Role of serotonergic neurons in the Drosophila larval response to light

Background Drosophila larval locomotion consists of forward peristalsis interrupted by episodes of pausing, turning and exploratory behavior (head swinging). This behavior can be regulated by visual input as seen by light-induced increase in pausing, head swinging and direction change as well as reduction of linear speed that characterizes the larval photophobic response. During 3rd instar stage, Drosophila larvae gradually cease to be repelled by light and are photoneutral by the time they wander in search for a place to undergo metamorphosis. Thus, Drosophila larval photobehavior can be used to study control of locomotion. Results We used targeted neuronal silencing to assess the role of candidate neurons in the regulation of larval photobehavior. Inactivation of DOPA decarboxylase (Ddc) neurons increases the response to light throughout larval development, including during the later stages of the 3rd instar characterized by photoneutral response. Increased response to light is characterized by increase in light-induced direction change and associated pause, and reduction of linear movement. Amongst Ddc neurons, suppression of the activity of corazonergic and serotonergic but not dopaminergic neurons increases the photophobic response observed during 3rd instar stage. Silencing of serotonergic neurons does not disrupt larval locomotion or the response to mechanical stimuli. Reduced serotonin (5-hydroxytryptamine, 5-HT) signaling within serotonergic neurons recapitulates the results obtained with targeted neuronal silencing. Ablation of serotonergic cells in the ventral nerve cord (VNC) does not affect the larval response to light. Similarly, disruption of serotonergic projections that contact the photoreceptor termini in the brain hemispheres does not impact the larval response to light. Finally, pan-neural over-expression of 5-HT1ADro receptors, but not of any other 5-HT receptor subtype, causes a significant decrease in the response to light of 3rd instar larvae. Conclusion Our data demonstrate that activity of serotonergic and corazonergic neurons contribute to the control of larval locomotion by light. We conclude that this control is carried out by 5-HT neurons located in the brain hemispheres, but does not appear to occur at the photoreceptor level and may be mediated by 5-HT1ADro receptors. These findings provide new insights into the function of 5-HT neurons in Drosophila larval behavior as well as into the mechanisms underlying regulation of larval response to light.


Background
Organisms possess a finite number of neuronal networks. Therefore, neurons and circuits must be multifunctional to provide individuals with a variety of behavioral outputs necessary for the adaptation to environmental and developmental changes. Neuromodulation is a powerful way to modify the function of an existing circuit without altering the 'hard-wiring' of such network (reviewed in [1]). In this regard, a large body of evidence indicates that neuromodulatory inputs cause short-term changes on neuronal network activity for adaptation to the environment (reviewed in [2,3]). Moreover, neuromodulators have been shown to play a crucial role in developmental tuning of neuronal circuit function, particularly important for ontogenetic plasticity (reviewed in [4]). In invertebrates, biogenic amines including serotonin and dopamine as well as neuropeptides are well studied neuromodulators regulating a diverse range of physiological, cellular and behavioral processes (reviewed in [5,6]).
Drosophila larval locomotion comprises rhythmic waves of forward peristalsis interrupted by episodes of pause, turning and occasional backward crawling [7]. This larval behavior is controlled by the activity of central pattern generators (CPG) [8] and may be modified by neuromodulators such as 5-HT [9] as well as by sensory information including visual input (e.g. [10]; [11]). Drosophila larval visual system consists of two bilateral clusters of 12 photoreceptors each (Bolwig's organs) [12]. Their axons form the larval optic nerve (LON), which projects into the brain toward the larval optic neuropil, also known as larval optic center (LOC) [13]. These photoreceptors express either Rhodopsin 5 (Rh5) or 6 (Rh6) [14], but only Rh5-expressing cells are required for the photophobic response displayed by larvae during foraging stage [15]. This response to light is characterized by increased pausing, head swinging, and direction change as well as reduced linear speed [11,16]. Interestingly, this aversion to light is down-regulated during development, achieving photoneutrality during wandering stage, at which time larvae are searching for a proper site to undergo metamorphosis [17]. The mechanisms underlying the modulation of this behavior remain unknown.
In order to identify neurons that provide neuromodulatory input for the regulation of larval photobehavior, we used genetic tools to study the impact of suppressing synaptic transmission in candidate neurons on the larval response to light. We began by silencing the activity of Ddc-expressing neurons by means of the Ddc-GAL4 driver [18]. Ddc catalyzes the last step in the synthesis of both serotonin and dopamine, and thus it is found in both serotonergic and dopaminergic neurons [19]. In addition, it has been reported that a third group of cells, the cora-zonin (CRZ)-releasing neurons are labeled by the Ddc-GAL4 construct during 3 rd instar stage [20].
Here, we report that 5-HT neurons located in the brain hemispheres but not those located in the VNC modulate the response to light during larval development. Furthermore, we show that 5-HT signaling is required for proper regulation of larval photobehavior, possibly through activation of 5-HT1A Dro receptors. Finally, our findings also suggest that corazonergic neurons may contribute to this modulation.

Harvesting and synchronization of larvae
Larvae were harvested following a protocol described previously [26]. Briefly, 4-7 day old parental flies were allowed to mate and lay eggs overnight in fly houses containing food plates (60 mm × 15 mm, Fisher Scientific, Houston, Tx) supplemented with vitamin A (Jamieson Laboratory, b carotene, 1.25 g/L). The next day, following a 2 hour pre-collection, a 1 hour-egg collection was performed. At 21 hours after egg laying (h AEL), all hatched larvae were removed from the collection plate under a dissection microscope. After incubating the remainder of the eggs for a period of 3 hours (corresponding to 21-24 h AEL, or 0-3 h after hatching (AH)), approximately 30-40 newly hatched larvae were collected and transferred to a fresh food plate and allowed to grow until 46-49 h AH ('late 2 nd instar'), 65-68 h AH ('early foraging 3 rd instar'), 72-75 h AH ('late foraging 3 rd instar'), 91-94 h AH ('early wandering 3 rd instar stage'), or 96-99 h AH ('late wandering 3 rd instar stage').

Verification of larval stages
Besides performing synchronized larval collections and timing their development at 25°C, several behavioral and anatomical characteristics of the larvae were used to confirm the expected larval stage. Anatomical features that can be used to distinguish the different larval stages include the morphology of the anterior spiracles, the shape of their mouth hooks and the number of teeth [27]. Therefore, these characteristics were checked after every larva was tested. In addition, especially to tell apart foraging from wandering 3 rd instar larvae, study of their behaviors such as digging into the food or wandering on the lid of the plate, reversion of spiracles and emptiness of the guts were performed. In this last case, empty guts were verified by disappearance of blue-colored food [28]. For this purpose, food coloring solution (0.05% bromophenol, Sigma) was dissolved in the regular fly medium. Egg collection and larval growth were conducted in the coloredfood plate following the same harvesting protocol used for behavioral assays. Animals were removed from their plates at early wandering stage (91-94 h AH) and rinsed with distilled water to remove any excess of food from their bodies. Verification of minimal residual blue staining remaining at the posterior tip of larval guts, characteristic of wandering stage, was performed under a Nikon SMZ1500 light microscope. Lastly, time of pupation was also observed.
Larval brains were viewed in a Nikon Eclipse Î 800 microscope. Confocal images were obtained with either a Bio-Rad Radiance MRC 600 Krypton/Argon laser confocal microscope using the LaserSharp software or a Zeiss confocal microscope using LSM510 software. Images were made of z-stack sections and their contrast and brightness were adjusted using Adobe Photoshop 5.0 software for Macintosh.

Photobehavioral assay and data collection
Photobehavioral assays were conducted using the ON/ OFF assay previously used in our laboratory [11]. Larvae were manipulated using a moist paintbrush under a dark room light (20 W lamp with Kodak GBX-2 filter), the same employed for studies of Drosophila circadian studies in free running conditions ('constant darkness') [31]. Previous larval photobehavior assays performed in our laboratory using the dark room light as the only light source confirmed earlier observations that Drosophila larvae do not respond to light of wavelengths above 650 nm [27].
Prior to the beginning of the photobehavioral assay, single larvae were removed from the food plate, carefully rinsed with distilled water to eliminate any excess of food, and placed on a pre-test non-nutritive agar plate for 1 minute to allow the larva to familiarize with the agar surface. To start the assay, individual larvae were placed then in a test agar plate and subjected to alternative 10 secondpulses of light and dark using a cool white bulb (20W Cool White, Philips) controlled by a tracking program.
Part of the quantitative analysis of larval photobehavior in the ON/OFF assay was conducted using a semi-automatic tracking system previously used in our laboratory [15]. This system allowed for stylus/tablet-based tracking of larval movement. The software (NIH Image 1.62f) automatically calculated a response index, RI = [(total distance traveled (pixels) in the dark period -total distance traveled (pixels) in the light period)/total distance traveled (pixels) in both the periods]. The duration of the assay consisted of a minimum of 60 seconds for foraging larvae and 40 seconds for wandering larvae, with a maximum testing time of 120 seconds in both cases, after which each larva was discarded. All data represented as RIs in figures are depicted as mean ± SEM.
When a detailed quantitative and qualitative analysis of larval behavior in the ON/OFF assay was required, locomotion of new larvae was captured using Pixelink Capture Software and analyzed by means of an advanced tracking software called Dynamic Image Analysis System (DIAS) (3.2; Solltech, Inc., Iowa, USA). This system has been recently used also in our laboratory for a kinematic description of larval locomotion during the ON/OFF assay [11]. Briefly, larval behavior in the ON/OFF assay was recorded for a total time of 60 seconds. The generated digital videos were analyzed in DIAS at a rate of 2 frames per second (2 f/s). Larval outlines were automatically determined using the 'Auto Trace DIC' function and their center positions (centroids) in each frame were automatically calculated. DIAS-based quantitative characterization of larval movement was conducted through analysis of direction change (deg), % of frames in linear locomotion, and centroid translocation (mm), all calculated as described previously [11,32]. Linear locomotion was defined as sequences of at least 5 frames with direction change less than 20 degrees [11,32]. For a qualitative analysis, centroid tracks (series of sequential centroids) and perimeter stacks (changes in larval outlines) of representative larvae were plotted according to their x, y coordinates over the course of the assay.

Locomotion in constant darkness
The larval response to light as measured in the ON/OFF assay depends on larval locomotion. Therefore, as a control, movement of all larvae used in this study was also examined in constant dark to verify that basic aspects of locomotion were not affected by the genetic background of the larvae. Thus, each larva tested in the ON/OFF assay was also subjected to a 30 second locomotory test using a similar manipulation protocol to the one mentioned above, but in this case under constant safe-light conditions. For quantitative analysis, the assay was performed using the semi-automatic system and data are shown when required as mean number of pixels traveled in 30 seconds ( ) ± SEM. Pixelink Capture Software and DIAS software were used for qualitative description of larval locomotion, and larval centroid tracks and perimeter stacks were generated as described before.

Touch sensitivity assay
The touch sensitivity test was performed as described in [10] with minor modifications. To avoid bias, this experiment was performed blind. During this assay, general handling of early foraging 3 rd instar larvae was the same as during the photobehavioral assay. In this case, larval behavior was observed under a dissection microscope illuminated by a red filter-adapted light source to ensure stimulus-free conditions ('constant darkness'). At the beginning of this assay, single larvae were placed on a non-nutritive agar plate identical to the ones used in the photobehavioral assay and allowed to initiate linear movement. Subsequently, each subject was gently touched with an eyelash on its anterior segments during free-run locomotion. Each larva was touched four times with an interval of 10-15 seconds between strokes. To quantify larval responsiveness to the stimulus, scores 0 to 4 were assigned to the different responses observed. A score of 0 was given to larvae that did not respond to the stimulus, whereas a full stop or hesitation was scored as 1. Larvae that retracted briefly but resumed their forward movement were scored as 2. In those cases in which larvae withdrew their anterior segments followed by a turn away from the stimulus with an angle < 90 degrees, their responses were scored as 3. Finally, when larvae retracted and turned away from the stimulus with an angle > 90 degrees, their behavior was scored as 4. The values obtained for each larva were added, and therefore individual larval scores ranged from 0 to 16. Values are shown as mean ( ) score for each group ± SEM.

Statistical analysis
Minitab 13.1 software for PC was used for statistical analysis. The statistical tests employed in the analysis of data included one-way analysis of variances (ANOVAs), and Tukey's-pairwise comparisons. Normality test on the residuals of the ANOVAs were conducted using the Rootogram test as well as the Ryan Joiner test. Verification of equal variances of the samples was performed by the F-test or Bartlett's test. In all statistical tests performed, the level of significance a was 0.05.

Silencing of Ddc neurons increases the response to light throughout larval development
In order to assess the role of specific neurons in the modulation of larval photobehavior, we used the GAL4/UAS system [33]. In this approach, specific enhancers or promoters are used to regulate the expression of the yeast transcription factor GAL4. A gene of interest, such as tetanus toxin light chain (TNT), is placed under the control of the GAL4-responsive upstream activating sequence (UAS), thereby, allowing its expression in a tissue-specific manner [33]. TNT cleaves the vesicle-associated protein synaptobrevin and its targeted expression disrupts evoked neurotransmitter release and decreases spontaneous release by ~50% [21].
The behavioral paradigm used was the ON/OFF assay, in which a single larva placed on a non-nutritive agar surface is subjected to intermittent 10 second-pulses of light. Larval behavior captured during the assay was analyzed by either the software DIAS [11,32,34] or the semi-automatic tracking system [15,26]. Larval behavior during the assay X X was assessed with DIAS by measuring changes in different locomotory parameters such as centroid translocation, change of direction and amount of linear locomotion that occurs when the larva is exposed to light and dark pulses [11,15]. Segments of linear locomotion have been previously defined as at least 5 consecutive frames with direction change less than 20 degrees per frame (deg/f) [11,32]. The semi-automatic tracking system was used to calculate a response index (RI), using the difference in distance traveled during the dark and light pulses (RI = [(total distance traveled (pixels) in the dark period -total distance traveled (pixels) in the light period)/total distance traveled (pixels) in both periods]).
Previous reports demonstrated that larval locomotion in a non-nutritive substrate is characterized by periods of linear locomotion interspersed by bouts of pause and exploratory behavior (head swinging) followed by a change of path direction [32,34]. Episodes of pause associated with head swinging behavior generate changes in path direction above 20 degrees [11,32,34]. Thus, control of larval locomotion oscillates between two states; one that promotes the peristaltic contraction of the larval musculature leading to linear locomotion and one that triggers episodes of pause and turning. Light dramatically influences the function of this control as seen by the increase in episodes of pause and turning during the light pulse when compared to what is observed during the dark pulse [11]. Interestingly, response to light is markedly reduced by the end of 3 rd instar larval stage [17]. The mechanisms responsible for this modulation are currently unknown.
In order to identify potential neuromodulatory inputs that play a role in regulating the larval response to light, we began by silencing the activity of Ddc neurons. To that end, we used the Ddc-GAL4 driver [18] to target TNT expression specifically to these cells. Fig. 1 shows centroid tracks and perimeter stacks depicting representative larval behavior during the ON/OFF assay, while Fig. 2 shows the corresponding quantification of locomotion during the assay. During the light pulse, control 3 rd instar foraging larvae expressing inactive TNT (TNT-VIF) in Ddc neurons (referred to as Ddc:TNT-VIF larvae) present the characteristic head swinging behavior, reduction in centroid translocation and change of direction that lead to reduction of linear movement (Figs. 1B and 2B, [11,32,34]). The behavior of 3 rd instar foraging larvae expressing active TNT (TNT-G) in Ddc neurons (henceforth referred to as Ddc:TNT-G larvae) during the light pulse is characterized by increased head swinging behavior and change in path direction (Figs. 1A and 2A). Analysis of wandering Ddc:TNT-VIF control larvae in contrast shows that, during the light (ON) pulses, these larvae exhibit markedly fewer episodes of head swinging behavior and direction change (Figs. 1D and 2D). Interestingly, wandering Ddc:TNT-G larvae respond to light in a manner similar to that of foraging larvae (Figs. 1 and 2, compare C with A and B). The results shown in Fig. 2 are consistent with previous reports that change of direction above 20 degrees are accompanied by sharp reduction in centroid translocation [11,32].
In order to better understand the changes in light induced-modulation of larval locomotion caused by inactivation of these neurons we used DIAS to measure direction change and amount of linear locomotion during the course of the assay. Table 1 shows that in both genotypes (Ddc:TNT-G and Ddc:TNT-VIF) linear movement is reduced during the light pulse relative to that measured during the dark pulse. Inactivation of Ddc neurons (Ddc:TNT-G) causes an overall reduction of linear locomotion which is much more pronounced during the light pulse. So, while linear locomotion of Ddc:TNT-VIF larvae during the light pulses, in comparison to that occurring during the dark pulses, is reduced by 1.3 (foraging) to 1.8 fold (wandering), in Ddc:TNT-G larvae this reduction is around 4.1 fold at both stages. Consistent with these results, when the response to light of Ddc:TNT larvae was measured in the semi-automated tracking system, a significant increase in the response to light was detected, throughout larval development, in Ddc:TNT-G larvae relative to control Ddc:TNT-VIF larvae (Fig. 3).
The values for direction change show a similar trend with a caveat. Upon expression of active TNT in Ddc neurons (Ddc:TNT-G), average direction change increases throughout the assay (Table 1). Consistent with previous findings, average direction change is always higher during the light pulses in all genotypes [11]. Interestingly, in foraging Ddc:TNT-VIF control larvae average direction change increases 4.2 fold during the light pulses as compared to the dark pulses. In contrast, direction change of Ddc:TNT-G foraging larvae is only 2.89 fold higher during the light pulses respect to the dark pulses. Inspection of qualitative and quantitative depiction of the behavior of an individual larva indicates that the increased change of direction during the light pulses caused by expression of active TNT in Ddc neurons may extend beyond the light period into the OFF pulses (e.g. see change of direction during the second ON/OFF transition in Figs. 1A and 2A), which may contribute to the average change of direction calculated for the dark pulse.
Taken together, these observations show that suppression of Ddc neuronal activity causes an increase in lightinduced change of direction and associated pause, and thus reduction of linear movement, characteristic of the larval photophobic behavior. Of note is the finding that during wandering stage, inactivation of Ddc-expressing neurons elicits a response to light (Figs. 1C, 2C and 3), whereas wild type larvae and control Ddc:TNT-VIF larvae Representative locomotor patterns during the ON/OFF assay of 3 rd instar Ddc:TNT larvae pulses. This suggests that, in the presence of light, Ddc:TNT-G larval locomotion is characterized by longer and/or more frequent pausing and change of direction and less linear movement. Inspection of corresponding perimeter stacks further supports these observations. During the light (ON) pulses, foraging Ddc:TNT-G larvae exhibited increased head swinging behavior and change of direction when compared with Ddc:TNT-VIF larvae (compare 1a-3a with 1b-3b). Although not as pronounced, similar behaviors were observed in Ddc:TNT-G larvae during early 3 rd instar wandering stage (91-94 h AH) (C), indicating that these larvae still respond to light (1c-3c). In contrast, early wandering Ddc:TNT-VIF larvae (D) mostly maintained linear movement and rarely head swung or changed direction over the course of the assay (1d-3d).
In order to determine whether the silencing of Ddc neurons causes a developmental delay that would explain the elevated larval response to light, all genotypes were tested for developmental timing (see Material and methods). The presence of specific morphological characteristics including the shape of their mouth hooks, the number of teeth, and the morphology of the anterior spiracles as well as behavioral characteristics such as crawling outside the food and emptying of the gut that occurs in the wandering stage demonstrated that suppression of Ddc neuronal activity does not change the timing of larval molts (data not shown) or the onset of the transition from foraging to wandering (see Additional file 1). Finally, we did not observe differences in pupation time between groups (data not shown).
It has been suggested that TNT expression may cause other phenotypes independently from its role as neuronal silencer [35]. Therefore, we carried out similar experiments using genetically modified Shaker and open rectifier K + channels, EKO and ORK1D-C respectively, both previously used to suppress neuronal excitability [36]. Larvae expressing either UAS-EKO or UAS-ORK1 -C construct in Ddc neurons show an increase in their response to light from late 2 nd to late 3 rd instar stage, similar to that displayed by Ddc:TNT-G larvae (see Additional files 2 and 3). Thus, we conclude that neuronal activity of Ddcexpressing neurons is required for regulation of the response to light during larval development.

Different subsets of Ddc neurons contribute to the modulation of the larval response to light
Ddc catalyzes the last step in the synthesis of both serotonin and dopamine, and thus it is found in both 5-HT and dopaminergic neurons (reviewed in [19]). A third group of cells, the corazonin (CRZ)-releasing neurons are also labeled by the Ddc-GAL4 construct [20]. In order to determine which subset (s) of Ddc neuron (s) (serotonergic, dopaminergic or corazonergic neurons) contribute(s) to the increase in the response to light observed in Ddc:TNT-G larvae, we took advantage of GAL4 driver constructs expressed exclusively in each neuronal type [37].
To target TNT in dopaminergic neurons, we used the TH-GAL4 driver [37]. Tyrosine hydroxylase (TH) performs the rate-limiting step in dopamine biosynthesis and is expressed specifically in dopaminergic cells ( [38]; data not shown). Targeted expression of active TNT using the TH-GAL4 driver does not cause any change in larval photobehavior (Fig. 4), suggesting that the increase in the response to light seen in Ddc:TNT-G larvae is not due to inactivation of dopaminergic neurons.
Tryptophan hydroxylase (TRH, known as TPH in mammals) catalyzes the biosynthesis of 5-hydroxytryptophan from the amino acid tryptophan and constitutes the ratelimiting step in 5-HT production. Drosophila has two enzymes able to synthesize 5-HT: neuronal tryptophan hydroxylase (DTRHn, referred here as to TRH), whose expression pattern in the CNS matches that of 5-HT ( [39]; data not shown), and phenylalanine hydroxylase (DTPHu), that functions as a non-neuronal or peripheral tryptophan hydroxylase [39][40][41]. Thus, in order to investigate the possible involvement of 5-HT neurons in modulation of larval photobehavior, we used the TRH-GAL4 driver [25].
Expression of TNT-G only in serotonergic neurons (TRH:TNT-G) causes a marked increase in the response to light of early foraging and wandering 3 rd instar larvae relative to that of control larvae (TRH:TNT-VIF) (Fig. 4). Of note, the level of this increase is comparable to that displayed by Ddc:TNT-G larvae, suggesting that the increase in the response to light of these larvae may be due mainly to suppression of serotonergic neuronal activity.
In order to evaluate the contribution of CRZ neuronal function to the regulation of larval response to light, we used the CRZ-GAL4 driver [24] to target the expression of TNT. CRZ:TNT-G larvae show a small but significant increase in the response to light when compared to control CRZ:TNT-VIF larvae in both foraging and wandering 3 rd instar stages (Fig. 4). Taken together, these findings indicate that 5-HT neurons but not dopaminergic neurons are involved in the regulation of the larval response to light. In addition, our results suggest that corazonergic neurons also contribute par-tially to the modulation of this larval behavior during development.

5-HT signaling is required for modulation of the larval response to light
The results above demonstrate that 5-HT-expressing cells play a role in the modulation of photobehavior during larval development. However, these experiments do not distinguish between this effect being due to decreased release of 5-HT or of other yet unknown neuromodulator also released by these neurons. One way to address this question is by studying the response to light of larvae with reduced 5-HT synthesis. To that end, we analyzed the response to light of homozygous mutant larvae carrying a putative null allele of the TRH locus (referred to as pBac-TRH). These mutant larvae are viable and show diminished 5-HT staining in the CNS [39]. Consistent with our previous observations, pBacTRH mutant larvae present an increase in their response to light when compared to that of heterozygous parental control larvae (Fig. 5). Taken together, our observations support the notion that serotonergic neuronal function is required for the modulation of the larval photobehavior. Furthermore, our findings reveal that this modulation is mediated at least in part by 5-HT signaling.

Silencing of 5-HT neurons does not disrupt larval locomotion
Locomotion represents a task-relevant behavior for the execution of the larval response to light in the ON/OFF assay. Therefore, it is important to determine whether inactivation of serotonergic neurons has an impact on locomotion in general. To that end, we measured the distance traveled by early foraging 3 rd instar TRH:TNT larvae in constant dark during 30 seconds. TRH:TNT-G and TRH:TNT-VIF larvae move equally well. Furthermore, no difference was found between the distance traveled by TRH:TNT-G and

Silencing of 5-HT neurons does not increase the response to mechanical stimuli
It is possible that silencing of 5-HT neurons causes increased response to other external stimuli besides light, X X Larvae expressing active TNT in Ddc neurons present increased response to light throughout development such as touch. Kernan and colleagues have shown that changes in larval sensitivity to mechanical stimuli can be measured using a touch sensitivity assay [42]. In their assay, wild type 3 rd instar larvae present a discrete set of stereotypical responses when stroked with the tip of an eyelash across the anterior body segments during linear locomotion. These responses range from withdrawing from the stimulus and turning away from it to no response at all.
We used a modified version of the touch sensitivity assay [10] to determine whether inactivation of serotonergic neurons also affect the larval response to mechanostimulation. Individual 3 rd instar foraging TRH:TNT larvae were touched four times during free crawling and the different responses observed were scored using the criteria of Caldwell and collaborators [10]

Modulation of the larval response to light requires 5-HT neurons located in the brain hemispheres
A total of 52 serotonergic neurons are found in the VNC in a segmental pattern, forming 14 bilaterally symmetrical clusters: 3 in the subesophageal region, 3 in the thoracic segments and 8 in the abdominal segments [43]. 5-HT projections in each segment, bifurcate ipsilaterally as well as contralaterally, innervating the entire neuropil [20]. Thus, it is possible that modulation of larval photobehavior is carried out by 5-HT neurons located in the VNC. The zinc-finger transcription factor Eagle (Eg) is required for differentiation of the VNC 5-HT neurons but not for those located in the brain hemispheres [44]. eg mutants carrying different alleles display different degree of ablation of VNC 5-HT neurons and the distribution of affected cells appears to be random ( [44]; our observations). For instance, larvae homozygous for the eg P289 hypomorphic allele show severe decrease in the number of 5-HT neurons in both the abdominal and thoracic segments, and to a lesser extent, in the subesophageal region ( [44]; see Additional file 5). In contrast, larvae carrying the heteroallelic eg 18B /eg-GAL4 combination present an overall less drastic reduction in the number of 5-HT-expressing cells of the VNC (see Additional file 5).
We used eg mutations to evaluate the relative requirement for the larval response to light of serotonergic neurons located in the brain hemispheres versus those located in the VNC. The response to light of eg P289 and eg 18B /eg-GAL4 mutants during both foraging and wandering 3 rd instar stages is indistinguishable from those of parental control larvae (Fig. 6), demonstrating that 5-HT neurons located in the VNC are not required for regulation of larval photobehavior. Together, these observations point to the 5-HT neurons located in the brain hemispheres as being critical for modulation of the larval response to light.

5-HT-mediated modulation of larval photobehavior does not occur at the photoreceptor level
In Drosophila larvae, circa 13 5-HT neurons can be seen projecting and arborizing in each brain hemisphere, innervating many different areas of the supraesophageal ganglion including the LOC where it overlaps with the photoreceptor termini [45]. Interestingly, a progressive increase in the innervation of the larval optic neuropil by 5-HT fibers from late 2 nd instar to late 3 rd instar larval stage coincides with the down-regulation of the larval response 3 rd instar pBacTRH mutant larvae display increased response to light to light during this period ( [17] and data not shown), suggesting that 5-HT neurons may be exerting their effect at the photoreceptor level.
We have previously reported that ablation of Rh6-specific photoreceptors prevents the appearance of the 5-HT arborization in the larval optic neuropil [29]. Similarly, over-expression of Slit in either all photoreceptors or in the Rh6 subset suppresses branching of the 5-HT processes in the LOC (see Additional file 6). Thus, in order to establish whether innervation of the larval optic neuropil by 5-HT neurons is required for the down-regulation of larval response to light seen during 3 rd instar wandering stage, we analyzed the response to light of 3 rd instar larvae in which the development of the optic neuropil 5-HT arborization was disrupted either by ablation of the Rh6 photoreceptors (UAS-hid/+;Rh6-GAL4/+) or by ectopic expression of Slit (GMR-GAL4/+;UAS-slit/+). The results shown in Fig. 7 demonstrate that lack or diminished 5-HT innervation of the LOC does not cause any significant dis-ruption in the larval response to light as measured in the ON/OFF assay. We conclude that 5-HT-mediated regulation of larval photobehavior does not occur at the photoreceptor level.
Knowledge of the 5-HT receptor involved in the downregulation of the larval response to light will aid the identification of neurons critical for the modulation of the larval response to light. Thus, we used a combination of upand down-regulation approaches in an attempt to identify the candidate receptor/s involved in this phenomenon.
Disruption of the optic neuropil 5-HT arborization does not affect the larval response to light For up-regulation, we took advantage of UAS constructs available for all receptors identified to date [46,47]. Down-regulation was limitedly achieved by targeted expression of a dsRNA construct available for 5-HT1B Dro [46], and a hypomorphic mutation in the 5-HT2 Dro gene (5-HT2 PL00052 allele) [48]. Although 5-HT1A Dro loss-of-function mutant larvae are viable, we were not able to test these larvae as they appear to display a developmental delay phenotype of variable penetrance (data not shown). Pan-neural expression of all UAS constructs was achieved by using the elav-GAL4 driver.
Based on the results obtained so far, we reasoned that increased 5-HT signaling achieved by up-regulation of 5-HT receptors (5-HT1A Dro , 5-HT1B Dro , 5-HT2 Dro , and 5-HT7 Dro ) might reduce the larval response to light during 3 rd instar foraging stage. In contrast, if down-regulation of 5-HT signaling by either expression of specific dsRNA constructs (5-HT1B Dro ) or a single gene mutation (5-HT2 PL00052 ) causes an increase in the response to light this would be likely more noticeable during 3 rd instar wandering stage, when normally larvae do not respond to the light stimulus in the ON/OFF assay.
Using the pan-neural driver elav-GAL4, forced expression of 5-HT1A Dro receptors, but not of any other 5-HT receptor subtype, causes a significant decrease in the response to light of foraging 3 rd instar larvae (Fig. 8). On the other hand, wandering larvae homozygous mutant for the 5-HT2 Dro gene (5-HT2 PL00052 ) shows the characteristic low response to light when compared with parental controls (data not shown). Similarly, targeted pan-neural expression of the dsRNA construct for the 5-HT1B Dro receptor does not affect the response to light of wandering 3 rd instar larvae (data not shown). Taken together, these observations point to the 5-HT1A Dro receptor subtype as a candidate receptor involved in 5-HT-mediated modulation of the larval response to light.

Discussion
In In Drosophila larvae, circa 100 5-HT neurons project toward different regions of the CNS, including the LOC, where they contact the LON [29]. Like in other organisms including the adult fly, their broad distribution in the nervous system suggests multiple roles for these cells. For instance, in addition to a suggested role in the modulation of larval heart rate [39], 5-HT neurons have been implicated in regulation of olfactory processing [53], and feeding behavior [39]. Here, we report on the role of sero-tonergic neurons in the modulation of Drosophila larval response to light, a paradigm used to study control of locomotion. regulation (Fig. 4). Differences in the strength of the promoters regulating the various GAL4 drivers used for targeted neuronal silencing must be taken into consideration when comparing the contribution of different neuronal groups. Thus, we cannot establish the relative contribution of 5-HT neurons and CRZ neurons to the modulation of the larval response to light. In addition, it has recently been shown that, during 3 rd instar larval stage, a fourth group of neurons immunoreactive for crustacean cardioactive peptide (CCAP) and myoinhibiting peptide (MIP) located in the ventral cord are also detected by Ddc-GAL4 [43]. Therefore, we cannot exclude the possibility that these cells may also play a modulatory role in the regulation of larval photobehavior.

5-HT neurons play a role in the modulation of the larval response to light
It is widely accepted that the activity of neuronal networks and the behavioral outputs controlled by them can be regulated by the action of different neuromodulators, which may or may not be co-released by the same terminal. Furthermore, neuronal co-localization of neuropeptides and 'classical' neurotransmitters including biogenic amines has been shown to be quite common in both vertebrates and invertebrates (reviewed in [54]). Therefore, it is important to consider that disruption of neuronal activity of 5-HT cells may affect not only the release of serotonin but also of other transmitter/neuromodulator potentially expressed by these neurons. Although it is currently unknown whether Drosophila serotonergic neurons express other neurotransmitters/neuromodulators, previous findings [43] as well as our personal observations (Camilletti and Campos unpublished results) indicate that CRZ and 5-HT do not co-localize.
Behavioral analysis of TRH null mutant larvae demonstrates that neuronal 5-HT signaling contributes to modulation of larval photobehavior (Fig. 5). It is important to mention that up-regulation of serotonin synthesis, and presumably therefore its release, by over-expressing TRH in Ddc-GAL4-expressing neurons, did not cause a decrease in the response to light of 3 rd instar foraging larvae (data not shown and [55]). Nevertheless, this observation does not argue against a role of serotonin in regulation of larval photobehavior. For instance, higher levels of released 5-HT during development might be over-compensated by increased up-take and/or inactivation of this amine as well as by down-regulation of 5-HT receptors, resulting in a wild type behavioral phenotype.
The increased response observed in pBacTRH mutants is not as high as that obtained after silencing of 5-HT neurons by TNT expression (compare Fig. 5 with Fig. 4). These results suggest that serotonin may not be the only signaling molecule released by 5-HT neurons and involved in modulation of larval photobehavior. Alternatively, this difference could be explained by residual serotonin release in pBacTRH mutants. It has been previously shown that pBacTRH larvae show decreased but not complete absence of 5-HT expression in the CNS [39]. These authors suggested that this is due, perhaps, to the reuptake of circulating 5-HT synthesized peripherally by DTPHu. Thus, it is feasible that, in these mutants, small amounts of 5-HT are still released from serotonergic neurons, thereby partially regulating the larval response to light.
It has been reported that neuronal 5-HT regulates larval feeding [39] and body size in adult flies [49]. Nevertheless, decreased 5-HT levels or release does not appear to affect larval growth, as the size of Ddc:TNT-G, TRH:TNT-G or TRH mutant larvae is within the range of wild type controls (data not shown). These observations are consistent with our conclusion that silencing of the 5-HT neurons did not cause a developmental delay.
As motor performance is crucial for analysis of photobehavior in our assay, it is important to consider the impact of diminished 5-HT synthesis or release on this task-relevant behavior. Previous observations indicate that the locomotion of TRH mutant larvae is normal as measured by the number of body wall contractions [39]. Our results agree with those of Neckameyer and collaborators, as Ddc:TNT-G (see Additional file 4), TRH:TNT-G as well as TRH mutant larvae (data not shown) showed normal locomotion in constant dark. In addition, our results demonstrate that inactivation of 5-HT neurons does not result in a generalized disruption of the larval response to external stimuli.

Modulation of the larval response to light requires 5-HT neurons located in the brain hemispheres
Mutations in the eg gene affect serotonergic neurons located in the subesophageal, thoracic and abdominal segments of the VNC but not those 5-HT neurons located in the brain lobes ([56]; see Additional file 5). The remaining VNC 5-HT neurons often show severe pathfinding defects ( [44]; see Additional file 5). Interestingly, eg mutant larvae respond to light indistinguishably from control larvae and show the expected reduction in this response as they reach the wandering stage, demonstrating that 5-HT neurons located in the VNC are not required in this process (Fig. 6).
It has been shown that CRZ neurons located in the VNC also express eg during 3 rd instar stage [56]. It is yet to be established whether VNC CRZ neurons are also affected in eg mutants. If so, our results suggest that this subset of CRZ cells may not be involved in the modulation of larval photobehavior.
The invasion of the LOC by 5-HT processes and their contact with the LON coincides with the gradual decrease in the larval aversion to light, suggesting that 5-HT neurons may be modulating this larval behavior at the photorecep-tor level ( [17] and data not shown). However, absence of the 5-HT arborization or disruption of its branching did not affect the response to light of either foraging or wandering larvae (Fig. 7), ruling out 5-HT-mediated modulation of this behavior at the photosensory level.
Neuromodulators may regulate rhythmic motor behaviors by acting at different levels within a specific neuronal circuit, that is, at the sensory and/or central level (reviewed in [3]). The latter may involve modulation within the CPG or at the level of the motorneurons (reviewed in [3]). It is worth noting that the CPG controlling Drosophila larval locomotion is thought to be located in the VNC [57]. Thus, one possibility is that the 5-HT neuromodulatory effect occurs within the brain at a central level other than the CPGs (e.g. higher order interneurons). Alternatively, 5-HT neuronal inputs descending from the brain hemispheres may act directly on the CPGs. Interestingly, early immunohistochemical studies have suggested that some 5-HT longitudinal fibers in the VNC may derive from brain lobe neurons [45].

5-HT-mediated modulation of larval photobehavior may involve 5-HT1A Dro receptors
In both vertebrates and invertebrates, 5-HT is widely expressed and is able to activate several 5-HT receptor subtypes, coupled to different signaling pathways (reviewed in [58]). Our results suggest that 5-HT1A Dro receptors may play a role in the modulation of the larval response to light (Fig. 8), further supporting the role of serotonin in this regulation. However, it is important to consider that over-expression of 5HT1A Dro receptors using the pan-neural driver elav-GAL4 most likely disrupt synaptic activity of the 5-HT neurons themselves. As a result, 5-HT1A Dro might act on these cells as an autoreceptor, thus modifying the larval response to light. Therefore, at the present time our observations allow us only to suggest this serotonin receptor as a candidate receptor involved in the regulation of photobehavior. Future experiments aimed at further investigating the possible involvement of this 5-HT Dro receptor in this phenomenon will help with the identification of the target cells on which larval 5-HT neurons act to modulate the larval response to light. These cells may in turn represent the critical neurons for the performance of this behavior.

Conclusion
In this paper we investigated the mechanisms underlying modulation of larval photobehavior and report a novel role for serotonergic and corazonergic neurons in Drosophila larva. Our data demonstrate that 5-HT neurons as well as corazonergic neurons contribute to the reduction in the response to light normally observed during larval stage. Study of the serotonergic system indicates that 5-HT-mediated modulation of this behavior is carried out by 5-HT cells located in the brain hemispheres. Further-more, our observations do not support the idea that this effect is a result of a direct role of 5-HT signaling on photoreceptor termini. Lastly, the suggestion that 5-HT1A Dro receptors are involved in this modulation may provide a tool to identify the target neurons of this 5-HT signaling and perhaps critical for the control of locomotion by light.