Neuroarchitecture of aminergic systems in the larval ventral ganglion of Drosophila melanogaster

Biogenic amines are important signaling molecules in the central nervous system of both vertebrates and invertebrates. In the fruit fly Drosophila melanogaster, biogenic amines take part in the regulation of various vital physiological processes such as feeding, learning/memory, locomotion, sexual behavior, and sleep/arousal. Consequently, several morphological studies have analyzed the distribution of aminergic neurons in the CNS. Previous descriptions, however, did not determine the exact spatial location of aminergic neurite arborizations within the neuropil. The release sites and pre-/postsynaptic compartments of aminergic neurons also remained largely unidentified. This study used gal4-driven marker gene expression and immunocytochemistry to map presumed serotonergic (5-HT), dopaminergic, and tyraminergic/octopaminergic neurons in the thoracic and abdominal neuromeres of the Drosophila larval ventral ganglion relying on Fasciclin2-immunoreactive tracts as three-dimensional landmarks. With tyrosine hydroxylase- (TH) or tyrosine decarboxylase 2 (TDC2)-specific gal4-drivers, the distribution of ectopically expressed neuronal compartment markers was examined in presumptive dopaminergic TH and tyraminergic/octopaminergic TDC2 neurons, respectively. The results suggest that thoracic and abdominal 5-HT and TH neurons are exclusively interneurons whereas most TDC2 neurons are efferent. 5-HT and TH neurons are ideally positioned to integrate sensory information and to modulate neuronal transmission within the ventral ganglion, while most TDC2 neurons appear to act peripherally. In contrast to 5-HT neurons, TH and TDC2 neurons each comprise morphologically different neuron subsets with separated in- and output compartments in specific neuropil regions. The three-dimensional mapping of aminergic neurons now facilitates the identification of neuronal network contacts and co-localized signaling molecules, as exemplified for DOPA decarboxylase-synthesizing neurons that co-express crustacean cardioactive peptide and myoinhibiting peptides (Vömel, 2008).

This study used gal4-driven marker gene expression and immunocytochemistry to three-dimensionally map presumed serotonergic, dopaminergic and tyraminergic/octopaminergic neurons within the Fas2 landmark system of the larval VG. Furthermore, several ectopically expressed pre- and postsynaptic markers were employed to reveal the in- and output compartments of presumptive dopaminergic TH and tyraminergic/octopaminergic TDC2 neurons. The results allow comparison of the segmental distribution patterns of aminergic neurons and to trace aminergic projections to defined neuropil areas within the VG. In the following, the morphology of aminergic neurons are related to known biogenic amine (BA) functions and describes putative neuronal network interactions with other VG neurons. This work also exemplifies how Fas2-based mapping can simplify the identification of co-localized signaling molecules, and allocate all neurons within the complex Ddc-gal4 expression pattern to distinct neuron subsets (Vömel, 2008).

Throughout the insects, similar neuron groups synthesize BAs. These groups typically comprise only few neurons with large branching patterns. In agreement with previous studies, 5-HT neurons in t1-a8 of the Drosophila larval VG represent interneurons with intrasegmental neurites. The 5-HT neurons of a8, however, appear to supply only the neuropil of a7, but not that of a8 and the adjacent 'terminal plexus'. Like 5-HT neurons, the presumptive dopaminergic TH neurons lack peripheral projections and appear to exclusively represent interneurons. In contrast, presumptive tyraminergic/octopaminergic TDC2 neurons mostly represent efferent vumTDC2 neurons. The vumTDC2 neurons obviously project to larval body wall muscles including M1 and M2 since these muscles showed TA- and OA-immunoreactive type II boutons. In a8, dorsally located dmTDC2 neurons send axons through the associated segmental nerves, and hence are efferent neurons as well. These dmTDC2 neurons probably innervate the reproductive tract in the adult female fly. Besides the dmTDC2 neurons of a8, typically two additional dmTDC2 neurons reside in the dorsal cortex between the last subesophageal neuromere and t1. These dmTDC2 neurons were not described in previous morphological studies on TA- and OA-/TβH-immunoreactive neurons. Nevertheless, all dmTDC2 neurons in the VG consistently showed strong Tdc2-gal4-driven mCD8GFP expression as well as TßH immunoreactivity. Thus, they likely synthesize both TA and OA. Although their neurites could not be traced, the dmTDC2 neurons resemble a pair of anterior medial neurons in locusts and crickets that localize to t1 and innervate the anterior connectives. Alternatively, dmTDC2 neurons may correspond to a single dorsal unpaired median neuron which resides in t1 of the locust and supplies the subesophageal nerves. Like dmTDC2 neurons, pmTDC2 neurons are probably interneurons as well. The soma position of pmTDC2 neurons highly resembles that of descending OA-immunoreactive interneurons detected in the subesophageal and thoracic neuromeres of bees, crickets, cockroaches, locusts, and moths (Vömel, 2008).

Within the larval VG of Drosophila, aminergic neurons typically show a segmentally reiterated distribution. The number of aminergic modules, however, often varies between different neuromeres. 5-HT neurons, for instance, typically occur as two bilateral pairs per neuromere. Yet, t1 comprises three 5-HT neuron pairs and a8 only one pair. The presumptive dopaminergic TH neurons also lack a strict serial homology since three ventral median TH (vmTH) neurons are present in t1, but only one in t2-a7. Furthermore, dlTH neurons locate to a1-7, but appear to be missing in t1-3. The neuromere a8 lacks TH neurons. The number of presumptive tyraminergic/octopaminergic TDC2 neurons differs between various neuromeres as well. Whereas t1 comprises one or two dmTDC2 neurons, comparable neurons are absent in t2-a7. Putative descending pmTDC2 interneurons localize to t1-a1, but appear to be missing in the remaining abdominal neuromeres. Taken together, the number of aminergic modules in t1 and a8 often deviated from that of t2-a7. This difference may-at least partially-reflect unique neuronal circuits in t1 and a8. While t1 specific physiological functions in larvae are unknown, a8 and the adjacent 'terminal plexus' are associated with the tail region, and hence contain a specific set of sensory neurons and motoneurons. The terminal neuromeres also supply several unique structures such as the spiracles or the anal pads (Vömel, 2008).

Besides the segmental differences in neuron number, the density of aminergic innervation and the amount of immunolabeling/marker gene expression varies between neuromeres as well. In particular, presumptive dopaminergic TH neurons show a striking neuromere-specific labeling pattern. Whereas a1-5 contain only few labeled TH projections, t1-3 and a6-7 comprise a comparably dense network of TH neurites. Similar to TH neurons, 5-HT neurons most densely innervate the neuropil of a7. Since a high extracellular concentration of 5-HT decreases the density of 5-HT-immunoreactive arborizations within the neuropil, a7 may represent a minor 5-HT release site. In contrast to a7, the neuropil of a8 and the adjacent 'terminal plexus' (which receive prominent peptidergic innervation) typically lack aminergic neurite arborizations. Consequently, larval aminergic neurons may play a subordinate role in tail-related physiological processes (Vömel, 2008).

To reveal putative synaptic in- and output zones of aminergic neurons, the neuronal compartment markers neuronal synaptobrevin-GFP, synaptotagmin 1-GFP, and Drosophila Down syndrome adhesion molecule [17.1]-GFP were employed. Neuronal synaptobrevin is a vesicle associated membrane protein that plays a role in the SNARE complex during vesicle transport and fusion with the plasma membrane. In accordance with this function, ectopically expressed neuronal synaptobrevin-GFP (SybGFP) accumulates at nerve terminals. SybGFP therefore served to define the presynaptic compartments of several Drosophila neurons, e.g. in the visual system. However, neuronal synaptobrevin is not restricted to small synaptic vesicles, but also locates to the membrane of large dense core vesicles, which contain BAs or neuropeptides. Consequently, in a7, SybGFP localized to putative release sites of presumptive serotonergic DDC neurons. SybGFP was also used to identify non-synaptic release sites in several peptidergic neurons. In aminergic neurons, the distribution of gal4-driven SybGFP highly resembled the corresponding mCD8GFP expression pattern. SybGFP localized in dotted patterns to aminergic neuron somata and associated neurites. It is therefore suggested that SybGFP does not exclusively label the presynaptic compartments of aminergic neurons. This fits to the assumption that ectopically expressed synaptic proteins can either localize to transport vesicles or non-synaptic compartments in peptidergic neurons. On the other hand, the ubiquitous distribution of SybGFP in aminergic neurites may suggest a widespread BA release/recycling from non-synaptic active sites. In mammals, BA release/recycling is not restricted to synapses. Vesicular monoamine transporters, which transport BAs into secretory vesicles, reside within neuron somata, axons, and dendrites. In Drosophila, the vesicular monoamine transporter DVMAT-A localizes to somata as well as neurites of several aminergic neurons both in the larval. Thus, the widespread distribution of SybGFP and DVMAT-A in aminergic neurons suggests that a considerable amount of aminergic vesicles resides at non-synaptic sites. Non-synaptic BA release/recycling might therefore play a major role for aminergic neuronal network signaling (Vömel, 2008).

Like neuronal synaptobrevin, synaptotagmins also represent integral membrane proteins of both small synaptic and large dense core vesicles. In Drosophila, the products of seven synaptotagmin genes localize to distinct neuronal compartments including the postsynaptic site. At the presynaptic site, synaptotagmin 1 does not participate in the SNARE complex, but acts as a Ca²⁺-sensor for synaptic vesicle fusion. Furthermore, synaptotagmin 1 appears to be the only crucial isoform for synaptic vesicle release. Consequently, a synaptotagmin 1-GFP fusion construct (SytGFP) was developed as a synaptic vesicle marker that specifically labels presynaptic sites. In aminergic neurons, the distribution pattern of SytGFP strikingly differed from the observed mCD8GFP and SybGFP labeling. Primary neurites of aminergic neurons always completely lacked SytGFP. Varicose neurite structures which were less evident in the mCD8GFP and SybGFP expression patterns showed strong SytGFP labeling. In agreement with the SytGFP distribution in other Drosophila neuron types, SytGFP hence appears to exclusively accumulate at the presynaptic sites of aminergic neurons. Thus, SytGFP represents a valuable marker to separate synapses from other neuronal compartments in aminergic neurons. However, since BA release is not restricted to synapses, SytGFP may not label all BA release sites of aminergic neurons. The sparse co-localization of SytGFP and SybGFP in aminergic neurites in fact suggests that aminergic vesicles-which are located distal to presynaptic sites-generally lack SytGFP. Consequently, non-synaptic BA release appears to be independent of synaptotagmin 1, but may depend on other synaptotagmin isoforms such as synaptotagmin α or β. The differing distribution of SytGFP and SybGFP also suggests that aminergic neurons contain several types of aminergic vesicles which are either associated with presynaptic or non-synaptic BA release. Alternatively, aminergic neurons may synthesize additional non-aminergic neurotransmitters like acetylcholine, GABA, or glutamate. Presumed octopaminergic efferent neurons, for instance, appear to release glutamate from type II terminals at the neuromuscular junction. In such neurons, SytGFP likely labels presynaptically located transmitter vesicles and may not reveal BA release sites (Vömel, 2008).

In contrast to SybGFP and SytGFP, ectopically expressed Drosophila Down syndrome adhesion molecule [17.1]-GFP (DscamGFP) localized to postsynaptic compartments and not to axons or presynaptic sites. Consequently, DscamGFP has served as dendrite marker in mushroom body lobe neurons. Aminergic neurons showed only weak DscamGFP labeling. DscamGFP primarily localized to neurites that lacked SytGFP labeling. Since SytGFP accumulates at presynaptic sites, DscamGFP appears to represent a valuable marker to define dendritic compartments in aminergic neurons (Vömel, 2008).

In 5-HT neurons, the distribution of ectopically expressed neuronal compartment markers was not examined since specific gal4 drivers are not available. The Ddc-gal4 driver induces marker gene expression not only in presumed serotonergic, but also in dopaminergic and additional peptidergic neurons. Consequently, neurites of different DDC neuron subsets overlap in specific neuropil areas. Presumptive serotonergic as well as dopaminergic DDC neurites, for instance, localize to the VG neuropil above the CI tracts. These conditions prevent an accurate description and interpretation of the compartment marker distribution in presumptive serotonergic DDC neurons. Thus, appropriate gal4 drivers (e.g. Dtph-gal4) are needed to further analyze 5-HT neuron morphology (Vömel, 2008).

5-HT neurons bifurcate strongly in the whole neuropil of t1-a7, and hence may influence various VG neurons including sensory, inter- as well as motoneurons. However, putative neuronal network contacts of 5-HT neurons were not examined since previous morphological studies on Drosophila 5-HT receptors did not describe the exact spatial location of the respective receptors in the larval VG (Vömel, 2008).

In TH neurons, the distribution of ectopically expressed mCD8GFP, SybGFP, SytGFP and DscamGFP differed only slightly. This might relate to the fact that the VG contains two different TH neuron groups, the vmTH and dlTH neurons, whose neurites contact each other at longitudinal projections. Consequently, pre- and postsynaptic compartments of both TH neuron groups appeared to overlap, e.g. at longitudinal projections next to the VL tracts. Since additional TH neurons located in the brain or subesophageal ganglia also innervate the VG, it was not possible to clarify which TH neuron group attributes to a particular neuronal projection. Several morphological findings, however, suggest that TH neurons possess distinct in- and output sites: Most strikingly, a1-5 contained less TH neurites labeled with mCD8GFP, SybGFP and DscamGFP, as compared to t1-3 and a6-7. In t1-a7, high amounts of SybGFP and SytGFP located to lateral longitudinal projections next to the VL tracts. These longitudinal TH neurites also contained a comparably high amount of DscamGFP, and hence likely represent synaptic in- as well as output compartments of different TH neuron groups. Besides lateral longitudinal TH projections, SybGFP and SytGFP also co-localized to the median neuropil between the DM/VM tracts. At least in a1-5, this neuropil area lacked DscamGFP, and hence probably represents a presynaptic output site of TH neurons. In a6-7, a comparably strong SybGFP and SytGFP labeling was observed in arborizations around transversal TH neurites. Whereas SybGFP mainly located to the dorsal branches of the transversal TH neurite loops, SytGFP and DscamGFP primarily labeled the ventral branches. Thus, the dorsal branches of the transversal TH neurite loops may represent non-synaptic DA release sites, while the ventral branches seem to comprise overlapping synaptic in- and output compartments of different TH neuron groups (Vömel, 2008).

Both vmTH and dorso-lateral TH (dlTH) neurons innervate distinct neuropil areas within the VG. The vmTH neurons send their primary neurites dorsally and then project through the dorsal part of the neuropil above Transversal projection (TP) 3. Since the dorsal neuropil comprises the dendritic compartments of most motoneurons, vmTH neurites are ideally located to modulate locomotor activity. This fits to the finding that DA application onto intact larval CNS-segmental preparations rapidly decreased the rhythmicity of CNS motor activity and synaptic vesicle release at the neuromuscular junction. Unlike vmTH neurons, dlTH neurons exclusively innervate the ventral part of the VG neuropil beneath TP 3. There, putative dendritic compartments of TH neurons mainly localize to lateral longitudinal and to transversal projections adjacent to the main output site of several afferent sensory neurons, e.g. tactile and proprioreceptive neurons. Thus, some TH neurons may receive synaptic input from specific sensory neurons. In contrast, TH neurons also seem to have output sites in the ventral part of the neuropil, and hence may influence the signal transmission between sensory neurons and interneurons. This fits to the finding that peptidergic apterous neurons, which appear to transmit sensory input from the VG to the brain, express DA receptors. Concomitantly, dendritic compartments of apterous neurons seem to reside adjacent to the putative DA release sites of TH neurons at the CI tracts. Besides the overlap between transversal TH neurites and sensory/interneuron projections in the ventral neuropil, TH neurons may influence several neuron groups at other locations within the VG. For instance, the putative synaptic output sites of TH neurons in the median neuropil between the DM/VM tracts overlap with presumptive input compartments of both interneurons and efferent neurons expressing peptides such as CCAP, corazonin, FMRFa, or MIP. Furthermore, the putative output sites at longitudinal TH projections next to the VL tracts lay adjacent to presumptive input compartments of e.g. efferent leucokininergic neurons (Vömel, 2008).

In the VG, most TDC2 neurons are efferent vumTDC2 neurons and showed a differential distribution of ectopically expressed SybGFP, SytGFP, and DscamGFP. The primary neurites and transversal projections of vumTDC2 neurons were labeled with DscamGFP, but lacked SytGFP. Therefore, these neurites likely represent dendritic input sites. This fits to the finding that vumTDC2 neurons possess output sites at larval body wall muscles. However, vumTDC2 neurites within the VG also contained high amounts of SybGFP, and hence may release TA/OA from non-synaptic sites. Besides vumTDC2 neurites, SybGFP strongly labeled longitudinal TDC2 neurites and associated arborizations in the dorso-lateral neuropil between TP 1 and 3. These TDC2 projections showed prominent SytGFP labeling and TßH immunoreactivity, but largely lacked DscamGFP. Thus, the dorsal part of the VG neuropil likely contains output compartments of TDC2 neurons. Since the larval brain seems to contain only tyramine- and no octopamine-immunoreactive neurons, these output sites likely derive from descending interneurons located in the subesophageal ganglia, dmTDC2 or pmTDC2 neurons. Noteworthy, the strong SybGFP and SytGFP labeling in TDC2 neurites projecting through the dorso-lateral neuropil of the VG overlapped with DscamGFP in transverse vumTDC2 neurites. Thus, descending TDC2 neurons may interact with vumTDC2 neurons (Vömel, 2008).

The VG comprises efferent vumTDC2 neurons as well as several putative TDC2 interneuron groups. Since all vumTDC2 neurons appear to have synapses at peripheral targets and dendrites in the dorsal neuropil, they show the typical motoneuron morphology. This corresponds to the finding that OA inhibited synaptic transmission at the neuromuscular junction by affecting both pre- and postsynaptic mechanisms. In addition, T?H mutant larvae, with altered levels of TA and OA, showed severe locomotion defects, which seemed to be linked to an imbalance between TA and OA signaling. Hence, vumTDC2 neurons likely regulate peripheral processes such as body wall muscle activity, whereas TDC2 interneurons centrally modulate the neuronal activity of motoneurons and interneurons involved in locomotor control. Interestingly, presumptive presynaptic compartments of descending TDC2 interneurons reside adjacent to transversal vumTDC2 dendrites. Thus, both TDC2 neuron groups may interact to modulate larval locomotor activity. Besides their function for locomotion, descending TDC2 neurons may also influence other neurons which project into the dorsal neuropil between TP 1 and 3. The putative output sites of TDC2 interneurons, for instance, lay adjacent to several peptidergic projections showing allatostatin-A, FMRFa, MIP or tachykinin immunoreactivity. However, nothing is known about TA/OA receptor distribution in the larval VG (Vömel, 2008).

During the morphological analysis of DDC neurons in the L3 larval VG, two DDC neuron groups were identified that obviously synthesize neither 5-HT nor DA. This corresponds to the previous finding that Ddc-gal4-driven marker gene expression is not restricted to presumptive serotonergic 5-HT and dopaminergic TH neurons. However, it cannot be excluded that the putative non-aminergic DDC neurons transiently synthesize BAs during other developmental stages. Ddc-gal4-driven mCD8GFP expression never revealed the dlTH neurons. This may relate to the fact that the onset of Ddc expression varies between different DDC neuron groups, and high DDC and TH levels do not temporally coincide. Taken together, these results suggest that-at least in the L3 larval VG-the Ddc-gal4 expression pattern 1) contains additional non-aminergic neurons, and 2) typically comprises most, but not all 5-HT and TH neurons. These particular characteristics of the Ddc-gal4 driver line should be carefully considered for the interpretation of studies that employed Ddc-gal4-driven expression to genetically manipulate serotonergic or dopaminergic neurons. Nevertheless, since all Ddc-gal4 expressing neurons within the VG showed at least faint DDC immunoreactivity, the Ddc-gal4 driver appears to restrict ectopical gene expression to DDC neurons. Noteworthy, the DDC neurons which lacked 5-HT and TH immunoreactivity showed corazonin and CCAP/MIP immunoreactivity respectively. In the moth Manduca sexta, these peptides play vital roles during ecdysis. At least the CCAP/MIP neurons are also necessary for the proper timing and execution of ecdysis behavior in Drosophila. Since dopaminergic DDC neurons regulate the titers of the molting hormones 20-hydroxyecdyson and juvenile hormone, both aminergic and peptidergic DDC neurons may interact to control ecdysis-related events. Recent findings indeed suggest that CCAP/MIP neurons modulate TH activity after eclosion to control the precise onset of tanning (Vömel, 2008).

References

Vömel, M. and Wegener, C. (2008). Neuroarchitecture of aminergic systems in the larval ventral ganglion of Drosophila melanogaster. PLoS ONE 3(3): e1848. PubMed Citation: 18365004

Zygotically transcribed genes

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.