The Olfactory Organ ModulatesGonadotropin-Releasing Hormone Typesand Nest-Building Behavior in theTilapia Oreochromis niloticus

Hiroshi Uchida,1,2 Satoshi Ogawa,1 Mina Harada,1 Masato Matushita,1

Munehico Iwata,2 Yasuo Sakuma,1 Ishwar S. Parhar1

1 Department of Physiology, Nippon Medical School, Tokyo 113-8602, Japan

2 Laboratory of Ecophysiology, Kitasato University, Sanriku, Iwate 022-0101, Japan

Received 18 August 2004; accepted 15 February 2005

ABSTRACT: Direct olfactory inputs to any of the

known gonadotropin-releasing hormone (GnRH) con-

taining neurons have not been demonstrated. Therefore,

the rationale of this study was to examine whether olfac-

tory inputs might in some way interact with the GnRH

system(s) to synchronize reproductive behaviors. In

order to establish this, we used anosmic mature male

tilapia to investigate changes in reproductive behaviors,

gonadal morphology, and GnRH1, GnRH2, and GnRH3

cellular morphology and change in GnRH mRNA levels

by real-time polymerase chain reaction. Bilateral

removal of the olfactory rosettes followed by occlusion

of the nasal cavity (ORX) inhibited nest-building behav-

ior, but had no effect on aggressive and sexual behaviors

or gonadal morphology. ORX failed to alter the mor-

phological features of GnRH1, GnRH2, and GnRH3

(cell number, size, GnRH optical density), but signifi-

cantly decreased copies of GnRH1 and GnRH2 mRNAs.

GnRH immunoreactive fibers were not evident in the

olfactory nerve and rosettes. DiI application to the olfac-

tory nerve labeled inputs primarily to the glomerular

layer of the olfactory bulbs and extrabulbar inputs to

the forebrain but not to GnRH neurons. These results

provide evidence that the olfactory rosette is crucial for

modulating nest-building behavior through second-

order olfactory pathways interacting with GnRH1 and

GnRH2 neuronal systems. ' 2005 Wiley Periodicals, Inc. J Neu-

robiol 65: 1–11, 2005

Keywords: terminal nerve; preoptic; midbrain; aggressive;

sexual behavior

INTRODUCTION

Fish do not possess a distinct vomeronasal organ

(Eisthen, 1992). Therefore, chemical cues in fish are

mediated primarily by the olfactory system express-

ing both the odorant and pheromone receptors that in

mammals are respectively segregated to the olfactory

epithelium–main olfactory bulb and the vomeronasal

organ–accessory olfactory bulb system (Dulka, 1993;

Bargmann, 1997; Westberry and Meredith, 2003).

Another neuronal system that has been proposed to

be chemosensory in nature in teleosts is the terminal

nerve (Demski and Northcutt, 1983); a bundle of

fibers and cells immunoreactive to gonadotropin-

releasing hormone (GnRH), which runs along the

olfactory nerve (Schwanzel-Fukuda and Silverman,

1980; Wirsig-Wiechmann et al., 2002) with cell

bodies located at the caudal-most part of the olfactory

Correspondence to: Ishwar S. Parhar (ishwar@nms.ac.jp).Contract grant sponsor: Ministry of Education, Culture, Sports,

Science, and Technology; contract grant numbers: 14580777 (I.S.P.)and 14370025 (Y.S.).

' 2005 Wiley Periodicals, Inc.Published online 7 July 2005 in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/neu.20156

1

bulbs (Parhar, 2002). Hence, there exists a possible

relationship between the olfactory chemosensory

inputs and the GnRH system(s), although the chemo-

sensory nature of the terminal nerve in teleosts

remains debatable (Fujita et al., 1991). In addition to

the terminal nerve GnRH (¼ GnRH3 in teleosts),

more recently derived teleosts possess two other

GnRH types in the brain: GnRH1 (preoptic area) and

GnRH2 (midbrain) whose precise functions remain

unclear (Parhar, 1997, 2002). Therefore, what

remains to be establish is whether olfactory inputs

might in some way interact with the GnRH system(s)

to synchronize reproductive behaviors.

In the present study, we examined the role of the

olfactory system in the control of reproductive behav-

iors and GnRH1, GnRH2, and GnRH3 types. For this

purpose, we used males of tilapia, Oreochromis nilot-cus, because they display distinctive reproductive

behaviors (nest-building and aggression). We surgi-

cally removed the olfactory rosettes and occluded the

nasal cavity, and observed reproductive behaviors

and gonadal morphology. Using immunocytochemis-

try and real-time quantitative reverse transcriptase

polymerase chain reaction (RT-Q-RT-PCR), we

examined morphological changes and mRNA levels

of GnRH1, GnRH2, and GnRH3 types.

MATERIALS AND METHODS

Animals

Experimental procedures were performed under the guide-

lines of the Animal Care Committee of Nippon Medical

School. Tilapia Oreochromis nilotcus were kept in a large

community tank (114 L) under controlled water tempera-

ture (27 6 18C) and photoperiods (10:14 h dark/light

cycle). Animals were fed, once per day, with commercial

fish pellets (Asakaze mini, Nippon-Haigoshiryo, Kanagawa,

Japan).

Sexually mature male fish that were aggressively

defending territories were selected for experiments. The

olfactory rosettes were heat lesioned for 15 min (electrode

diameter ¼ 0.7 mm; Radiofrequency lesion generator sys-

tem, RFG-4A, Radionics, Inc., MA, USA) and the resulting

dissociated cells and tissue were drawn using a Pasteur

capillary pipette (Corning Glass Works, Corning, NY,

USA). The nasal cavity was filled with dental cement

(Repairsin, GC Corporation, Tokyo, Japan) and sealed

using an adhesive (Aron� 201; Toua-gousei, Tokyo, Japan).

All operations were performed on fish anesthetized in

0.01% solution of tricaine methane-sulfonate (MS222;

Sigma, St. Louis, MO, USA). After the operation, fish were

kept individually in 40-L tanks for 7 days to recover. Fish

with bilaterally removed olfactory rosettes followed by

occlusion of the nasal cavity (ORX) and control fish with

intact olfactory rosettes were segregated into three experi-

mental groups.

Experiment 1 . ORX males (standard length, SL ¼ 14.4

6 0.9 cm; body weight, BW ¼ 76.46 13.7 g, n ¼ 5), intact

males (SL ¼ 13.8 6 1.0 cm, BW ¼ 77.1 6 6.0 g, n ¼ 6),

and ovariectomized prostaglandin injected females (SL

¼ 13.7 6 0.3 cm; BW ¼ 76.7 6 5.5 g) were kept for a 70-

dayssurvival period and observed for nest-building, aggres-

sive, and sexual behavior (see below for details about ovar-

iectomy and behavioral test). To minimize bias in behavio-

ral testing, each female or intact male had an equal opportu-

nity to be paired with a different ORX male. Each ORX

male was tested for sexual behavior with a female and

aggressive behavior with an intact male. Observations for

nest-building behavior were conducted daily, however,

observations for aggressive and sexual behaviors were con-

ducted on different days but at a particular time of the day

(1400–1700 h), which needed more time; therefore animals

were kept for a 70-day survival period. Because the olfac-

tory epithelium has the unique capacity to reconstitute near-

normal its population of neurons and non-neuronal cells

after near-complete destruction (Schwob et al., 1999), ORX

was performed three times (days 0, 30, and 45) during the

70-day survival period to ensure remnants of the olfactory

rosettes did not regenerate. At the end of the 70-day experi-

mental period, the brains were dissected out and immunor-

eacted with specific GnRH antibodies, and changes in cell

morphological features were analyzed. The fact that fish

actively building nests also aggressively defend their terri-

tories suggests nest-building and aggressive behaviors

might be regulated through a common neuronal pathway.

However, to our surprise the results in Experiment 1

showed a significant decrease in the number of ORX fish

making nests without change in their aggressive behavior,

which prompted us to conduct Experiments 2 and 3 to

examine if we could repeat these results; test for sexual

behavior was omitted. Therefore, Experiments 2 and 3 had

a short survival period (14 days).

Experiment 2. ORX males (SL ¼ 12.6 6 0.9 cm, BW

¼ 65.0 6 12.0 g, n ¼ 5) and intact males (SL ¼ 13.1 6 0.6

cm, BW ¼ 70.0 6 11.7 g, n ¼ 6) were kept for a 14-day

survival period and observed for nest-building and aggres-

sive behavior. The brains were immunoreacted with spe-

cific GnRH antibodies, and changes in GnRH1, GnRH2,

and GnRH3 cell morphological features were analyzed.

Experiment 3. ORX males (SL ¼ 12.6 6 0.9 cm, BW

¼ 61.2 6 12.2 g, n ¼ 6) and intact males (SL ¼ 13.1 6 0.7

cm, BW ¼ 63.6 6 6.9 g, n ¼ 5) were kept for a 14-day sur-

vival period and observed for nest-building and aggressive

behavior. The brains were analyzed for GnRH1, GnRH2,

and GnRH3 mRNA levels.

Behavioral Observations

Behavioral tests were conducted in twelve 40-L (liters)

observation aquaria each containing a gravel bed (�2 cm

2 Uchida et al.

thick). Observation aquaria (45 � 30 � 30 cm) containing

40 L of water were covered along the sides with opaque

sheets of paper to prevent visual contact between aquaria,

and arranged in two rows of six on a stand so that activity

in two aquaria could be recorded simultaneously by one

video camera. To observe aggressive behaviors in tilapia

niloticus (BW ¼ 50–70 g; SL ¼ 10–15 cm), 40–60-L tanks

are ideal (personal observations). Nest-building, aggressive,

and sexual behaviors were observed between 1400 and

1700 h, i.e., the time when aggressive behavior is at the

peak under natural daylight conditions in tilapia niloticus(unpublished observations) and tilapia mossambicus(Munro and Singh, 1987). To minimize bias in testing, each

female or intact male had an equal opportunity to be paired

with a different ORX male. Each ORX male was tested for

sexual behavior with a female or for aggressive behavior

with an intact male.

Nest-Building Behavior. The gravel bed in each tank

allowed the male fish to make nest. During nest-building,

the male fish picks up large amounts of gravels with the

mouth and deposits them away from the cleaned surface,

resulting in a clear depression in the gravel substrate near

the chosen spawning site. The nest is generally circular and

the glass bottom of the tanks becomes clearly visible. The

nest size was scored as 3 points (>15 cm), 2 points (5–

15 cm), 1 point (1–5 cm), or 0 point (no nest). The tanks

were observed for the presence of nest once a day (2 PM),

after which the nests were destroyed.

Aggressive Behavior. In Experiment 1, to avoid bias in

testing, aggressive behavior was conducted in the ORX

male’s tank and again on a different day in the intact male’s

tank; in Experiments 2 and 3, aggressive behavior was con-

ducted in an unfamiliar tank (neutral tank) to both the ORX

and the intact male. Behavioral observations were limited

to 15 min duration, during which the dominant male (hold-

ing spawning site) was pinkish/red in coloration and the

defeated fish changed its body coloration to dark pigmenta-

tion (‘‘zebra-like’’ pattern) and became motionless at the

surface of the water. For each fish, behavior was observed

on alternate days. N ¼ 5/6 observations/fish. Aggressive

behavior was scored on a 3-point scale: winner (2 points),

equal (1 point) and loser (0 point).

Male Sexual Behavior. Sexually mature females (SL

¼ 13.76 0.3 cm; BW ¼ 76.76 5.5 g) were ovariectomized

and kept for a week to recover. Prostaglandin F2� [10 �g/2 �L; a dose that should induce female spawning behavior

for at least 1 h postinjection in the goldfish (Stacey and

Kobayashi, 1996) and in a cichlid fish (Cole and Stacey,

1984) was injected intraperitoneally into ovariectomized

females 30 min before being exposed to ORX males. Male

sexual behavior such as nudging (‘‘gentle bites’’: male pla-

ces his mouth close to the female’s anal region and gently

nudges the female) or leading (the male swims ostenta-

tiously in front of the female, leading her to the spawning

site), courtship (the male quivers sideways and circles

around the female), and foraging behavior (the fish picks up

small amounts of gravel, which it ‘‘chews’’ awhile and

spits out again, appears searching for food) were video

recorded for 10 min, and the number of times each act was

performed was counted. Male sexual behavior was

observed only in Experiment 1.

Tissue Fixation

At the end of the survival period, male fish from Experi-

ments 1 and 2 were anesthetized by immersing in 0.01%

solution of tricaine methane-sulfonate (MS222; Sigma,

St. Louis, MO, USA) and sacrificed by rapid spinal transec-

tion. The brains and testis tissues were removed and fixed

in Bouin’s solution for 20 h at room temperature. The tis-

sues were then dehydrated through a graded series of etha-

nols, cleared in n-butanol, and embedded in Paraplast Plus

(Oxford Labware, St. Louis, MO, USA). Serial brain sec-

tions were cut in the coronal plane (15 �m) and processed

for GnRH1, GnRH2, and GnRH3 immunocytochemistry;

testis sections (5 �m) were processed for routine histology

using cresyl violet staining.

GnRH Peptides: Immunocytochemistry

The brain sections were deparaffinized in xylene, rehy-

drated through graded ethanols to phosphate-buffered saline

(0.01 M PBS; pH 7.6), and incubated in 0.03% solution of

H2O2 in 20% methanol for 15 min in the dark. Primary anti-

serum (anti-mammalian GnRH, no. 635.5: 1:3500; anti-

chicken GnRH II, ISP II: 1:3000 in 0.01 M PBS) with 0.3%

Triton X-100 was applied to each section, and slides were

incubated in a closed moist chamber for 36 h at 48C. Thesections were then incubated in biotinylated anti-rabbit

immunoglobulin G (IgG) for 1 h and in avidin-biotinylated

horseradish peroxidase complex for 1 h (Vectastain ABC

Elite kit, Vector Laboratories, Burlingame, CA, USA). Sec-

tions were rinsed in 0.05 M Tris, pH 7.6 (Sigma) and devel-

oped with 0.05% 3,30 diaminobenzidine tetrahydrochloride

(Sigma) with 0.03% H2O2 in 0.05 M Tris buffer for 15 min.

To stop the reaction, sections were rinsed in tap water.

Dehydrated sections were cleared in xylene and coverslips

were applied with Permount (Fisher Chemical, NJ, USA).

The specificities of GnRH antisera used in this study have

been reported previously (Parhar et al., 1996).

GnRH Morphological Data Analysis

GnRH cell numbers, cell area, and cell optical density were

determined by digitizing images of the cells with a micro-

scope (X 400; Leica Microsystems, Model DM-RXA)

equipped with a CCD video camera (Model 1300-Y/HS;

Roper Scientific, Tucson, AZ, USA). Two-dimensional area

was calculated for all stained cells in the plane of the sec-

tion, whose perimeter and nucleus was discernible. Outlines

of stained cells were traced by moving a mouse-controlled

cursor along the digitized image to remove all cell proc-

esses. The cell size and optical density of staining were ana-

Olfactory Organ Modulates GnRH 3

lyzed using an image-processing program (Leica Quantimet

500). The imaging system was set to predetermined settings

on the microscope, so that the objective, the light intensity,

the openings of the condenser, and base diaphragms center-

ing the light for the condenser on the specimen and the con-

denser height were all used at constant settings. The optical

density of staining for each GnRH cell was corrected by

subtracting the background value. The average optical den-

sity of staining per cell was calculated from values taken

from all GnRH-immunoreactive cells in alternate sections

throughout the brain. Although semiquantitative, the esti-

mates of optical density were highly reproducible. To

reduce variability in immunocytochemical results, the con-

ditions of the immunocytochemical reactions were con-

trolled and kept homogeneous on all parameters. The

results were analyzed using Student’s t test.

GnRH mRNAs: Real-Time PCR

ORX male fish from Experiments 3 were anesthetized by

immersing in 0.01% solution of MS222 (Sigma) and sacri-

ficed by rapid spinal transaction at the end of the 14-day

survival period. The brains were removed and placed on

dry ice and stored at �808C. Total RNA was isolated from

whole brain of tilapia O. niloticus using Isogen (Nippon

Gene, Tokyo, Japan). First-strand cDNA was synthesized

from 5 �g of total RNA using Superscript III Reverse Tran-

scriptase (Invitrogen Corp., Carlsbad, CA, USA) and

50 pmol of oligo(dT)18–21 primer (Invitrogen) according to

the manufacturer’s instructions.

RT-Q-RT-PCR was performed in duplicate in 10 �Lreaction volumes consisting of 1X TaqMan Universal PCR

Master Mix (Applied Biosystems, Foster City, CA), 300

nM of primers (G1, G2, G4, G5, G7, G8, A1, A2; Table 1),

200 nM of hybridization probe (G3, G6, G9, A3; Table 1),

and 1/20 of a reverse transcribed sample cDNA or standard

cDNA using the ABI PRISM 7700 Sequence Detection

System (TaqMan PCR, PE Applied Biosystems). The PCR

conditions were 958C for 10 min, followed by 75 cycles at

958C for 15 sec, 608C for 1 min. Hybridization primers and

fluorogenic probes for RT-Q-RT-PCR were optimized

using the ABI PRISM Primer Express Software (Applied

Biosystems). The GenBank accession numbers of the

primers and probes of GnRH1, GnRH2, and GnRH3

types and �-actin in tilapia are as follows: GnRH1,

AB101665; GnRH2, AB101666; GnRH3, AB101667; �-actin, AB037865.

For quantification of transcripts by RT-Q-RT-PCR,

standard curves were generated following the method of

Fronhoffs et al. (2002). Briefly, cRNA of target genes

(GnRH1, GnRH2, GnRH3, and �-actin) were synthesized

by in vitro transcription. After linearization and removal of

the plasmid vector, the total cRNA was diluted in 20 �L of

RNase-free water and quantified using a spectrophotometer

at 260 nm (GeneQuant pro ‘‘S,’’ Amersham Pharmacia

Biotech, NJ, USA). Serial dilutions of cRNA of the target

genes (104–108 molecules), reverse-transcribed to cDNA,

were used to determine the threshold cycle at which the flu-

orescence intensity became delectable. Using an analysis

software (TaqMan PCR, ABI PRISM 7700 Sequence

Detection System, PE Applied Biosystems, Foster City,

CA, USA), a standard curve was generated by plotting the

threshold cycles against the serially diluted known copy

numbers of transcripts. The unknown copies of transcripts

in the samples were calculated from the standard curve

using the number of threshold cycles at which the tran-

scripts became detectable. Copy numbers of GnRH tran-

scripts were normalization with �-actin. Quantification data

are given as mean 6 SEM followed by Student’s t test.

Carboxycyanine Dye Application to theOlfactory Nerve

Sexually mature male and female tilapia (BW ¼ 45.56 3.3 g;

SL ¼ 11.4 6 0.2 cm; n ¼ 7) were anesthetized in 0.01%

solution of MS222 (Sigma). Carbocyanine dye 1,10-diocta-decyl-3,3,3,30-tetramethylindocarbocyanine perchlorate (DiI,

Molecular Probes, Inc., Eugene, OR) crystals were applied

to the right olfactory nerve and the nasal cavity was filled

with dental cement (GC Corporation) in order to keep the

DiI crystal from dislocating, and sealed using an adhesive

(Toua-gousei). After DiI application, animals were sacri-

Table 1 Sequence of GnRH1, GnRH2, and GnRH3 Primers and Fluorogenic Probes Used in TaqMan Quantitative PCR

Primer Sequence Code

Q-GnRH1-F 50-CTCGCAGGGACGGTGTTT-30 G1

Q-GnRH1-R 50-TCTTCCCTCCTGGGCTCAGT-30 G2

Q-GnRH1 Probe 50-CACAGGGCTGCTGTCAACACTGGTCATA-30 G3

Q-GnRH2-F 50-TGGTCCCATGGTTGGTATCC-30 G4

Q-GnRH2-R 50-CCCTGCTTCACACAGCTTAATCT-30 G5

Q-GnRH2 Probe 50-AAATCTCTGATGTCCCAAAGGAGTCCAGCT-30 G6

Q-GnRH3-F 50-TGCTGGCGTTGGTGGTT-30 G7

Q-GnRH3-R 50-CCTCAAGCTCTCCCACACTTCT-30 G8

Q-GnRH3 Probe 50-CAGCACTGGTCCTATGGATGGCTACC-30 G9

Q-�-actin F 50-CCTGACAGAGCGTGGCTACTC-30 A1

Q-�-actin R 50-TCTCTTTGATGTCACGCACGAT-30 A2

Q-�-actin Probe 50-TTCACCACCACAGCCGAGAGGGA-30 A3

All probes were labeled with 50 FAM reporter dye and 30TAMARA quencher dye.

4 Uchida et al.

ficed by rapid spinal transection. The brains were kept

in situ in the brain cases, but the skulls were carefully

opened to give better access to the fixative. Brains were

exposed in 4% paraformaldehyde and kept in the dark at

428C for 2–12 weeks.

Before cryostat sectioning, the heads were cryoprotected

by overnight infiltration with 20% sucrose in phosphate

buffer (pH 7.6). The brains were excised, embedded in Tis-

sue Tek OCT Compound (Sakura Finetechnical, Tokyo,

Japan), and serial 25-�m sagittal sections were cut and thaw

mounted on gelatin-coated slides. The sections were air

dried, evaluated for DiI fluorescence either with or without

coverslipping with PBS, and further processed for GnRH1,

GnRH2, and GnRH3 fluorescence immunocytochemistry.

RESULTS

Behavior Observations

Nest-Building Behavior. In all three experiments

there was a significant decrease in the number of

ORX fish making nests and the nest size when com-

pared to intact males (P < 0.01 and P < 0.001) [Fig.

1(A,B)].

Aggressive and Sexual Behavior. There was no dif-

ference in aggressive and foraging behavior or any of

the parameters of sexual behavior (nudging, court-

ship, leading) when compared between ORX and

intact males (Fig. 2).

Testis Morphology

Histologically, testes sections of ORX fish had germ

cells at all stages of spermatogenic activity and the

lobules were filled with spermatozoa. There was an

abundance of primary and secondary spermatocytes,

spermatids, and mature spermatozoa in the testis of

ORX fish similar to those in intact males [Fig. 3(A)].

GnRH Peptides: Immunocytochemistry

Three distinct GnRH cell populations [GnRH1,

GnRH2, GnRH3; Fig. 3(B–D)] were observed in the

brain of tilapia. GnRH1 cell bodies were fusiform in

shape, approximately 5–15 �m in diameter, and

present in the preoptic area (nucleus anteriorisrecesses). GnRH2 cell bodies were ovoid in shape,

approximately 15–40 �m in diameter, and present

immediately caudal to the posterior commissure at

the level of the midbrain. GnRH3 cell bodies were

round in shape, approximately 10–30 �m in diameter,

and localized in the transitional area between the

olfactory bulb and the telencephalon. In Experiment

1, GnRH cells were lightly immunoreactive due to

processing problems and were therefore excluded

from the analysis. In Experiment 2, GnRH immunor-

eactive fibers were seen within the brain but not in

the olfactory nerve and olfactory rosettes (data not

shown). There was no difference in GnRH1, GnRH2,

and GnRH3 cell numbers, cell size, and cell optical

density when compared between ORX and intact

males [Fig. 4(A–C)].

GnRH mRNAs: Real-Time PCR

Using RT-Q-RT-PCR, significantly lower copy num-

bers of GnRH1 (p < 0.05) and GnRH2 (p < 0.01)

transcripts were observed in ORX compared to intact

males. No change in GnRH3 transcripts was observed

(Fig. 5).

DiI Application to the Olfactory Nerve

Application of DiI into the right olfactory nerve

resulted in a bright red DiI fluorescence, which could

Figure 1 Histograms displaying (A) the number of fish

making nest and (B) the nest size in ORX (see Methods for

description of ORX) and intact males (black bars). The

asterisk indicates significant statistical difference from

intacts. *P < 0.01 and *P < 0.001, Student’s t test.

Olfactory Organ Modulates GnRH 5

be traced through the olfactory nerve into the olfac-

tory bulb. Heavily stained fibers were seen in the

glomerular layer of the olfactory bulbs, ventral tele-

ncephalon, preoptic area, basal hypothalamus, dorsal

thalamus and the habenular [Fig. 6(A–C)]. DiI-labeled

projections were not visible in the proximity of GnRH

neurons (data not shown).

DISCUSSION

Primary and Secondary Inputs from theOlfactory Epithelium

In the present tract-tracing study, primary olfactory

inputs were observed projecting from the olfactory

epithelium mainly to the glomerular layer of the

olfactory bulb, but a significant number of extrabul-

bar olfactory inputs extended past the olfactory bulbs

to various telencephalic, diencephalic, and mesence-

phalic regions as in other teleost (Honkanen and

Ekstrom, 1990; Riddle and Oakley, 1992; Becerra

et al., 1994; Anadon et al., 1995; Hofmann and

Meyer, 1995; Nieuwenhuys, 1998; Huesa et al.,

2000). The extrabulbar inputs have been considered

as the terminal nerve components immunoreactive to

GnRH1 (Schwanzel-Fukuda and Silverman, 1980;

Wirsig-Wiechmann et al., 2002) or GnRH3 in teleost,

which are closely associated with the olfactory sys-

tem and the forebrain (Demski and Northcutt, 1983;

Demski, 1984; Parhar, 2002). However, in the case of

tilapia, the distribution pattern of GnRH3 immunor-

eactive fibers was different from the DiI-labeled pro-

jections in the brain. Therefore, the possibility that

the GnRH3 system is part of extrabulbar olfactory

inputs seems unlikely in the tilapia.

Role of Olfactory Inputs in theRegulation of Reproductive Behaviors

Although nest-building behavior is an integral part

of aggressive behavior, ablations of the olfactory

Figure 2 Histograms displaying male aggressive behavior (left panel), foraging behavior (middle

panel), and sexual behavior (right panel, A–C). There was no statistical difference in aggressive,

foraging, and sexual behavior between ORX and intact males (black bars).

Figure 3 (A) Testis of ORX tilapia stained with cresyl

violet. ORX fish have cells at all stages of spermatogenic

activity—primary spermatocytes (psc), secondary sperma-

tocytes (ssc), and spermatids (st)—and the lobules are filled

with mature spermatozoa (sz) similar to those in intact

males. (B) GnRH immunoreactive neurons seen in the pre-

optic area (GnRH1), (C) midbrain tegmentum (GnRH2),

and (D) terminal nerve (GnRH3). Scale bar ¼ 20 �m.

6 Uchida et al.

rosettes abolished nest-building but not aggressive

behavior. Therefore, it is likely that in tilapia differ-

ent pathways regulate these two behaviors and that

the olfactory inputs are responsible for the regulation

of motivational rather than the consummatory aspects

of male behaviors.

ORX fish showed no sign of deficiency in any

aspect of male sexual behaviors. Olfactory signals

may not be important for mechanical aspects of

sexual behaviors (e.g., nudging, leading, courtship),

consistent with anosmic male goldfish, which fol-

lowed ovulated females as much as they followed

nonovulated females (Partridge et al., 1976). Fur-

thermore, olfactory signals did not influence any

physiological aspects of sexual behavior such as

spermatogenesis in ORX males, which provides

morphological evidence to support that ORX males

were reproductively fertile. Olfactory tract manipu-

lations have been reported to disrupt ovulation

(musk shrew: Rissman and Li, 2000), spermiation

and cause infertility with reduced levels of gonado-

tropins in the goldfish (Dulka et al., 1987). How-

ever, the present study was not design to observe

sperm release and gonadotropin release. Therefore,

Figure 4 Histograms displaying the cell number and the percentage of labeled cells of different

size and different optical density of GnRH1 (A), GnRH2 (B), and GnRH3 (C). The mean size and

mean optical density (6SE) of GnRH1, GnRH2, and GnRH3 per brain are shown in the inset. No

statistical difference was seen in any of these measures between ORX and intact males (black

bars).

Olfactory Organ Modulates GnRH 7

there remains the possibility that different pathways

regulate different components of male sexual

behavior (Weltzien et al., 2003), which is supported

by the fact that anosmia has little effect on the

spawning success of female blue gourami that are

already in visual contact with males (Pollack et al.,

1978). Thus, it is likely that the olfactory system

together with other sensory systems (visual, tactile)

play an integrated role in the control of sexual

behaviors in male tilapia.

Role of Olfactory Inputs in theRegulation of GnRH Types

GnRH1, GnRH2, and GnRH3 immunoreactive effer-

ents were not detectable in the olfactory rosettes or

the olfactory nerve. On the contrary, olfactory inputs

from the olfactory epithelium were seen projecting to

the olfactory bulbs, ventral telencephalon, the pre-

optic area, basal hypothalamus, dorsal thalamus, and

the habenular. These brain regions are known to con-

tain GnRH cell bodies and/or GnRH terminal fields

(Silverman et al., 1994; Parhar, 1997). Since olfac-

tory projections were not visible in the proximity of

GnRH neurons, the possibility that primary olfactory

inputs project to GnRH terminal fields or second-

order olfactory pathways to GnRH neurons/terminal

fields could account for the down-regulation of

GnRH1 and GnRH2 transcripts levels in ORX fish.

However, ORX significantly reduced copies of

GnRH1 and GnRH2 transcripts but failed to influence

GnRH1 and GnRH2 cell numbers, cell size, and cell

optical density. These unexpected paradoxical find-

ings between transcripts and peptide levels are prob-

ably due to the sensitivities of the techniques; RT-Q-

RT-PCR being more sensitive to detect small changes

in mRNA but cell optical density, a semiquantitative

method, does not detect small changes in GnRH pep-

tide.

The regulation of GnRH3 is probably independ-

ent of the olfactory organ, since destruction of the

olfactory rosettes did not change GnRH3 mRNA

and peptide levels. This came as a surprise, since

the terminal nerve (¼GnRH3 in teleosts) is associ-

ated with the olfactory organ (Parhar, 2002) and

removal of the olfactory rosettes would damage the

GnRH3 system. Since adult tilapia lacked GnRH

fibers along the olfactory nerve/rosettes, GnRH3

neurons in tilapia may not be the terminal nerve

ganglia by definition (Wirsig-Wiechmann et al.,

2002); rather, it is the nucleus olfactoretinalis

(Szabo et al., 1991) that has projections to the ret-

ina (Munz et al., 1982) and play an integrated role

with the visual system. In addition, the activity of

the GnRH3 system does not change when exposed

to odorants or chemical stimuli (Bullock and North-

cutt, 1984); instead, the olfactory system (Fujita

et al., 1991; Huesa et al. 2000) and/or a population

of neurons, present in the vicinity of GnRH3 neu-

rons, might be chemosensory in teleost (Laberge

and Hara, 2003).

Role of GnRH Neurons inReproductive Behaviors

GnRH1 neurons in the forebrain (Pfaff, 1973) or

administration of GnRH1 into the midbrain elicit sex-

ual behavior in female rats (Sakuma and Pfaff, 1980).

More recently, GnRH2 has been reported in the mid-

brain of many vertebrate species, which is more

potent than GnRH1 in the regulation of gonadotro-

pins and reproductive behaviors in the goldfish and in

other vertebrates (Parhar, 2002; Millar, 2003; Volkoff

and Peter, 1999; Rissman, 1996). In the present study,

ablations of olfactory rosettes inhibited GnRH1 and

GnRH2 transcripts and nest-building behavior possi-

bly through primary or second-order olfactory path-

ways to GnRH neurons/terminal fields that have been

implicated in reproductive behaviors in vertebrates

(Silverman et al., 1994; Rissman, 1996; Parhar,

1997). Interestingly, chronic social stress (Ogawa

et al., 2003) or ablations of olfactory rosettes down-

regulate GnRH1 and GnRH2 in tilapia, which suggest

that olfactory inputs and the hypothalamic–pituitary–

adrenal axis might be part of a common integrated

circuitry that regulates GnRH1 and GnRH2 neuronal

system. In another recently derived teleost, the dwarf

gourami, lesions of GnRH3 neurons inhibited nest-

building behavior (Yamamoto et al., 1997); whether

these lesions include destruction of secondary olfac-

Figure 5 Histograms displaying copies of GnRH1,

GnRH2, and GnRH3 transcripts. The asterisk indicates sig-

nificant statistical difference in GnRH1 and GnRH2 tran-

scripts between ORX and intact males (black bars). Stu-

dent’s t test *P < 0.01 and *P < 0.05.

8 Uchida et al.

tory inputs to GnRH1 and GnRH2 remains a possibil-

ity. GnRH3 neurons regulate aggressive behavior in

the tilapia (Akiyama et al., 2002) independent of

olfactory inputs, as seen in the present study but

might involve the visual system to which GnRH3

neurons send projections (Munz et al., 1982).

Figure 6 (A) Diagrammatic representation of the brain and the olfactory system of tilapia.

Note the terminal nerve (TN in yellow) is absent in tilapia. DiI projections (red) along the

olfactory nerve (ON) terminate in the golmerular layer (GL) of the olfactory bulbs (OB) and in

the forebrain and midbrain areas containing GnRH neurons and terminal fields. (B) Schematic

drawing of a sagittal section of the brain showing projections of DiI to the forebrain and mid-

brain. The dash lines indicate uncertain connections or second-order projections. (C) Bright red

DiI fluorescence seen along the olfactory nerve (ON) with heavily stained fibers in the glomeru-

lar cell layer (rectangle) of the olfactory bulb (OB) and extrabulbar projections seen in the tele-

ncephalon (TEL), preoptic area (POA), hypothalamus (HYP), and habenular (H). Arrows indi-

cate DiI projections along the basal hypothalamus. OC, optic chiasma; OE, olfactory epithelium;

MB, midbrain. Scale bar ¼ 20 �m.

Olfactory Organ Modulates GnRH 9

The Leica Image Analysis System was purchased with a

grant from the Japanese Ministry of Education and Science

for Independent Colleges and Universities (2000).

REFERENCES

Akiyama G, Kato S, Soga T, Tamano K, Kawai T, Sakuma

Y, Parhar IS. 2002. Social stress controls terminal nerve

GnRH neurons. Society for Behavioral Neuroscience,

Amherst, MA, USA, Abstract 179.

Anadon R, Manso MJ, Rodriguez-Moldes I, Becerra M.

1995. Neurons of the olfactory organ projecting to the

caudal telencephalon and hypothalamus: a carbocyanine-

dye labelling study in the brown trout (Teleostei). Neuro-

sci Lett 191:157–160.

Bargmann CI. 1997. Olfactory receptors, vomeronasal

receptors, and the organization of olfactory information.

Cell 90:585–587.

Becerra M, Manso MJ, Rodriguez-Moldes I, Anadon R.

1994. Primary olfactory fibres project to the ventral tele-

ncephalon and preoptic region in trout (Salmo trutta): adevelopmental immunocytochemical study. J Comp Neu-

rol 342:131–143.

Bullock TH, Northcutt RG. 1984. Nervus terminalis in dog-

fish (Squalus acanthias, Elasmobranchii) carries tonic

efferent impulses. Neurosci Lett 44:155–160.

Cole KS, Stacey NE. 1984. Prostaglandin induction of

spawning behavior in Cichlasoma bimaculatum (Pisces

cichlidae). Horm Behav 18:235–248.

Demski LS. 1984. The evolution of neuroanatomical sub-

strates of reproductive behavior: sex steroid and LHRH-

specific pathways including the terminal nerve. Am Zool

24:809–830.

Demski LS, Northcutt RG. 1983. The terminal nerve: a new

chemosensory system in vertebrates? Science 220:435–

437.

Dulka JG. 1993. Sex pheromone systems in goldfish: com-

parisons to vomeronasal systems in tetrapods. Brain

Behav Evol 42:265–280.

Dulka JG, Stacey NE, Sorensen PW, Van der Kraak GJ,

Marchant TA. 1987. A sex pheromone system in gold-

fish: is the nervus terminalis involved? Ann NY Acad Sci

519:411–420.

Eisthen HL. 1992. Phylogeny of the vomeronasal system

and of receptor cell types in the olfactory and vomero-

nasal epithelia of vertebrates. Micro Res Tech 23:

1–21.

Fronhoffs S, Totzke G, Stier S, Wernert N, Rothe M, Brun-

ing T, Koch B, Sachinidis A, Vetter H, Ko Y. 2002.

A method for the rapid construction of cRNA stan-

dard curves in quantitative real-time reverse transcrip-

tion polymerase chain reaction. Mol Cell Probes 16:

99–110.

Fujita I, Sorensen PW, Stacey NE, Hara TJ. 1991. The

olfactory system, not the terminal nerve, functions as the

primary chemosensory pathway mediating responses to

sex pheromones in male goldfish. Brain Behav Evol

38:313–321.

Hofmann MH, Meyer DL. 1995. The extrabulbar olfactory

pathway: primary olfactory fibers bypassing the olfactory

bulb in bony fishes? Brain Behav Evol 46:378–388.

Honkanen T, Ekstrom P. 1990. An immunocytochemical

study of the olfactory projections in the three-spined

stickleback, Gasterosteus aculeatus, L. J Comp Neurol

292:65–72.

Huesa G, Anadon R, Yanez J. 2000. Olfactory projections

in a chondrostean fish, Acipenser baeri: an experimental

study. J Comp Neurol 428:145–158.

Laberge F, Hara TJ. 2003. Non-oscillatory discharges of an

F-prostaglandin responsive neuron population in the

olfactory bulb-telencephalon transition area in lake

whitefish. Neuroscience 116:1089–1095.

Millar RP. 2003. GnRH II and type II GnRH receptors.

Trends Endocrinol Metab 14:35–43.

Munro AD, Singh I. 1987. Diurnal changes in the territorial

behavior of the tilapia Oreochromis (¼Sarotheradon)mossambicus. J Fish Biol 30:459–464.

Munz H, Claas B, Stumpf WE, Jennes L. 1982. Centrifugal

innervation of the retina by luteinizing hormone-releasing

hormone (LHRH)-immunoreactive telencephalic neurons

in teleostean fishes. Cell Tissue Res 222:313–323.

Nieuwenhuys R. 1998. Chondrostean fishes. In: Nieuwen-

huys R, Ten Donkelaar HJ, Nicholson C, editors. The

Central Nervous System of Vertebrates, Vol. 1. Berlin:

Springer, p 701–758.

Ogawa S, Soga T, Sakuma Y, Parhar IS. 2003. Modulation

of GnRH subtypes by social stress and aggressive behav-

ior. Fish Physiol Biochem 28:49–50.

Parhar IS. 1997. GnRH in tilapia: three genes, three origins

and their roles. In: Parhar IS, Sakuma Y, editors. GnRH

neurons: gene to behavior. Tokyo: Brain Shuppan, p 99–

122.

Parhar IS. 2002. Cell migration and evolutionary signifi-

cance of GnRH subtypes. Prog Brain Res 141:3–17.

Parhar IS, Pfaff DW, Schwanzel-Fukuda M. 1996. Gonado-

tropin-releasing hormone gene expression in teleosts.

Mol Brain Res 41:216–227.

Partridge BL, Liley NR, Stacey NE. 1976. The role of pher-

omones in the sexual behavior of the goldfish. Anim

Behav 24:291–299.

Pfaff DW. 1973. Luteinizing hormone-releasing factor

potentiates lordosis behavior in hypophysectomized

ovariectomized female rats. Science 182:1148–1149.

Pollack EL, Becker LR, Haynes K. 1978. Sensory control

of mating in the blue gourami, Trichogaster trichopterus(Pisces, Belontiidae). Behav Biol 22:92–103.

Riddle DR, Oakley B. 1992. Immunocytochemical identifi-

cation of primary olfactory afferents in rainbow trout.

J Comp Neurol 324:575–589.

Rissman EF. 1996. Behavioral regulation of gonadotropin-

releasing hormone. Biol Reprod 54:413–419.

Rissman EF, Li X. 2000. Olfactory bulbectomy blocks mat-

ing-induced ovulation in musk shrews (Suncus murinus).Biol Reprod 62:1052–1058.

Sakuma Y, Pfaff DW. 1980. LH-RH in the mesencephalic

central grey can potentiate lordosis reflex of female rats.

Nature 283:566–567.

10 Uchida et al.

Schwanzel-Fukuda M, Silverman AJ. 1980. The nervus ter-

minalis of the guinea pig: a new luteinizing hormone-

releasing hormone (LHRH) neuronal system. J Comp

Neurol 191:213–225.

Schwob JE, Youngentob SL, Ring G, Iwema CL,

Mezza RC. 1999. Reinnervation of the rat olfactory

bulb after methyl bromide-induced lesion: timing and

extent of reinnervation. J Comp Neurol 412:439–

457.

Silverman AJ, Livne I, Witkin JW. 1994. The gonadotro-

pin-releasing hormone (GnRH) neuronal systems: immu-

nocytochemistry and in situ hybridization. In: Knobil E,

Neill JE, editors. Physiology of Reproduction, 2nd ed.

New York: Raven Press, p 1683–1709.

Stacey N, Kobayshi M. 1996. Androgen induction of male

sexual behaviors in female goldfish. Horm Behav 30:

434–445.

Szabo T, Blahser S, Denizot J-P, Ravaille-Veron M. 1991.

The olfactoretinalis system ¼ terminal nerve? Neuro-

report 2:73–76.

Volkoff H, Peter RE. 1999. Actions of two forms of gona-

dotropin releasing hormone and a GnRH antagonist on

spawning behavior of the goldfish Carassius auratus.Gen Comp Endocrinol 116:347–355.

Weltzien FA, Hoglund E, Hamadani el H, Døving KB.

2003. Does the lateral bundle of the medial olfactory

tract mediate reproductive behavior in male crucian

carp? Chem Senses 28:293–300.

Westberry J, Meredith M. 2003. The influence of chemo-

sensory input and gonadotropin releasing hormone on

mating behavior circuits in male hamsters. Brain Res

974:1–16.

Wirsig-Wiechmann CR, Wiechmann AF, Eisthen HL.

2002. What defines the nervus terminalis? Neurochem-

ical, development, and anatomical criteria. Prog Brain

Res 141:45–58.

Yamamoto N, Oka Y, Kawashima S. 1997. Lesions of gona-

dotropin-releasing hormone-immunoreactive terminal

nerve cells: effects on the reproductive behavior of male

dwarf gouramis. Neuroendocrinology 65:403–412.

Olfactory Organ Modulates GnRH 11