The function of mature neural networks is absolutely dependent upon their development, which requires a complex interplay of numerous morphogenetic events including neurogenesis, migration, neurite outgrowth and synaptogenesis.
Disturbances in the development of neural networks likely underlie the etiology of neuropsychiatric diseases such as schizophrenia, and pervasive developmental disorders such as autism. It is therefore critical to identify factors necessary for normal CNS development, since they may serve as biomarkers for pathological disease states as well as possible avenues for therapeutic intervention. To facilitate elucidation of these factors, we have characterized a simple neural circuit, the 5-HT stomatogastric feeding circuit in the Drosophila larva. We demonstrated an inverse relationship between neuronal 5-HT levels during development of the circuit and the complexity of the 5-HT axonal fibers projecting from the larval brain to the foregut, which correlate with perturbations in feeding, the functional output of the circuit. In mammals, the neurotransmitter actions of DA modulate the motivation or anticipation of feeding behavior, as opposed to the behavior itself [e.g., [34–36].] Rats in which mature DA neurons have been lesioned with 6-hydroxydopamine display severe aphagia, but can recover from this treatment to eat normally . Our data show that induction or reduction of neuronal DA synthesis in the larval stage has no effect on feeding. However, perturbations in DA levels during late embryogenesis have a significant effect on the function of the 5-HT feeding circuit.
The homology between DTH and its sister enzyme DTRH at the DNA level is sufficiently limited that neuronal expression of the DTH RNAi and UAS constructs should not affect DTRH expression [30, 38], Figure 9]. While DTRH protein was ectopically observed in the CNS when DTRH expression was induced using a UAS transgene, the 5-HT pattern was unchanged except for intensity, implying that all the necessary factors for 5-HT synthesis (substrate, cofactor) were only available in 5-HT neurons . The analogous experiment for DA could not be assessed in elav
/UASDTH animals due to the lack of an anti-dopamine antibody competent for our immunohistochemical studies. However, it is reasonable to assume that DA synthesis is also not induced in non-DA neurons, as these cells would lack dopa decarboxylase, the second enzyme in DA synthesis, as well as the cofactor tetrahydrobiopterin, and sufficient amounts of the substrate tyrosine. Since DTH also forms a complex with GTP-cyclohydrolase, the rate-limiting enzyme in tetrahydrobopterin synthesis, and this complex is necessary for DTH activity , it is likely that while DTH protein may be expressed in other neurons, its activity would be significantly compromised. This is consistent with the observation that late-stage embryos exposed to exogenous DA recapitulate the feeding and gut fiber phenotypes observed with elav
/UAS DTH animals, since the DA transporter is only expressed in DA neurons . Thus, our approach permitted controlled and specific manipulation of neuronal DA synthesis.
Dopamine plays a critical role in development of the 5-HT feeding circuit although it does not act as a transmitter to modulate feeding behavior
Neuronal expression of two independent DTH RNAi transgenes singly as well as in combination demonstrated that the level of DTH knockdown could be titrated; this correlated with increased complexity of the 5-HT axonal gut fibers as well as with deficits in feeding rate. When these transgenes were expressed after the circuit was mature, in 2nd instar larvae, there were no effects observed on either feeding or gut fiber appearance. More importantly, the defects resulting from constitutive knockdown of DTH expression were fully rescued by exposure to DA during the last 6 hours of embryogenesis, when the 5-HT feeding circuit is developing, confirming a role for DA in the circuit's normal development and mature function. It can be argued that high levels of exogenous DA could be taken up by the serotonin transporter and metabolized via monoamine oxidase activity, resulting in toxicity of the 5-HT neurons. However, Drosophila do not catabolize DA via this route since they lack these enzymes . Excess dopamine is modified and cross-linked into the chitinous exoskeleton, and thus the cuticle is the dopamine "sink." Therefore, not only is it unlikely that sufficient amounts of DA would be taken up by dSERT into 5-HT neurons, DA would not be degraded into a toxic metabolite. Additionally, since reduction in developmental 5-HT levels increases complexity of the fibers , one would expect, if DA was toxic to 5-HT neurons, that the complexity would be further increased, rather than reduced to normal levels. Constitutive overexpression of DTH, or exposure of wild-type embryos to exogenous DA during the last six hours of embryogenesis, also resulted in reduced feeding and more complex gut fiber architecture. Lastly, our observation that the actions of DA on the feeding circuit occur via a D2 receptor expressed in 5-HT neurons results in enhanced complexity rather than reduced complexity of the fibers, also argues against this mechanism. Thus, unlike 5-HT, the development of the circuit is sensitive to a threshold for DA levels, above or below which affects developmental signaling pathways. A similar response for DA has been observed in mammals , as well as in zebrafish [reviewed in ], and may reflect conserved signaling mechanisms across species.
In mammals, neonatal depletion of DA results in an array of behavioral perturbations in relatively simple motor tasks such as locomotion , as well as in higher order cognitive function . It has also been shown to affect cortical morphogenesis , implying a direct correlation between changes in neural circuitry caused by altered neuronal DA levels during a vulnerable and plastic developmental period, and the behavioral deficits observed in the animal. Depletion of cortical DA in neonatal mice results in changes in gene expression in factors necessary for axon guidance and dendritic growth, as well as those required for folding actin and tubulin . Lesioning neonatal DA neurons via 6-hydroxydopamine is also known to induce sprouting of 5-HT axonal fibers in the rat cerebral cortex . Our results suggest that the morphogenic role for DA in neural circuit development, and its interactions with 5-HT signaling pathways, may be an evolutionarily conserved mechanism.
Neuronal DA and 5-HT affect pupal development, but not as a consequence of reduced feeding
In order to leave the food medium in preparation for pupariation, Drosophila melanogaster larvae must reach a minimum mass; in general, this occurs early in the 3rd larval instar . Slower feeding larvae take longer to reach this critical weight . When neuronal DTH levels are either decreased or increased during late embryogenesis, larval feeding rate is reduced, but while there is an increase in lethality, neither pupal size nor time to pupal formation is affected (Figure 8). This cannot be attributed to a reduced feeding rate, however, since animals carrying two copies of the DTRH RNAi transgene display a more robust survival rate than controls; pupal size and time to pupation are the same. The feeding rate is depressed to approximately 85% of normal in the both elav/THA;THK and elav/TRHE;TRHA animals. Thus, while both neuronal DA and 5-HT modify the larval feeding circuit and thus its functional output in the larva, the survival outcomes must arise as a consequence of perturbations of other neural circuits. A role for DTRH in the developing brain has long been established , and DA has been shown to play a critical role in development in Drosophila [31, 51]. Therefore, a reduction or increase in neuronal DA synthesis, or a reduction in neuronal 5-HT synthesis, in late embryogenesis affects the development and function of the feeding circuit, but alterations in this circuit arising from these perturbations have no direct effect on general growth parameters.
DA exerts its neurotrophic effects on the 5-HT circuit via the D2R receptor expressed in 5-HT neurons during embryonic CNS development
An in vitro role for D2R in neuronal development has been established, since it was identified in a screen for RNAi phenotypes that altered growth of primary neural cells in culture; in these studies, reduced D2R expression resulted in increased neurite branching . Our results demonstrate an in vivo developmental role for D2R in CNS development. In mice, blockade of the D2 receptor during striatal development has been shown to result in increased sprouting of axonal fibers . The D2 receptor in larval zebrafish transiently regulates a swim circuit, which may be critical for the development of the mature circuit ; it also modulates levels of the protein kinase Akt, resulting in perturbations in development of GABAergic neuronal pathways . Thus, our results are consistent with vertebrate studies demonstrating a key role for the D2 receptor in neural circuit development. There are three D2-like receptors in mammals: D2, D3 and D4, but while there is only a single gene encoding a D2 receptor in Drosophila, there are 8 protein isoforms that arise as a consequence of alternative splicing, with differences in the length and sequence of the third intracellular loop . While D2R is expressed at high levels during the latter stages of embryogenesis, the spatial localization has been determined only in the larval and adult CNS , and it is unknown which isoform(s) mediate the developmental effects of DA on the 5-HT feeding circuit.
Both dopamine and serotonin have been shown to promote as well as inhibit axonal outgrowth [13, 14, 57–59]. Released 5-HT can induce growth cone collapse, resulting in reduced neurite branching in molluscan cerebral giant cells, acting via 5-HT receptors located in the growth cone as well as along the connective [reviewed in . Since the altered axonal fibers in our study arise from 5-HT neurons, and since DA exerts its effect via a D2 receptor expressed in serotonergic neurons, dopamine may directly regulate the extent of neurite outgrowth from the central 5-HT neurons in the stomatogastric circuit. However, classic studies in Lymnaea have demonstrated that DA can act as a diffusible substance to promote neurite outgrowth and synapse formation [reviewed in [60, 61]. Our studies do not exclude this possibility in Drosophila, but suggest that mechanisms are in place for the DA signals arising from DA neurons to both directly and indirectly affect the sprouting of 5-HT axonal fibers from specific central 5-HT neurons.
Neuronal DA and 5-HT interact during CNS development to generate the mature feeding circuit
6-hydroxydopamine lesions in neonatal rats affect 5-HT axonal sprouting in the striatum, which does not occur in similarly lesioned adults [48, 62, 63]. While there is significant loss of DA neurons after 6-hydroxydopamine lesioning, which is not observed in the Drosophila transgenic lines in which tyrosine hydroxylase has been specifically reduced within central DA neurons, the net effect in both systems is a loss of DA release. Therefore, the neurotropic effects of DA on 5-HT axonal fibers that we have observed in our studies are consistent with mammalian studies, and again suggest that the developmental interactions between DA and 5-HT are likely to be evolutionarily conserved.
Our results differ from those of Budnik et al.  who examined the 5-HT stomatogastric feeding circuit in Drosophila larvae unable to synthesize dopa decarboxylase (DDC), the second step in the biosynthetic pathway for both dopamine and serotonin. In this study, they found that the fibers projecting from the proventriculus into the midgut displayed greater branching and increased numbers of varicosities; in our study, which assessed the fibers as they entered the proventricular foregut, we found that reduction of both neuronal DA and 5-HT synthesis during development of the circuit had no effect on the axonal fiber architecture, consistent with the normal feeding behavior observed in these animals. It is possible that the developing axonal projections could be exposed to different morphogenic environments as the neurite length extended during development from the foregut to the midgut. Since Budnik and colleagues eliminated DDC in every tissue, while our transgenics (elav/TRHE;THK and elav/THA;TRHA) specifically reduced central DA and 5-HT synthesis, the two studies may not be directly comparable.
How do dopamine and serotonin interact to effect changes in axonal arborization and vesicle populations? A number of studies have shown that neurotransmitters, including dopamine and serotonin, regulate dendritic transport along microtubules [reviewed in , and their activities in interacting with factors that stabilize or destabilize microtubules would also affect axonal stability. Dysfunctional microtubules result in aberrant neurotransmission. It is possible that that dopamine and serotonin may interact directly with different factors necessary for microtubule stabilization, via expression and/or post-translational modifications of microtubule-associated proteins. Changes in stabilization of the microtubule would then affect axonal outgrowth and branch development. These changes in the microtubule network would result in aberrant signaling of the 5-HT stomatogastric circuit, with perturbed feeding as the functional outcome. Past studies have demonstrated a direct role for DA in altering the phosphorylation state of MAP2, which is critical for dendritic maturation . Alternatively, DA-receptor-mediated signaling might influence expression of neurotrophins by transactivation of other G protein coupled receptors. Activation of the D1 receptor in cultured rodent embryonic striatal neurons results in increased TrkB cell surface expression; TrkB is a high-affinity receptor for several neurotrophic factors including brain-derived-neurotrophic factor .
The DA neurons first appear 1-2 hours after the 5-HT neurons, suggesting there may be temporal constraints in the developmental pathway - that is, DA may exert its influence during a limited window of the time required for development of the feeding circuit. The identification of the factors by which dopamine and serotonin affect axonal maturation in this circuit is currently under investigation.