The analysis of dopaminergic (DA) systems in mammalian model organisms has focused on the mesdiencephalic DA systems of the substantia nigra and ventral tegmental area based on the relevance of the mesostriatal, mesolimbic, and mesocortical systems for Parkinson's disease and other disorders. However, the desire to understand the development, evolution, and function of catecholaminergic (CA) systems has in parallel resulted in a detailed analysis of DA and noradrenergic (NA) neurons and their projections in all major vertebrate groups (Smeets and González,2000). Tyrosine hydroxylase (TH), the rate-limiting enzyme that catalyzes the first step of catecholamine biosynthesis, has been classically employed as the marker of choice to visualize both DA and NA neuronal systems in all animal models analyzed so far. NA neurons express Dopamine beta Hydroxylase (DBH) in addition to TH. The expression of the reuptake carrier Dopamine transporter (DAT) positively identifies many, but not all DA neurons. The combined analysis of markers revealed that DA neurons are typically located in defined groups of the fore- and midbrain, while NA neurons are restricted to the hindbrain and peripheral nervous system in most vertebrates. Zebrafish (Danio rerio) promised to be a particularly useful model organism to understand CA development, because the small size and transparency of early larvae allows visualization of the complete central nervous system, and a variety of experimental approaches can be combined, including genetic screens, experimental manipulations, pharmacological treatments, and transgenesis for fluorescent tagging or gain-of-function analysis. These tools have revealed specific highly conserved as well as other evolutionary more dynamic CA systems.
Catecholaminergic Systems Anatomy and Development
The organization and projectivity of catecholamine-producing neurons have been extensively analyzed in the adult zebrafish brain primarily by TH immunohistochemistry (Ma,1994a, b, 1997, 2003; Kaslin and Panula,2001; Rink and Wullimann,2001). These studies highlighted the presence of TH immunoreactive (THir) neuronal somata in basically every division of the zebrafish brain, with the important exception of the midbrain (see Fig. 1). The analysis of DBH expression revealed that NA neurons are confined to the hindbrain (Ma,1994a, b, 1997), and all the THir neurons developing in the forebrain were assumed to be DA because they lacked DBH immunoreactivity. From rostral to caudal, heterogeneous groups of CA populations were described and referred to specific areas using the atlas of the zebrafish brain as anatomical reference (Wullimann et al.,1996). However, clear boundaries could not always be identified between cells residing in different nuclei. In the telencephalon, DA neurons were detected in the olfactory bulb (OB) and in dorsal, central and ventral nuclei of the ventral telencephalic area (Vd, Vc, Vv) (Kaslin and Panula,2001; Rink and Wullimann,2001). In the diencephalon, DA neurons were detected in the anterior and posterior parts of the parvocellular preoptic area (PPa, PPp), in the suprachiasmatic nucleus (SC), in the periventricular pretectum (PPr) and ventral thalamus (VL, VM), in the periventricular nucleus of the posterior tuberculum (TPp) and paraventricular organ (PVO), in the posterior tuberal nucleus (PTN) and in the caudal hypothalamus (Hc) (Kaslin and Panula,2001; Rink and Wullimann,2001; Ma,2003). In the hindbrain, a rostral NA group was described in the locus coeruleus (LC), and three caudal clusters were identified in the interfascicular area, vagal area, and area postrema of the medulla oblongata (MO) (Ma,1997; Kaslin and Panula,2001).
Figure 1. Catecholaminergic groups and major tracts in the 4-day old zebrafish larva. Schematic drawing of DA (blue) and NA (red) neuron groups as well as TH-immunoreactive CA tracts (gray) in 4 dpf zebrafish larvae (from Kastenhuber et al.,2010; Tay et al.,2011). (A) Lateral view of fore- and midbrain; (C): dorsal view of CNS. The different shades of blue indicate the relative dorsoventral positioning of CA groups: Dark blue—dorsal-most, light blue ventral-most. Diencephalic DA groups: DC1—ventral thalamus and periventricular posterior tuberculum; DC2, DC4—large neurons in posterior tuberculum; DC5, DC6—medium-sized neurons in posterior tuberculum and hypothalamus; DC3—anterior hypothalamus liquor contacting neurons; DC7—caudal hypothalamus; PO—preoptic region; POa—anterior preoptic region; Pr—dorsal pretectum; RAC—retinal amacrine cells. Telencephalic DA groups: SP—subpallium; OB—olfactory bulb. NA groups: LC—locus coeruleus; MO—medulla oblongata interfascicular zone and vagal area and AP—area postrema. Abbreviations of major CA axonal tracts: ac, anterior commissure; act, anterior CA tract; eht, endohypothalamic tract; mlct, medial longitudinal CA tract; pc, posterior commissure (not shown in C); poc, postoptic commissure; poht, preopticohypothalamic tract; prp, pretectal projections; prtep, pretectotectal projections. (B): Summary of birthdating analysis and time of differentiation for catecholaminergic groups. For each neuronal group, the time windows during which most of the dopaminergic neurons of the 4dpf larvae become postmitotic (Mahler et al.,2010) is shown, together with the age from which onward terminal differentiation occurs, as judged from tyrosine hydroxylase expression (Holzschuh et al.,2001). (D): Schematic lateral view of catecholaminergic systems of the rodent brain (modified from Björklund and Dunnet, 2007). Major dopaminergic projections are indicated. The locus coeruleus noradrenergic projections target all major areas of the brain (not shown). The likely correlates in zebrafish are: A11—DC2,4-6; A12, 14—hypothalamic DC 3,7; A13—ventral thalamic DC1; A15—preoptic groups; A16—olfactory group; A8-10 have no anatomical correlates in zebrafish. Abbreviations: hpf—hours post fertilization.
Download figure to PowerPoint
A more complete picture of the CA organization in the zebrafish brain could be drawn based on the identification of a second tyrosine hydroxylase gene (th2) in teleosts (Candy and Collet,2005). The expression of zebrafish th2 was independently analyzed by three laboratories in embryonic/larval (Chen et al.,2009; Filippi et al.,2010), juvenile (Filippi et al.,2010) and adult (Yamamoto et al.,2010) brains. The results consistently showed that th2 expression is restricted to the diencephalon. There, th2 is expressed in four main clusters: the most rostral group was detected in the preoptic region, whereas three hypothalamic groups were found in the paraventricular organ and in two caudal nuclei lining the lateral and posterior recesses, respectively (Chen et al.,2009; Filippi et al.,2010; Yamamoto et al.,2010). Western blot analyses and comparison of th2 expression with TH protein distribution revealed that commercially available antibodies directed against mammalian TH do not efficiently recognize the TH2 protein (Filippi et al.,2010; Yamamoto et al.,2010). Double labeling experiments could show that th2 expression partly overlaps with TH in preoptic region and PVO, but is mostly exclusive in the two caudal hypothalamic domains (Filippi et al.,2010; Yamamoto et al.,2010). Notably, two populations of cells in the caudal hypothalamus had been shown to be strongly dopamine-immunoreactive but not or little THir in other teleosts (Yoshida et al.,1983; Meek et al.,1989; Ekström et al.,1990, 1992; Sas et al.,1990; Batten et al.,1993; Meek and Joosten,1993). This apparent discrepancy is now solved in the light of the recent zebrafish th2 expression data, which suggest a major role for TH2 (instead of TH) in the synthesis of dopamine in the caudal hypothalamus (Yamamoto et al.,2011).
TH2 appears to prominently contribute to dopamine synthesis in the hypothalamus at juvenile/adult stages (Filippi et al.,2010; Yamamoto et al.,2010, 2011), and also at late larval stages (Chen et al.,2009). In contrast, during embryonic and early larval stages th2 expression is much weaker as compared to th (Chen et al.,2009; Filippi et al.,2010). Thus, during embryonic and early larval stages, dopamine activity appears to be provided mostly by th. Based on the expression analysis of th and th2, our knowledge on the time of development and distribution of CA neuronal groups in the embryonic brain is likely to be quite complete. The combined analysis of th and th2 expression has also revealed that there is definitely no mesencephalic DA system in zebrafish.
The formation and distribution of CA groups in the early zebrafish brain has been studied by TH immunohistochemistry (Rink and Wullimann,2002; McLean and Fetcho,2004a; Sallinen et al.,2009) and expression analysis of th (Holzschuh et al.,2001), as well as DA and NA selective markers, like dat and dbh, respectively (Holzschuh et al.,2001; Holzschuh et al.,2003a). Already by 3-days post-fertilization (dpf), most of the groups described in the adult brain could already be detected (Rink and Wullimann,2002). While some CA groups were named based on their anatomical location (OB, subpallial group, pretectal group, PO, LC and MO), the DA neurons developing in the ventral diencephalon (DC) were assigned numbers (from DC1 to DC7) on the basis of their morphology and position along the rostro-caudal axis (Rink and Wullimann,2002) (Table 1): small round group DC1 DA neurons arise in the prethalamus at the border with the posterior tuberculum; group DC2 and 4 cells are large pear-shaped neurons located in the posterior tuberculum; groups DC3, 5, and 7 are cerebrospinal fluid (CSF)-contacting cells lining the ventricle in anterior, intermediate and posterior hypothalamus, respectively; group DC6 neurons develop in the posterior part of the posterior tuberculum and will constitute the population of small round cells populating the posterior tuberal nucleus in the adult.
Table 1. Catecholaminergic Groups in the Zebrafish Larval and Adult Brain
| ||Brain Areas Containing CA Groups in the Larval Brain (72 hpf)||CA Groups [Nomenclature Introduced by (Rink and Wullimann,2002)]||CA Groups [Nomenclature Introduced by (Sallinen et al.,2009)]||Corresponding Areas in the Adult Brain|
|Telencephalon||Olfactory bulb||OB||population 1||OB|
|Subpallium||SP (subpallial group)||population 2||Vd, Vc, Vv|
|Diencephalon||Preoptic area||PO (preoptic group)||populations 3,4(1)||PPa, PPp, PM, SC|
|Prethalamus||DC1 group (3)||populations 5,6,11(2)||VL, VM|
|Posterior tuberculum/hypothalamus||DC2 group||population 12||TPp|
|DC3 group||population 8||PVO|
|DC4 group||population 12||TP|
|DC5 group||population 9||LR|
|DC6 group||population 13||PTN|
|DC7 group||population 10||PR|
|Pretectum||DC8 or Pr - pretectal group||population 7||PPr|
|Rhombencephalon||Rhombomere 1||LC (Locus coeruleus)||population 14||LC|
|Rhombomere 7||MO (Medulla oblongata)||populations 15,16||vagal motor area, superior reticular formation|
|AP (Area postrema)||population 17||AP|
More recently, an alternative nomenclature has been suggested (Sallinen et al.,2009), in which all CA groups were assigned numbers (1–17) according to their rostral to caudal distribution. A comparison of the two distinct nomenclatures is reported in Table 1, also correlating the larval and adult CA groups. In this review, we will use the original nomenclature introduced by Rink and Wullimann (2002).
During the last two decades, the zebrafish has emerged as new model for CA systems development. Using the zebrafish to unravel genetic networks underlying CA neuron development has two major advantages: First, the large evolutionary distance between teleosts and mammals might provide an opportunity to identify core regulatory molecules of CA cell specification/differentiation which have been conserved throughout vertebrate evolution. Second, the zebrafish is the preferred vertebrate model for forward genetic screens and thus offers a possibility to identify novel genes involved in vertebrate CA cell specification/differentiation. In the following, a short overview of studies will be presented which contributed to the general understanding of CA specification and differentiation in zebrafish with a specific focus on DA neuron development.
Birth-dating analysis using BrdU (5-bromo-2-deoxyuridine) or EdU (5-ethynyl-2-deoxyuridine) incorporation together with TH-immunohistochemistry revealed the timing of cell cycle exit of zebrafish CA neurons [Russek-Blum et al.,2008; Mahler et al.,2010; Fig. 1(B)]. The first CA precursors becoming postmitotic are group DC2 DA neurons and NA neurons of the LC already before 16 hpf, suggesting that these neurons directly derive from neural plate cells. DC1 and DC4/5 groups display a less uniform pattern: the earliest postmitotic precursors for DC1 and DC4/5 are born before 16 hpf as well, however precursors of these clusters exit cycling precursor pools also later during the first day of development. Other DA and NA groups including those in the telencephalon, DC3/6/7, Pr, and PO rather display continuous neurogenesis from precursors over longer developmental time. Instead, DA precursors of the retina become postmitotic in a clearly delimited time window between 36 and 42 hpf. The same study revealed specific and distinct time windows during which the Delta/Notch neurogenic switch initiates DA neurogenesis for each of the various DA groups (Mahler et al.,2010). Despite the broad developmental time during which DA neurons form, very specific Notch ligands appear to control their neurogenesis; genetic analysis of Notch signaling indicated that DeltaA and DeltaD, the two zebrafish orthologs of the mammalian Notch ligand Delta1 are required for maintenance of DA progenitor pools, and in addition control the neurogenic switch of essentially all early forming DA neurons (Mahler et al.,2010).
Signaling Mechanisms in Catecholaminergic Specification
Given the essential role of Shh and Fgf8 in induction of mesencephalic DA neurons in mammals, their function during DA development in zebrafish was analyzed (Holzschuh et al.,2003b). In zebrafish, mutants for shh itself (syu) or its co-receptor smoothened (smu) did not display major defects in formation of ventral diencephalic DA groups but lacked late forming pretectal and amacrine DA neurons. In zebrafish ace mutants, in which Fgf8 function is disrupted, all DA groups form normally, though eventually with a slight delay. The reasons for the differential requirement of Shh and Fgf8 signals for the differentiation of DA neurons in zebrafish and mammals are presently not known. However, more recent findings in mice have also challenged the dominant roles of Shh and Fgf8 in mesencephalic DA neuron differentiation (Prakash and Wurst,2006), indicating requirement for additional signals.
These additional signals include TGF (Lints and Emmons,1999; Ohyama et al.,2005) and Wnt (Castelo-Branco and Arenas,2006) signals. In zebrafish, analysis of mutant embryos devoid of distinct TGF/Nodal pathway components (cyc/ndr2, oep and sur/foxH1/fast1) revealed strong reduction or loss of specific DA groups, especially those located in the ventral diencephalon (Holzschuh et al.,2003b). However, Nodal signaling is generally required for development of ventral cell fates in the early embryo (Sampath et al.,1998), implicating that Nodal might not be directly involved in DA cell specification. Recently, a role of canonical Wnt signaling on DA development in zebrafish has been reported (Russek-Blum et al.,2008). Ubiquitous overexpression of the Wnt signaling antagonist dkk1 resulted in an increase in number of ventral diencephalic DA neurons likely via modulation of a Wnt8b/Lef1 signaling cascade. Other cell types adjacent to DA neurons did not display changes in cell number upon dkk1 overexpression. The action of dkk1 is temporally restricted to early developmental (gastrulation) stages since late induction of ubiquitous dkk1 expression did not lead to an increase in DA cell number. By analyzing effects of distinct Wnt signals, inhibitors, and receptors on formation of DA groups, the authors of this study suggest that canonical Wnt signaling selectively limits the initial pool of DA progenitors.
Transcriptional Mechanisms of Catecholaminergic Specification
In addition to signaling molecules, various transcription factors have been identified, which participate in specification of DA neurons in mammals (for review see Smidt and Burbach,2007). In mammals Lmx1a/b is involved in specification of early precursors, while Nurr1 and Pitx3 appear to contribute to later aspects of differentiation of mesdiencephalic DA neurons. To identify conserved functions of mammalian midbrain DA determinants in zebrafish DA neuron formation, a potential requirement for lmx1b and pitx3, (Filippi et al.,2007) as well as nr4a2/nurr1 (Filippi et al.,2007; Blin et al.,2008; Luo et al.,2008) in zebrafish DA development has been analyzed.
Zebrafish lmx1b paralogs, lmx1b.1 and lmx1b.2 are expressed adjacent to posterior tubercular/hypothalamic DA neurons. Co-expression of lmx1b factors with THir neurons is only detectable for lmx1b.1 in NA neurons of the LC and MO. However, knockdown of lmx1b.1/2 function did lead to reduction of DA neurons of the ventral thalamus. Whether the reported reduction of ventral thalamic DA cells is due to a requirement for Lmx1b.1/2 in DA progenitors or caused by a non-autonomous effect of lmx1b.1/2 remains elusive. Loss of function studies on pitx3, which is not expressed in any DA group but might be expressed in a DA progenitor domain, did not affect formation of any DA group. For nr4a2 genes, different studies have reported different results, based on whether coexpression was analyzed at confocal single cell level and morpholino knockdown experiments controlled for non-specific apoptosis with p53 morpholinos (Filippi et al.,2007), or not (Blin et al.,2008; Luo et al.,2008). The zebrafish nr4a2a and nr4a2b paralogs are not coexpressed with posterior tubercular/hypothalamic DA groups either, but with preoptic (only nr4a2a), pretectal and retinal DA cells. One study (Filippi et al.,2007) reports that knockdown of nr4a2 genes did lead to a selective loss of nr4a2-expressing DA groups only. These findings are consistent with the notion that Nr4a2 is part of a conserved transcriptional module for DA differentiation, which during evolution is dynamically used in different DA populations.
Further insights into CA specification were derived from systematic genetic screens that led to the isolation of mutant alleles affecting CA development (Guo et al.,1999; Holzschuh et al.,2003a). Several of these mutations affected specific steps of DA neuron development, while others were shown to be involved in more general aspects of development, and thus are only indirectly involved in DA specification. Three new mutations led to identification of genes which had not been previously associated with DA development, namely the forebrain embryonic zinc finger-like protein Fezf2/Fezl (Levkowitz et al.,2003; Rink and Guo,2004), the homeodomain transcription factor Orthopedia (Otp) (Del Giacco et al.,2006; Blechman et al.,2007; Ryu et al.,2007), and the PAS-domain protein Arnt2 (Löhr et al.,2009).
Fezf2/Fezl is specifically expressed in the prospective forebrain field starting during gastrulation, and is required for proper development of DA groups DC2-6 (and eventually DC7). Transplantation experiments suggest that it functions in a non cell-autonomous fashion (Levkowitz et al.,2003; Rink and Guo,2004). Analysis of Fezf2/Fezl deficient mice did not reveal any defects in formation of DA systems (Hirata et al.,2004), however these mice displayed general changes in establishment of diencephalic subdivisions suggesting that Fezf2/Fezl may act in a rather global way during vertebrate forebrain patterning. Fezf2/Fezl has been shown to regulate expression of the basic helix-loop-helix factor neurogenin1 (ngn1) in diencephalic DA precursors and knockdown of ngn1 resulted in reduction of differentiated forebrain DA neurons judged by lack of th expression (Jeong et al.,2006). Moreover, overexpression of ngn1 is sufficient to induce supernumerary DA neurons in forebrain and non-neural ectoderm, which is consistent with a role of ngn1 in neurogenesis of ventral diencephalic DA neurons.
In zebrafish, Otp is selectively required for development of posterior tubercular and hypothalamic DA groups DC2 and DC4-6 in zebrafish (Del Giacco et al.,2006; Blechman et al.,2007; Ryu et al.,2007). The function of Otp in DA cell specification is conserved, since mice mutant for Otp specifically lack group A11 DA neurons in the diencephalon. Therefore DC 2 and DC4-6 can be considered as A11-type DA neurons in zebrafish. The combined loss of function of both paralogous otp genes achieved by otpb morpholino knockdown in otpa mutant embryos eliminates all DA neurons of DC2, 4-6 (Ryu et al.,2007). However, Otp is not specific for DA neurons, as it is also required for development of specific populations of neuroendocrine neurons.
Interestingly, a heterodimer of the helix-loop-helix PAS domain transcription factors Arnt2 and Sim1a is required for development of the same subset of DA neurons, which are absent in the otpa mutants (DC2 and DC4-6) (Löhr et al.,2009). arnt2 is broadly expressed throughout the embryonic and larval zebrafish nervous system. Instead, sim1a was found to be specifically localized in otp-expressing DA cell groups. Genetic analyses revealed that, similar to the specification of specific neurosecretory cells in the mammalian hypothalamus (Michaud et al.,1998; Acampora et al.,1999), Arnt2/Sim1a and Otp act in parallel pathways during zebrafish DA development (Löhr et al.,2009). Ubiquitous overexpression of Otpa and Sim1a alone can induce supernumerary neurons with DA cell characteristics (th and dat expression) in the posterior tuberculum and hypothalamus, a phenotype that is strongly enhanced by combined Otpa/Sim1a mis-expression (Ryu et al.,2007; Löhr et al.,2009). The fact that Otpa and Sim1a can efficiently induce supernumerary DA neurons only in a defined ventral diencephalic territory indicates that additional local factors are required for DA specification.
So far it is not well understood which factors act upstream of Sim1a and Otpa/b. Genetic analyses revealed that otpb is regulated by Fezf2/Fezl and the G-protein-coupled receptor PAC1 acting in parallel pathways (Blechman et al.,2007). Fezf1/Fezl regulates otpb at the transcriptional level, whereas PAC1 is involved in post-transcriptional modification of otpb thereby controlling the levels of Otpb protein. Moreover, a negative effect of FGF8 by suppressing Otp expression in the preoptic area has been observed (Del Giacco et al.,2006). Recent data provide evidence that Olig2 regulates expression of sim1a (Borodovsky etal.,2009). Further, during neurogenesis, Notch signaling via DeltaA and DeltaD ligands is involved in maintaining olig2- and ngn1-expressing DA precursors thereby regulating the size of the sim1a/otp positive cell population (Mahler et al.,2010). Further systematic analysis of expression of transcription factors in areas of DA differentiation may lead to a more detailed understanding of transcriptional regulation in the DA systems (Li et al.,2010). A detailed understanding of transcriptional control will also involve the identification of specific response elements in the control regions of DA differentiation genes, which has only recently come to fruition with the identification of control elements of otp genes (Fujimoto et al.,2011) and th gene (Meng et al.,2008).
Circuit Formation: The Catecholaminergic “Projectome”
The axonogenesis of CA systems during early developmental stages between 1 and 3 dpf has been analyzed by anti-TH immunohistochemistry (McLean and Fetcho,2004a; Sallinen et al.,2009; Kastenhuber et al.,2010). The first THir neurons in the zebrafish appear between 16 and 24 hpf in the ventral diencephalon (Holzschuh et al.,2001; McLean and Fetcho,2004a; Kastenhuber et al.,2010). By 24 hpf, the first THir axons have already passed the area of the noradrenergic LC neurons and reached the spinal cord. These axons project longitudinally and grow at a defined distance towards the midline, thereby constituting the medial longitudinal CA tract (mlct) (McLean and Fetcho,2004a; Kastenhuber et al.,2010). The THir mlct axons in the spinal cord grow closely to the dendritic processes of primary and secondary motor neurons (McLean and Fetcho,2004b) suggesting a potential function in spinal motor control. At 2 dpf the first commissural THir axons can be detected which cross the midline at the level of the rostral hindbrain (Kastenhuber et al.,2010). At 3 dpf more CA tracts can be discerned. In the forebrain the TH positive anterior- and postoptic- commissures are visible. The endo-hypothalamic tract connecting DA neurons in the posterior tuberculum and hypothalamus, the hypothalamic-hypophyseal projection and the preoptico-hypothalamic tract connecting the preoptic region with the ventral diencephalon have been established. The anterior CA tract connects the ventral diencephalon and ventral telencephalon. Furthermore, THir fibers connecting the dorsal most diencephalic DA neurons in the pretectum with the ventral diencephalon (pretectal projections) and the tectum (pretecto-tectal projections) are formed. The THir cells in the retina and olfactory bulb only project locally (see Fig. 1). Thus, the almost complete axonal scaffold of CA circuits described in the adult zebrafish (Ma,1994a, 1997, 2003; Kaslin and Panula,2001) is already present in 3-day-old larvae (McLean and Fetcho,2004a; Kastenhuber et al.,2010). Genetic dissection of the DA and NA contributions of CA tracts in 3 dpf old embryos revealed that DA neurons constitute predominantly the hypothalamic and spinal CA tracts whereas NA neurons, quantitatively, make only a minor contribution (Kastenhuber et al.,2010).
The knowledge of the projection and connection repertoire of the neurons in each of the CA groups would represent the ideal basis to understand DA and NA contributions to other circuits in zebrafish. While connections have not been mapped in zebrafish, the systematic genetic labeling of single CA neurons and their projections has revealed the complete set of projections (“projectome”) for each CA group in 4-day old zebrafish larvae (Tay et al.,2011; see Table 2). Among the DA groups, the Otp-dependent DA neurons stand out based on the complexity and range of their projection patterns. The large Otp-dependent DC2 and 4 DA neurons are the only source of ascending projections to the subpallium at this stage. The same DA neurons that send ascending projections usually also target other areas of the CNS, prominently including the hypothalamus, the tectum, the hindbrain and the spinal cord. DC5 neurons share this latter list of target areas, while DC6 preferentially project within the hypothalamus and into the hindbrain. Otp-dependent DA neurons are thus also the only major source of DA projections into hindbrain and spinal cord. Thus, confirming earlier genetic analysis (Kastenhuber et al.,2010), the Otp- dependent A11-type DA neurons are the major far- projecting DA system in zebrafish larvae. Among the other DA groups, the hypothalamic DC3 and DC7 groups project locally within the hypothalamus. Preoptic DA neurons project locally and into the hypothalamus. Pretectal DA neurons project locally within the diencephalon as well as into the tectum, where they arborize extensively. Ventral thalamic DC1 also project locally and predominantly into the hypothalamus, but also into the midbrain. The subpallial DA neurons surprisingly showed an extensive arborization within the subpallium, but also extended processes into the contralateral subpallium, the thalamus and hypothalamus (Tay et al.,2011). Quantitatively, there are only a few number of ascending DA projections from Otp-dependent DA neurons into the subpallium, indicating that the majority of DA input into the subpallium likely does not derive from ascending projections, but from endogenous subpallial DA neurons.
Table 2. Projection Target Areas of Catecholaminergic Neurons in 4-Days Old Zebrafish Larvae
| ||CA Group||Projection Target Area|
|Dopaminergic||Retinal amacrine cells||local (retina)|
|Olfactory bulb||local (OB)|
|Subpallium||local (SP), ac, dTHA, HYP|
|Anterior preoptic||local (PO), poc, eht, HYP|
|Preoptic||local (PO), eht, HYP|
|Pretectum||local (PT), pc, vTHA, TEC|
|DC1||local (vTHA), poc, PT, HYP, TEC,|
|DC2—post. tuberculum||SP, poc, lat. DC, PT, eht, HYP, TEC, HB, SC|
|DC4—post. tuberculum||SP, poc, PT, eht, HYP, TEC, HB, SC|
|DC5—hypothalamus||poc, PT, eht, pc, HYP, TEC, LC, HB, SC|
|DC6—hypothalamus||local (HYP), eht, HB|
|DC3—hypothalamus||local (HYP), eht|
|DC7—hypothalamus||local (HYP), posterior recess|
|Noradrenergic||Locus coeruleus||local (LC), OB, SP, ac, poc, lat. DC, PT, pc, POT, HYP, TEC, HB, SC|
|Medulla oblongata—interfascicular zone||local (MO), HB, SC|
|Medulla oblongata—vagal area||local (MO), SP, ac, vTHA, POT, HYP, HB|
|Medulla oblongata—area postrema||local (AP), OB, SP, ac, poc, lat. DC, PT, pc, POT, HYP, TEC, HB, SC|
The impact of this pattern of DA projections for the understanding of circuits controlling behavior still has to be determined. Given the absence of a mesencephalic DA system in zebrafish, one can currently only speculate that endogenous subpallial DA neurons may provide DA modulation of subpallial/striatal circuits in zebrafish (Tay et al.,2011). Whether input from other areas, e.g., the thalamus, may be relayed to subpallial DA neurons, which thus may execute similar modulatory function in the fish subpallium as the nigrostriatal system in mammals, needs to be determined.
In contrast to several DA groups, which appear evolutionary highly dynamic, the analysis of the NA projections indicates a high degree of conservation with respect to NA axonal tracts and target areas from fish to mammalian systems (Smeets and González,2000; Tay et al.,2011). Interestingly, essentially all areas of the larval fish brain that are targets of DA projections also receive NA projections. Among the new findings in this work are ascending NA projections from the medulla oblongata into the subpallium and other telencephalic areas (Tay et al.,2011). Thus, not only the locus coeruleus, but also more caudal NA groups provide telencephalic input. To determine the relative DA and NA contributions to CA neuromodulation in the telencephalon is an exciting challenge for the future.
Mechanisms of Axonogenesis
In contrast to the extensively analyzed CA projections, the cellular and molecular mechanisms underlying axonal pathfinding of DA as well as NA neurons in fish are less well understood. The only studied CA projection system with regard to axon guidance mechanisms so far is the formation of the mlct (see Fig. 2A). While growing from the diencephalon towards the spinal cord, the mlct growth cones do not navigate along the primary neuron axonal scaffold, but establish their own projections at defined medio-lateral positions (Kastenhuber et al.,2010). These findings suggest that formation of the mlct depends on a particular set of guidance molecules. During mlct pathfinding the DA neurons in the ventral diencephalon co-express the two axon guidance receptors dcc and robo2. slits and netrins, which are the ligands for the Robo2 and DCC receptors respectively, are expressed at the ventral midline (Kastenhuber et al.,2009). The DCC/Netrin system primarily mediates axon attraction, whereas Robo/Slit signaling mediates repulsion (reviewed in Rajasekharan and Kennedy,2009; Ypsilanti et al.,2010). Lateral positioning of longitudinal DA axons in robo2 mutants is defective. Instead of selecting a defined medio-lateral position, THir axons grow toward and eventually across the midline. Knocking down either dcc or netrin1 function in robo2 mutant background leads to a partial rescue of the observed pathfinding errors (Kastenhuber et al.,2009). These findings suggest that simultaneous integration of repulsive Slit and attractive Netrin signals from the midline act in a concerted manner to define lateral positions of the mlct. Heparan sulfate proteoglycans, which are important for axon guidance (Lee and Chien,2004), are also required for lateral positioning of the mlct (Kastenhuber et al.,2009). Thus, multiple guidance cues cooperate to orchestrate lateral positioning of the far-reaching mlct axons (see Fig. 2B). Future work on axon guidance mechanisms of CA neurons will benefit from recently identified enhancers driving expression in DA neurons (Fujimoto et al.,2011). These enhancers may be used to visualize DA axons, and in combination with gain and loss of function approaches will facilitate in vivo analysis of DA axon guidance.
Figure 2. Establishment of the medial longitudinal catecholaminergic tract. (A) Dorsal view of a confocal z-projection of a 3 dpf embryo of brain and hindbrain region is shown. Anterior to left. Immunohistochemistry for TH reveals diencephalic DA neurons, the NA locus coeruleus neurons in the hindbrain and their respective axons. The mlct is primarily derived from diencephalic DA neurons. (B) Schematic drawing of a region comparable to A is shown. Coordinated action of repulsion by Robo/Slit (red arrows), attraction by DCC/Netrin (green arrows) and function of HSPGs (blue ovals) specify formation and lateral positions of longitudinal DA axons (indicated by black lines). LC indicates noradrenergic locus coeruleus neurons in the hindbrain, mlct; medial longitudinal catecholaminergic tract. Groups 1–5 according to the nomenclature of Rink and Wullimann (2002). Scale bar in A = 50 μm.
Download figure to PowerPoint
DA Systems Function
Analysis of DA circuit function in zebrafish would require knowledge not only of DA projections, but also of synapses and DA receptor expression in postsynaptic neurons. Expression of zebrafish DA receptors has only been characterized at an anatomical level so far (Boehmler et al.,2004; Boehmler et al.,2007; Li et al.,2007). The only circuits characterized at cellular level so far are local DA circuits that closely resemble those in higher vertebrates, including the olfactory system (Byrd and Brunjes,1995) and the retina, for which the ease of pharmacological manipulations has helped analysis of the role of local dopamine input (Li and Dowling,2000; Yu and Li,2005). DA contribution to modulation of behaviors has been investigated in a series of studies, but the circuits underlying these functions are largely unknown. Specific behaviors for which involvement of DA input has been shown include aggression (Filby et al.,2010), addiction (Darland and Dowling,2001; Bretaud et al.,2007b), and control of motor behaviors (Burgess and Granato,2007; Thirumalai and Cline,2008). Similar to tetrapod vertebrates, dopamine modulates prepulse inhibition (Burgess and Granato,2007). Further, endogenous dopamine may transiently suppress swim circuits in developing zebrafish larvae (Thirumalai and Cline,2008). Further, effects of the application of DA neurotoxins have been reported with respect to motor behavior of larvae (Anichtchik et al.,2004; Lam et al.,2005; Sallinen et al.,2009), although the cell populations affected and the circuits mediating behavioral changes need further characterization.
Zebrafish experimental features have generated a strong interest as a model for neurodegenerative and neuropsychiatric diseases (Panula et al.,2010). Several recent studies demonstrate that loss- or gain-of- function for proteins that affect mammalian DA neurons and have been linked to Parkinson's disease also affect DA neurons in zebrafish (Bretaud et al.,2007a; Anichtchik et al.,2008; Flinn et al.,2009; Fett et al.,2010; Sheng et al.,2010; Xi et al.,2010). Based on these studies, zebrafish appear to be a good model with respect to the cellular functions and cell physiology of DA neurons related to neurotoxicity, and may provide a good model to develop and screen for neuroprotective approaches. However, when analyzing pathways of neural specification, one has to keep in mind the evolutionary differences between fish and mammals: there is no direct cellular correlate to mammalian substatia nigra/ventral tegmental area DA neurons in zebrafish. It is clear now that the prominent zebrafish posterior tubercular Otp-dependent DA neurons that send ascending projections to the subpallium both with respect to their genetic specification as well as their projection patterns do not represent a rostrally shifted mes-diencephalic system, but are the equivalent of the mammalian A11 DA system. Thus, a comparison of behavioral phenotypes between mammals and fish has to be undertaken with great care, and needs to consider the A11 and other DA systems, which are also less well understood in mammals.
The extraordinary range and quantitative prominence of projections of A11-type, Otp-dependent DA neurons excites interest in their roles in control of behavior throughout vertebrate evolution. Several findings point towards functions of the A11 system: (1) The evolutionary ancient shared Otp- and Sim1/Arnt2 dependent transcriptional mechanisms specifying A11-type DA neurons as well as specific neuroendocrine neurons, including CRH neurons (Ryu et al., 2007; Löhr et al., 2009), argue for a close link with the neuroendocrine systems. (2) In addition similarities exist between descending diencephalospinal projections of Otp-dependent DA and those of CRH as well as other neuroendocrine neurons (Krisch,1981; Sawchenko,1987; Hallbeck and Blomqvist,1999), which have been implied with somatomotor function (Rose and Moore,2002; Kerman,2008). These findings led to the hypothesis that the Otp- dependent DA system may be involved in control of basic motor activity patterns, for example switching between active (fight/flight) and passive coping (freezing) (Löhr et al.,2009; Tay et al.,2011). Interestingly, the A11 system has been linked to the motor disorder Ekbom Syndrome/restless legs syndrome (Ondo et al.,2000; Clemens et al.,2006). Whether the A11 system is also affected in dopamine neurodegenerative diseases still needs to be investigated.
The endogenous zebrafish subpallial DA neurons further stimulate thoughts about dopamine function in the subpallium/striatum from fish to mammals. While DA neurons have also been reported in the primate striatum (Betarbet et al.,1997), they quantitatively may represent a minor DA input compared to the endogenous subpallial DA system in zebrafish. It remains to be determined whether subpallial DA neurons in zebrafish provide the major DA input into the subpallium/striatum, and whether their depletion may provoke similar behavioral phenotypes as depletion of nigrostriatal input in mammals.
In summary, it appears that developmental, cellular, and behavioral analysis in the zebrafish model should contribute to a better understanding of CA circuit function and evolution in vertebrates.