Gonadotropin-releasing hormone (GnRH) is a decapeptide originally isolated from pig and sheep hypothalami as a physiologic regulator of luteinizing hormone (LH) release from the pituitary (Matsuo et al., 1971; Burgas et al., 1972). At present, nine GnRH molecules have been characterized and named for the vertebrate species in which they were first identified (Fig. 1): mammalian GnRH, chicken GnRH-I (King and Millar, 1982a,b; Miyamoto et al., 1982, 1983), chicken GnRH-II (Miyamoto et al., 1984), salmon GnRH (Sherwood et al., 1983), lamprey GnRH-I (Sherwood et al., 1986), lamprey GnRH-III (Sower et al., 1993), catfish GnRH (Bogerd et al., 1992; Ngamvongchon et al., 1992), dogfish GnRH (Lovejoy et al., 1992), and seabream GnRH (Powell et al., 1994). They have been identified in vertebrate brains and constitute a family of structurally-related decapeptides. During the 500 million years of vertebrate evolution, the primary structure of GnRH has been remarkably conserved.
It is generally accepted that GnRH regulates synthesis and release of pituitary gonadotropin (GTH) (see King and Millar, 1992; Sherwood et al., 1993). In addition, GnRH can act as a neuromodulator, and administration of exogenous GnRH facilitates sexual behavior in many species (see Pfaff et al., 1987).
In mammalian, avian, reptilian, and amphibian animals, GnRH is conveyed to the pituitary via the hypothalamohypophyseal portal vessels. In mammals, pulsatile release of mammalian GnRH (mGnRH) by hypothalamic neurons stimulates GTH secretion from the pituitary. Teleost fishes lack the median eminence. Instead GnRH neurons are found to directly innervate the pituitary. It is therefore interesting to examine GnRH systems in teleost fish in view of comparative endocrinology.
Immunocytochemistry and radioimmunoassay (RIA) have been utilized for studying the distribution and function of GnRHs in the brain. In addition, recent progress in molecular biological methods has enabled us to study expression of GnRH genes using hybridization techniques (see Urano and Hyodo, 1990).
This review will focus on the roles of the GnRHs in the teleost brain, particularly in salmonid fishes.
IDENTIFICATION OF GnRH
In teleost fish, the primary structure of salmon GnRH (sGnRH) was first determined from chum salmon, Oncorhynchus keta (Sherwood et al., 1983), followed by subsequent determination of catfish GnRH (Ngamvongchon et al., 1992) and seabream GnRH (Powell et al., 1994).
The presence of two types of GnRH, sGnRH and chicken GnRH-II (cGnRH-II), in the teleost brain was first demonstrated in the goldfish, Carassius auratus, by employing high performance liquid chromatography (HPLC) in conjunction with RIA (Yu et al., 1988). Recent studies have shown that in teleosts, more than one type of GnRH molecule exist even within the same species, and all of the teleosts examined thus far have cGnRH-II, whereas the second form is either sGnRH, mGnRH, catfish GnRH, or seabream GnRH. Existence of three forms of GnRH, i.e. sGnRH, cGnRH-II, and seabream GnRH, was recently reported in the brain of the seabream, Sparus aurata (Powell et al., 1994) and African cichlid, Haplochromis burtoni (White et al., 1995). The existence of multiple forms of GnRH in the brain suggests that each GnRH has a different distribution and function.
Differential distribution of multiple forms of GnRH in discrete brain areas was examined by RIA in several fishes in order to clarify their functions in the brain. Okuzawa et al. (1990) first measured sGnRH and cGnRH-II contents in the discrete brain regions of the rainbow trout, Oncorhynchus mykiss, using specific RIAs. These authors found that the contents of both types of GnRHs varied in different brain regions. The levels of sGnRH were higher than those of cGnRH-II in the olfactory bulbs, the telencephalon, the hypothalamus, the optic tectum-thalamus and the pituitary, whereas the cerebellum and the medulla oblongata contained much more cGnRH-II than sGnRH. Especially of note, cGnRH-II was undetectable in the pituitary. This is also true for masu salmon, Oncorhynchus masou (Amano et al., 1992, 1993). These results suggest that, of the two GnRHs, only sGnRH is involved in GTH secretion. As an example, sGnRH contents and cGnRH-II contents in each region of the brain of ovulated female masu salmon are shown in Fig. 2.
In the goldfish, sGnRH was distributed in a larger amount in the olfactory bulbs, the telencephalon, the hypothalamus, and the pituitary than in the other regions, whereas cGnRH-II was distributed widely throughout the brain with highest concentrations in the medulla oblongata. The major difference between salmonid fishes and goldfish is that goldfish pituitary contains cGnRH-II (Kobayashi et al., 1992, 1994).
In the European eel, Anguilla anguilla, mGnRH levels were higher than cGnRH-II levels in the pituitary, the olfactory lobes together with the telencephalon, and the diencephalon together with the mesencephalon, while the opposite results were obtained for the posterior part of the brain. Of interest, cGnRH-II levels in the pituitary were slightly above the detectable limit (Dufour et al., 1993).
These studies demonstrated that sGnRH (or mGnRH) and cGnRH-II are differently distributed in the brain, and also necessitated to investigate the localization of GnRH neurons and the changes in GnRH levels during gonadal maturation.
IMMUNOCYTOCHEMICAL DISTRIBUTION OF GnRH NEURONS
There have been many immunocytochemical studies relating to the GnRH system in teleost brains. Serious problems here are that most of these studies used antiserum against mGnRH, for example, in the rainbow trout (Goos and Murathanoglu, 1977; Schäfer et al., 1989), in the goldfish (Stell et al., 1984; Kah et al., 1984), in the platyfish, Xiphophorus maculatus (Münz et al., 1981; Halpern-Sebold and Schreibman, 1983), in the eel (Nozaki et al., 1985; Grober et al., 1987), in the catfish, Clarias batrachus (Subheader and Krishna, 1988), in the labrid fish (Grober and Bass, 1991). Some researchers used antiserum against sGnRH, without any characterization of cross reactivity to cGnRH-II, i.e., in the goldfish (Kah et al., 1986), in the green molly, Poecilia latipinna (Batten et al., 1990), and in the rainbow trout (Bailhache et al., 1994). Since, as is mentioned above, more than one type of GnRH molecule exists even within the same species, several authors have immunocytochemically examined the distribution of multiple GnRH forms in the teleost fish brains (Amano et al., 1991; Montero et al., 1994; Yamamoto et al., 1995; Kim et al., 1995).
The distribution of sGnRH immunoreactive (ir) neuronal somata and fibers, and that of cGnRH-II-ir neuronal somata and fibers in the masu salmon, are summarized in Fig. 3 (Amano et al., 1991). sGnRH-ir neuronal somata were scattered in the olfactory nerve (ON), the olfactory bulb (OB), between the olfactory bulb and the part of telencephalon which corresponds to the terminal nerve ganglion (TNG), the ventral telencephalon (VT), and the preoptic area (POA). sGnRH-ir fibers were distributed in various brain regions from the OB to the spinal cord. sGnRH-ir fibers directly innervated the pituitary. cGnRH-II-ir neuronal somata were found in the midbrain tegmentum located rostral to the motoneurons of the oculomotor nerve. The distribution of cGnRH-II-ir fibers was basically similar to that of sGnRH-ir fibers except for the absence of cGnRH-II-ir fibers in the pituitary. The number of cGnRH-II-ir fibers in the brain were much fewer than those of sGnRH. The distribution of sGnRH-ir neuronal somata in chum salmon was consistent with that in masu salmon (Kudo et al., 1996). These results suggest that, in salmonid, sGnRH not only regulates GTH secretion but also functions as a neuromodulator, whereas cGnRH-II functions only as a neuromodulator. Nonetheless, both GnRHs can stimulate release of GTH I and GTH II from the pituitary in vitro in sockeye salmon, Oncorhynchus nerka (Amano, 1993).
Differential localization of two types of GnRH, mGnRH and cGnRH-II, were reported in the European silver eel (Montero et al., 1994). mGnRH-ir neuronal somata were observed in the OB, the nucleus olfactoretinalis, the VT, the POA, and the mediobasal hypothalamus. mGnRH-ir fibers were distributed in many parts of the brain and also in the pituitary. cGnRH-II-ir neuronal somata were detected in the midbrain tegmentum, and very few cGnRH-II-ir fibers were observed in the pituitary.
In the goldfish, sGnRH-ir neuronal somata were localized in the TNG area between the ON and the OB, the VT, the POA, and the hypothalamus. cGnRH-II-ir neuronal somata were observed in the TNG, the VT, the POA, the hypothalamus and the midbrain tegmentum. Both sGnRH-ir and cGnRH-II-ir fibers were distributed not only in the hypothalamus and the pituitary but also in various brain areas from the OB to the spinal cord (Kim et al., 1995). These results suggest that both sGnRH and cGnRH-II not only regulates GTH secretion but also function as neuromodulators. Indeed, both forms of GnRH have been shown to induce in vitro release of GTH II and growth hormone (GH) in goldfish (Marchant and Peter, 1989; Marchant et al., 1989; Chang et al., 1990; Habibi et al., 1992). Although both GnRHs bind to the same pituitary receptors (Habibi, 1991; Habibi et al., 1992), they appear to induce hormone release via somewhat dissimilar mechanisms (Chang et al., 1991, 1992). Moreover, stimulation of sGnRH and cGnRH-II on GTH II synthesis was shown in goldfish (Khakoo et al., 1994); sGnRH, but not cGnRH-II, increased the accumulation of GTH IIβ and GTH α mRNA in sexually regressed fish, whereas both sGnRH and cGnRH-II stimulated accumulation of GTH IIβ and GTH α mRNA in sexually mature fish.
An approach requisite for clarification of the function of each GnRH-ir neuronal group is to examine its projection area in the brain. Recently, three different projection patterns were demonstrated in dwarf gourami, Colisa laria. GnRH neurons in the TNG which showed strong sGnRH and weaker cGnRH-II immunoreactivity projected to wide areas of the central nervous system from the OB to the spinal cord; sGnRH neurons in the POA projected only to the pituitary, and cGnRH-II neurons in the midbrain tegmentum projected mainly to brain regions posterior to the hypothalamus and the spinal cord (Yamamoto et al., 1995). These different projection patterns indicate functional differentiation of GnRH neuronal groups.
CHANGES IN GnRH CONTENTS DURING GONADAL MATURATION
If GnRH is involved in gonadal maturation, GnRH contents would be expected to change during gonadal maturation. Several studies were conducted to measure brain GnRH contents by RIA in relation to gonadal maturation in eel (Dufour et al., 1982), platyfish (Schreibman et al., 1983), brown trout, Salmo trutta (Breton et al., 1986), caribe colorado, Pygocentrus notatus (Gentile et al., 1986), goldfish (Yu et al., 1987), and chinook salmon, Oncorhynchus tschaawytsca (Lewis et al., 1992). However, the results were discordant. A clear correlation between brain GnRH contents and gonadal maturity was observed only in caribe colorado (Gentile et al., 1986). Such discrepancies may be, in part, due to the use of RIAs employing nonspecific antibodies.
We investigated changes in sGnRH and cGnRH-II contents in the brain and pituitary of masu salmon from hatching through gonadal maturation for three years. During gonadal maturation, sGnRH levels in the telencephalon including the POA and the pituitary correlatively increased with the elevation in GTH II levels in the pituitary and the plasma (Fig. 4). cGnRH-II were undetectable in the pituitary. Further, no significant changes in the concentration were found in discrete brain areas during vitellogenesis and ovulation (Amano et al., 1992). These results together with those of immunocytochemical studies suggest that sGnRH neurons in the telencephalon-POA are involved in gonadal maturation in masu salmon.
Recently, it was demonstrated that GnRH fibers originating from the TNG is not involved in GTH secretion and further in gonadal maturation in goldfish. The sGnRH contents in the brain except the OB markedly decreased by olfactory tract sectioning (OTX), whereas the cGnRH-II contents in the brain showed no clear changes. Despite large decreases in the brain sGnRH contents, gonadal maturation was not inhibited (Kobayashi et al., 1992, 1994). In dwarf gourami, GnRH neurons in the TNG, possibly sGnRH neurons, may be the most extensively projecting GnRH neurons in the brain except in the pituitary (Oka and Matsushima, 1993). These results indicate that most of the sGnRH fibers in the brain originates from the TNG, and that sGnRH neurons in the TNG do not project to the pituitary. Thus, it is possible that changes in the sGnRH contents measured by RIA reflect the activities of GnRH neurons in the TNG. It should be noted again that the levels of GnRH measured by RIA must be considered as a summation of synthesis, release and degradation of GnRH at any point in development. Therefore, it is necessary to examine the expression of GnRH gene in order to clarify function of GnRH in reproduction.
MOLECULAR CLONING OF GnRH GENES
There are several methods for the examination of hormonal gene expression. Among them, in situ hybridization (ISH) can be used to analyze mRNA expression of tissues of interest directly on histological sections (see Urano and Hyodo, 1990). Since sGnRH neuronal somata are widely scattered in the brain, ISH in combination with immunocytochemistry is a strong tool to yield information on the dynamic aspects of the synthesis of GnRH in a certain neuron or region. However, the ISH method require a hybridization probe whose nucleotide sequence is complementary to the mRNA to be examined.
The cDNA coding for mGnRH was initially isolated from human placenta (Seeburg and Adelman, 1984), and subsequently, mGnRH genes have been cloned from several species (Adelman et al., 1986; Mason et al., 1986). Chicken GnRH-I (cGnRH-I) gene has been cloned from chicken (Dunn et al., 1993). The cDNA for cGnRH-II has been isolated from teleost fish (White et al., 1994; Bogerd et al., 1994; Lin and Peter, 1996), the cDNA for sGnRH from several teleost fish (Bond et al., 1991; Suzuki et al., 1992; Klungland et al., 1992; Okuzawa et al., 1994; Grober et al., 1995; Ashihara et al., 1995; Suetake et al., 1995; Lin and Peter, 1996), cDNA for catfish GnRH from African catfish, Clarias gariepinus (Bogerd et al., 1994), and cDNA for seabream GnRH from African cichlid, Haplochromis burtoni (White et al., 1995) and from red seabream (Okuzawa et al., 1996).
In general, a GnRH precursor is composed of a signal peptide (SP), GnRH and a GnRH-associated peptide (GAP) which is connected to GnRH by a Gly-Lys-Arg sequence. The number of amino acid residues in SP and GAP in teleost GnRH precursors are shown in Table 1. The Gly-Lys-Arg sequence is considered to serve as a signal for proteolytic processing and C-terminal amidation (Nikolics et al., 1988). It was reported that GAP showed GTH-releasing activity and prolactininhibiting activity in rat (Nikolics et al., 1985), although prolactininhibiting activity of GAP is controversial. There have been no reports on the physiological functions of GAP in teleost fishes.
Amino acid numbers of signal peptide (SP) and GnRH associated peptide (GAP) of GnRH precursors identified in teleost fishes
The distribution of GnRH ISH positive neurons has been examined in several teleost fishes (Suzuki et al., 1992; Bailhache et al., 1994; White et al., 1995; Bogerd et al., 1994). The distribution of sGnRH ISH positive neurons is consistent with that of sGnRH-ir neuronal somata in masu salmon (Suzuki et al., 1992), rainbow trout and Atlantic salmon (Bailhache et al., 1994). In African catfish, catfish GnRH ISH positive neurons were scattered in the ON, along both sides of the midline of the telencephalon, the POA, and in the infundibular stalk close to the pituitary, whereas cGnRH-II ISH positive neurons were observed in the midbrain tegmentum (Bogerd et al., 1994). In African cichlid, sGnRH ISH positive neurons were detected in the TNG, whereas seabream GnRH ISH positive neurons were detected in the POA, and cGnRH-II ISH positive neurons were detected in the mesencephalon (White et al., 1995). In red seabream, sGnRH ISH positive neurons were detected in the TNG, and seabream GnRH ISH neurons were detected in the POA (Okuzawa et al., 1996). All these studies support the data obtained from immunocytochemistry.
REGULATION OF sGnRH mRNA EXPRESSION IN SALMONID FISHES
The cDNA sequences for sGnRH, which is involved in gonadal maturation through GTH secretion, have been isolated from several salmonid fishes (masu salmon, Suzuki et al., 1992; Atlantic salmon, Klungland et al., 1992; sockeye salmon, Ashihara et al., 1995), making it possible to examine sGnRH synthetic activity using ISH. Hence, we examined the changes of sGnRH mRNA expression in context of several aspects of reproduction in salmonid fish.
(a) Changes in sGnRH mRNA expression during gonadal maturation
The widespread distribution of sGnRH-ir neuronal somata in the ventral part of the brain suggests that sGnRH has several functions in the brain in addition to the stimulation of GTH secretion (Amano et al., 1991); however, it still remains obscure which sGnRH neurons are involved in gonadal maturation via GTH secretion. We therefore examined the changes of sGnRH mRNA expression during ovulation in 2-year-old female masu salmon (Amano et al., 1995a). In this study, sGnRH synthetic activity were estimated by the number of neurons expressing sGnRH mRNA, the number of silver grains per neuron, and the total number of silver grains per unit area. The activity was increased in the VT and the POA during maturation in accordance with increases in GSI and plasma GTH II concentrations (Fig. 5). On the contrary, no significant changes were observed in the OB and the TNG. These results indicate that sGnRH neurons in the VT and the POA, but not in the OB and the TNG, are involved in the regulation of gonadal maturation possibly through GTH secretion. These results are consistent with the previous data that sGnRH concentrations in the telencephalon including the POA increased with gonadal maturation (Amano et al., 1992, 1993).
In the dwarf gourami, GnRH originating from the TNG does not affect GTH secretion but rather has a function as a neuromodulator in the brain (Oka and Ichikawa, 1990; Oka, 1992; Yamamoto et al., 1995). sGnRH originating from the TNG is also not considered to be essential for gonadal development in goldfish (Kobayashi et al., 1992, 1994). It is also possible that sGnRH neurons in the TNG function to only modulate neuronal activity in masu salmon.
(b) Changes in sGnRH mRNA expression induced by photoperiod manipulation
Gonadal maturation in salmonid fish occurs in autumn, since the decreasing photoperiod in autumn accelerates gonadal maturation (Billard et al., 1978). Testicular maturation of underyearling precocious male masu salmon could be experimentally manipulated by changing the length of the light-dark photoperiod. Maturation was accelerated by changing the photoperiod from natural to short (8L 16D) in June, and was delayed by keeping the photoperiod long (16L8D) from June (Amano et al., 1994a). sGnRH mRNA levels in the VT and the POA increased when the fish spermiated; the activity increased in August in the 8L 16D group, and in September in the 16L8D group, respectively. Moreover, the increase of sGnRH mRNA levels was in accordance with the increase of pituitary GTH II contents (Fig. 6). No significant changes in sGnRH mRNA levels in relation to gonadal maturation were observed in the OB and the TNG (Amano et al., 1995b). These results indicate that sGnRH neurons in the VT and the POA are influenced by photoperiod, and are involved in the regulation of gonadal maturation through GTH secretion.
(c) Changes in sGnRH mRNA expression induced by sex steroids
Accumulation of GTH in the pituitary can be stimulated by aromatizable androgen or estrogen in juvenile fish; this is a known positive feedback system (see Goos, 1987). The involvement of GnRH in this mechanism is speculated, because sex steroid administration increased the amount of GnRH in the hypothalamus of the European silver eel (Dufour et al., 1985), and rainbow trout (Goos et al., 1986). On the other hand, the stimulatory effects of testosterone on the expression of pituitary GTH IIβ gene was demonstrated in vitro in juvenile rainbow trout (Xiong et al., 1993). Thus, whether steroids have direct actions on the pituitary, or act indirectly via GnRH was unclear.
We examined the effects of 17α-methyltestosterone (MT) on sGnRH mRNA expression in yearling masu salmon (Amano et al., 1994b). Oral MT application markedly increased pituitary GTH IIβ, but not GTH Iβ concentrations in both sexes. In future precocious males, MT treatment further increased the number of neurons expressing sGnRH mRNA in the POA but not in the OB and the VT (Fig. 7), whereas sGnRH mRNA levels were not changed by MT in immature females. These results suggest that sGnRH is involved in the positive feedback system at least in future precocious males, and that the difference in the responsiveness of preoptic sGnRH neurons to MT is based on the maturational stage of the fish. The result that sGnRH mRNA levels in the POA was increased by the administration of MT in 2-year-old females which are just before the initiation of gonadal maturation supports this hypothesis (Amano et al., 1997). Of note, studies on pubertal development have been conducted in a number of teleost fish species such as the rainbow trout (Gielen et al., 1982; Goos et al., 1986), the eel (Dufour et al., 1985; Counis et al., 1987), the platyfish (Schreibman et al., 1986), and African catfish (Schulz et al., 1994a,b). These studies suggest that sex steroids initiate and/or accelerate the development of the brain-pituitary-gonadal axis. Taken together, the response of preoptic sGnRH neurons to steroids is a necessary condition for gonadal maturation or the development of the brain-pituitary-gonadal axis.
(d) Changes in sGnRH mRNA expression during seaward migration
sGnRH neurons in the ON and the OB may be involved in the seaward migration of salmonid fish, as the importance of olfactory functions during homing migration has been confirmed. Interestingly, sGnRH mRNA expression in cribriform ganglion in the ON increased during downstream migration in juvenile chum salmon (Parhar et al., 1994). Meanwhile, changes of sGnRH-producing neurons in chum salmon during homing migration from coastal seas to the spawning ground of the maternal river was examined using ISH (Kudo et al., 1996). sGnRH mRNA levels in the ON and the ON-OB were high in fish in the coastal seas, but activity decreased in the spawning ground. These results suggest that sGnRH neurons in the ON and probably some in the ON-OB may be involved in seaward migration.
(e) Existence of two differing precursor genes for sGnRH
Existence of two different genes for sGnRH have been demonstrated in sockeye salmon (Ashihara et al., 1995). These authors found two different cDNAs for sGnRH precursors (pro sGnRH-I and pro sGnRH-II); pro sGnRH-I cDNA is 436 base pair long (short type) and pro sGnRH-II cDNA is 482 base pair long (long type). A question arising here is whether the two genes encoding sGnRH are expressed equally or differently. We are now attempting to examine the distribution of sGnRH neurons expressing short type and/or long type sGnRH precursors in the sockeye salmon brain using ISH.
At present, nine GnRH molecules have been characterized throughout the vertebrates, and the existence of multiple forms of GnRH in the brain was observed also in teleost fish. All of the teleosts examined thus far have cGnRH-II, whereas the second form is either sGnRH, mGnRH, catfish GnRH, or seabream GnRH. Distribution of GnRH neurons was examined using immunocytochemistry. In general, sGnRH neurons were distributed from the ON to the POA or the hypothalamus, whereas cGnRH-II neurons were detected in the midbrain tegmentum. GnRH-ir fibers were distributed in various brain regions, indicating multiple neuroendocrine functions in the brain.
The brain GnRH contents were measured by RIA and related to gonadal maturation, although results are discordant. It was demonstrated that most of the sGnRH fibers in the brain originate from the TNG, and that sGnRH derived from the TNG is not essential for gonadal maturation.
Recent determination of the cDNA encoding GnRH precursors has enabled the investigation of GnRH synthetic activity using ISH. The distribution of GnRH ISH positive neurons was examined in several teleost fishes and the results corresponded to those obtained by immunocytochemical investigation. We examined sGnRH mRNA expression of salmonid fishes, and found that sGnRH neurons in the different brain regions have different functions according to their location.
The function of sGnRH in the brain of teleost fish, especially in salmonids, has been partly elucidated by use of RIA, immunocytochemistry and ISH. However, the function of cGnRH-II, which is distributed in almost all of the teleost fishes examined so far, remains unclear, except for the stimulation of GTH and GH release in vitro.
It has been proposed that GnRH-ir neurons are derived from the olfactory placode and migrate to the forebrain during prenatal development in the mouse (mGnRH; Schwanzel-Fukuda and Pfaff, 1989; Wray et al., 1989), the chicken (probably cGnRH-I, Murakami et al., 1991), and the newt, Cynops pyrrhogaster (probably mGnRH, Murakami et al., 1992). In chum salmon and sockeye salmon, GnRH (probably sGnRH) neurons originate in the olfactory placode and then migrate into the brain along the ON (Chiba et al., 1994; Parhar et al., 1995). Recently, it has been reported that cGnRH-II are not olfactory placode origin in the axolotl, Ambystoma mexicanum (Northcutt and Muske, 1994). cGnRH-II may not originate from the olfactory placode in teleost fish as well. Studies on the ontogeny of the GnRH system may be helpful in understanding the function of GnRH in the brain.
Our research cited in this review was obtained at the Department of Aquatic Bioscience, Graduate School of Agricultural and Life Sciences, The University of Tokyo, and the Nikko Branch of the National Research Institute of Aquaculture. It was supported in part by grants from the Ministry of Education, Culture and Science, and the Fisheries Agency, Japan. We thank Professor Emeritus Seiichiro Kawashima and Associate Professor Yoshitaka Oka, The University of Tokyo, for instruction in immunocytochemistry. We are indebted to Dr. Susumu Hyodo, The University of Tokyo, for giving us the opportunity to study in situ hybridization. We are also thankful to Professor Yoshihisa Hasegawa and Professor Hiroshi Kawauchi, Kitasato University, for providing us the antisera against cGnRH-II and GTH (GTH Iβ and GTH IIβ). We also thank Dr. Marcy N. Wilder, Japan International Research Center for Agricultural Sciences, for reading the manuscript.