Kisspeptin-1 regulates forebrain dopaminergic neurons in the zebrafish

The habenula is a phylogenetically conserved epithalamic structure, which conveys negative information via inhibition of mesolimbic dopamine neurons. We have previously shown the expression of kisspeptin (Kiss1) in the habenula and its role in the modulation of fear responses in the zebrafish. In this study, to investigate whether habenular Kiss1 regulates fear responses via dopamine neurons in the zebrafish, Kiss1 peptides were intracranially administered close to the habenula, and the expression of dopamine-related genes (th1, th2 and dat) were examined in the brain using real-time PCR and dopamine levels using LC–MS/MS. th1 mRNA levels and dopamine levels were significantly increased in the telencephalon 24-h and 30-min after Kiss1 administration, respectively. In fish administered with Kiss1, expression of neural activity marker gene, npas4a and kiss1 gene were significantly decreased in the ventral habenula. Application of neural tracer into the median raphe, site of habenular Kiss1 neural terminal projections showed tracer-labelled projections in the medial forebrain bundle towards the telencephalon where dopamine neurons reside. These results suggest that Kiss1 negatively regulates its own neuronal activity in the ventral habenula via autocrine action. This, in turn affects neurons of the median raphe via interneurons, which project to the telencephalic dopaminergic neurons.

Effect of Kiss1 on dopamine levels in macro-dissected brain. Dopamine (DA) levels were measured in three macro-dissected brain regions following Kiss1 administration (Fig. 3A). All groups including Kiss1-treated and controls treated with saline, absolute DA levels fluctuated time-dependently in all three brain regions ( Fig. 3B-D). In the telencephalic and preoptic-pretectal regions, basal levels of DA seen at 30-min and 1-h post saline administration were decreased at 3-h (telencephalon: P < 0.0001 vs 30-min and 1-h; preoptic area, P < 0.001 vs 30-min, P < 0.05 vs 1-h) but returned to similar levels at 6-h post saline administration (telencephalon: P < 0.0001; preoptic are: P < 0.001 vs 3-h). On the other hand, in the midbrain-hypothalamic region, DA levels continued to increase from 3-h until 6-h post saline administration (P < 0.0001 vs 30-min and 1-h; P < 0.001 vs 3-h).
Kiss1 administration differentially altered DA levels in time-dependent and brain region-dependent manner ( Fig. 3B-D).
Telencephalic region: Administration of Kiss1 at the dose of 10 -12 mol/fish significantly increased DA levels 30-min after the administration, which further continued to increase until 1-h post administration (P = 0.0018 and P = 0.0035 for 30-min and 1-h, respectively) (Fig. 3B). There was no difference in DA levels between Kiss1 at the dose of 10 -9 mol/fish and control at any time point.
Preoptic-pretectal region: There was no effect of Kiss1 administration on DA levels at 30-min, 3-h and 6-h post administration, except at 1-h post administration, where DA levels were significantly higher in fish treated with Kiss1 at the doses of 10 -12 mol/fish as compared to fish treated with 10 -9 mol/fish of Kiss1 (Fig. 3C).
Midbrain-hypothalamic region: Unlike the telencephalic or preoptic-pretectal regions, DA levels were significantly increased only 6-h after administration of Kiss1 at the dose of 10 -12 mol/fish (Fig. 3D).

Effect of Kiss1 administration on the habenula activity.
To confirm the effect of Kiss1 administration on the habenula activity, kiss1 gene expression was measured in the brain of fish at 30-min, 1-h and 24-h after Kiss1 administration. In fish treated with Kiss1 with dose of 10 -9 mol/ fish, kiss1 mRNA levels in 30-min post administration were significantly decreased as compared to vehicle controls (P = 0.0263, Fig. 4A). There was no effect of Kiss1 on kiss1 mRNA expression in 1-h and 24-h post administration. We also examined effect of Kiss1 on neural activity in the habenula using a neural activity marker gene, npas4a expression. In the fish treated with Kiss1 (10 -9 mol/ fish), number of cells expressing npas4a mRNA was significantly lower in the ventral habenula as compared to vehicle treated controls (P = 0.0008, Fig. 4B-D).
Neuronal activity in dopaminergic neurons followed by Kiss1 administration. To confirm whether and which dopaminergic neuronal population (Fig. 5a) is activated by Kiss1 administration, we examined the effect of Kiss1 administration on the expression of activity-dependent neural marker, npas4a in dopaminergic neurons using the transgenic Tg(dat:EGFP) zebrafish at 30-min post administration (Fig. 5b-e). Fluorescent in situ hybridization showed expression of npas4a mRNA in several brain regions including the brain regions containing dopaminergic neurons (Fig. 5b'-e'). However, there was no expression of npas4a in EGFPlabelled dopaminergic neurons in any brain regions including the olfactory bulb ( Fig. 5b")  Neural tracing from the median raphe. To reveal the potential link between habenular Kiss1 neurons and dopaminergic neurons, a thin-strip of nylon membrane coated with neural tracer was selectively applied to the MR where habenula Kiss1 neuronal terminals are located in transgenic Tg(kiss1:mCherry) zebrafish (Figs. 6a-l and 7A). As negative control, we used a brain without tracer implantation ( Fig. 6a-f) and a brain in which tracer was applied to the IPN region ( Fig. 7A'). In the brain implanted with a tracer-coated strip in the MR ( Fig. 6g-i), tracer-labelled neural tracts were seen in the fasciculus retroflexus towards the habenula (Fig. 6j-l).
In addition, observation of series of the brain sections ( Fig. 7B) revealed the presence of tracer-labelled nerve fibers in the medullary/brain stem regions, mesencephalic regions including the optic tectum and tegmentum, and in the optic chiasma ( Fig. 7c-e). Further, distinct neural tracts were labelled in the diencephalon, which are extended towards the telencephalon. This tract is known as the medial forebrain bundle (MFB, Fig. 7c,f). In the brain implanted with a tracer into the IPN (Fig. 7A'), tracer-labelled nerve fibers were seen in the mesencephalic and rhombencephalic regions including the optic tectum, tegmentum, and cerebellum, and in the medullary regions ( Fig. 7c'-f ') including the griseum centrale (GC, Fig. 7e').

Discussion
The role of LHb as a centre for processing negative stimuli and its-related decision-making through the modulation of midbrain dopaminergic neurons is well established in mammals 22 . Although the habenula structure and the dopamine system in the central nervous system are well conserved across the vertebrate species, the link between them remains to be identified in non-mammalian vertebrates. We have previously shown the expression of Kiss1 and Kiss1R in the vHb-MR 3,9 , and their role in fear responses 11 . In the present study, we elaborate further the possible link between habenular Kiss1 and dopaminergic neurons in the zebrafish. Administration of Kiss1 at the dose of 10 -12 mol/fish elevated the expression of dopamine-related genes 24-h post administration, while Kiss1 at the dose of 10 -9 mol/fish effected only th1 expression 3-h post administration in the whole brain. However, at the brain regional levels, expression of dopamine-related genes appeared www.nature.com/scientificreports/ to be differentially altered in different brain regions and different concentrations of Kiss1 solutions, suggesting that not all dopaminergic neurons are sensitive to Kiss1. th1 gene expression was upregulated only in the telencephalon and the hindbrain, and th2 gene expression was downregulated in the telencephalon, optic tectum and the hindbrain despite of the absence of dopaminergic neurons in some of the brain regions. On the other hand, the dopamine levels increased in the telencephalon, preoptic area and the hypothalamus after Kiss1 administration with peaks at different time points, which is inconsistent with gene expression results. At present, it is unclear how Kiss1 modulates dopamine-related genes and how these gene expression levels are associated with dopamine content levels at different times in different brain regions. Nevertheless, differential effects of Kiss1 on dopamine-related genes and dopamine content levels indicate that Kiss1 may have neuromodulatory effect on dopaminergic neurons. As in mammals, tyrosine hydroxylase encoded by th1 gene is responsible for dopamine production 23 and dopamine transporter encoded by dat is responsible for reuptake of extracellular dopamine in fish 24   www.nature.com/scientificreports/ Since all brain specimens were collected at a fixed time (1300 till 1500) of the day, the alteration of dopamine levels is not likely due to the circadian rhythmicity of dopamine release and turnover 25,26 . However, it is possible that the injection itself or the associated handling stress may have affected the dopamine content levels as the dopamine system is known to be influenced by stress 27,28 . On the other hand, in the zebrafish, th2 has been shown to modulate the synthesis of serotonin (5-HT) in vivo and has tryptophan hydroxylase activity in vitro 29 .
In fact, neurons immunoreactive for zebrafish TH2 are highly co-localized with 5-HT in the hypothalamus in the zebrafish 30 . Therefore, downregulation of th2 expression might have affected 5-HT rather than dopaminergic levels in the telencephalon, optic tectum and hindbrain regions. In addition, because tyrosine hydroxylase is also expressed in noradrenergic neurons 31 , Kiss1 administration may have also stimulated synthesis and release of norepinephrine.
To date, there has been no study that has shown morphological or physiological interaction between habenular Kiss1 and the dopaminergic system in any species. In addition, functional Kiss1Rs are not expressed in brain regions containing dopaminergic neurons in the zebrafish 3 . Although an association between hypothalamic kisspeptin and dopamine has been demonstrated in some species [32][33][34][35] , in the present study, administration of the Kiss1 paralogous peptide, Kiss2, which is mainly expressed in the hypothalamus in the zebrafish 9 , had no effect on the expression of dopamine-related genes. Therefore, the effect of Kiss1 on dopamine levels is likely due to the activation of habenular Kiss1-Kiss1R signalling and not Kiss1-Kiss2R signalling. In the brain of zebrafish, kiss1 and kissr1 mRNA are expressed in the vHb neurons, although a small amount of kissr1 mRNA was also detected in microdissected ventral IPN region 9,10 . Kiss1R protein is also expressed along Kiss1 neuronal projections derived from the vHb towards presynaptic terminals at the MR 3,9 . Zebrafish kissr1 (also referred to as kiss1rb) gene has been shown to produce 4 additional alternative splice variants encoding truncated forms of Kiss1R, which are expressed outside of the habenula 3,36 . However, none of the truncated Kiss1R forms have Kiss1 peptides-binding capabilities 36 . Since there is no expression of functional Kiss1R outside of vHb-MR pathway (postsynaptic zone of Kiss1 terminals or interneurons within the MR), Kiss1 may act on Kiss1 neurons in a closed-loop autocrine fashion 10 , which corresponds with the reduced kiss1 expression in the fish administered with Kiss1. We also found a significant reduction of npas4a expression in the vHb following exogenous Kiss1 administration 10 , indicating that administered Kiss1 peptides would have hyperpolarized vHb neurons 13 through Kiss1R. Hence, the inhibition of habenula Kiss1 neural activities and Kiss1 levels are likely to be induced by negative feedback of exogenous Kiss1 on habenula Kiss1 neurons. However, the exact location of action sites of Kiss1 within Kiss1 neurons (cell coma, axon or presynaptic terminals) and how Kiss1 neural terminal at the MR further transmit their neural activities downstream remain unknown.
To determine the neural activity of dopaminergic neurons, we utilised Npas4, as it has recently been recognized as a sensitive and rapid neuronal activity marker for both excitatory and inhibitory (e.g. glutamatergic and GABAergic) neurons in mammals and zebrafish 21,37-40 . However, we failed to detect co-expression of npas4a in dopaminergic neuronal populations following Kiss1 administration. Although Npas4 expression is selectively induced by neuronal activity, it is however, required when activity-dependent synapses are newly formed 21,38 . Therefore, although Kiss1 administration stimulates dopaminergic synthesis, but it may not have induced any synaptic formation in dopaminergic neurons. In addition, in rats, basal levels (without any stimuli) of Fos-related antigen expression is seen only in 10-20% of hypothalamic dopaminergic neurons 33 , and similarly in the zebra finches, basal levels of expression of Fos is seen only in 0.1-2% of preoptic/hypothalamic dopaminergic neurons 41 . Therefore, absence of npas4a in dopaminergic neurons could also be due to low levels of expression of neural activity markers in these neurons.
Although dopamine levels in all three macro-dissected brain regions increased following Kiss1 administration, they showed different pattern of responses (time and amplitude) against Kiss1 administration. The increment of th1 mRNA and dopamine levels in the telencephalic regions at 30-min and 1-h post administration of Kiss1 was the most apparent, while effect of Kiss1 on dopamine levels in the diencephalic region was only seen at 6-h post administration. In mammals, LHb neurons mainly innervate the tail of the ventral tegmental area (VTA, also known as rostromedial tegmental nucleus), which project heavily to midbrain dopamine neurons through the GABAergic pathway 42,43 . In zebrafish, dopaminergic neurons in the periventricular nucleus of posterior tuberculum in the diencephalic cluster are considered a homologue of mammalian midbrain dopaminergic population 44,45 . However, there is neither a direct innervation by Kiss1 fibers nor Kiss1R expression in the diencephalic regions 3,46 . In addition, there was no prominent and rapid effect of Kiss1 on dopamine levels in the diencephalic regions, suggesting that the telencephalic dopaminergic clusters could be the primary action target for habenular Kiss1 neurons. In the present study, our tracer experiment showed putative connections from the MR region to the forebrain as proposed previously 8 . Tracer-labelled nerve fibres were also seen in the mesencephalic and rhombencephalic regions. However, similar distribution of tracer-labelled fibres, except for those in the telencephalon, were also observed when the tracer was implanted into the IPN region, an adjacent region to the MR. Hence, it can be speculated that the fibres labelled in the mesencephalon and rhombencephalon are primarily derived from the IPN due to leakage of the tracer from the MR, but the fibres/tracts labelled in the diencephalon towards the telencephalon are derived from the MR. The tract linking the diencephalon and the telencephalon is known as the medial or lateral forebrain bundle (MFB/LFB), that starts in the telencephalon and extends into two distinct tracts; lateral to the dorsal preglomerular area and medial to the posterior tuberal nucleus in the fish brain 47 . Direct connections between the MR and MFB/LFB have not been shown in the brain of teleosts. However, in rats, the MR-forebrain tract lies in the ventromedial aspect of the MFB and projects to medial forebrain areas 48 . Similarly, in a cartilaginous fish (Platyrhinoidis triseriata), retrograde neural tracer injection into the forebrain bundle showed tracer-labelled cells in the superior raphe 49  Although the physiological role of Kiss1-Kiss1R signalling in habenula neurons remains unclear, expression of Kiss1R (= GPR54) in the habenula has been identified in several fish species and also in mammals 52,53 . In mammals and some non-mammalian species, Kiss1 primarily acts as modulator of reproductive functions via regulation of the hypothalamus-pituitary axis 54 . In mammals, Kiss1 suppresses the activity of the tuberoinfundibular dopaminergic neurons, which has been implicated in prolactin secretion 33,34 . Recently, the potential involvement of Kiss1-Kiss1R signalling in cognitive functions has been reported. In mice, Kiss1 peptide (kisspeptin-13) enhances memory including passive avoidance memory consolidation and also mitigates memory impairment induced by Amyloid-β 55,56 , suggesting possible link between Kiss1 and dopaminergic signalling. However, neither direct association of Kiss1 with nor expression of Kiss1R in midbrain dopaminergic neurons have been reported. Recent accumulated evidences have implicated the role of habenula in spatial and aversive memory processing [57][58][59] . Further, it is well known that midbrain dopamine system is necessary for associative learning and temporal stability of memories 60,61 . Thus, it can be postulated that Kiss1-Kiss1R signalling could influence cognitive functions via indirect action through the habenula-VTA pathway in mammals.
In summary, the present study demonstrates stimulatory effect of centrally administered Kiss1 peptides on dopamine levels in the telencephalic regions. In fish administered with Kiss1, neural activity and kiss1 gene expression were significantly decreased, suggesting autocrine and negative feedback regulation of Kiss1 neurons. Since the habenular Kiss1 neurons form terminals in the MR 11,46 and our tracer study showed potential link between the MR and the telencephalon via the MFB, we speculate that habenular Kiss1 negatively regulates forebrain dopamine neurons through interneurons in the MR (Fig. 8).

Intracranial administration of Kiss1 peptides and sample collection. Central administration of
Kiss1 peptides was performed as described previously in Kizil and Brand (2011) and Ogawa et al. (2012) 10,64 . For whole brain gene expression assay, AB fish were anesthetized by immersion in 0.01% benzocaine (Sigma, USA) solution and intracranially administered with 1 μL of either kisspeptin1-15 (Kiss1; pyroglut-NVAYYNLNSF-GLRY-NH 2 ; Open Biosystems) or kisspeptin2-10 (Kiss2; FNYNPFGLRF-NH 2 ; BioGenes) with two different doses (10 -12 and 10 -9 mol/fish in distilled water). For injection, the cranial bone close to the habenula above the left side of the anterior part of the optic tectum was incised using a sterilized barbed-end needle (BD Precision Glide 30G × 0.5″, BD Medical, New Jersey) (Fig. 1A), and through this incision, Kiss1 solution was administered using a heat-pulled glass capillary micropipette attached with a microinjector (IM-9B, Narishige, Tokyo, Japan) (Fig. 1B). For control, the same volume of distilled water was administered. Fish in sham group received the puncture procedure without any administration.
Real-time PCR for dopamine-related genes and kiss1 gene. Expression of dopamine-related genes: th1, th2 and dat, and kiss1 mRNA levels were measured in whole and macro-dissected brains using real-time PCR. For whole brain, following intracranial injection, fish were euthanized by rapid-cooling in ice cold water 30-min, 1-h, 3-h, 6-h or 24-h after administration (n = 7/group), while for the macro-dissected samples, the brains were dissected 24-h after administration (n = 10/group). For the fish treated with sham and Kiss2, the brains were collected 24-h after administration. All sample collections were conducted consistently between 13:00-14:00 for every experimental protocol. For regional gene expression assay, the whole brain was further macro-dissected into five different brain regions: telencephalic region, preoptic-pretectum region, optic tectum, midbrain-hypothalamic region, rhombencephalic region ( Fig. 2A) under a stereoscopic microscope. Whole or macro-dissected brain samples were homogenised in TRIzol reagent (Life Technologies), and total RNA were isolated and dissolved in 10 μL of RNase-free water. The total RNA was reverse transcribed to cDNA using High Capacity cDNA Reverse Transcriptase kit (Applied Biosystems) in a 20 μL reaction mixture according to the manufacturer's instruction. Real-time PCR were performed using the Applied Biosystems StepOne (PE Applied Biosystems, CA, USA) in a total volume of 10 μl reactions contained 1 × POWER SYBR Green PCR Master Mix (PE Applied Biosystems, CA, USA), 0.2 μM each gene-specific forward and reverse primers ( Table 1) and 1 µl cDNA. The reaction was programmed for 50 °C for 2-min, 95 °C for 10-min and 40 cycles of 95 °C for 15-s and 60 °C for 1-min, followed by a dissociation stage. The threshold cycle (Ct) of each gene of interest was obtained and normalised against mRNA expression levels of eukaryotic translation elongation factor 1 alpha 1, like 1 (eef1a1l1). The data were then analysed according to 2 -ΔΔCt relative gene expression quantification.
Dopamine quantification. Dopamine levels in macro-dissected brain samples were determined using liquid chromatogram (LC)-double mass spectrometry (MS/MS) (LC-MS/MS) following the procedures previously developed by Kim et al. (2016) 65 with some modifications. AB fish were first administered with Kiss1 at the doses of 10 -12 and 10 -9 mol/fish or saline (n = 6 /group), and the brains were collected at 30-min, 1-h, 3-h, and 6-h after Kiss1 administration. The dissected brain was further macro-dissected into three different brain regions (Fig. 3A)   Sample preparation. The macro-dissected brain was homogenized in 80 µL of buffer B supplemented with 20 pg/ µL of the internal standard solution using micropestle for 60s. For the investigation of selectivity, matrix effects and sensitivity, authentic homogenized filtered samples were used. The sample preparation was processed briefly in an ice bath to prevent the possible degradation of the analytes. Five microliter of the samples were injected into the LC-MS/MS system. Quality control was done by protein quantification against bovine serum albumin using BioRad Protein Assay dye (Cat#: 5000006, BioRad) according to the manufacturer's instructions. Dopamine levels in the brain tissue homogenate were normalized by total protein content levels. The assay was read on Labsystems multiskan 352 plate reader (Thermo Fisher Scientific) at a wavelength of 595 nm. In situ hybridization of npas4a. Effect of Kiss1 treatment on neural activities in the habenula and dopaminergic neurons were examined by expression of neural activity marker, npas4a expression, which were examined in the brain of AB-wild type or Tg(dat:EGFP) fish, respectively. The whole brain (n = 6 fish/group) was dissected 30-min after the Kiss1 administration and fixed in buffered 4% paraformaldehyde for overnight at 4 °C and then expression of npas4a were examined by in situ hybridization. For npas4a mRNA expression in the habenula, coronal sections (10 µm) (n = 6 for control and Kiss-treated) of wild-type AB male were hybridized with DIG-labelled npas4a riboprobes (737nt, GenBank accession number: NM_001045321) for 16 h at 55 °C. DIG-labelled napas4a mRNA was detected with either anti-DIG-AP or anti-DIG-POD antibody (Roche) followed by chromogenic development with NBT/BCIP or amplification using TSA Plus Cyanine 3 System (Perkin Elmer/AKOYA Biosciences), respectively. After TSA amplification of npas4a mRNA signals, EGFP signals were further enhanced with a rabbit anti-GFP antibody (1:500 dilution, Millipore Cat# AB3080, RRID:AB_91337) followed by incubation with Alexa Fluor 488-labeled anti-rabbit IgG (1:500 dilution, Thermo Fisher Scientific).
Neural tracer application into the median raphe. To examine the possible target of the Kiss1-terminus region in the median raphe (MR), a lipophilic tracer, NeuroVue Jade dye (green emitting dye; excitation maximum, 478 nm; emission maximum, 508 nm)-coated filter (Polysciences, Inc., Warrington, PA) was applied to the MR region in the brain of Tg(kiss1:mCherry) fish (n = 3). A nylon filter coated with dye was cut in a small pieces in triangular thin-strips (approx. 0.1 mm width × 1 m length) using an agitate scissors under a stereoscopic microscope, which allows to apply the tracer into the target brain region more precisely with less leakage of tracer from the target area 66 . As negative controls for tracer applications, the brain without tracer implantation (n = 1) and the brain with tracer application into the IPN region (n = 3) were prepared. The brain of Tg(kiss1:mCherry) fish was dissected and fixed in buffered 4% paraformaldehyde for overnight at 4 °C. The whole brain was cut into half longitudinally using a surgical knife and a dye-coated thin-strip was inserted to the mCherry-labelled MR region via sharp forceps under a florescence stereoscopic microscope (Nikon, SMZ1500). After dye application, the brain specimen were placed in buffered 4% paraformaldehyde and incubated at 37 °C for 3-days. For observation, the brain specimen was cryoprotected in 20% sucrose and embedded in Tissue Tek OCT compound (Sakura Finetechnical, Tokyo, Japan). The sagittal brain sections (15 μm thickness) were cut using a cryostat and were thaw-mounted onto 3-aminopropylsilane-coated glass slides.