Specification of hypothalamic neurons by dual regulation of the homeodomain protein Orthopedia

The hypothalamus is a complex brain structure that regulates endocrine,behavioral and autonomic functions by means of its ability to affect both central and peripheral activities (Iversen et al., 2000). Accordingly, developmental impairments in hypothalamic differentiation are associated with defects in energy balance,and in neuroendocrine and psychiatric disorders(Swaab, 2004). For example,loss of hypocretin neurons leads to narcolepsy, and patients with Prader-Willi syndrome have a deficit in oxytocinergic (OT) neurons that is accompanied by hyperphagia and severe obesity (Peyron et al., 2000; Swaab et al.,1995). The medial area of the hypothalamus contains magnocellular and parvocellular neurons that control pituitary activities. The magnocellular neurons project to the posterior pituitary where they release oxytocin and arginine-vasopressin directly into the general circulation(Landgraf and Neumann, 2004). The parvocellular neurons affect the anterior pituitary by releasing hypophysiotropic hormones such as dopamine and somatostatin into the hypophysial-portal-vascular system(Markakis, 2002).

Hypothalamic development poses a challenging model for understanding neural patterning and specification because the hypothalamus contains multiple nuclei, each composed of several neuronal cell types that form connections with many parts of the nervous system(Markakis, 2002). Uncovering critical molecules regulating neural diversification of the hypothalamus is essential to understand how this elaborate brain region is formed. Insights into the differentiation of certain hypothalamic neurons has been contributed by targeted gene knockouts of the transcription regulators Sim1, Brn2,Arnt2, Hmx2/3 and Otp(Acampora et al., 1999; Michaud et al., 2000; Michaud et al., 1998; Schonemann et al., 1995; Wang et al., 2004; Wang and Lufkin, 2000). The homeodomain-containing protein Orthopedia (Otp) is a key determinant controlling the specification of neuroendocrine hypothalamic neurons(Acampora et al., 1999; Wang and Lufkin, 2000). However, the signaling pathway(s) that regulate Otp and eventually lead to synchronized hypothalamic differentiation have not been elucidated.

Here, we focus on studying the development of dopaminergic (DA) and OT-like neurons [termed isotocinergic (IT) neurons] representing mammalian parvocellular and magnocellular cell types, respectively. We report the mode of regulation of zebrafish Otpb during the development of these two prominent neuronal clusters. We show that regulated expression of Otpb by two novel converging pathways coordinate the development of IT and DA neurons.

Fish breeding and maintenance were performed as previously described(Levkowitz et al., 2003). Experiments were performed in accordance with the Weizmann Institute IACUC protocol number 770104-1.

Full-length otpb (see ZFIN ID: ZDB-GENE-990708-7 for nomenclature history), pac1, and pacap1b cDNAs were amplified by PCR from RNA that was isolated from embryos at 48 hours post fertilization. Identification of the pac1 translation start site is detailed (see Fig. S8 in the supplementary material). Activated pac1^*(E239Q) was generated by PCR-based site-directed mutagenesis. To construct theΔ1-5-pacap1b mutant, site-directed mutagenesis was used to delete the cDNA nucleotides encoding amino acids 121-125 of the PACAP1b precursor protein. cDNA was subsequently subcloned into either the pCS2⁺ plasmid or the heat-shock response element (HSE)-driven expression vector, pSGH2 (Bajoghli et al.,2004) and confirmed by nucleotide sequencing. Oligonucleotide primers that were used to amplify DNA templates for all digoxigenin(DIG)-labeled probe synthesis reactions are described in Table 1.

Whole mount immunostaining with either polyclonal or monoclonal anti-tyrosine hydroxylase (TH) antibody (Chemicon, Temecula, CA) and in situ hybridization were performed as described(Levkowitz et al., 2003). Otp antibody was raised against a C-terminal Otp peptide and purified by affinity chromatography as described (Lin et al.,1999). Following in situ hybridization, embryos were embedded into 1.5% agarose, dehydrated and embedded in paraffin. Paraffin blocks were sectioned (6 μm) on a microtome, mounted on slides and subjected to deparaffinization. Microwave-induced Otp antigen retrieval was performed in 10 mM citric acid (pH 6) for 10 minutes. Sections were blocked in PBS with 20%goat serum and 0.5% Triton X-100, and then incubated overnight at room temperature with affinity-purified rabbit Otp antisera (at 2 μg/ml). Sections were then washed with PBS and bathed with a goat anti-rabbit biotinylated antibody (at 1:200 dilution) for 1.5 hours at RT and visualized using a Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA).

HEK293 cells were grown in 12-well plates and transfected (at 60%confluence) with a total amount of 1.0 μg/well of the indicated pCS2-based expression vectors using a standard calcium phosphate method. In some experiments, cells were incubated for the indicated time periods in the presence of different concentrations of a synthetic PACAP₃₈ peptide(Sigma-Aldrich, Rehovot, Israel). Proteins were harvested 24 hours post transfection in 150 μl of hot SDS sample buffer and 15 μl of the crude protein extract was fractionated by 10% SDS-PAGE followed by immunoblotting with an affinity-purified anti-Otp antibody. Thereafter, PVDF membranes were acid stripped and reprobed with a monoclonal anti-β-actin antibody (clone AC-74, Sigma-Aldrich, Rehovot, Israel).

PC12 cell line (clone number CRL-1721, 7-10 passage) was obtained from the ATCC Bioresource Center and propagated according to ATCC instructions. Nuclear-enriched protein extraction as well as total RNA preparation were previously described (Schreiber et al.,1989). For western blot analysis of Otp from either PC12 or zebrafish proteins, 20 μg total protein from each treatment was fractionated on 8.5% SDS-PAGE and immunoblotted with anti-Otp antibody as described above. Goat antisera directed against the nuclear regulator of chromosome condensation 1 (Rcc1; Santa-Cruz Biotechnology, Santa Cruz, CA) was used as an internal reference. Protein bands were quantified using an imaging densitometer and analyzed with Multi-Analyst Software (Bio-Rad Laboratories,Rishon Le Zion, Israel). [³⁵S]methionine incorporation and pulse-chase labeling procedures followed by anti-Otp immunoprecipitation and gel-autoradiography were performed according to published methods(Sambrook and Russell,2001).

The level of endogenous otp RNA in PACAP₃₈-treated PC12 cells was determined by quantitative real-time PCR kit (DyNAmo HS SYBR Green qPCR Kit, Finnzymes, Finland) using 7300 Real-Time PCR system (Applied Biosystems, Foster City, CA). Two micrograms of total RNA was reverse-transcribed using Moloney murine leukemia virus reverse transcriptase(Promega, Madison, USA) in a 20 μl reaction volume. PCR was performed in a 20 μl total reaction volume according to the kit instructions using rat-specific primers (Table 1)for either otp or βactin. Serial dilutions of the standard templates were also used for parallel amplification. The threshold cycles (Ct) were calculated and standard curves were plotted with Ct versus log template quantities. The quantities of samples were determined from the standard curves. otp levels were then normalized to those ofβ -actin in each corresponding sample.

Capped RNAs were synthesized with mMESSAGE mMACHINE kit (Ambion, Austin,TX) from linearized pCS2⁺ plasmids. The sequences of antisense morpholino oligonucleotides (Gene Tools, LLC, Corvallis, OR) targeted to fezl, otpb and pac1 are listed in Table 2. Splice-blocking morpholinos to fezl (Exon2-Intron2), otpa (Exon2-Intron2)and pacap1b translation start blocking morpholino were previously described (Jeong et al., 2006; Ryu et al., 2007; Wu et al., 2006). fezl otpb and pac1 RNAs were injected at the concentrations indicated in Table 3 into embryos at the two- to eight-cell stage. Injected embryos were allowed to develop at 28.5°C.

Effects of the various genetic perturbations on DA and IT neurons were analyzed by counting the number of cells in the respective neuronal cluster of a given treated population of embryos and thereafter one-way ANOVA test was performed followed by the Tukey method for comparison of means using JMP software (SAS Institute, Cary, NC). The non-treated controls of the various treatments were pooled together after we determined (Tukey analysis) that there was no significant difference between the various control samples.

To address the mechanism by which coordinated generation of different hypothalamic neurons is achieved, we focused on the DA and IT neuronal populations, which play major roles in neuroendocrine modulation of the anterior and posterior pituitary, respectively. IT neurons express isotocin-neurophysin (IT-NP), which is the zebrafish ortholog of the mammalian oxytocin-neurophysin (OT-NP) precursor protein(Unger and Glasgow, 2003). In wild-type (WT) embryos, IT neurons were readily detected in the neurosecretory preoptic nucleus (NPO), which is analogous to the mammalian supraoptic nucleus(SON) (Peter and Fryer, 1983),whereas hypothalamic DA cells developed in several clusters located at the basal plate posterior tuberculum (PT), adjacent to IT neurons(Fig. 1A). These DA clusters express the previously described (Puelles and Rubenstein, 2003; Rohr et al., 2001) hypothalamic markers nk2.1a/titf1a,nk2.1b/titf1b and sim1 (N.B., H.N.-R. and G.L., unpublished material).

We have previously characterized the zebrafish mutant, too few(tof^m808), in which the development of hypothalamic DA neurons is significantly impaired because of a recessive mutation in the gene encoding the Fezl zinc-finger-containing protein (fezl, also known as fezf2 - ZFIN) (Levkowitz et al.,2003). Simultaneous examination of IT and DA neurons revealed that the development of IT cells was attenuated in tof^m808embryos, which also displayed a clear deficit in DA neurons(Fig. 1B). However,hypothalamic hypocretin- and somatostatin-secreting neurons differentiated properly in the tof^m808 mutant(Fig. 1C-F). Thus, proper development of both IT and DA neurons requires Fezl activity.

We sought to identify the molecular events underlying fezl/tof^m808 activity. As IT and DA are generated in the NPO and PT,respectively, we performed an RNA in situ hybridization screen to identify candidate hypothalamic molecules that are expressed in these two adjacent hypothalamic nuclei (J.B. and G.L., unpublished material). We identified two genes, the homeodomain-containing gene, otpb, and the G-protein-coupled receptor, pac1 (also known as adcyap1r1 -ZFIN), that displayed a discernible expression pattern in the developing zebrafish hypothalamus (Figs 2, 3 and see Fig. S1 in the supplementary material). The zebrafish otpb gene is one of the two predicted orthologs of the mammalian Otp, and PAC1 is the predominant high-affinity receptor for the pituitary adenylate cyclase activating polypeptide (PACAP) neuropeptide (Vaudry et al., 2000). We observed that both Otp and PAC1 are expressed in the developing PT and NPO adjacent to or at the Fezl-expressing domains,respectively, suggesting that these molecules may be involved in differentiation of the two hypothalamic nuclei(Fig. 2A-D and see Fig. S1 in the supplementary material). Otp protein was detected later in terminally differentiated DA and IT neurons (Fig. 2E,F). Although PAC1 was readily detected in TH⁺ DA neurons at 24-36 hpf, its expression in the later appearing (44-48 hpf) IT cells was nearly undetectable by the time these neurons underwent terminal differentiation (see Fig. S2 in the supplementary material and data not shown).

As both Otp and Fezl are critical determinants of hypothalamic development we examined whether Fezl might regulate otpb by comparing its expression in WT embryos to tof^m808 mutants and to fezl knockdown embryos. tof^m808 embryos displayed a complete loss of otpb gene expression in the PT area, and a slight reduction in otpb⁺ cells of the NPO(Fig. 3B, n=70/70). The expression of otpb in the ventral diencephalon and hindbrain was unaltered in tof^m808 mutants. Complete inactivation of Fezl by injecting two independent splice-blocking morpholino oligonucleotides into tof^m808 (denoted tof ^fezlMO) resulted in a stronger effect; markedly diminished otpb expression was observed in the PT, NPO and ventral diencephalon, but not in the hindbrain(Fig. 3C, n=10/15). The latter result suggests that the tof^m808 allele of fezl is a hypomorph that retains residual transcriptional activity. In agreement with the stronger effect of fezl knockdown on otpb expression, the tof^m808 mutant allele displayed a delay in IT development whereas tof ^fezlMO embryos displayed a sustained loss of IT neurons (Fig. 3G). Finally, the expression of the G-protein-coupled receptor pac1 in the PT, NPO and pituitary was not affected by gene perturbations of fezl/tof(Fig. 3D-F, n=60). Hence, the Fezl transcription factor specifically regulated the expression of otpb at two distinct hypothalamic nuclei that produce IT and DA neurons.

To address the role of otpb in DA and IT development we examined the phenotype of these neurons following genetic perturbation of otpb. Consistently with the regulated expression of otpb in the NPO and PT, injection of a splice-blocking morpholino directed against otpb (denoted otpbMO) impaired the development of IT and DA neurons (Fig. 4B,C and see Fig. S3 in the supplementary material). We then analyzed the effect of otpbMO on discrete DA clusters of the PT in 2- to 3-day-old embryos. These neuronal clusters can be readily identified in the PT of 2- to 5-day-old zebrafish embryos (Rink and Wullimann,2002). Interestingly, otpbMO treatment had a stronger effect on the posterior hypothalamic DA groups 3-6, whereas group 2 was less affected (Fig. 4C and see Fig. S3 in the supplementary material).

The G-protein-coupled receptor, pac1, was detected in the zebrafish NPO and PT before and during the period of DA and IT differentiation(Figs 2 and 3 and see Fig. S2 in the supplementary material). This suggested that pac1 might be a good candidate transducer of an extracellular cue, which may be involved in DA and IT development. Indeed, knockdown of pac1 gene activity impaired the development of both IT and DA neurons (Fig. 4F and see Fig. S3 in the supplementary material). As in the case of otpbMO, inactivation of pac1 mainly affected hypothalamic DA groups 3-6, implying a common genetic pathway shared by otp and pac1 (Fig. 4B,C,F). Similarly to the reported tof^m808 phenotype(Levkowitz et al., 2003; Rink and Guo, 2004), DA cell groups in the retina, telencephalon and pretectal diencephalon, as well as TH⁺;Otp⁺ noradrenergic neurons of the locus coeruleus(LC), were not affected by otpb or pac1 inactivation (data not shown). Consistently with the pac1 knockdown phenotype, the high-affinity ligand to PAC1, PACAP1b, was expressed in the NPO and PT, and was found to be necessary for IT and DA development (see Figs S1,S3 in the supplementary material). The pacap1b morphant, however, had a more pleiotropic effect including diminution of mid-hindbrain boundary and cerebellar structures, suggesting that PACAP1b may bind to additional G-protein-coupled receptors other than PAC1(Vaudry et al., 2000).

Our results thus far suggest that differentiation of IT neurons as well as of discrete clusters of hypothalamic DA cells is regulated by the transcription factors Fezl and Otp and by the G-protein-coupled receptor,PAC1. We next performed a genetic epistasis analysis to reveal the hierarchical interaction between fezl, otpb and pac1. For gain of function of Otpb we first determined the dose of otpb RNA(7-15pg) that could rescue the otpb morphant without affecting patterning (Fig. 4D and data not shown). In order to activate the PAC1-mediated signaling pathway, we generated a constitutively active form of the PAC1 receptor(Cao et al., 2000), denoted PAC1^*. This construct had no obvious effect on patterning even at the highest dose used (Fig. 4Hand data not shown). Injection of pac1^* mRNA rescued the pac1 knockdown phenotype, but could not complement otpbdeficiencies, suggesting that Otp is acting downstream of PAC1(Fig. 4E,H). Consistently, the deficiencies in DA and IT neurons that were caused by injection of pac1MO could be rescued by co-injection of mRNA encoding to the Otpb protein (Fig. 4G). Rescue of pac1 morphant by otpb mRNA was not due to upregulation of endogenous PAC1 or its ligand PACAP1b (see Fig. S5 in the supplementary material).

Although otpb expression was diminished in the DA-deficient tof^m808 mutant (Fig. 3), overexpression of otpb RNA could not rescue the tof phenotype (data not shown). We hypothesized that this was due to insufficient Otpb protein expression levels in hypothalamic precursors. We thus attempted to rescue tof^m808 by injection of heat-shock-inducible expression vector, which drives either Otpb or PAC1^* together with a GFP tracer(Bajoghli et al., 2004),followed by selection of mosaic embryos expressing high GFP levels in the hypothalamus (see Fig. 5F). Using this system, we were able to obtain complete rescue of the tofphenotype by gain of function of Otpb (Fig. 5C; n=9/11), but not of PAC1^*(Fig. 5D; n=0/12). Taken together the above epistatic analyses show that otpb acts downstream of both fezl and pac1 and that proper development of DA and IT neurons requires the activation of two signaling pathways that converge to coordinate otpb function at different hypothalamic nuclei.

To determine the critical period of competence, in which Otpb is able to control DA development we temporally expressed otpb in zebrafish embryos using the aforementioned heat-shock-inducible Otpb construct. Induction of ectopic expression of the Otpb protein was detected at 90 minutes after shifting the temperature from 28°C to 38°C (data not shown). Gain of function of Otpb by such temporal temperature shift resulted in a supernumerary DA phenotype (∼twofold increase) when Otpb was induced at 7 and 10 but not at 14 hpf (Fig. 5E-G). Accordingly, Otpb could rescue the tof phenotype when its expression was induced at 7 but not 14 hpf(Fig. 5C and data not shown). As the majority of DA precursors exit the cell cycle by 14-16 hpf (N. Russek-Blum and G.L., unpublished results) we conclude that Otpb is required in hypothalamic DA progenitors and not in post-mitotic neurons.

Although we demonstrated that otpb is epistatic to pac1,knockdowns of pac1 and of pacap1b had no effect on otpb transcript levels (Fig. 6C and supplementary material Fig. S4). Therefore, unlike the evident effect of fezl/tof on the levels of otpb RNA(Fig. 3A-C), regulation of otpb transcription could not account for the genetic interaction between pac1 and otpb. We then monitored Otp protein in whole embryos following knockdown of either pac1 or its ligand, pacap1b (Fig. 6 and see Fig. S4 in the supplementary material). The levels of Otp protein were moderately reduced in the hypothalamus and hindbrain following injection of pac1MO (Fig. 6D; n=25/30). Western blot analysis of Otp, which allows more accurate quantification of the effects of pac1MO, resulted in a 50% decrease in total Otp protein levels (Fig. 6D). As the anti-Otp antibody we used was raised against a C-terminal epitope of Otp (see Materials and methods) it recognizes both Otpb and its paralog Otpa. These two genes have nearly complete overlapping expression domains (Ryu et al.,2007) (data not shown). To demonstrate the net effect of PAC1 on Otpb protein we compared Otp immunoreactivity in otpa morphants(Fig. 6F; n=22) versus pac1+otpa double morphants(Fig. 6H). Otp immunoreactivity was markedly reduced in the double morphant with no significant change in the levels and in expression pattern of otpb RNA, suggesting that PAC1 might modulate otpb post-transcriptionally(Fig. 6G,H; n=29/30). To further examine this possibility we expressed zebrafish PAC1, PACAP1b and Otpb in a heterologous cell culture system. In agreement with Lin et al.(Lin et al., 1999), transient expression of zebrafish Otpb in the human HEK293 cell line followed by western blot analysis detected an Otp-immunoreactive protein band with an apparent molecular mass of 50 kDa (see Fig. S4 in the supplementary material). Otpb protein levels were increased two- to threefold after coexpression of Otpb with PAC1^*, the constitutively activated form of the receptor(Fig. 7A). Higher induction(sixfold) of Otpb protein was detected following coexpression of Otpb and PAC1 together with the PACAP1b precursor protein, but not with an N-terminally truncated form of PACAP1b, denoted Δ1-5-PACAP, which acts as a PAC1 antagonist (Robberecht et al.,1992) (Fig. 7A). Hence, the in vivo genetic interaction between zebrafish pac1 and otpb could be reconstituted in a mammalian system in vitro.

We next examined whether stimulation of PAC1 with a synthetic ligand could affect Otp protein levels. Treatment of otpb;pac1double-transfected cells with increasing concentrations of a synthetic PACAP₃₈ peptide led to a four- to fivefold increase of Otpb protein(Fig. 7B). We then tested the effects of PACAP₃₈ on otp gene products in the rat PC12 cells, which express endogenous Otp (Fig. 7C,D), and in which the PACAP-PAC1 pathway has been extensively studied (Vaudry et al., 2000). Treatment of PC12 cells for up to 120 minutes with different concentrations of a synthetic PACAP₃₈ neuropeptide led to a fivefold increase of endogenous levels of Otp protein (Fig. 7C,E). In accordance with our in vivo results, quantitative real-time PCR analysis indicated that the level of otp RNA in PC12 cells was unaffected by PACAP stimulation, suggesting that a biochemical pathway triggered by PACAP and its receptor PAC1 regulates Otp at the post-transcriptional level (Fig. 7E). To determine the nature of this post-transcriptional control we examined the ability of PACAP to affect Otp protein stability and synthesis. Otp stability was examined by treating PC12 cells with the translation inhibitor cycloheximide (CHX) and thereafter monitoring the levels of Otp in the absence and presence of PACAP. This analysis showed that Otp protein was relatively stable throughout the time of the experiment and that PACAP-induced accumulation of Otp was blocked by CHX(Fig. 7D,E). This result was corroborated by [³⁵S]methionine pulse-chase kinetics analysis of Otp protein (see Fig. S9 in the supplementary material). To examine Otp synthesis, we measured [³⁵S]methionine incorporation into Otp in the presence or absence of PACAP. The rate of CHX-sensitive[³⁵S]methionine incorporation into Otp was significantly increased upon application of PACAP to PC12 cells(Fig. 7F). Taken together the above results indicate that PACAP controls the levels of Otp by promoting Otp synthesis without affecting the stability of both protein and mRNA.

In conclusion, we identified a novel regulatory network of cell-intrinsic and cell-extrinsic cues that act together to maintain coordinated development of hypothalamic DA and OT-like neurons.

In this study, we examined the genetic and biochemical basis for the apparent coordinated fashion in which DA and IT/OT neurons are generated within the relatively small hypothalamic territory. We show that the integration of transcriptional and post-transcriptional inputs that modulate the levels of the homeodomain-containing protein Otp in distinct hypothalamic nuclei (i.e. PT and NPO) may trigger specific differentiation programs that promote DA and IT identities in time and space(Fig. 8).

It has been suggested that the intersection between the secreted molecules Shh, Fgf8 and Bmp7 creates an induction site for hypothalamic DA identity(Ohyama et al., 2005; Ye et al., 1998). In spite of these reports, the regulation of cell-autonomous determinants, which presumably convert patterning signals into precise control of hypothalamic development, is poorly understood. Otp is a critical cell-intrinsic determinant, which controls the fates, migration and terminal differentiation of mammalian hypothalamic neuroendocrine cells(Acampora et al., 1999; Wang and Lufkin, 2000). We show that regulation of otp levels is important for the spatial and temporal development of IT and DA neurons. We suggest that the tight regulation of Otp is achieved by two sequential manners: first transcription of otp mRNA is induced by Fezl, then Otp protein levels are modulated by PAC1 (Fig. 8). In support of this model, we show that the transcript levels of otpb were markedly affected in the absence of fezl/tof gene function and that the levels of Otp protein were controlled by PAC1 and its ligand PACAP (Figs 3 and 6). Our in vitro analyses show that PACAP affects the rate of Otp protein synthesis, providing a mechanism for the post-transcriptional control of otp, which was observed in vivo (Fig. 7). Interestingly,PACAP exerts a persistent post-transcriptional effect on the steady-state levels of tyrosine hydroxylase in PC12 cells(Corbitt et al., 1998).

fezl, otp and pac1 are coordinately expressed in the developing hypothalamus (Fig. 2). We report here that all three molecules are necessary for proper development of hypothalamic DA and IT neurons (Figs 4 and 5). Our epistatic and gene expression analyses are consistent with two parallel pathways that co-regulate IT and DA development by acting upstream of otpb.

The levels of otpb expression may also determine the time of appearence of IT neurons. Thus, the expression of otpb in the NPO was strongly affected in a null tof ^fezlMO mutant(i.e. tof^m808 injected with fezlMO) and only mildly affected in the tof^m808 hypomorph(Fig. 3A-C). These levels of otpb transcripts correlated with the sustained absence of IT in tof ^fezlMO and delayed IT development in the tof^m808 hypomorph allele(Fig. 3G). Conversely, the deficit in DA neurons in tof^m808 is maintained throughout embryogenesis and in adult mutant animals(Rink and Guo, 2004). The varied sensitivities of otp, IT and DA to Fezl activity correlate with fezl and otp expression patterns: whereas fezland Otp colocalize in the NPO where IT cells develop, only otp is found in DA neurons of the PT (Fig. 2). Lack of Fezl-Otp colocalization in the PT is consistent with our previously published mosaic analysis showing that fezl/tofregulates hypothalamic DA development in a non cell-autonomous manner(Levkowitz et al., 2003). Hence, coordination of IT and DA cells may be attained by different sensitivities of the otpb promoter to Fezl-mediated signal at different hypothalamic territories.

Recent studies described the phenotype of Fezl-deficient mice. Similarly to zebrafish, mammalian Fezl is expressed in the telencephalon and diencephalon(Hirata et al., 2004; Mutsuga et al., 2005). In the mouse telencephalon, Fezl is required for fate specification and axonal projections of cortico-spinal motor neurons, subplate cortical neurons and deep-layer pyramidal neurons (Chen et al.,2005a; Chen et al.,2005b; Hirata et al.,2004; Molyneaux et al.,2005). A close homolog of Fezl, denoted Fez, is expressed in the mouse hypothalamus in partially overlapping domains with Fezl and fez-fezl double deficient mouse displays defects in diencephalic subdivisions (Hirata et al.,2006a; Hirata et al.,2006b). A similar role for zebrafish Fezl in diencephalic patterning was also reported (Jeong et al., 2007).

The precise regulation of Otp, DA and IT by the tof^m808 hypomorph (Figs 1 and 3) allows us to separate the role of Fezl in regional patterning from its more selective role in cell specification. The fact that Otpb could rescue the fezl/tof phenotype indicates that Otpb is a critical target of Fezl that underlies its effect on hypothalamic differentiation. In this respect, the proneural gene neurogenin1 (neurog1) is regulated by Fezl and is both necessary and sufficient for zebrafish DA development(Jeong et al., 2006). Similarly, dlx2, which controls specification of ventral thalamic DA progenitors in the mouse, is regulated by Fezl(Andrews et al., 2003; Yang et al., 2001). Neither neurog1 nor dlx2 is affected in either otpb or pac1 morphants (data not shown), suggesting that Fezl may control both early regional diencephalic commitment, which is mediated by Neurog1 and Dlx2, and later cell-type specification, which is mediated by Otp(Fig. 8). Notably, there are two zebrafish orthologs of the mammalian Otp gene (denoted otpa and otpb). Although this study demonstrates the function and mode of regulation of otpb, a similar deficit in diencephalic DA neurons was recently found in a null mutant allele of otpa, suggesting that the activity of both otpa and otpb is required for hypothalamic development(Ryu et al., 2007). We found that the transcription of otpa was affected in the absence of the fezl/tof gene, reinforcing the significance of Fezl in regulating hypothalamic cell fate decisions (see Fig. S5 in the supplementary material).

Although the development of diencephalic neurons was not analyzed in detail, hypothalamic neurons appear to be present in Fezl^-/- mice(Chen et al., 2005a; Hirata et al., 2004). However,in fezl/tof^m808 mutants, otpb morphants and in otp-deficient mice, selective groups of hypothalamic DA neurons are reduced or missing, whereas other DA groups develop normally(Rink and Guo, 2004; Ryu et al., 2007)(Fig. 4). Further analysis is necessary to clarify whether subsets of hypothalamic neurons are affected in fez/fezl-deficient mice.

Finally, activation of the Otp regulatory network might be relevant to adult physiological states as fezl, pac1 and oxytocin are upregulated in the rat SON in response to hyper-osmotic conditions and fezl and oxytocin are downregulated following sustained hypo-osmolality(Gillard et al., 2006; Mutsuga et al., 2005). Moreover, PACAP is enriched in rat mesencephalic DA neurons and protects DA neurons from neurotoxin-induced death(Chung et al., 2005; Grimm et al., 2004; Reglodi et al., 2004).

In sum, our data reveal two novel genetic pathways, which control Otp activity during differentiation of hypothalamic DA and OT-like neurons and may be relevant to hypothalamic developmental defects that cause metabolic and psychiatric clinical disorders.

Thanks are due to Thomas Czerny for kindly providing the pSGH2 expression vector; E. Peles, O. Reiner, M. Fainzilber and T. Volk for comments on this work; Shifra Ben-Or and Amos Gutnick for help with the bioinformatics and statistical analyses; Philippe Vernier and members of the Levkowitz lab for stimulating discussions. This study was supported by the Israel Science Foundation grant number 944/04, Parkinson Disease Foundation, Benozyio Center for Neurological Disorders, The Helen and Martin Kimmel Stem Cell Research Institute and Minna James Heineman Foundation. G.L. is an incumbent of the Tauro Career Development Chair in Biomedical Research.