The Drosophila larval feeding circuit
The general structure of the larval stomatogastric nervous system is highly conserved across insect species [28
]. The 5-HT feeding circuit is an integral part of the stomatogastric pattern generator, consistent with the fact that 5-HT modulates appetitive behavior in all species including Drosophila
]. Unlike adults, Drosophila
larvae feed continuously if placed in an appropriate food source [29
], in anticipation of the enormous energy demands for metamorphosis. The larval mouth hooks (forming the most anterior part of the cephalopharyngeal plates) shovel food into the gut at a relatively constant rate. Bundled 5-HT nerve fibers connect neurons within the subesophageal ganglion to the cephalopharyngeal plates via the frontal nerve (denoted by the arrowhead in Figure ) as well as to the proventriculus (foregut) via the recurrens nerve (denoted by arrow, Figure ). The frontal nerve is responsible for the generation of feeding-related motor patterns [28
]. The recurrens nerve fasciculates in the proventriculus into individual axonal fibers containing 5-HT-containing vesicles (denoted by arrowheads in Figure ); additional fibers branch off when they reach the midgut (denoted by arrows in Figure ). Synaptotagmin fused to green fluorescent protein, under the control of the pan-neuronal elav
promoter, demonstrates 1:1 colocalization with Drosophila
neuronal tryptophan hydroxylase (DTRH, the rate-limiting step in 5-HT synthesis) within the fibers, evidence that the fibers are axons projecting from central presynaptic neurons (DTRH is responsible for synthesis of neuronal 5-HT [30
]) (Figure ). No tyrosine hydroxylase-immunoreactive fibers innervate the gut (Figure ), suggesting that, unlike 5-HT, DA neurotransmission does not modulate larval feeding. This is consistent with the observation that larvae systemically depleted of DA as 2nd
instars display normal feeding behavior [31
Reduced TH levels in the embryonic CNS affect development of the 5-HT feeding circuit
To reduce levels of neuronal DA, we generated transgenic RNAi lines to knockdown levels of Drosophila tyrosine hydroxylase (DTH), which is the first and rate-limiting enzyme in DA synthesis. We identified two transgenic lines (THA, on chromosome 2, and THK, on chromosome 3) and manipulated them singly, and in combination, to effect a titration of DTH knockdown in the CNS. Since expression of DTH is limited to less than 100 neurons in the larval CNS, demonstration of reduced neuronal DTH could not be assessed by changes in whole body DTH enzyme activity or protein levels. Quantitating DA levels strictly within the CNS would also be difficult, since DA is found in the circulating hemolymph as well as in the fat body and ring gland, which are in close proximity to the larval brain. Therefore, the intensity of DTH immunofluorescence of the two most distal dorsolateral neurons in the ventral ganglion (denoted by arrows in Figure ) was compared between control animals (w1118, the parental line for generation of the transgenics), animals carrying one copy of the transgene (THA or THK), and animals carrying both THA and THK. These neurons were chosen because they were distinct and easily identifiable, and essentially free of the TH-immunoreactive neuropil within the ventral nerve cord. To confirm extent of the knockdown, two independent drivers were used: elavC155, which drives expression throughout the CNS (Figure ), and THGal4, which drives expression in the majority of DA neurons, including the distal dorsolateral pair, as well as in other, non-neuronal tissues (Figure ). Brains were dissected from each genotype and assessed in parallel under identical conditions. Each neuron was photographed at the same exposure and magnification, and the relative fluorescence of each was determined by quantitating the pixel intensity of a circular sampling region placed at seven different locations within the cytoplasm of each cell. This number was averaged to give a score for each neuron. Under the control of either driver, both THA and THK are capable of reducing DTH levels to a roughly similar extent; the combination of THA and THK reduces expression even further, demonstrating that the degree of knockdown can be titrated. The decrease in DTH levels correlates with reduction in larval feeding in animals where TH has been constitutively reduced in the CNS during development of the circuit (Figure ); all feeding assays were accomplished using late 2nd - early 3rd instar larvae. Note that elav/DTHA results in reduced feeding relative to elav/DTHK animals, consistent with the stronger knockdown of DTHA using the elav driver (Figure ). To demonstrate that the genetic manipulations did not affect the larva's ability to extend and retract the mouth hooks, body wall contractions for each animal were measured. On an agar surface, the larvae extend and then retract their mouth hooks into the agar to initiate the body wall contractions; the imprint from their mouth hooks is left in the agar substrate. Since this behavior was unaffected (Figure ), the defect in feeding cannot be due to motor deficits or other generalized perturbations in physiology.
Figure 2 Constitutive reduction in neuronal DTH results in depressed feeding behavior. Two independent RNAi transgenic lines (DTHA, inserted on chromosome 2, and DTHK, on chromosome 3) were used to titrate constitutive knockdown of DTH levels throughout CNS development. (more ...)
Reduced neuronal 5-HT during development of this circuit results in increased complexity of the architecture of the 5-HT axonal fibers projecting to the proventriculus: these fibers displayed increased branching, greater numbers of 5-HT-containing vesicles, and significantly increased vesicle size [17
]. To determine whether the reduction in neuronal DA similarly correlates reductions in feeding behavior with increased complexity of the 5-HT axonal projections, these parameters were assessed in control and TH knockdown animals (Figure ). All immunohistochemical analyses were performed using late 3rd
instar larvae. While not significant, there is a trend towards increased branching of the 5-HT axonal fibers (Figure ). However, the number (Figure ) and size (Figure ) of the 5-HT-containing vesicles are significantly increased with increasing knockdown of DTH; overall, the reduction in DTH levels (and therefore DA synthesis) correlates with increased vesicle area (Figure ). Varicosity number is also increased (data not shown). This can be visually observed by the increasing complexity of the fiber appearance when comparing w1118
/THA;THK (Figure ) with elav
/THK (Figure ), elav
/THA (Figure ) and elav
/THA;THK (Figure ) guts. Thus, increasing reductions in DTH levels directly correlate with reductions in feeding (Figure ), as well as with increases in the number and size of the 5-HT-containing vesicles. These results demonstrate that neuronal DA affects development, and thus mature function, of the 5-HT feeding circuit.
Figure 3 Constitutive reduction in neuronal DTH results in results in aberrant gut fiber architecture. Analysis of proventricular tissues from 3rd larval instars dissected and incubated with anti-5-HT. A. Neurite branching. B. Number of large varicosities (> (more ...)
To conclusively demonstrate that this was a developmental effect, as opposed to the actions of DA as a neurotransmitter, 16-hour old elav/THA;THK embryos (the double transgenic line, providing the greatest reduction in DTH) were exposed for the last 6 hours of embryogenesis either to serum-free media or to serum-free media plus 10-6 M DA-HCl. Once hatched, these animals were removed from the medium, placed on normal food, and assayed for feeding behavior as late 2nd - early 3rd instar larvae. They were compared with w1118/THA;THK animals exposed only to serum-free media during this stage of embryonic development (control). The exogenous DA rescued the feeding defects in elav/THA;THK larvae, resulting in behavior indistinguishable from that of the control (Figure ). As before, mouth hook contractions during locomotion were normal (data not shown). We then assessed the gut fiber architecture of these animals. The increase in fiber branching (Figure ), total varicosity number (Figure ), number of varicosities greater than 5 μm2 (Figure ), and varicosity area (Figure ) of elav/THA;THK axonal fibers projecting to the proventriculus were rescued to control levels by exposure to DA during the last 6 hours of embryogenesis. This can be observed visually when comparing the 5-HT axonal fibers from fibers from elav/THA;THK (F), elav/THA;THK + DA (G), and w1118/THA;THK (H) guts. This result conclusively demonstrated that the development of the 5-HT feeding circuit, and thus larval feeding behavior, is sensitive to DA levels during development of the larval feeding circuit.
Figure 4 Exposure of elav/THA;THK late-stage embryos to exogenous DA rescues both the feeding and axonal fiber defects. A. Feeding. elav/THA;THK 16 hr old embryos were dechorionated, devitinellated, and placed in serum-free media until hatching (control), or in (more ...)
Increased TH levels in the embryonic CNS also affect development of the 5-HT feeding circuit
To assess the effects of constitutive overexpression of neuronal DTH on the feeding circuit, we generated transgenic lines to induce DTH levels (35.1 on chromosome 2, and 27.6 on chromosome 3). DTH immunofluorescent staining of control (w1118/27.6) larval brains revealed the normal DA dorsolateral (Figure ) and medial (Figure ) patterns; as expected, there was no DTH immunofluorescence detected in the proventriculus (Figure ). However, in elav/27.6 larvae, DTH was expressed in essentially every cell in the CNS (Figure ), as well as in the 5-HT axonal proventricular fibers (Figure ). There is a 1:1 correspondence between the fibers revealed via 5-HT immunofluorescence and with anti-DTH (data not shown). Analysis of the progeny from elavGal4 crossed with the 35.1 UAS transgene maintained over a larval marker, Black cell (Bc1) revealed the normal DA dorsolateral (Figure ) and medial (Figure ) patterns in elav/Bc1 larval brains, but their elav/35.1 siblings displayed the same ubiquitous DTH immunofluorescence as did elav/27.6 animals (Figure ). This was quantitated by analyzing pixel intensity of the last pair of dorsolateral neurons in THGal4/35.1 and THGal4/27.6 animals (Figure ), since expression would then be limited to DA neurons. This analysis confirmed the increase in DTH levels in these transgenic animals.
Figure 5 Constitutive induction in neuronal DTH or increased levels of systemic DA during late embryogenesis results in depressed feeding behavior in larvae. Two independent UAS transgenic lines (35.1 and 27.6, on chromosomes 2 and 3, respectively) were used to (more ...)
Feeding behavior in animals with constitutive overexpression of neuronal DTH was reduced, similar to the DTH knockdowns (Figure ). To conclusively establish that this resulted from increased DA synthesis, 16-hour old control embryos (w1118;THA;THK) were exposed for the last 6 hours of embryogenesis either to serum-free media, or to serum-free media plus 10-6 M DA-HCl, and assayed for feeding behavior as larvae. Only embryos exposed to exogenous DA during CNS development displayed a reduction in feeding (Figure ). As before, locomotor behavior of the larvae was normal (Figure ). Thus, levels of developmental neuronal DA above or below the normal level resulted in depressed feeding.
Consistent with the perturbations in feeding, the 5-HT axonal fibers projecting to the gut from larvae with increased neuronal DA levels during development of the circuit display greater complexity relative to controls: increased fiber branching (Figure ), increased number of 5-HT-containing vesicles along the axon length (Figure ), and greater numbers of larger vesicles (Figure ). This can be observed visually when examining axonal fibers from elavC155/w1118 (Figure ), elavC155/THA;THK (Figure ), w1118/THA;THK -DA (Figure ), and w1118/THA;THK + DA (Figure ). These data suggest that perturbations in developmental neuronal DA levels above or below a certain threshold can affect development of neural circuitry.
Figure 6 Constitutive induction in neuronal DTH or increased systemic DA during late embryogenesis results in aberrant larval gut fiber architecture. elavC155/UAS27.6, its parental control, elavC155/w1118, and 16 hr old control (w1118;THA;THK) embryos were dechorionated, (more ...)
Perturbations in neuronal DTH levels after larval CNS development do not affect feeding or fiber architecture
Although previous studies demonstrated that systemic feeding of a TH inhibitor or of L-DOPA (the product of the TH reaction) to 2nd
instar larvae did not affect feeding behavior [31
], we used an inducible elav
driver to reduce (GSelav
/THA;THK) or increase (GSelav
/27.6) neuronal DTH expression in 2nd
instar larvae, after the larval nervous system had fully developed. As expected, there were no changes in feeding behavior (Figure ) or in locomotion (Figure ), and gut fiber architecture was also unchanged. Branching (Figure ), varicosity number (data not shown), varicosity area (data not shown) and number of large varicosities (Figure ) were unaffected by manipulation of neuronal DA levels after the nervous system had developed. Thus, whether DA levels are manipulated systemically in mature larvae, as opposed to during late embryogenesis, by pharmacological agents, or via targeted expression in the CNS, there is no effect on larval feeding. These results conclusively demonstrate that although DA neurotransmission does not modulate larval feeding behavior, neuronal DA is required during late embryogenesis for the normal development of the serotonergic feeding circuit.
Figure 7 Manipulation of neuronal DA levels after the larval CNS has developed has no effect on feeding behavior or gut fiber architecture. Control - uninduced; RU486 - induced. A. Feeding. B. Locomotion. n = 40 for behavioral analyses, from 4-6 independent experiments. (more ...)
Reduction in feeding does not affect time to pupariation or pupal size. Drosophila
larvae must reach a critical size before pupariation can be initiated. Larvae with higher feeding rates are able to reach this critical size more quickly, and thus pupate faster [32
]. We therefore assessed whether the reduced feeding rate resulting from impairment of the 5-HT feeding circuit as a consequence of reduced or increased embryonic DTH levels affected developmental or growth rates. 50 embryos from each genotype (elav/w1118
/THA;THK and elav
/UAS27.6) were established in parallel in 16 independent vials containing standard food medium. At the same time each day the number of pupal cases in each vial was recorded, and pupae from each vial were removed and measured. There was no change in pupal size for animals with reduced or increased TH levels in the CNS (Figure ), and no change in time to pupariation (Figure ), but there was a direct correlation between the extent of neuronal TH knockdown or upregulated TH levels and the number of pupae (Figure ). The vials were observed over a period of several days, and all pupae that were viable were recorded; no pupae were ever observed after day 10. To determine whether the increase in lethality arose from reduced feeding, the same experiment was performed using animals with targeted knockdown of neuronal 5-HT synthesis (elav/w1118
[control] and elav
/TRHE;TRHA [a transgenic line containing two copies of the DTRH RNAi transgene]), since these animals display a similarly reduced feeding rate when compared with elav
/THA;THK animals (Figure ). Again, there was no change in pupal size (Figure ) or time to pupariation (Figure ), but in this case, the reduced feeding rate in elavC155
/TRHE;TRHA larvae correlated with increased
pupal survival (Figure ). Therefore, changes in developmental rate are unlikely to be the consequence of reduction in feeding rate to 83%, the rate observed in both elavC155
/THA;THK and elavC155
Reduction in feeding does not affect organismal development.
DA exerts its neurotrophic effects on the 5-HT circuit via the D2R receptor expressed in 5-HT neurons during embryonic CNS development
To confirm that these developmental effects occurred via the actions of DA released from DA neurons, rather than any reduction in 5-HT via knockdown of DTRH, which shares some homology with DTH, the THA and THK transgenes were expressed in serotonergic neurons (using the DTRHGal4 driver) as well as in dopaminergic neurons (using the DTHGal4 driver) (Figure ). As expected, reduction in feeding was observed only when the transgenes were expressed in DA neurons (Figure ). Similarly, the effects on gut fiber architecture occurred only when the transgenes were expressed in DA neurons. Increased branching (Figure ), increased numbers of 5-HT containing vesicles (Figure ), and increased numbers of large 5-HT-containing vesicles (Figure ) were observed only in fibers from elavC155/THA;THK and DTHGal4/THA;THK, and not in w1118/THA;THK and DTRHGal4/THA;THK, animals. This was observed visually when comparing w1118/THA;THK (Figure ), elavC155/THA;THK (Figure ), DTRHGal4/THA;THK (Figure ) and DTHGal4/THA;THK (Figure ) 5-HT axonal gut fibers.
Figure 9 The DA neurotrophic signal is released from DA neurons. Knockdown of neuronal DTH synthesis in 5-HT neurons (DTRHGal4) and DA neurons (DTHGal4) was compared with pan-neuronal (elavC155) expression. A. Feeding. ***p < 0.001, one-way ANOVA followed (more ...)
Since DA must signal through a DA receptor, we used the elavC155 driver to express dsRNA transgenic constructs for the two Drosophila D1 DA receptors (DopR and DopR2) as well as the single Drosophila D2 receptor, D2R. Only reduction in D2R expression had any effect on feeding (Figure ) when compared with the elavC155/pattP2 parental control. To confirm that D2R function was required during development of the circuit, we used the inducible elav driver GSelav in 2nd instar larvae to reduce D2R expression, and as expected, saw no effect on feeding (Figure ). More critically, when using the DTRHGAL4 and DTHGal4 drivers to reduce expression of the D2R receptor in serotonergic or dopaminergic neurons, respectively, the effect on feeding was only observed when D2R expression was reduced in 5-HT neurons (Figure ), suggesting that dopamine exerts its effects on the 5-HT feeding circuit via signaling through the D2R receptor expressed in 5-HT neurons during development of the circuit. Again, the changes in feeding correlated with changes in the proventricular 5-HT axonal fiber architecture. Reduced DA signaling resulted in increased branching (Figure ) and varicosity area (Figure ), due to increased numbers of larger (greater than 5 μm2 in diameter) 5-HT-containing vesicles (compare elavC155/pattP2, Figure with elavC155/D2R, Figure , and RH/pattP2, Figure with DTRH/D2R, Figure ).
Figure 10 DA exerts its effects via a D2R receptor expressed in 5-HT neurons during development of the circuit. A - C. Feeding behavior. A, Only the D2 receptor, D2R, affects feeding behavior (elavC155/pattP2, parental control; DopR and DopR2, D1 receptors). B. (more ...)
Disparate relative levels of neuronal DA and 5-HT during development of the feeding circuit affect the function and architecture of the mature 5-HT feeding circuit
To assess whether neuronal DA and 5-HT interact with each other for normal development of the feeding circuit, we generated lines with different combinations of the DTH and DTRH transgenic RNAi constructs, and placed them under the control of the elavC155, DTHGal4 and DTRHGal4 drivers (Figure ). Both THK and TRHE, when independently driven by elav, effect a reduction in larval feeding; however, there is no reduction in feeding in elav/TRHE;THK larvae (Figure ). When the same transgenic combination is placed under the control of either the DTHGal4 or DTRHGal4 driver, feeding is reduced, consistent with knockdown of only DA or 5-HT synthesis, but not both. Similar results were observed using the THA;TRHA transgenic combination, which results in stronger knockdown of TH and TRH (Figure ). These results suggest that when both neuronal DA and 5-HT levels are reduced, there is no net effect on feeding; feeding rate is perturbed only when DA levels are reduced, and normal 5-HT levels are maintained, or vice versa.
Figure 11 Neuronal knockdown of both TRH and TH during development of the circuit results in normal feeding. elavC155, pan-neuronal driver; DTRHGal4, drives expression in 5-HT neurons; DTHGal4, drives expression in DA neurons. A. TRHE, THK, weaker RNAi transgenes. (more ...)
These effects were also observed for the gut fiber architecture: when either neuronal DA or 5-HT synthesis was reduced, depending upon the strength of the knockdown, the fibers displayed increased branching, which was rescued to control levels in the presence of both transgenes (Figure ). Total varicosity number along the neurite length (Figure ), number of large vesicles (Figure ) and number of vesicles exceeding 5 μm2 in diameter (Figure ) were also increased when either DTH or DTRH levels were reduced. However, when both DTH and DTRH levels were reduced, the gut fiber architecture was indistinguishable from that of controls.
Figure 12 Neuronal knockdown of both DTRH and DTH during CNS development results in normal gut fiber architecture. elavC155, pan-neuronal driver; TRHE, THK, weaker RNAi transgenes; TRHA, THA, stronger RNAi transgenes. A. Branching. B. Total varicosity number per (more ...)
Increased neuronal 5-HT during development of the circuit results in increased larval feeding and reduced complexity of the 5-HT axonal projections to the gut [17
], and increased developmental DA has the opposite effect (Figures , ). One would thus expect that increasing levels of both biogenic amines would result in normal feeding and gut fiber appearance, and this is what is observed (data not shown). This is also true for elav
/THA;THK embryos exposed to exogenous 5-HT during the last 6 hours of embryogenesis (data not shown). These data provide evidence for interactions between DA and 5-HT during development of the 5-HT feeding circuit, perhaps to serve as inhibitory "checks" on their respective signaling pathways to generate a functional stomatogastric circuit.