Wild type and GnRH3:EMD transgenic lines of brass zebrafish were used in this study. Generation of gene constructs and transgenic fish have been described previously (19). Animals were maintained in a zebrafish aquarium system on a 14L:10D photoperiod at 28°C. Fish were fed twice daily with flake food and live brine shrimp.

Timed breeding was organized in order to collect embryos with a known fertilization time. The day before breeding, a pair of adult GnRH3:EMD male and female fish were divided in a tank with a physical barrier. After lights on the next morning, the divider was removed allowing the fish to mate – which happens very quickly. Fertilized eggs were immediately collected and maintained in a Petri dish filled with embryonic water at 28.5°C. The timing of hours post fertilization (hpf) was the time that the dividers were removed for mating. All procedures were carried out in accordance with and approved by the Animal Care and Use Committee of UCLA.

Live embryos were dechorionated manually, embedded in a drop of 0.8% agarose (Fisher Scientific, Pittsburgh, PA, USA) in glass-bottom culture dishes (35 mm diameter with 14 mm glass, No. 1.5 thickness; MatTek Corp., Ashland, MA, USA) and oriented using blunt forceps (lateral-side or ventral-side up, depending on which populations of GnRH3 neurons were being imaged). Once the agarose solidified, fish saline was added to keep the agarose moist. Embryos were then viewed under an upright Olympus confocal microscope and imaged using the Fluoview imaging system (Olympus America Inc., Center Valley, PA, USA). Images were taken using a water immersion 20× or 40× objective. EMD fluorescence was observed using an Argon laser (488 nm) with an emission barrier filter of 510 nm. Optical sections were made every 0.5 μm along the z-axis from the dorsal to the ventral side of the embryo. Images were achieved through projections of the z-stack. Timelapse images were acquired by scanning the embryos every 15 or 30 min. To follow the development of a specific group of GnRH3 neurons, we used a 40× objective with digital zoom (additional 2× or 3×) and took z-stack images every 15 min. The projection images were compiled in chronological order to form timelapse movies using Windows Movie Maker (Microsoft Inc., Redmond, WA, USA). Z-stack images were processed for projection, three-dimensional animations, movies, and analysis. To control the quality of images, parameters of the Fluoview program and microscope were adapted, but kept constant for the same set of experiments.

In some experiments (shown in Figures 3B and 4B), embryos were anesthetized with ethyl 3-aminobenzoate methanesulfonate salt (MS-222) and fixed at different stages with 4% paraformaldehyde as describe below, washed with PBS, and embedded in 0.8% agarose in lateral or ventral positions so as to image in the same way as the live embryos. Imaging fixed embryos provided a “snap-shot” view of different populations of GnRH3 neurons at known critical developmental times based on the live imaging, and allowed for more accurate analysis of changes in neuron number.

GnRH:EMD neurons were counted through the use of ImageJ processing software (ImageJ, free software from National Institutes of Health). Using ImageJ, contrast and brightness of the confocal z-stack images were adjusted in order to best visualize the individual cell bodies in a given neuronal population. Because it was difficult to determine the total number of cells in a 2-D projection image, each 0.5 μm optical slice in the z-stack was also individually analyzed. After the number of observed cell bodies was counted in each optical slice, the 3-D animation feature was used to confirm the number of neurons.

SV2 immunohistochemistry is widely used in zebrafish as a presynaptic marker for monitoring synaptogenesis at early stages of development (20–22). Embryos at 25 and 50 hpf were fixed in 4% paraformaldehyde at 4°C overnight. Fixed embryos were washed with PBS, permeabilized in ice cold acetone for 8 min, rinsed in 0.3% Tween solution in PBS, then blocked for 2 h at room temperature in the dark with a solution of 0.3% Tween and 10% goat serum (in PBS). Following blocking, sections were washed and incubated overnight at 4°C with the primary antibody (anti-SV2, synaptic vesicle protein 2, mouse, 1:200) diluted in the blocking solution (23). Anti-SV2 antibody was purchased from Developmental Studies Hybridoma Bank, University of Iowa (Iowa City, IA, USA). After washing with 0.3% Tween in PBS, embryos were incubated in secondary antibody (Alexa Fluor 594 anti-mouse, 1:1000; Invitrogen, Carlsbad, CA, USA) at room temperature for 1 h, followed by rinsing with PBS five times (10 min for each rinse). Fixed embryos were mounted in 0.8% agarose at lateral or ventral positions, covered with PBS, then imaged with the confocal imaging system described above. Images were taken using a 40× water objective. EMD fluorescence was observed using an Argon laser (488 nm) with an emission barrier filter of 510 nm. Alexa Fluor 594 (secondary antibody staining) was visualized using a HeNe laser (543 nm) with an emission filter of 560–600 nm. All images were captured using the sequence mode to eliminate bleed-through of fluorescent signals. Images were taken in the z-plane at 0.5 μm steps. Specificity of immunostaining was confirmed using control embryos that were processed the same as experimental animals, but without the primary antibody.

Loose-patch extracellular electrophysiological recordings of terminal nerve GnRH3 neurons from 2 to 3 dpf GnRH3:EMD live embryos were carried out as previously described (19). Briefly, the embryos were dechorionated and anesthetized by immersion in MS-222 (150 mg/l) and d-Tubocurarine (10 μM). They were glued ventral-side up to a glass coverslip at the bottom of a flow-through recording chamber (P1; Warner Instrument Corp., Hamden, CT, USA). Fine-tip forceps are used to gently remove the skull and expose the brain. Terminal nerve GnRH3:EMD neurons were identified using a 40× objective on an upright microscope (BX50W, Olympus, Melville, NY, USA). EMD-expressing neurons were targeted for loose-patch electrophysiology and recorded with a low-resistance seal (30–100 MΩ) in current clamp mode. After 5 min of stable baseline recording, 100 nM kisspeptin1(10) was bath applied for 5 min and spontaneous action potential firing was continuously recorded. Kisspeptin1(10) is a biologically active 10-amino-acid sequence (synthesized by Bachem Inc., Torrance, CA, USA) previously described for treatment of fish neural tissue by Zhao and Wayne (24). Both loose-patch and whole-cell electrophysiology following the procedure described by Ramakrishnan et al. (19) was used to record membrane potential (Voltmeters) and spontaneous action potential firing from trigeminal GnRH3:EMD neurons from live, intact embryos (25–30 hpf). Embryos were glued in the recording chamber in a lateral position so as to best expose the GnRH3:EMD-expressing neurons in the trigeminal ganglion. Neuron electrical activities were monitored and recorded by using both AxoGraph software (Axon Instruments, Foster City, CA, USA) and PowerLab data acquisition and analysis instrumentation and software (ADInstruments Inc., Colorado Springs, CO, USA). Subsequent data analysis of spike frequency was performed using AxoGraph software. Temperature in the recording chamber was maintained at 21–22°C throughout the experiments.

Values are shown as the mean ± SEM. Data were analyzed using Prism (version 5.0; GraphPad Software, Inc., San Diego, CA, USA). Differences were considered significant if p < 0.05. Experiments in which there were only two groups, statistical analysis was performed using Student’s t-test. Experiments in which there were multiple groups, statistical analysis was performed using ANOVA followed by post hoc Student’s t-test.

Previous work using GnRH3:eGFP and GnRH3:EMD transgenic zebrafish showed multiple populations of neurons expressing green fluorescence. Validating that these neurons are bone fide GnRH3 neurons depends on co-localization of green fluorescence with either GnRH3 peptide (e.g., immunohistochemistry) or mRNA (e.g., in situ hybridization). Using these approaches, it was confirmed that neurons in the terminal nerve/olfactory region, hypothalamus, POA, ventral telencephalon, and trigeminal ganglion express endogenous GnRH3 (11, 19).

Figure 1A shows live images of a ventral view of the whole head of two zebrafish embryos at 40 hpf. Both images are shown with the merging of brightlight and GFP fluorescence stimulation so that the positioning of GnRH3:EMD-expressing neurons can be visualized relative to the eyes and nose. The image on the right shows populations of GnRH3:EMD-expressing neurons in the central nervous system of a transgenic zebrafish: terminal nerve, POA, and hypothalamus. The image on the left shows a wildtype embryo, exposed to the same UV light as the transgenic embryo – with little to no green autofluorescence at this age. Although there was GnRH3:EMD expression emerging in retina at around 30 hpf, we have so far not been able to verify that these neurons contain endogenous GnRH3. Failure to show endogenous GnRH3 mRNA may be due to these neurons showing such low expression of message that it is undetectable using in situ hybridization, or that the transgene is being ectopically expressed in these retina neurons. Nevertheless, it is notable that the retina is a target of GnRH3 neurons (11, 25, 26), expresses GnRH receptors (27, 28), and its development is disrupted when GnRH3 gene expression is knocked down (11, 29).

Figure 1B shows live transgenic zebrafish embryos imaged in different positions so that different populations of GnRH3:EMD neurons can be more easily viewed. The panel on the left shows a lateral view at 20 hpf, with emergence of a group of GnRH3:EMD neurons in the trigeminal ganglion of the peripheral nervous system [first described and validated by (11)]. The panel on the right shows a ventral view of the emergence of the terminal nerve, POA, and hypothalamic populations of GnRH3:EMD neurons at 30 hpf. Figure 1C shows a timeline of the first emergence of different populations of GnRH3:EMD neurons, based on timelapse analysis of development of five transgenic embryos. The first population to emerge were those in the trigeminal ganglion at 15 hpf, then the terminal nerve and hypothalamic populations emerged about 5 h later at 20 hpf, followed by the POA population at about 30 hpf. This is an earlier time course than previously described by Abraham et al. (11). This can be explained by differences in the intensity of fluorescence between eGFP used by the Zohar group in their studies (11, 30, 31) and EMD used by the Wayne laboratory (19). Because EMD provides a stronger fluorescent signal than eGFP (32), it is a more sensitive marker of early GnRH neuron development.

Figure 2A shows a lateral view of the head of a live transgenic embryo from 17 to 25 hpf. GnRH3 neurons in the bilateral trigeminal ganglion are the first population to emerge, and very quickly send projections posterior into the spinal cord and anterior along the surface of the head. Movie S1 in Supplementary Material shows a timelapse movie with images taken every 30 min from 17 to 24 hpf. The movement of the projections is linear, with no back-and-forth meandering. Although there was GnRH3:EMD expression emerging in the spinal cord at around the same time as that of the trigeminal ganglion (data not shown), as with the retina, we have so far not been able to verify that these spinal cord neurons contain endogenous GnRH3.

To investigate if there was spontaneous electrical activity during early development, both loose-patch and whole-cell electrophysiological recordings were performed on trigeminal GnRH3:EMD neurons from 24 to 30 hpf embryos. Recordings were performed on seven neurons from six embryos (three whole cell and four loose-patch recordings) for approximately 10 min. Three neurons were electrically silent without any action potential firing during the recording period, one neuron showed two bursts (two to three spikes) of action potentials, and three neurons showed two to five spikes during the entire recording period. An earlier study also showed that trigeminal neurons in general show little to no spontaneous action potential firing in zebrafish embryos (33). Figure 2B shows electrophysiological recordings from two representative GnRH3 neurons in the trigeminal ganglion in 24 hpf (cell 1) and 27 hpf (cell 2) embryos. Whole-cell recordings revealed an average resting membrane potential (Vm) of −50 mV. In cell 2, there were three action potentials during a 9-min recording period. Although cell 1 did not fire action potentials, it did show depolarizing oscillations in Vm. This could be a reflection of an endogenous pacemaker potential as observed with terminal nerve GnRH3 neurons (34–36), or synaptically driven oscillations in Vm.

Figure 3A shows images of the development of one cluster of terminal nerve GnRH3 neurons between 24 and 31 hpf. During this time, there is an increase in the number of neurons expressing EMD, and neural projections are becoming more extensive. By 50 hpf (shown in the right panel with a view of the bilateral clusters), the number of GnRH3 neurons has stabilized, and the posterior projections between the bilateral clusters are in close apposition to each other (also shown in Figure 5A). Movie S2 in Supplementary Material shows a timelapse movie with images taken every 15 min from 24 to 31 hpf of a single cluster of terminal nerve GnRH3 neurons. During this time, anterior neural projections are meandering back-and-forth – very different from the much more linear movements of the projections from the trigeminal GnRH3 neurons shown in Movie S1 in Supplementary Material. Figure 3B shows the numbers of terminal nerve GnRH3 neurons between 25 and 50 hpf. There is a linear increase in cell number, which plateaus by 40 hpf. This is very similar to the time course of terminal nerve GnRH3 neuronal proliferation described by Gopinath et al. (37) using immunohistochemistry and in situ hybridization. We have done timelapse imaging of the terminal nerve GnRH3 neurons from a total of five embryos between 14 and 50 hpf, and have never seen any evidence of migration of neurons from this population to the hypophysiotropic populations as suggested by Abraham et al. (9).

We tested the hypothesis that kisspeptin, a neuropeptide that stimulates puberty and fertility through its actions on GnRH neurons (38–42), is also acting to stimulate the electrical activity of developing GnRH3 neurons during embryogenesis. Figure 3C shows the electrophysiological response of developing GnRH3 neurons in the terminal nerve to 100 nM kisspeptin1. Our earlier work showed that between 2 and 3 dpf, there is a maturation process by which zebrafish terminal nerve GnRH3 neurons become increasingly active (increase in number of neurons showing spontaneous action potential firing), and segue from an immature form of electrical activity (bursting pattern of firing) to a mature form of activity (tonic pattern of firing) (19). Figure 3C1 shows that there is no effect of kisspeptin1 on firing frequency of the GnRH3 neurons exhibiting the immature bursting form of action potential firing (baseline: 1.2 ± 0.2 Hz; kisspeptin: 0.9 ± 0.2; n = 9 neurons). However, the neurons exhibiting the mature tonic form of firing were stimulated by kisspeptin, showing a significant increase in firing frequency (baseline: 1.7 ± 0.5 Hz; kisspeptin: 3.0 ± 0.6 Hz, p < 0.05; n = 10 neurons) (Figure 3C2). This is very similar to the increase in firing rate of adult medaka terminal nerve GnRH3 neurons in response to treatment with 100 nM kisspeptin 1 (24).

Figure 4 demonstrates the progressive development of the POA and hypothalamic populations of GnRH3 neurons between 25 hpf and 40–50 hpf. Figure 4A shows a ventral view of the head of a live embryo at different time points. Images of the POA and hypothalamic populations suggest that hypothalamic GnRH3 neurons emerge primarily in situ, and not from migrating neurons originating from the terminal nerve population as suggested by Abraham et al. (9). However, in 100% of timelapse images of this region, it was noted that some GnRH3 neurons migrated from the hypothalamic to the POA beginning at 28–29 hpf. This migration can be viewed in a timelapse movie shown in Movie S3 in Supplementary Material.

Figure 4B shows the rate of proliferation of GnRH3 neurons in the POA (right panel) and hypothalamus (left panel). By 40–50 hpf, the numbers have not yet reached a plateau – unlike what was shown with the terminal nerve neurons (Figure 4B). Also, the hypothalamic population has by far the greatest number of GnRH3 neurons during embryogenesis compared to the other populations.

Live imaging studies indicate that neural processes from one population of GnRH3 neurons form close appositions to processes and soma of GnRH3 neurons from other populations during embryogenesis (Figure 5). Ventral view of the developing head using confocal microscopy showed close apposition between: the neural processes of bilateral clusters of terminal nerve GnRH3 neurons at 40 hpf (Figure 5A1), and terminal nerve GnRH3 neural processes and POA GnRH3 soma at 45 hpf (Figure 5A2; Movie S4 in Supplementary Material showing three-dimensional rotating image).

SV2 is a membrane glycoprotein present in all synaptic vesicles and regulated secretory vesicles of endocrine cells, and is a marker of synaptic transmission (43–46). In the present study, immunohistochemical analysis of SV2 shows the expression of SV2 on GnRH3 neurons, suggesting that they are capable of synaptic communication during early embryogenesis (Figure 5B). Figure 5B1 shows a lateral view of the head to visualize GnRH3 neurons in the trigeminal ganglion at 25 hpf, and demonstrates co-localization of GnRH3:EMD and SV2 in the trigeminal neurons. Figure 5B2 shows a ventral view to visualize GnRH3 neurons in the forebrain at 50 hpf, and demonstrates co-localization of GnRH3:EMD and SV2 in neural processes originating from the terminal nerve population. There were less SV2 immunoreactive contacts on somas of terminal nerve GnRH3 neurons than on trigeminal GnRH3 neurons (SV2-ir positive somas: 25% of terminal nerve GnRH3 neurons; 100% of trigeminal GnRH3 neurons).