF295761); goldfishA (AB009336); goldfishB (U18974); catfishA (S75715); catfishB (AF417239) and human (P11511). Goby P450aromA was more closely related (60–73.5% identity) to isoform A of other teleosts than isoform B (58–68% identity). The two aromatases of goby shared only 55.8% overall identity. Identity to human aromatase is even lower (48, 51% respectively). Fig. 1 also depicts that putative functional domains, including the I-helix, the aromatase-specific region, and the heme-binding regions are highly conserved among vertebrates. Phylogenic analysis of the full-length goby P450aromA and B with other reported aromatases indicated that the teleost aromatases clearly segregated into P450aromA and B branches (Fig. 2). Goby P450aromA and B are grouped in teleost P450aromA and B branch, respectively.
Transient expression of P450arom cDNAs in HEK293 cells
Fig. 3 shows the production of estradiol-17β in HEK293 cells transfected with both gfAA/pc and gfAB/pc constructs using testosterone as substrate. Estradiol-17β was converted from testosterone in a linear fashion over 48 h of incubation. These results confirmed that goby cDNA encoded P450aromA and B.
Tissue distribution pattern of P450aromA and B
Goby P450aromA was distributed in brain, spleen, ovary and testis. No signal could be detected in the liver and kidney (Fig. 4). P450aromB was expressed in brain, liver, ovary and testis. In contrast to P450aromA, P450aromB was undetectable in the spleen (Fig. 4). The expression of β-actin (Fig. 4) indicated the functional integrity of the cDNA template.
In situ hybridization of gonadal tissue
P450aromA transcripts were found in both cell layers of the ovarian follicle. In previtellogenic follicles, signals were seen only in the thecal cells (Fig. 5; AI and BI). In vitellogenic follicles, signals were more abundant in the granulosa cells, as compared to the thecal cells (Fig. 5; CI, DI and EI). P450aromA transcripts were also detected in interstitial cells of testis (Fig. 5; FI). Sense probes for P450aromA showed no signal (Fig. 5; AII–FII). In contrast to P450aromA, P450aromB mRNA was not detected in either testis or ovary (data not shown).
Quantitative analysis of P450aromA and B transcripts in ovarian follicles
Expression levels of P450aromA and B mRNA during the spawning cycle were examined (Fig. 6). Both forms (P450aromA and B) were detected in the ovary; however, P450aromA transcripts were at least more than 100-fold greater than P450aromB transcripts. P450aromA transcript levels changed markedly with progression of oocyte growth. Maximum levels of P450aromA transcripts were detected during late vitellogenesis on Day 3. Thereafter, levels sharply declined with maturation on Day 4. In contrast, the levels of P450aromB did not change during oocyte development.
Most teleosts have at least two CYP19 genes (P450arom), which encode structurally and functionally distinct P450arom isozymes in ovary (CYP19a/P450aromA) and brain (CYP19b/P450aromB) (Callard and Tchoudakova, 1997; Chiang et al., 2001; Gelinas et al., 1998; Kwon et al., 2001). In this study, we isolated two forms of the aromatase cDNA from the goby, T. okinawae. Sequence comparison and phylogenic analysis indicate that two isoforms of fish P450arom are orthologs of the previously identified mammalian and avian aromatase and members of paralogous clades within teleost A- and B-lineages. The A- and B-iso-forms of goby are only about 58% identical, indicating a long evolutionary history as separate genes.
In addition to sequence analysis, the ability to catalyze the transformation of androgen to estrogen was checked in HEK293 cells transfected with two CYP19 genes. We confirmed that both isoforms have aromatization ability as is the case with goldfish (Tchoudakova et al., 1998) and zebrafish (Chiang et al., 2001).
The non-quantitative RT-PCR analysis of aromatase expression in various adult tissues of T. okinawae revealed that P450aromA expressed in spleen and P450aromB in liver similar to other species; Japanese flounder (Kitano et al., 1999), Nile tilapia (Kwon et al., 2001) and human fetus (Price et al., 1992) in spleen whereas Xenopus (Miyashita et al., 2000), Nile tilapia (Kown et al., 2001) and human (Harada et al., 1993; Toda et al., 1994; also review by Simpson et al., 2002) in liver. These results suggest that the spleen and liver are capable of producing estrogen from precursors. However, the potential function of locally-produced steroids in these organs remain unknown. Our RT-PCR analysis also detected both P450arom transcripts in testis, ovary and brain. The overlapping expression of both forms in gonads and brain is consistent with the results obtained in zebrafish (Kishida and Callard, 2001; Trant et al., 2001), Nile tilapia (Kwon et al., 2001) and rainbow trout (Valle et al., 2002). The expression of aromatase gene in brain suggests an important role for sexual behavior (Matsumoto et al., 2003) and development of the central nervous system (Maclusky et al., 1981). Changes in the expression of both forms of P450arom genes in brain during sex change in T. okinawae is of particular interest.
Our in situ hybridization studies determined the cellular location of P450arom within the goby ovarian tissues. The most abundant transcripts for P450aromA could be seen in the granulosa cells of mid- and late-vitellogenesis and weak signals were seen in thecal cells throughout vitellogenesis. Aromatase activity was reported to be present in ovarian follicles of several teleosts including Nile tilapia (Chang et al., 1997), amago salmon (Young et al., 1983) and medaka (Fukada et al., 1996). However, these studies did not examine which type of aromatase is localized in ovarian follicles. Thus, the present study was the first to show that A-type of aromatase, but not B-type, is localized in granulosa cells in teleosts.
Real-time quantitative RT-PCR revealed that P450aromB transcripts were very low in ovary and showed no marked changes during ovarian development. The low levels of P450aromB transcripts may explain why our in situ hybridization did not detect the transcripts in ovary. In fact, using the same probes, we were able to detect P450aromB transcripts in brain (unpublished data). In any case, the expression profile of P450aromB during the spawning cycle suggests that this form of P450arom does not play a major role in regulating ovarian development.
The present study clearly showed the differential expression of two forms of P450arom transcripts in ovary. The amounts of P450aromA transcripts increased significantly during vitellogenesis and declined during late/post vitellogenesis. In contrast, P450aromB transcripts were very low (RT-PCR) or hardly detectable (in situ hybridization) throughout vitellogenesis. Study of the 5′-flanking region of both P450arom genes may help to determine the mechanism of this tissue-specific expression. We recently showed that in the Nile tilapia, Oreochromis niloticus, the expression of P450aromA and Ad4BP/SF-1 increased in parallel with ovarian growth (Yoshiura et al., 2003). These findings suggest that the form of P450arom responsible for estrogen synthesis during vitellogenesis is P450aromA, and Ad4BP/SF-1 plays an important role in regulating the expression pattern of P450aromA gene in ovarian follicles. It was also shown that in mammals, decrease in P450arom transcripts in granulosa cells was associated with a reduction in the mRNA of SF-1 (Fitzpatrick et al., 1997). Further studies are required to analyze the role of Ad4BP/SF-1 in relation to ovarian development and also with reference to P450arom in T. okinawae.
Surprisingly, our in situ hybridization analysis revealed that P450aromA transcripts could be detected in the interstitial cells of mature testis. Expression of aromatase in the testis agrees with earlier findings in dogfish (Callard et al., 1985), channel catfish (Trant et al., 1997), rainbow trout (Belvedere et al., 1998), European sea bass (Valle et al., 2002) and Atlantic stingray (Ijiri et al., 2000). However, the situation between these species and T. okinawae in terms of the mode of gonadal formation is different. T. okinawae is unique in having both ovary and testis regardless of sexual phase (Sunobe and Nakazono, 1997). The functional significance of P450aromB in the testis of the sex changing goby remains to be studied. In mammals, estrogens have recently been shown to play a physiological role in the regulation of spermatogenesis (see a review by Carreau et al., 2003). The expression profiles of P450aromA in the testis during the spermatogenesis as well as sex change need to be determined. It is also important to determine whether expressions of ovarian and testicular P450aromA genes are regulated by a similar or different mechanism.
In summary, we cloned, sequenced and characterized two P450arom cDNAs from T. okinawae. P450aromA transcripts, but not P450aromB, exhibited distinct changes during ovarian development, suggesting an important role for P450aromA in the synthesis of estrogen during active vitellogenesis. These cDNA probes will be useful for future studies on the roles of aromatase in gonadal development and even sex behavior during the serial-sex change in Trimma okinawae.
We would like to thank Dr. Hisaya Manabe, Kagoshima Univ., for collecting fish. We would also like to thank Dr. Toshitaka Ikeuchi, Nagahama Institute of Bio-science and Technology, for technical assistance. This work was in part by Grants-in-Aid for Research for CREST of JST and Bio-Design Program from the Ministry of Agriculture, Forrestry, and Fisheries, Japan.