Specific hypothalamic gonadotropin-releasing hormone (GnRH) -secreting cells control the onset and maintenance of the reproductive axis in vertebrates (Gore,2002). These neurons undergo a unique odyssey during embryonic development. They are first detected in association with the olfactory placode, then, by neurophilic migration, they enter the forebrain along olfactory, terminal and veromonasal nerves, and further continue to the preoptic area in the hypothalamus (Schwanzel-Fukuda and Pfaff,1989; Wray et al.,1989). This pathway of developmental migration has been shown in a variety of vertebrate species (Dubois et al.,2002), including zebrafish (Palevitch et al.,2007; Abraham et al.,2008). In mammals, the gene that encodes the hypophysiotropic form is termed GnRH1 (White and Fernald,1998). In the zebrafish genome, which has been almost completely sequenced, the gnrh1 gene is most definitely absent (Palevitch et al.,2007), probably due to selective loss during evolution (Kuo et al.,2005). This gene loss has most likely occurred also in other teleost species belonging to various orders (reviewed in Okubo and Nagahama,2008). Nevertheless, two other genes encoding different forms of GnRH—gnrh2 and gnrh3 (formerly cGnRH-II and sGnRH, respectively)—have been isolated and characterized in zebrafish (Torgersen et al.,2002; Steven et al.,2003). In adult zebrafish, gnrh2 is localized to the midbrain tegmentum, as described in all other investigated jawed vertebrates (Steven et al.,2003). The gnrh3 transcript is located in adults in the terminal nerve ganglion (TNg). A gnrh3-expressing subset group resides in the ventral telencephalon-preoptic area. The GnRH3 axons project to various parts of the brain including the pituitary gland (Abraham et al.,2008). Furthermore, combined high performance liquid chromatography/radioimmunoassay analyses have indicated that the GnRH3 peptide is the predominant GnRH peptide in the adult zebrafish pituitary (Powell et al.,1996). Therefore, GnRH3 is considered the hypophysiotropic form of GnRH in zebrafish. Furthermore, the ontogeny of GnRH3 neurons in zebrafish resembles the development of GnRH1 neurons in mammals: In both cases, the expression of GnRH is initially detected in the olfactory placode, later in association with the olfactory/terminal nerves, and then migrating through the ventral telencephalon to the hypothalamic preoptic area (Schwanzel-Fukuda,1999; Palevitch et al.,2007; Abraham et al.,2008). Of interest, in zebrafish, GnRH3 projections were shown to precede the migration of their cell soma along the terminal nerve, delineating the pathway for the migration of a subset of GnRH3 cells from the olfactory region to the hypothalamus (Palevitch et al.,2007; Abraham et al.,2008). Moreover, projections extended to the pituitary gland even before GnRH3 perikarya entered the forebrain (Abraham et al.,2008).
The importance of an appropriate migration of hypothalamic GnRH neurons for normal reproductive function is well demonstrated by a rare human disease called Kallmann's syndrome (KS; Kallmann et al.,1944). KS is characterized by anosmia coupled with hypogonadotropic hypogonadism that is secondary to deficiency in hypothalamic GnRH (Naftolin et al.,1971). Impaired development of the olfactory nerves in a KS human fetus specimen resulted in failure of GnRH neurons to undergo normal migration (Schwanzel-Fukuda et al.,1989).
The migration of GnRH neurons in vertebrates is modulated by a plethora of factors, such as receptors, secreted molecules, and adhesion molecules (Tobet and Schwarting,2006). However, the exact mechanism that controls these developmental events is poorly understood (Schwarting et al.,2007). One approach that was implemented toward identifying additional regulatory molecules was a cDNA subtraction screen in migrating vs. nonmigrating GnRH neurons in mice (Kramer and Wray,2000). This analysis revealed a new factor, named Nasal Embryonic LHRH factor (NELF), which was shown to be involved in the outgrowth of olfactory axon projections and to affect the migration of GnRH neurons in mice (Kramer and Wray,2000). Consequently, human NELF was considered as a candidate gene for the etiology of idiopathic hypogonadotropic hypogonadism and KS. Indeed, genetic studies in human implicated mutations in the NELF gene in two cases of idiopathic hypogonadotrophic hypogonadism and KS (Miura et al.,2004; Pitteloud et al.,2007). Here, we report on the isolation of the first nonmammalian nelf gene, describe its expression pattern during development, and demonstrate its involvement in the development of the GnRH3 system in zebrafish.
Characterization of Zebrafish nelf
The zebrafish nelf cDNA sequence was deposited in the GenBank (accession no. EU328156). The arrangement of this gene is shown in Figure 1. The entire length of the isolated nelf cDNA is 2,803 bp comprising 15 exons. The 5′ untranslated region (UTR; 316 bp) is composed by the majority of the first exon. The open reading frame includes 1,713 bp coding a 570 amino acids predicted protein, with the start codon located in the 1st exon and the stop codon in the 15th exon. The 3′ UTR fragment contains 745 bp followed by the poly A tail. The full-length nelf cDNA is mapped to chromosome 5 (nt. 17,052,092–17,106,355) in the zebrafish genome (Zv7, USCS browser). Comparison of the deduced zebrafish Nelf protein to the orthologous human protein demonstrates 70% similarity (Fig. 2). Transmembrane helices analysis using the TMHMM v2 software (Krogh et al.,2001) indicates that zebrafish Nelf does not contain hydrophobic regions, and suggests that this is an extracellular protein. This topology is in accordance with that of mice NELF (Kramer and Wray,2000).
nelf Expression Pattern
Whole-mount in situ hybridization (ISH) analysis for nelf mRNA in zebrafish embryos revealed a dynamic expression pattern within the central nervous system (CNS; Fig. 3). The expression of nelf was initially detected at 16 hours postfertilization (hpf) in the presumptive telencephalon (data not shown). At 24 hpf, nelf transcript is robustly expressed in the telencephalon and hypothalamus, the midbrain tegmentum, and along the hindbrain and the spinal cord (Fig. 3A,B). At this time, nelf is not expressed in the olfactory epithelium but rather in the adjacent domain of the telencephalon (Fig. 3C). The first detection of nelf transcript in the olfactory epithelium and olfactory bulb was at 32 hpf (Fig. 3D–F). At 48 hpf, nelf expression expanded to include the epiphysis and the retina, while expression in the hypothalamus decreased (Fig. 6G,H). The nelf signal in the retina and epiphysis significantly increased at later developmental stages, 72–120 hpf (Fig. 6J,K; only 72 hpf is shown). Altogether, this spatiotemporal expression pattern of nelf transcript in zebrafish generally resembles that described in rodents (Kramer and Wray,2001).
Double Labeling of GnRH3 Neurons and nelf mRNA
The domain of nelf expression in relation to GnRH3 neurons was determined by double-labeling assays in Tg(gnrh3:EGFP) embryos. Fluorescence ISH was used to label nelf mRNA, and immunofluorescence staining was used to detect enhanced green fluorescent protein (EGFP) protein driven by gnrh3 promoter (Fig. 4). The first detection of GnRH3 neurons was in the olfactory epithelium at 26 hpf. These neurons extended projections toward the telencephalon where nelf is robustly expressed (Fig. 4A–C). By 32 hpf, GnRH3-producing neurons, still present in the olfactory epithelium, adjacent to the developing olfactory organ, co-expressed nelf transcript (Figs. 4D–F, 5A–C). Thereafter, at 48 hpf, the cluster of GnRH3 neurons located at the origin of the terminal nerve continued to express nelf mRNA in an area with strong expression of nelf (Figs. 4G–I, 5D–F). At this time, GnRH3 neurons sent projections along the terminal nerve toward the optic chiasm and the retina (Fig. 4H). Other GnRH3 projections extended caudally toward the midbrain where nelf was robustly expressed (Fig. 4I). At 72 hpf, when GnRH3 neurons are localized to the terminal nerve adjacent to the olfactory organ–olfactory bulb boundary, nelf signal was not detected in GnRH3 cells but rather in the surrounding cells in the olfactory bulb and telencephalon (Figs. 4J–L, 5G–I). This spatiotemporal expression pattern of nelf in forebrain GnRH cells resembles that described in mice (Kramer and Wray,2000).
Aberrant GnRH3 Neural Migration in Nelf-knockdown Fish
To test the possible involvement of Nelf in GnRH3 development, an in vivo functional analysis was done in Tg(gnrh3:EGFP) embryos by Nelf knockdown using specific morpholino-modified antisense oligonucleotides (MOs). After injection, the GnRH3 phenotype in live embryos and larvae was monitored daily for 10 days. Standard control-MO injection did not cause any visible disturbances in the GnRH3 migration pattern as compared to that of noninjected transgenic fish (Abraham et al.,2008). Time-lapse analysis of typical control-MO injected transgenic larva is shown in Figure 6. At 3 dpf, GnRH3 neurons were seen as bilateral dots in the olfactory organ–olfactory bulb boundary (Fig. 6A). During the next days, bilateral projections of the GnRH3 neurons extended caudally through the telencephalon (Fig. 6B,C), intersected at the anterior commissure and the optic chiasm (Fig. 6D–F). A subpopulation of GnRH3 cells somata followed this pathway during their migration and further targeted the hypothalamus (Fig. 6F). Both of Nelf[AUG]MO and Nelf[Int2Ex3]MO injections resulted in a similar abnormal phenotype of the GnRH3 neurons in 65–67% of the injected embryos, respectively (Fig. 7). The abnormal GnRH3 phenotype included absence or mistargeting of axonal GnRH3 projections, arrested migration of GnRH3 somata inside the nasal area, and migration of cells to random areas of the brain. Representative Nelf-MO injected larva with abnormal GnRH3 phenotype is shown in Figure 6H–M. At 3 dpf, Nelf[Int2Ex3]MO-injected larvae had higher numbers of EGFP-positive signals that were randomly scattered in the olfactory area (Fig. 6H), as was in the case with Nelf[AUG]-MO injection (data not shown). At later developmental stages, the usual GnRH3 axonal pathway was absent, projections did not extend lateromedially, and no GnRH3 axons were detected at the anterior commissure nor in the optic chiasm. Interestingly, the randomly scattered EGFP signal seen at earlier stages had disappeared. It is possible that these cells had died due to their arrival at an incorrect and nonsupportive environment. Thus, Nelf knockdown affected both axonal projections and somata migration of GnRH3 neurons.
The specific effect of Nelf(I2E3)MO on nelf splicing was validated by reverse transcriptase-polymerase chain reaction (RT-PCR). This analysis demonstrated the addition of 249 bp of intron 2 to the nelf mRNA in Nelf[Int2Ex3]MO-injected embryos (Fig. 6O,P). Such addition is predicted to cause a truncation of the Nelf protein (77 aa protein instead of the predicted 567 aa intact protein).
The effects of Nelf-MO were further validated by a rescue experiment. Co-injection of splicing-MO and capped nelf mRNA resulted in normal GnRH3 system phenotype in 75% of the larvae (Fig. 7). Injection of nelf mRNA alone had not aberrant effect; 100% of the embryos had normal GnRH3 phenotype. Overall, these knockdown experiments demonstrate that Nelf has a crucial role in the guiding of GnRH3 somata migration and in their axonal outgrowth.
The appropriate development of the forebrain GnRH system is critical for attaining reproductive function (Whitlock et al.,2006; Wray,2002). Thus, considering the involvement of NELF in this process in mice (Kramer and Wray,2000), we hypothesized that NELF sequence and function is evolutionary conserved. Indeed, we found in the zebrafish genome an orthologous nelf with a high similarity to the mammalian NELF, suggesting that the NELF gene is conserved between fish and mammals. Characterization of nelf mRNA expression in developing zebrafish revealed a dynamic expression in the CNS, including the olfactory bulb, telencephalon, hypothalamus, midbrain tegmentum, retina, hindbrain and the spinal cord. Interestingly, many of these nelf expression domains—olfactory bulb, telencephalon, hypothalamus, and retina—depict the pathway of the terminal nerve and the GnRH3 system in zebrafish; the primary GnRH3 projections follow this route to the retina (Fig. 4H; Abraham et al.,2008). Moreover, double-labeling analyses revealed the expression of nelf transcripts in GnRH3 neurons within the olfactory epithelium and terminal nerve. Finally, knockdown of Nelf in Tg(gnrh3:EGFP) embryos caused aberrant development of GnRH3 cells and projections, reinforcing the notion of an evolutionary conserved role for NELF. The specificity of the nelf splicing-MO was proved by a rescue experiment and by RT-PCR analysis. Taken together, these results suggest that zebrafish Nelf is needed for GnRH3 axonal outgrowth and the subsequent migration of GnRH3 neurons from the olfactory area to the telencephalon and possibly also to other GnRH3 targets.
The migration of forebrain GnRH neurons offers a distinctive example of axonophilic tangential migration (Marin et al.,2003). Studies in mammals have shown that GnRH1 neurons migrate in close association with the olfactory and terminal-vomeronasal nerves (reviewed in Wray,2002). Manipulated disruption of these nerve tracks caused alteration in GnRH1 migration (Yamamoto et al.,1996; Murakami et al.,1998; Gao et al.,2000). Consequently, the molecular mechanism that regulates the development of the nasal pathways appears to affect, either directly or indirectly, the development of the GnRH1 system. The action of NELF provides a good example for both possibilities, as it was shown in mice to act as a guidance cue for olfactory axon projections and the subsequent migration of GnRH1 neurons (Kramer and Wray,2000). Knockdown of mouse NELF in embryonic nasal culture resulted in perturbation of olfactory axon outgrowth and inhibition of GnRH1 cell movement along these axons (Kramer and Wray,2000). The expression pattern, topology and function of zebrafish Nelf, revealed in the current study, are in accordance with the findings in the mice study. The question of whether the disruption of GnRH cells migration is a direct consequence of the decrease in NELF (i.e., cell-autonomous action), a result of the absence of the axonal track substrate, or both, warrants further studies in both the zebrafish and mice models.
The recent finding of mutations in the human NELF gene in cases of idiopathic hypogonadotrophic hypogonadism and KS (Miura et al.,2004; Pitteloud et al.,2007) highlights the clinical relevance of NELF in the development of the GnRH system and reproduction. GnRH neurons navigate through various cellular environments, requiring multiple attractant and repellant molecular cues at specific sites and time points (Cariboni et al.,2007). However, to date only a few genes have been implicated in the etiology of KS (for recent reviews, see Cadman et al.,2007; Bhagavath and Layman,2007). Discovering additional KS-linked genes requires a better comprehension of the molecular mechanism of GnRH neuronal migration.
The zebrafish has become one of the most attractive and popular vertebrate models for developmental genetic studies, offering a plethora of genetic and molecular techniques and imaging capabilities. As demonstrated in the current study, the Tg(gnrh3:EGFP) line enables the investigation of two related developmental processes: axonal outgrowth and cell migration. This in vivo model will thus facilitate the study of the role of additional candidate genes in the process of the development of the hypophysiotropic GnRH system, and may serve as a model to investigate the etiology of hypogonadotrophic hypogonadism.
A transgenic line, Tg(gnrh3:EGFP) was used to visualize GnRH3 neurons and fibers. Extensive characterization of this line indicates that EGFP is exclusively expressed in GnRH3 cell bodies and processes (Abraham et al.,2008). Wild-type and transgenic zebrafish embryos were generated by natural mating. Embryos were raised and maintained under standard conditions (Westerfield,1995). In some cases, pigmentation was prevented by adding 0.2 mM phenylthiourea (PTU) to the water at 24 hpf. All procedures were conducted in accordance with the International Guiding Principles for Biomedical Research Involving Animals as promulgated by the Society for the Study of Reproduction.
Zebrafish nelf Gene Isolation
Based on alignment of mouse Nelf sequence to the zebrafish genome, specific primers (nelf F: 5′-aacttccggaaacatctgag-3′; nelf R: 5′-ggcacacatcatta-gcaatg-3′) were designed, and a 462 bp fragment was PCR-amplified from an embryonic zebrafish cDNA library. The PCR product was cloned into pGEM-T-easy vector (Promega, Madison, WI) and sequenced using ABI PRISM 3100 automated sequencer (Applied Biosystems, Foster City, CA). Gene-specific primers were designed based on this initial DNA sequence for further cloning and sequencing of the entire gene using 5′ and 3′ rapid amplification of cDNA ends (RACE)-PCR kit according to the manufactures' instructions (Clontech, Palo Alto, CA).
ISH and Immunocytochemistry (ICC) analysis
A DNA fragment corresponding to nt 57-1031 of zebrafish nelf mRNA (GenBank accession no. EU328156) was PCR-amplified from an embryonic zebrafish (48 hpf) cDNA library. This fragment was then linearized and used as a template for the transcription of the digoxigenin-labeled RNA probes (RNA labeling kit, Roche Diagnostics Ltd, Basel, Switzerland). Whole-mount ISH analyses were conducted as described previously (Palevitch et al.,2007).
Double staining of GnRH3 and nelf was achieved by applying fluorescence ISH-ICC procedure to Tg(gnrh3:EGFP) fish. Embryos which had been previously stained by ISH for nelf mRNA using the FastRed fluorescence substrate (FastRed tablets, Sigma-Aldrich, St. Louis, MO) were washed 4 × 15 min in PBST, and then blocked in 2% lamb serum/PBST for 1 hr. Embryos were then incubated overnight at 4°C in 2% lamb serum/PBST containing a monoclonal rabbit anti-GFP IgG (Invitrogen Co., Carlsbad, CA) at 1:500 dilution. Next, the embryos were washed in PBTw with 0.3% Triton X for 4 × 30 min at room temperature, and a fluorescent secondary antibody, cyanine Cy2 goat-anti-rabbit was added (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA) for 2 hr incubation at room temperature. Finally, the embryos were rinsed 3 × 15 min in PBST with 0.3% Triton X and then gradually transferred to 75% glycerol for mounting on slides. Images were captured with an LSM 510 confocal laser scanning microscope (Carl Zeiss MicroImaging Inc., Thornwood, NY) together with Laser-Sharp and LSM imaging software. Images were color balanced using PhotoShop 8.0 (Adobe Systems, Inc., CA).
Morpholino Knockdown Experiments
To test the effect of knockdown of Nelf on the development of the GnRH3 system, Morpholino-modified antisense oligonucleotides (MOs) were used (Nasevicius and Ekker,2000). Two nanoliters of MOs (= 15 ng /embryo) were injected into Tg(gnrh3:EGFP) embryos, immediately after fertilization. This was followed by a daily monitoring of the EGFP signals in live individual animals, under a fluorescent dissecting microscope (SZX12, Olympus, Japan) for 10–12 days. Time lapse studies were carried on representative individual larvae. A qualitative assessment of the GnRH3 neuronal development was performed individually on each injected larvae. The following MOs (Gene Tools, Philomath, OR) were applied: One designed to block Nelf translation, (Nelf[AUG]MO, 1 mM, 5′-TTCTTCTTTT-TGGACACGGCAGTTC-3′), and another designed toward the intron 2-exon 3 boundary to interfere with splicing (Nelf[Int2Ex3]MO; 1 mM, 5′-CAAGTGATCTGAAATGAGAGAC-AGA-3′). Gene-Tools standard control-MO was used as a negative control (1 mM, 5′-CCTCTTACCTCAGTTACAATTTATA-3′). The effect of Nelf knockdown on the GnRH3 phenotype was statistically analyzed by Pearson's χ2 test (χ2). Efficacy of Nelf[Int2Ex3]MO was evaluated by RT-PCR. Nelf(I2E3)MO-injected and uninjected embryos were sampled at 24 hpf and total RNA was extracted (EZ RNA Total RNA Isolation kit, Biological Industries, Beit Haemek, Israel). RNA samples were then treated by DNase (Promega) to remove residues of genomic DNA. M-MLV reverse transcriptase (Promega) and random hexamer primers were used to prepare the cDNA. Specific primers corresponding to zebrafish nelf exons 1 and 4 (nelf-E1F 5′-ACAGACCCTGAGCAGCAGTT-3′ and nelf-E4R 5′-GAAACCAGCTGTTCGTGTGA-3′) were used for PCR amplification. PCR products were separated on an agarose gel, subcloned into pGEM-T easy vector (Promega) and sequenced (ABI PRISM 3100 Genetic Analyzer, AME Bioscience). Rescue experiments were conducted by co-injection of a constant dose of splicing-MO (15 ng/embryo) and capped nelf mRNA. Initially, nelf CDS was subcloned into pCS2+ vector. Then, capped mRNA was generated by transcription of NotI (Invitrogen)-linearized plasmid (1 μg/l μ) using the SP6 mMessage mMachine kit (Ambion, TX).
We thank Professor Susan Wray, NIH, for her critical reading of the manuscript.