A Neuronal Pathway that Commands Deceleration in Drosophila Larval Light-Avoidance

Abstract

When facing a sudden danger or aversive condition while engaged in on-going forward motion, animals transiently slow down and make a turn to escape. The neural mechanisms underlying stimulation-induced deceleration in avoidance behavior are largely unknown. Here, we report that in Drosophila larvae, light-induced deceleration was commanded by a continuous neural pathway that included prothoracicotropic hormone neurons, eclosion hormone neurons, and tyrosine decarboxylase 2 motor neurons (the PET pathway). Inhibiting neurons in the PET pathway led to defects in light-avoidance due to insufficient deceleration and head casting. On the other hand, activation of PET pathway neurons specifically caused immediate deceleration in larval locomotion. Our findings reveal a neural substrate for the emergent deceleration response and provide a new understanding of the relationship between behavioral modules in animal avoidance responses.

Electronic supplementary material

The online version of this article (10.1007/s12264-019-00349-w) contains supplementary material, which is available to authorized users.

Introduction

Avoidance responses are generally thought to result from the activation of neurons that program avoidance behavior in response to external stimuli. When making an avoidance response during locomotion, animals such as Drosophila generally choose from two options: (1) stop on-going movement and perform a backward movement to leave [1–5], or (2) stop on-going movement, change the heading direction by making a turn, and continue forward movement to leave [6–9]. In both modes, it is necessary to stop on-going forward movement to successfully escape. Although the neural circuits for initiating or maintaining locomotion have been relatively well studied [10–14], the neural mechanisms involved in stopping behavior have received relatively little investigation. Most studies related to this topic focus only on stopping behavior – simply the cessation of movement without further action [15–18]. However, the neural mechanisms underlying the stopping or deceleration response that is followed by further behavioral maneuvers such as turning or backward movement, and their role in typical avoidance behavior, are still unclear. We hypothesized that the deceleration response is controlled by a specific neural pathway to serve as a critical initial step for avoiding danger.

Widely used in avoidance studies, phototaxis in Drosophila larvae is a well-suited model for testing such a hypothesis. When encountering a light area in a dark environment, a larva casts its head from side to side (i.e., sweeps its head back and forth) and then makes a turn to change the direction of movement [8, 19]. Although not regularly mentioned, larvae consistently slow down prior to head casting and turning behavior [6, 20]. The neural circuitry underlying larval light avoidance has not been fully characterized. Larvae sense light either by Bolwig’s organs (BOs), which are located in the anterior tip and are responsive to light of low or intermediate intensity, and multidendritic neurons in the body wall that are only sensitive to high-intensity light [21]. BOs send projections into the brain and synapse on lateral neurons, such as pigment-dispersing factor neurons and fifth lateral neurons, which are involved in phototaxis [22–24]. Prothoracicotropic hormone (PTTH) neurons, which are required for normal phototaxis, are downstream of the lateral neurons [25–27]. However, the way in which visual signals are transmitted to the motor system to generate avoidance behavior is not clear.

Therefore, we set out to determine the downstream neurons of PTTH neurons in phototaxis-regulating pathway and how the neurons in the pathway control light avoidance.

Materials and Methods

Fly Culture and Strains

All flies were raised at 25°C on standard medium in a room with a 12 h:12 h light/dark cycle. For details of fly stock sources, see Supplemental Experimental Procedures.

Behavioral Tests

In the behavioral assay for light/dark preference, 20 third-instar larvae were allowed to choose between light and dark for 10 min. In the light spot assay, individual larvae were allowed to start from darkness and enter a blue light spot on an agar plate. The whole process of larval entry and exit of the light spot was recorded and analyzed using SOS software, a computer-vision tracking system written in Matlab (Mathworks) [6] and custom-written scripts. For optogenetics and larval locomotion, larvae of the proper genotypes raised on food supplied with 5.0 mmol/L trans-retinal were stimulated with 620-nm (red) light during straight forward locomotion. Video recordings of larval locomotion were processed with SOS and custom-written software (see Supplemental Experimental Procedures for details).

Calcium Imaging and Confocal Microscopy

For Ca²⁺ imaging, individual second-instar larvae were immobilized with a coverslip. To assess responses to optogenetic stimulation, Ca²⁺-dependent fluorescent intensity in neurons was monitored with an Olympus FV1200 MPE multi-photon microscope (Tokyo, Japan). Images were processed using Image J (https://imagej.nih.gov/ij/) (see Supplemental Experimental Procedures for details). For confocal microscopy, dissected larval brains were mounted and viewed using an Olympus FV1000 confocal laser scanning microscope.

Statistics

Statistics were performed with Prism6.0 (Graphpad Inc., https://www.graphpad.com/). For SOS-derived locomotor speed and head cast angle data, the nonparametric Mann-Whitney test and Kruskal-Wallis followed by Dunn’s post hoc test were used. For all other data that were assumed to conform to a Gaussian distribution, we used Student’s t-tests or ANOVA followed by Tukey’s post hoc test.

Results

Deceleration Was Necessary for Head Casting and Light Avoidance

Because avoidance behavior typically involves changing the heading direction, we analyzed the time series of larval movement-related parameters during head casting in light-avoidance using a light-spot assay [19]. We found two interesting phenomena: (1) deceleration was almost always initiated before head casting, and (2) the head cast was initiated during deceleration. The former finding implied that deceleration is not a result of head casting, and the latter indicated that a larva does not have to fully stop to turn its head. As expected, the maximal body bending appeared after the angular velocity peak (Fig. S1A, Movie S1). As the sequential occurrence of deceleration, head casting, and maximal body bending appeared to be stereotypical, we named it the DOT pattern (deceleration starts → headomega peaks → headtheta peaks, see “Methods” for details). As shown in Figure S1B, larval locomotor speed normally dropped within 1 s before the initiation of body bending. The negative correlation between minimal speeds and maximal sizes of head casting showed that the size of the head cast tended to be greater when the speed after deceleration was lower (Fig. S1C, D). Remarkably, the size of the head cast decreased with increasing speed of movement. As a result, large head casts (>50°) only occurred at low speeds (Fig. S1C). Comparing with the case during free movement, the speed after deceleration in the light response was significantly lower and the size of the head cast following deceleration was significantly larger (Fig. S1E, F). Given that PTTH neurons are required for larval light avoidance [25, 26, 28], we determined whether they play a role in mediating the light-induced deceleration. Larvae with blocked PTTH neurons exhibited smaller head casts when encountering light in the light-spot assay (Fig. S1G). Meanwhile, the minimal speed after light-induced deceleration was significantly higher than that in the parental controls (Fig. S1H), while the locomotor speeds in the experimental and control groups during free movement were all >1 mm/s and did not significantly differ (Fig. S2). Thus, a lack of light-avoidance may be caused by a defect in light-induced deceleration, at least in the case of inhibiting PTTH neurons.

It should be noted that the degree of head casting could be either large or small at low speed (Fig. S1C), such that it did not correspond to the speed value in a one-to-one manner. Therefore, the relationship between speed and head casting was not like the velocity-curvature relationship of a projectile as defined by physical laws, or the stereotyped velocity-curvature relationship when human or animal moves in a curved trajectory [29, 30]. There must be neural mechanisms that control the larval deceleration and head casting separately, but in a coordinated manner.

EH Neurons are Innervated by PTTH Neurons in Larval Light-Avoidance

We next determined where PTTH neurons could send output. One pair of EH-expressing neurons that are known to regulate eclosion in response to light-on at dawn [31] became candidates. We used EH-Gal4 to drive the expression of the presynaptic marker syt-GFP [32, 33] and the postsynaptic marker Denmark [33]. The syt-GFP signal was mainly located in the corpora cardiaca of the ring gland and along the midline of the ventral nerve cord (VNC), suggesting that these regions were presynaptic sites. The Denmark signal was located mainly in the medial anterior part of each hemisphere, close to the dendrites of PTTH neurons (Fig. 1A). Direct comparison of the morphologies of PTTH and EH neurons showed that the dendrites of EH neurons intermingled with the medial dendritic arborizations of PTTH neurons (Fig. 1B, C). Further confirmation was established with GFP reconstitution across synaptic partners (GRASP) [34, 35]. We found significant recombinant GFP signals only in the region of overlapping dendrites, while no signal was seen in the controls (Fig. 1D, F). It is noteworthy that no GRASP signal was found in the ring gland where the axonal termini of PTTH and EH neurons are located (Fig. 1D). Thus, there might be a physical interaction between the dendrites of PTTH and EH neurons.

Fig. 1.

EH neurons receive visual input from PTTH neurons. A Expression of the pre- and post-synaptic markers sytGFP and Denmark, respectively, in EH neurons using EH-Gal4 (green, sytGFP; magenta, Denmark; arrows indicate dendrites; arrowheads indicate axonal termini). B Stacked image of co-staining of PTTH-Gal4 and anti-EH in larval central brain (green, PTTH-Gal4 marked by GFP; magenta, anti-EH signal; arrows indicate co-localization). C Single-layer view of (B). Arrows indicate co-localization. D–F GRASP between PTTH and EH neurons. D The GFP signal occurred where the dendrites of PTTH and EH neurons overlap (arrows). E, F GRASP control in which split GFP expression was driven by EH-LexA (E) or PTTH-Gal4 (F). G Optogenetic activation of PTTH neurons stimulated a Ca²⁺ response in EH neurons (n = 13 for PTTH-Gal4/+; EH > GCAMP6s, 10 for UAS-Chrimson/+; EH > GCAMP6s, and 8 for PTTH > Chrimson; EH > GCAMP6s. H Quantification of the peak Ca²⁺ response in (G) (***P < 0.001). I Optogenetic activation of EH neurons inhibited the Ca²⁺ signal in PTTH neurons (n = 12 for EH-LexA/+; PTTH > GCAMP6m, 16 for LexAop-Chrimson/+; PTTH > GCAMP6m, and 5 for EH > Chrimson; PTTH > GCAM6m. J Quantification of peak Ca²⁺ response during red light stimulation in (I) (**P < 0.01). Red lines in (G) and (I), periods of optogenetic stimulation. Scale bars, 50 μm in (A–F). One-way ANOVA followed by post hoc Tukey’s multiple comparison test was used for (H) and (J).

Next, we combined Ca²⁺ imaging with optogenetics to further confirm the existence of a functional connection between EH and PTTH neurons. We found that, as expected, exciting PTTH neurons induced a robust increase in fluorescence in EH neurons. The peak response of a >100% increase was reached within 4 s of the onset of red light stimulation (Fig. 1G, H). Interestingly, optogenetic activation of EH neurons inhibited PTTH neurons by >40% as measured by fluorescence intensity (Fig. 1I, J). These results supported the existence of dendrodendritic synapses between PTTH and EH neurons; recurrent inhibition is commonly seen in dendrodendritic interactions [37, 38]. Taken together, our results suggested that EH neurons receive visual input via PTTH neurons.

EH Neurons Mediate Deceleration in Larval Light-Avoidance

We then asked if EH neurons are required for larval light-avoidance. Larvae expressing UAS-TNTG [38], or UAS-dORKΔC [40] with EH-Gal4 [31] exhibited impaired phototaxis, and this effect was reversed by the introduction of elav-Gal80 (Fig. 2A). This finding was confirmed using R28E02-Gal4, which labeled the same two neurons as EH-Gal4 (Fig. S3). A possible developmental effect of inhibition could be excluded using the temporal and regional gene expression targeting (TARGET) system [40] in which Tub-Gal80^ts was combined with EH-Gal4 and UAS-dORKΔC. After being heat-shocked at 32°C for 12 h to allow instant expression of dORKΔC and thus inhibition of EH neurons; mid-late third-instar larvae that had been raised at 18°C showed no light-avoidance. Sibling larvae that did not receive this heat-shock treatment demonstrated normal light avoidance (Fig. 2B). Furthermore, hyperactivating EH neurons by expressing the depolarizing Na⁺ channel NaChBac [41] enhanced the larval preference for darkness over light (Fig. S3D). Together, these results suggested that EH neurons are required for larval light-avoidance.

Fig. 2.

Larval EH neurons were part of the visual pathway that commanded the deceleration in light avoidance. A Inhibiting EH neurons abolished the preference for darkness (**P < 0.01, ***P < 0.001; n = 19, 16, 17, 16, 19, and 19 from left to right). B Acute inhibition of EH neurons abolished the preference for darkness (Heat shock, with restrictive temperature; No heat shock, with permissive temperature; ***P < 0.001, n.s., not significant; n = 17 for all groups). C, D Inhibiting EH neurons decreased the degree of head casting in response to light-spot entry (C) and increased the minimal speed after light-evoked deceleration (D) (*P < 0.05, n = 99, 67, and 70 in (C) and n = 100, 69, and 72 in (D) for EH-Gal4/+, UAS-TNTG/+, and EH > TNTG, respectively). E Ca²⁺ imaging of cell bodies of EH neurons expressing GCAMP6s in response to optogenetic activation of GMR-Gal4-labeled photoreceptor neurons (n = 11 for GMR > Chrimson; EH > GCAMP6s, 9 for GMR-Gal4/+; EH > GCAMP6s, and 10 for UAS-Chrimson/+; EH > GCAMP6s. F Quantification of peak Ca²⁺ imaging responses in (E) (***P < 0.001). One-way ANOVA followed by post hoc Tukey’s multiple comparison test was used for (A, B) and (F); the Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test was used for (C, D).

To determine whether EH neurons are required for the light-induced deceleration and the head casting, we tested the light-avoidance using a light-spot assay during which detailed larval behavior was recorded [19]. Compared with the control parental lines, the degree of the light-induced head casting was reduced when the EH neurons were blocked with UAS-TNTG (Fig. 2C). This was to be expected as defects in head casting and turning behavior appeared to be directly linked to the loss of light-avoidance. Notably, the minimal speed after light-induced deceleration was significantly higher in the larvae with inhibited EH neurons than in control lines (Fig. 2D). The higher minimal speed was not due to increased locomotor speed because that of the experimental line was not higher than that of the control lines. Note that the locomotor speed in all three lines was >1 mm/s, the normal speed of locomotion (Fig. S2). Thus, EH neurons appear to be required for light-evoked deceleration, in addition to light-induced head casting in light-avoidance.

To verify that EH neurons indeed receive input from larval “photoreceptors” – the BOs – we combined optogenetics with Ca²⁺ imaging to directly activate their photoreceptors and record the response of EH neurons [42]. Larvae with UAS-Chrimson expressed in larval photoreceptors with GMR-Gal4 and GCAMP6 [43] expressed in EH neurons were cultured on food supplied with all-trans-retinal. Upon stimulation of the photoreceptors by red light, we recorded an immediate and strong Ca²⁺ response of ~100% increase in fluorescence intensity in the cell bodies (Fig. 2E, F).

Taken together, EH neurons are part of the larval visual pathway required for the light-evoked deceleration. These neurons might indirectly affect head casting by regulating deceleration.

EH Neurons Innervate Tdc2 Motor Neurons in Larval Light-Avoidance

As the axons of EH neurons project through the larval VNC, we examined whether they directly innervate motor neurons that are involved in processing light-avoidance behavior. Close proximity was easily detected at the level of single confocal sections when motor-neuron-specific D42-Gal4 was counterstained with anti-EH (data not shown). As D42-Gal4 was comprehensively expressed, we asked specifically which motor neurons were innervated by EH neurons. We tested Tdc2-Gal4-labeled motor neurons [44] by comparing Tdc2-Gal4 [45] with anti-EH in the larval nervous system. In addition to sporadic overlapping in the brain, the anti-EH signal overlapped well with the Tdc2-Gal4 signal in abdominal segments of the VNC (Fig. 3A, B). We next used GRASP to confirm the physical interaction between EH neurons and Tdc2 motor neurons in the VNC. Again, we observed a GRASP signal between EH neurons and Tdc2-Gal4-labeled neurons in the VNC (Fig. 3C–E). Thus, EH neurons directly innervate Tdc2 motor neurons.

Fig. 3.

EH neurons innervate Tdc2 motor neurons. A Stacked image of co-staining of Tdc2-Gal4 marked by GFP and anti-EH in the larval VNC (green, GFP signal; magenta, anti-EH signal; arrows indicate co-localization. B Single layer view of (A). Arrows indicate co-localization. C–E GRASP between EH neurons and Tdc2 motor neurons (C). The GFP signal is seen where the axonal termini of EH neurons and the dendrites of Tdc2 motor neurons overlap, as indicated by arrows. D, E GRASP control in which split GFP expression was driven by Tdc2-Gal4 (D) and EH-LexA (E). No GFP signal for GRASP was seen (arrows). F Statistics of the peak Ca²⁺ response in cell bodies of Tdc2 motor neurons upon optogenetic activation of EH neurons (**P < 0.01, ***P < 0.001, one-way ANOVA and post hoc Tukey’s multiple comparison test; no other significant differences between the experimental and control groups. See Fig. S4 for details. Scale bars, 50 μm in A–E.

We further confirmed the EH-Tdc2 neuronal interaction using a combination of optogenetics and Ca²⁺ imaging. Upon stimulation of EH neurons with 620-nm light, we observed a >50% increase in fluorescence intensity within 5 s in the avm3 and avm7 clusters of Tdc2 neuron cell bodies in abdominal segments (Fig. 3F, Fig. S4). The Ca²⁺ response was not significant in Tdc2 neurons in other abdominal segments. Thus, we confirmed that EH neurons innervate Tdc2-Gal4 neurons, at least those in the avm3 and avm7 clusters.

Tdc2 Motor Neurons Mediate Deceleration in Larval Light-Avoidance

Although we had determined that Tdc2-Gal4-labeled neurons were targeted by EH neurons, whether larval Tdc2 motor neurons were required for the light-evoked deceleration and light-avoidance was unclear. Thus, we blocked Tdc2-Gal4-labeled neurons by expressing TNTG and dORKΔC. Larvae with inhibited Tdc2-Gal4 neurons lost light-avoidance (Fig. 4A). Introduction of VNC-specific tsh-Gal80 [46] that prevented Tdc2-Gal4 expression in the VNC restored normal light-avoidance, suggesting that the Tdc2 motor neurons in the VNC are required for the light preference (Fig. 4A). When the TARGET system was used to exclude possible developmental effects, light-avoidance in larvae subjected to 12 h of 32°C heat shock was abolished, while the control larvae raised at 18°C throughout had intact light-avoidance (Fig. 4B). Thus, Tdc2 motor neurons are required for normal light-avoidance. We next conducted the light-spot assay in larvae in which Tdc2 neurons were blocked with TNTG. These larvae demonstrated a lesser degree of head casting when encountering light (Fig. 4C). However, their minimum speed after light-induced deceleration was not significantly greater than that in the parental controls (Fig. 4D). Considering that the inhibition of Tdc2 motor neurons led to slower locomotion in free movement (Fig. S2) [44], the lack of a change in minimal speed after deceleration might reflect less deceleration in response to light stimulation. Thus, inhibiting Tdc2 neurons appeared to undermine the light-induced deceleration and head casting.

Fig. 4.

Tdc2 motor neurons mediated light-induced deceleration. A Inhibiting Tdc2 motor neurons abolished larval preference for darkness (**P < 0.01, ***P < 0.001; n = 22, 23, 17, 19, 23, and 23 from left to right). B Acute inhibition of Tdc2 neurons abolished the preference for darkness (Heat shock, with restrictive temperature; No heat shock, with permissive temperature; ***P < 0.001, n.s., not significant; n = 29, 19, 47, 15, 19, and 28 from left to right). C, D Effect of inhibiting Tdc2 neurons on larval head casting and deceleration in response to light-spot entry. C The degree of head casting was decreased. D The minimum speed after deceleration did not differ from controls (**P < 0.01, n.s., not significant; n = 81, 52, and 92 in (C) and 82, 53, and 96 in (D) for Tdc2-Gal4, UAS-TNTG/+ and Tdc2 > TNTG, respectively). One-way ANOVA followed by post hoc Tukey’s multiple comparison test was used for A and B. Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test was used for C and D.

Activation of PET Pathway Neurons Induced Deceleration

We attempted to verify the function of PET pathway neurons in deceleration by activating them artificially using optogenetics. We drove the expression of UAS-Chrimson [47] with PTTH-Gal4, EH-Gal4, and Tdc2-Gal4. Upon stimulation with 620-nm light, larvae raised on food containing all-trans-retinal showed more speed reduction than controls, though the deceleration response was mild (Fig. 5A, B for PTTH neurons, 5C, D for EH neurons, 5E, F for Tdc2 motor neurons, Movie S2–5). This finding was confirmed using NP423-Gal4 and R28E02-Gal4 that respectively labeled PTTH neurons and EH neurons (Fig. S3E–L and Fig. S5). It should be noted that head casting was not always seen after deceleration, suggesting that the activation of PET pathway neurons specifically induces deceleration.

Fig. 5.

Optogenetic activation of PET pathway neurons induced deceleration. A, B Optogenetic activation of PTTH neurons using PTTH-Gal4 led to strong deceleration. A Time curves of percentage change in larval tail speed. B Maximum decrease in tail speed within 2 s after optogenetic activation of PTTH neurons (n = 15, 23, and 20 for PTTH-Gal4 > Chrimson, UAS-Chrimson/+, and PTTH-Gal4, respectively; red line, period of optogenetic stimulation). C-D Optogenetic activation of EH neurons using EH-Gal4 led to a deceleration in larval locomotion. C Time curves of percentage change in larval tail speed. D Maximum decrease in tail speed within 2 s after optogenetic activation. n = 40, 37, and 29 for EH-Gal4 > Chrimson, UAS-Chrimson/+, and EH-Gal4/+, respectively. E, F Optogenetic activation of Tdc2 neurons led to strong deceleration. Time curves of percentage change in larval tail speed (E). Maximum decrease in tail speed within 2 s of optogenetic activation of Tdc2 neurons labeled with Tdc2-Gal4 (F) (n = 44, 45, 43, 37, and 15 for Tdc2-Gal4, UAS-Chrimson, Tdc2-Gal4 > Chrimson, Tdc2-Gal4/tsh-Gal80 > Chrimson, and tsh-Gal80/+; UAS-Chrimson/+, respectively; red lines, period of optogenetic stimulation in (A), (C) and (E)). *P < 0.05, **P < 0.01, ***P < 0.001, Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test was used for (B), (D), and (F).

Discussion

In this study, we found that larvae need to decelerate before making a head cast in light-avoidance. This represents an additional stage of deceleration between the locomotion and head casting that have received the most attention in studies of larval avoidance behavior. The results showed that the light-induced deceleration response is mediated by the PET pathway and facilitates subsequent head casting in larval light-avoidance behavior.

PET pathway neurons can inhibit locomotion in different ways. First, Tdc2 motor neurons may use tyramine as a neurotransmitter, which can play an inhibitory role by inhibiting muscle cells and thus stopping larval locomotion [48, 49]. Another possibility is that a light stimulus may disturb the regular firing of Tdc2 motor neurons or their downstream neurons required for the normal rhythm of locomotion. Because motor neurons or pre-motor neurons need to fire in a specific sequence for coordinated locomotion [45, 50–52], disturbing the sequence may be a natural way of reducing locomotor speed. If Tdc2 motor neurons themselves do not produce firing sequences to drive periodic locomotion, they may regulate locomotor central pattern generator (CPG) neurons in the larval VNC, for example, via gap junctions between motor neurons and premotor neurons [52–54]. In either case, activating Tdc2 neurons in a larva by an external stimulus can interfere with its locomotor rhythm and reduce its crawling speed.

Most stop responses in previous reports, such as the gentle touch in Drosophila larvae, touch in tadpoles, and the stop mediated by reticulospinal V2a neurons in mice, appear to result in a full stop with no subsequent turning behavior [15–17]. The PET-mediated light-induced deceleration response seems to be an emergent stop response that is generally followed by a head cast. As the head cast is actually initiated during the process of deceleration, deceleration can be considered as a preparatory stage for head casting.

In the PET pathway, the long axonal projections of EH neurons resemble those of the reticulospinal neurons that command stop responses in vertebrates [16–18]. These neurons also have cell bodies and dendrites located in the brain to receive input from higher order neurons, while sending long axonal projections to innervate locomotor CPG neurons in the spinal cord and produce a stop response. It is likely that sensory stimulus-induced deceleration responses in vertebrate avoidance behavior are mediated by neuronal pathways similar to the PET pathway.

Therefore, in support of the hypothesis that there is a separate neural pathway for controlling deceleration prior to avoiding danger, the discovery of the PET pathway gives a new insight into survival-critical avoidance behaviors of animals in general.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Compliance with Ethical Standards

Conflict of interest

The authors declare no competing interest.