After emergence, adult flies and other insects select a suitable perch and expand their wings. Wing expansion is governed by the hormone bursicon and can be delayed under adverse environmental conditions. How environmental factors delay bursicon release and alter perch selection and expansion behaviors has not been investigated in detail. Here we provide evidence that in Drosophila the motor programs underlying perch selection and wing expansion have different environmental dependencies. Using physical manipulations, we demonstrate that the decision to perch is based primarily on environmental valuations and is incrementally delayed under conditions of increasing perturbation and confinement. In contrast, the all-or-none motor patterns underlying wing expansion are relatively invariant in length regardless of environmental conditions. Using a novel technique for targeted activation of neurons, we show that the highly stereotyped wing expansion motor patterns can be initiated by stimulation of NCCAP, a small network of central neurons that regulates the release of bursicon. Activation of this network using the cold-sensitive rat TRPM8 channel is sufficient to trigger all essential behavioral and somatic processes required for wing expansion. The delay of wing expansion under adverse circumstances thus couples an environmentally sensitive decision network to a command-like network that initiates a fixed action pattern. Because NCCAP mediates environmentally insensitive ecdysis-related behaviors in Drosophila development before adult emergence, the study of wing expansion promises insights not only into how networks mediate behavioral choices, but also into how decision networks develop.
A primary function of the nervous system is to execute behavioral choices, that is, to alter an animal's behavior in response to changes in environment or internal demands. The neural mechanisms that underlie such choices have principally been investigated in reduced, invertebrate preparations, where work has focused largely on behaviors that are elicited by simple environmental stimuli and have motor outputs that are easily monitored in the absence of actual behavior (Kristan and Gillette, 2007). The introduction of genetically encoded effectors, which permit the targeted manipulation of specific neurons in intact animals (Miesenböck and Kevrekidis, 2005; Luo et al., 2008), is increasingly allowing the investigation of behavioral decisions that are based on more complex environmental stimuli and on neuromodulatory changes in internal state (Wu et al., 2003; Adamantidis et al., 2007; Dickson, 2008; Yang et al., 2008). Because the new methods are easily deployed in genetic model organisms, such as Drosophila (Holmes et al., 2007), which are also amenable to developmental and comparative studies, they promise progress in addressing the broader issues regarding the development, evolution, and adaptive value of the processes and circuits underlying behavioral choice first articulated by Tinbergen (2005).
An extensively analyzed neuromodulatory circuit, for which physiological, developmental, and comparative data are available, is the network underlying ecdysis in insect molting (Ewer, 2005; Truman, 2005; Zitnan et al., 2007). This circuit orchestrates the motor patterns (the so-called “ecdysis sequence”) needed to shed the old exoskeleton and expand and harden the new one. This sequence must be executed flawlessly, even in the absence of previous experience, and is therefore largely governed by endogenous factors, but often some components of the sequence are sensitive to external conditions to allow synchronization of hormonal release with environmental cues.
Such coordination is frequently required for the adapted ecdysis sequence used by adult insects to expand their wings, a process which in newly emerged flies requires a suitable perch (Cottrell, 1962; Fraenkel and Hsiao, 1965). If prevented from perching by perturbation or confinement, flies will choose to delay wing expansion for hours. The mechanisms underlying this choice are unknown, as are the neural determinants of the environmentally sensitive perch selection program. However, in Drosophila the network underlying wing expansion has been shown to include a set of neurons implicated in both larval and pupal ecdysis in Drosophila and in other insects (Gammie and Truman, 1997; Park et al., 2003; Clark et al., 2004; Luan et al., 2006). These neurons (i.e., NCCAP) express crustacean cardioactive peptide, and a subset expresses the hormone bursicon (Dewey et al., 2004; Luo et al., 2005), which governs two motor patterns underlying wing expansion (Baker and Truman, 2002).
Here we combine environmental perturbations with targeted manipulations of NCCAP activity to show that environmental factors prolong the perch selection program and delay activation of the expansion program. We also introduce a method for acutely activating neurons by temperature decrements to demonstrate that NCCAP activation is sufficient to abrogate environmental inhibition and elicit the expansion program.
Materials and Methods
All flies were grown on corn meal-molasses medium and maintained at 25°C in a constant 12 h light/dark cycle. Wild-type flies (Canton-S strain) were from the Bloomington Stock Center (Indiana University, Bloomington, IN). The Gal4 driver lines used in this study were generous gifts of the following: yw; CCAP-Gal4; + and w; +; CCAP-Gal4 (Park et al., 2003, John Ewer); w; c929-Gal4; + (O'Brien and Taghert, 1998, Paul Taghert). The elavc155-Gal4;+;+, yw;2×UAS-EGFP; 2×UAS-EGFP, and 1×, 2×, and 3× EKO lines, which carry one, two, and three copies of the EKO transgene have been described previously (White et al., 2001).
Generation of UAS-TRPM8 constructs and fly lines
The coding sequence of the TRPM8 cDNA (McKemy et al., 2002) was amplified by PCR using primers that introduced an EcoRI restriction site and an optimized translation initiation motif [with sequence CAAA, Cavener (1987)] immediately before the ATG start codon and a KpnI restriction site just after the stop codon. The amplified fragment was then subcloned into the pUAST plasmid for P-element transformation, using the unique EcoRI and KpnI restriction sites in the multiple cloning site. P-element injections and isolation of transformants was performed by Genetic Services. Flies with inserts of UAS-TRPM8 on the second (yw; C4-D; +) and third (yw; +; C4-A and yw; +; C1-A2) chromosomes were used in this study, separately or in combination.
Behavioral observations and analysis
Behavior of newly eclosed flies was observed by one of three methods as described in the Results. In all cases, flies were observed for at least 90 min after eclosion and wing expansion phenotypes were assessed at least 24 h after eclosion.
For observations in the low-perturbation condition, the vials used were either cultured at low density and cleared of any adults before observation of the first fly to eclose, or third instar larvae in the wandering stage were transferred individually to new culture vials to pupate and were observed after eclosion. Most observations were made by eye, although in a few cases flies were videotaped and behavior was scored from the video. The onset and offset of walking and abdominal flexion were scored to determine the three principal phases defined below.
For observations in the medium-perturbation condition, individual flies were transferred within 1–3 min of eclosion to a cylindrical chamber (1.4 cm diameter, 0.4 cm thick) formed by placing a plastic ring between two glass cover slides, which were held together by magnets glued to each slide. In some cases, an acetate sheet was attached to one of the glass slides to provide a better substrate for perching. Flies were videorecorded in this chamber using a Sony DCR-PC115 Digital Video Recorder, mounted on an Olympus SZX-12 stereomicroscope, with the video signal streamed to an external hard drive using a Firestore FS-1 (Focus Enhancements) for storage (see supplemental Fig. S1, available at www.jneurosci.org as supplemental material, for a full description of the setup). Videorecords for 40 animals were analyzed for sustained “state” behaviors (e.g., walking, grooming, abdominal contraction, proboscis extension, etc.) using Observer Video-Pro Version 5.0 software (Noldus Information Techology). Detailed ethograms were constructed for 10 flies, scoring individual movements of the legs, wings, abdomen, and proboscis. Cibarial pumping, which appeared as pulses of light reflected from the base of the proboscis could not be resolved in all flies depending on their orientation relative to the camera.
For the high-perturbation condition, flies were transferred after eclosion into glass tubes (0.3 cm diameter) plugged on each end to form a small chamber 0.7 cm in length. Flies confined within the tubes were videorecorded using a Sony HDR-FX7 digital videocamera, and records were scored for walking, grooming, abdominal flexion, abdominal contraction, and wing expansion.
The behavioral data in the text and in Figures 1 B and 5 A,B are presented as mean values ± SEM. For box-and-whisker plots, the boxes represent the interquartile range of the data separated by the median value, and the “whiskers” extend to the minimum and maximum values of the data, with the exception of outliers. Outliers are defined as values that lie more than three times the interquartile range above the upper quartile value (i.e., top of the upper box). Data were analyzed using nonparametric statistics, due to significant heteroskedacity based on the Levene test. Comparisons of more than two groups were performed using the Kruskal–Wallis test, followed by repeated post hoc Mann–Whitney U tests with Bonferroni corrections for multiple comparisons. Pairs of groups were compared with Mann–Whitney U tests. All statistical analysis was performed with SPSS version 13.0.
Definitions of behavioral phases
From eclosion (or placement in the videorecording chamber) to first perch. A perch is defined as the cessation of walking for more than 5 min. Flies sometimes commenced walking and reperched, but such episodes were typically brief compared with the initial walking bout.
From first perch to the start of sustained abdominal flexion, a condition defined by the elongation and downward flexion of the abdomen. Although short pulses of abdominal flexion could be observed in phase II, sustained abdominal flexion typically lasted 10–15 min.
From the onset to the cessation of abdominal flexion.
Quantification of air swallowing
Air swallowing was quantified by measuring the volume of air in the gut of fly. To do so, flies were briefly immersed in 100% ethanol to free the cuticle and bristles of any air, then pinned down through the head and anus in a Sylgard-treated culture dish filled with glycerol. The legs and wings were removed under a SZX-12 Olympus dissection microscope outfitted with a Nikon camera, and the gut was exposed by an incision extending up the midline of the fly. The air inside the gut was then released by gently tearing the gut membrane with forceps and, once liberated, formed a spherical bubble that rose slowly in the glycerol. This bubble was then photographed and its volume calculated from the value of the diameter measured using ImageJ software (W. S. Rasband, National Institutes of Health, Bethesda, MD, http://rsb.info.nih.gov/ij/) calibrated with a 2 mm micrometer. Typically, air volume was measured at the end of phase III or, in cases where this phase was absent, 90 min after the eclosion time.
Testing UAS-TRPM8 and electrophysiology
The efficacy of the TRPM8 channel was tested using a pan-neural driver (elavc155-Gal4;+;+) to express one, two, or three copies of UAS-TRPM8. Up to 10 flies were placed in a large chamber (either a 35 mm culture dish or for the menthol experiments an empty food vial) and then either subjected to a temperature shift or exposed to saturating menthol vapor. The number of flies no longer standing (i.e., “flies down”) was scored every 2.5 min, typically from videorecords or in some cases by eye. Electrophysiological recordings from adult flight muscle were made as follows: Flies were anesthetized with CO2 just long enough to remove the wings and place them onto a temperature-controlled stage, where they were held in place with vacuum. To record UAS-TRPM8 activated motoneuron activity, electrolytically sharpened tungsten recording electrodes were placed in dorsal longitudinal muscles (DLM) and reference electrodes in the abdomen. Signals were amplified with Dagan IX2–700 (Dagan Corporation) or A-M Systems 1700 (A-M Systems) differential amplifiers. The temperature-controlled stage consisted of a hollow brass disc through which water of the appropriate temperature was circulated. Temperature changes of >10°C/min could be achieved by switching between two water baths maintained at target temperatures. Temperature was monitored with a thermocouple probe (T-type, Physitemp Instruments) placed adjacent to the fly, with the signal transduced by a temperature controller (CNI-3252-DC, Omega Engineering). Electrophysiological and temperature data were recorded using pClamp v. 8.2 (Molecular Devices).
Animals were exposed to saturating menthol vapors in plastic vials (94 mm long × 27 mm in diameter) as follows: flies were introduced into a vial capped by a foam plug in which menthol crystals were embedded. To test the effects of menthol on animals expressing UAS-TRPM8 pan-neuronally, 1-d-old adults were transiently immobilized by CO2 and permitted 1 h recovery in an empty vial before menthol exposure. Animals were observed and scored every two and a half minutes for 25 min. To test the effects of menthol on wing expansion time, animals were collected within 5 min of eclosion and immediately transferred into a menthol-saturated vial for 15 min. After 15 min, the plug was gently replaced with one lacking menthol crystals. Individual flies were scored for wing expansion phenotypes at 1 min intervals.
Hemolymph collection and immunoblotting
For experiments on wild-type flies in the low-perturbation condition, hemolymph was extracted (as described below) from flies transferred into individual culture vials directly after eclosion and continuously monitored until they had spent 5 min in phases I, II, or III, or were at least 15 min past completion of phase III. For experiments in which flies expressing UAS-TRPM8 in NCCAP had been subjected to a temperature shift in the high-perturbation condition, hemolymph was taken 1 h after the onset of the temperature shift (i.e., 45 min after return of the flies to 25°C). Hemolymph was extracted in both cases by briefly anesthetizing flies with CO2, piercing the thorax with a needle and centrifuging as described previously (Luan et al., 2006). Collected samples were mixed with HE buffer (100 mm KCL, 20 mm HEPES-pH 7.5, 5% glycerol, 0.5M EDTA, 0.1% Triton X-100) containing 2× Halt protease inhibitor (Pierce) and then frozen on dry ice. They were then thawed in equal portions of Laemmli sample loading buffer containing 5% β-mercaptoethanol, and boiled for 5 min before electrophoresis on a 12% Tris-HCl gel (Bio-Rad). Gels were transferred to 0.2 μm nitrocellulose membranes using a Tris-Glycine-20%MeOH buffer, and immunoblotted. The primary rabbit anti-bursicon α-subunit antibody (Luan et al., 2006) was used at a dilution of 1:5000 and a goat anti-rabbit secondary antibody (10 μg/ml, Pierce) at 1:2000. Blots were incubated with West Femto chemiluminescent substrate (Pierce) for 5 min before development on BioMax film (Eastman Kodak) for 10 min.
Immunohistochemistry and confocal analysis
For analysis of whole-mount nervous system preparations, freshly emerged adults (i.e., within 15 min of eclosion) or stage P15i pharate adults (Bainbridge and Bownes, 1981) were dissected in PBS, and the excised nervous systems were fixed in 4% paraformaldehyde (Electron Microscopy Sciences) in PBS for ∼20 min, followed by postfixation in 4% paraformaldehyde/PBS plus 0.5% Triton X-100 (Sigma) for 15 min. Procedures for immunostaining were as previously described (Luan et al., 2006). Rabbit anti-TRPM8 antibodies (Novus Biologicals) were used at 1:200 dilution. Secondary antibodies (AlexaFluor 594 goat anti-rabbit from Invitrogen) were used at 1:500 dilution. Confocal imaging was performed using a Nikon C-1 confocal microscope. Z-series through either the brain or ventral nerve cord of each sample were acquired in 1 μm increments using a 20× objective unless otherwise noted, using 488 and 543 nm laser emission lines for EGFP and fluorophore excitation, respectively. Unless otherwise noted, the images shown are maximal projections of the volume rendered Z-stacks of confocal sections taken through the entire nervous system.
The consensus pattern of CCAP-Gal4>UAS-TRPM8 expression was determined by analyzing the intensity and frequency of labeling of identified CCAP-expressing neurons in multiple wholemount preparations. CCAP-expressing neurons in each preparation were identified in confocal sections by expression of UAS-EGFP and the intensity of TRPM8-associated fluorescence was scored for each neuronal soma on a scale of 0–3. The consensus intensity value (I) for a given neuron was calculated by averaging all nonzero values for this neuron across preparations. The frequency (v) with which a given neuron was labeled was calculated by dividing the number of preparations in which that neuron had a nonzero labeling intensity by the total number of preparations.
Environmental perturbation delays wing expansion
The behavioral pattern required for wing expansion in Drosophila has been described previously as consisting of two principal phases: perch selection, in which the fly walks to identify a site for wing expansion, and expansion itself, during which the stationary fly swallows air and simultaneously contracts its abdomen to drive hemolymph (i.e., blood) into the wings (Baker and Truman, 2002). These behavioral phases are executed immediately after emergence from the pupal case (a process called eclosion) and also have been extensively characterized in blowflies (Zdarek et al., 1984). Genetic data demonstrate that the expansion phase is absent in Drosophila bearing null mutations in the rickets gene, which encodes the receptor for the hormone bursicon (Baker and Truman, 2002; Luo et al., 2005). However, apart from the observation of Kimura and Truman (1990) that fruit flies will delay wing expansion if forced to eclose into an empty puparium from which they cannot escape, there has been little investigation of how the behavioral programs underlying perch selection and wing expansion respond to environmental stimuli.
To analyze the behavioral consequences of different environmental conditions, we therefore investigated perch selection and wing expansion in wild-type flies in three different paradigms. In the first, which involved minimal perturbation, the flies were kept in the culture vial in which they had pupated (volume, ∼30 ml) and were observed by eye after eclosion. This condition required intermittently approaching the vial with a magnifying lens or gently manipulating the vial to optimize viewing as a fly moved. In the second paradigm, flies were gently transferred into a thin cylindrical chamber (volume, 0.616 ml) immediately after eclosion. This configuration (supplemental Fig. S1, available at www.jneurosci.org as supplemental material) was optimized for videomicroscopy to obtain comprehensive behavioral records, as described in Materials and Methods. Flies in the videochamber were exposed to a constant visual environment. In the third paradigm, flies were tapped into an even smaller cylindrical minichamber with cotton endplugs (volume, 4.9 × 10−2 ml) and their behavior was videorecorded. In this paradigm, multiple chambers were monitored simultaneously and flies in adjacent chambers were exposed to each others' movements. The three paradigms thus differed in the degree of handling experienced by the fly after eclosion, the size of—and type of environment within—the chamber, and the visual environment to which the fly was exposed. We did not examine the contributions of each of these variables individually to the behavior of the fly, but each is likely to exert some effect on posteclosion behavior (J. Pohl, N. Peabody, and B. White, unpublished observations). In the following, we refer to the different paradigms as the “low-perturbation,” “medium-perturbation,” and “high-perturbation” conditions to reflect the increasing degrees of stimulation and confinement experienced by the flies after emergence.
We found that the time required for flies to complete wing expansion varied considerably in the three different conditions, increasing progressively with the level of perturbation. In the low-perturbation condition, flies typically expanded their wings in approximately half an hour (32.8 ± 12.8 min, n = 20), whereas in the medium- and high-perturbation conditions they took nearly three and four times this long (83.8 ± 30.7 min, n = 40 and 119.3 ± 42.7 min, n = 44, respectively). Our preliminary analysis indicated that this difference was largely due to an extension of the perch selection phase, as flies in the smaller chambers persisted in walking for extended periods before finally engaging in wing expansion. This difference was most pronounced for flies in the high-perturbation condition, which sometimes walked for several hours. In addition, these flies frequently executed a distinct “digging” program (Fraenkel, 1935; Reid et al., 1987) when they encountered the cotton plugs at the ends of the chambers. They would attempt to push through the plug by expanding their ptilinum into the gap between the chamber wall and the plug while trying to propel themselves forward by abdominal, and sometimes leg, movements. Animals eclosing in the culture vials never displayed this behavior, and it was observed in the videochamber only on rare occasions when a small gap was accidentally left between the hard plastic ring and the glass plates that formed the chamber.
Definition of posteclosion behavioral phases
To quantify the differences between the behavioral patterns observed in the three paradigms, we required a baseline description of the post-emergence behavioral program and behavioral endpoints that could be used to define the phases. We therefore analyzed in detail the records of 10 flies observed in the videochamber where movements could be observed at the highest resolution. The pattern of behavior for all the flies was quite stereotyped, as previously observed by Baker and Truman (2002), and followed the progression shown in the ethogram of Figure 1 A. Emerging flies, which have compact, folded wings (Fig. 1 Ai) first walked extensively and then perched. Sustained walking was rarely resumed after perching, although flies occasionally pivoted or walked short distances. While perched (Fig. 1 Aii), the flies groomed intermittently and occasionally extended their abdomens or probosces. Eventually, the extension of both the proboscis and the abdomen became sustained, with the extended abdomen contracted radially and flexed downward (Fig. 1 Aiiia). During this interval, the wings expanded, initially adopting a downwardly cupped shape, but gradually flattening while angled slightly away from the body axis. As the flattened wings were brought together over the body, the abdomen ceased to be flexed, although the proboscis often remained extended for some time. Grooming continued throughout the period of wing expansion, and persisted for some time afterward.
The overall behavioral pattern divided naturally into three principal phases, which we defined with respect to easily observable behavioral endpoints. As shown in Figure 1 B, each phase is typically distinguished by a primary category of behavior. The perch selection phase, which we refer to as phase I, is dominated largely by walking and ends when the animal stops and then remains sedentary for at least 5 min. The expansion phase, which we refer to as phase III, is defined by the onset and offset of tonic abdominal contraction. The proboscis also is extended during this phase and in favorably oriented flies (n = 7/10), the cibarial pump, which drives air swallowing, was seen to be active (Fig. 1 B). In the low-perturbation condition, flies dissected immediately after this phase had large bubbles in the abdomen (Fig. 1 Aiiib; Table 1), confirming the ingestion of air. Although expansion of the wings typically occurs during phase III, this event is thought to be mediated solely by movement of the blood into the wings rather than by exertions of wing-attached muscles. As described below, wing expansion can under some circumstances fail even when phase III and its associated motor patterns (i.e., abdominal contraction and air swallowing) appear normal. Interposed between phases I and III is a period dominated by grooming (Fig. 1 A,B). We call this period, which also has been noted in blowflies (Cottrell, 1962), phase II. Twenty animals from the low-perturbation condition dissected during this behavioral phase were devoid of air in the gut, indicating that air ingestion is largely or completely restricted to phase III.
We also examined bursicon release as a function of behavioral phase for flies eclosing in their culture vials (i.e., the low-perturbation condition), where we could most readily collect sufficient numbers of similarly staged animals. Hemolymph samples collected from flies during phases I, II, and III, and ∼15 min after the end of phase III were analyzed by Western blot (Fig. 1 C). Bursicon was not detectable in the hemolymph of animals during phase I. It first appeared in phase II, and rose to its highest level during the expansion phase (phase III). Thus, in the low-perturbation condition, bursicon release appears to be initiated within 5 min of perch selection and to peak during wing expansion.
Environmental perturbation extends the perch selection phase
Having defined the behavioral phases, we compared their durations under the three environmental conditions (Fig. 2). This analysis confirmed that the differences in wing expansion time resulted primarily from a progressive increase in the duration of the perch selection phase. Mean phase I durations differed significantly between the three conditions (p < 0.001 by the Kruskal–Wallis test), rising 20-fold between the low- and high-perturbation conditions. In contrast, the mean duration of the expansion phase (III) was ∼15 min in all cases and did not differ significantly between perturbation conditions (p > 0.7 by Kruskal–Wallis test). The stereotyped duration of phase III suggests that the motor patterns underlying expansion, once triggered, run for a fixed time. Bursicon is the likely trigger, and, as noted above, its release begins during phase II. Although phase II durations in the three different conditions were significantly different (p ≤ 0.007 in pairwise comparisons using the Mann–Whitney test), they showed no obvious trend with respect to the level of perturbation. The observed variability in phase II durations, both within any given condition and across conditions, suggests that the timing or rate of bursicon release may differ considerably among individual animals.
The perch selection phase is insensitive to NCCAP suppression
The above data demonstrate that environmental perturbations delay wing expansion primarily by extending the perch selection phase. The perch selection program is thus modulated by the sensory pathways that mediate environmental perturbation, but it is possible that this modulation is indirect and requires the circuitry underlying the expansion program. For example, the expansion program may be the primary target of environmental modulation, and the perch selection program could simply be active until inhibited by initiation of the expansion program. If the expansion program never starts up, perch selection should continue indefinitely. To directly test the possible involvement of the expansion program circuitry, we examined the effects of inhibiting activity in NCCAP in the relative absence of environmental stimuli, i.e., in the low-perturbation condition.
We previously showed (Luan et al., 2006) that graded levels of NCCAP suppression can be achieved by using the CCAP-Gal4 driver to express increasing copy numbers of the UAS-EKO suppressor transgene. Expression of three copies of the UAS-EKO transgene (i.e., 3× UAS-EKO) in NCCAP completely blocks wing expansion and inhibits bursicon release into the hemolymph (Luan et al., 2006). As expected, we found that in the low-perturbation condition incremental NCCAP suppression causes wing expansion deficits of increasing severity and frequency (Fig. 3 A, top panel). Consistent with the known requirement for bursicon in expansion behaviors (Baker and Truman, 2002), suppression of NCCAP by 3× UAS-EKO completely eliminated both the tonic abdominal contraction that defines phase III (Fig. 3 A, bottom panel) and air swallowing (Table 1; see also supplemental Movie S2, available at www.jneurosci.org as supplemental material). Despite the absence of the expansion phase, however, the duration of phase I in these animals was indistinguishable from that of the control flies (Fig. 3 B). Indeed, there was no significant difference in the length of phase I for any of the experimental groups (p > 0.2 by Kruskal–Wallis test), and almost all the flies completed perch selection in <10 min regardless of the level of NCCAP inhibition. Our results demonstrate that NCCAP neurons are not involved in the termination of perch selection, and therefore in the transition to phase II, but that they are essential for the subsequent activation of the expansion program.
NCCAP activity regulates initiation of the expansion program
The behavioral effects of NCCAP suppression have implications not only for the regulation of the perch selection program, but also for the mechanisms governing the wing expansion-related motor patterns. In this regard, the results for animals expressing 2× UAS-EKO are of particular interest. These flies can be divided into two distinct categories: Approximately half resembled flies expressing 1× UAS-EKO, which exhibited a robust expansion phase that included tonic abdominal contraction (Fig. 3 A, bottom panel) and air swallowing (Table 1), while the rest resembled 3× UAS-EKO animals, which completely lack phase III behaviors (Fig. 3 A, bottom panel; Table 1). This observation implies that there is a threshold level of NCCAP activity which must be exceeded if the expansion phase is to be initiated. Because the two behaviors required for wing expansion either occur or fail in tandem, our results also imply that the two motor patterns share a common mechanism of initiation.
The common mechanism of initiation of the phase III motor patterns most likely reflects their shared requirement for bursicon (Baker and Truman, 2002), and their coincident failure at higher levels of NCCAP suppression presumably results from impaired hormone release. Impaired bursicon release is also likely to underlie the observation that animals expressing 1× UAS-EKO, as well as approximately one-half of animals expressing 2× UAS-EKO, initiate phase III only after a significant delay, as reflected in the extension of phase II (Fig. 3 B). Further work will be required to determine whether the extension of phase II, which appears to result from longer quiescent episodes between grooming bouts, is caused by the slowing of bursicon release or by its delay (or both), but the former possibility would conveniently explain the observed threshold in NCCAP activity for wing expansion. If EKO-mediated suppression attenuates bursicon release, the threshold for NCCAP activity would simply reflect a threshold requirement for bursicon, with expansion triggered only when bursicon reaches a critical level.
A final implication of the results shown in Figure 3 is that NCCAP modulates not just the expansion program, but also the nonbehavioral processes necessary for wing expansion. This is again most obvious for flies expressing 2× UAS-EKO. Approximately half of these flies executed phase III (Fig. 3 A, bottom panel), but none of these expanded their wings fully, and 70% did not expand them at all (Fig. 3 A, top panel). In addition, most flies expressing 1× UAS-EKO in NCCAP only partially expanded their wings, despite executing normal phase III behavior. While we cannot rule out subtle behavioral deficits in the animals that failed to expand their wings, our results suggest that NCCAP regulates nonbehavioral processes involved in wing expansion, such as plasticizing the wing cuticle to render it pliable. These nonbehavioral processes may well be mediated by bursicon released into the hemolymph, as the blood-borne hormone is known to be required for somatic changes at the level of the wing cuticle (Reynolds, 1976, 1977; Dai et al., 2008). In contrast to motor-pattern deficits resulting from NCCAP inhibition, which are all-or-none, wing expansion defects are often partial (Fig. 3 A, top panel) and change in a graded manner. The nonbehavioral deficits associated with NCCAP suppression thus appear to lack a threshold.
Targeted expression of rat TRPM8 can be used to acutely activate neurons
The suppression experiments described above indicate that perch selection normally terminates without input from NCCAP and that the bursicon released from NCCAP during phase II initiates the expansion program. Initiation of expansion thus requires activation (or disinhibition) of NCCAP, and these neurons are probable targets of modulation by circuits that mediate the perch selection program or environmental inputs. To determine whether NCCAP neurons are the sole targets of such modulation, or whether downstream neurons that mediate the execution of the expansion program are also subject to regulation, we sought to artificially activate NCCAP during perch selection. We reasoned that if neurons downstream of NCCAP are normally inhibited during perch selection, either by the circuitry underlying this program or by negative environmental signals, activation of these neurons would fail to initiate the expansion motor patterns. If, on the other hand, artificial activation of NCCAP readily induced expansion, this would provide strong evidence that neural substrates of the expansion program downstream of NCCAP were unlikely regulatory targets of the circuits underlying perch selection and/or environmental inputs. To determine which of these two possibilities might be the case, and more generally to determine whether NCCAP activity is sufficient to initiate expansion behavior, we sought to acutely stimulate these neurons in post-eclosion flies.
Efforts to use the light-sensitive Channelrhodopsin-2 protein (Nagel et al., 2003), which has been used successfully before in Drosophila for acute neuronal activation (Schroll et al., 2006; Suh et al., 2007), proved ineffective in newly emerged adults, perhaps due to poor light penetration through the cuticle or depletion of the retinal-A cofactor during metamorphosis. We therefore sought to develop an alternative, and simpler, system for stimulating neurons in the live animal. To this end, we generated transgenic flies that can express the rat TRPM8 gene, which encodes a nonselective cation channel expressed in mammalian sensory neurons (McKemy et al., 2002; Peier et al., 2002). This channel is responsible for the sensation of mild cold (Colburn et al., 2007; Dhaka et al., 2007) and activates in response to decrements in temperature in the range of ∼25°C to 10°C. The thermal sensitivity of the TRPM8 channel, which is compatible with the temperatures at which flies are normally active, makes it particularly suitable for manipulating activity in Drosophila. Because cold, unlike light, normally attenuates motor activity, neuronal activation using TRPM8 also has an advantage over optical techniques in minimizing the confounding behavioral effects of the activating stimulus.
We made transgenic fly lines that express the TRPM8 channel under the control of Gal4, and tested their efficacy by assaying for abnormal behavior when the channel was expressed pan-neuronally. Several lines expressed the TRPM8 channel at sufficiently high levels to give robust phenotypes (such as seizing, jumping, and eventually falling down) upon transfer to 15°C from room temperature. We combined the strongest inserts to make lines that bore from one to three copies of the UAS-TRPM8 transgene. As shown in Figure 4 A, flies expressing UAS-TRPM8 pan-neuronally fell down quickly upon transfer to 15°C, with animals expressing 3× UAS-TRPM8 displaying a more rapid and robust response. Electrophysiological recordings from flight muscles of these animals revealed sustained action potential generation in motoneurons in response a downward temperature ramp (Fig. 4 B). This effect was readily reversible. Males typically showed greater sensitivity than females to a given temperature shift, as can be seen from the responses of animals expressing 2× UAS-TRPM8 shifted to 18°C (Fig. 4 C). At this temperature, 60% of males eventually fell over and could not right themselves, but none of the females did so, even after 10 min. We also tested the sensitivity of flies expressing UAS-TRPM8 to menthol. Menthol potentiates the activity of TRPM8 by shifting its temperature sensitivity to more positive temperatures (McKemy et al., 2002). Accordingly, saturating menthol vapor administered to flies expressing UAS-TRPM8 pan-neuronally at room temperature (24°C) caused the animals to seize and fall over as they did in response to cold (Fig. 4 D). The response was slow and poorly penetrant with a single copy of the transgene, but most animals expressing 2× UAS-TRPM8 fell down within 5 min of menthol exposure and all were down by 10 min.
Activation of NCCAP using TRPM8 is sufficient to initiate wing expansion
To activate NCCAP with stimuli of various magnitudes, we subjected flies expressing 1× and 2× UAS-TRPM8 in NCCAP (see Fig. S3 for anti-TRPM8 immunostaining) to temperatures shifts of various magnitudes. Under conditions similar to our low-perturbation condition, newly eclosed flies were transferred into food vials at room temperature and then placed at temperatures ranging from 18°C to 12°C. Even the mildest stimulus tested gave robust results: CCAP-Gal4 > 1× UAS-TRPM8 animals shifted briefly to 18°C immediately after eclosion rapidly expanded their wings relative to controls bearing the 1× UAS-TRPM8 transgene without the CCAP-Gal4 driver (Fig. 5 A, left). Both the speed with which the experimental group expanded their wings and the low variance in expansion times implied strong and uniform stimulation of NCCAP in these animals, effects that were also observed when wing expansion was induced with a 15 min exposure to saturating menthol vapor at room temperature (Fig. 5 A, right).
To determine whether strong induction of the perch selection program attenuated the ability of UAS-TRPM8 to drive wing expansion, we examined the response of flies in the high-perturbation condition. Newly eclosed animals were placed in minichambers and exposed to a 15 min temperature shift from 24°C to 18°C. We found that NCCAP activation by UAS-TRPM8 resulted in a substantial acceleration of wing expansion in all animals (supplemental Movie S4, available at www.jneurosci.org as supplemental material). On average, these animals took 25 ± 3 min (n = 12) to expand their wings compared with 219 ± 25 min (n = 10) for control flies bearing the 1× UAS-TRPM8 transgene, but lacking the CCAP-Gal4 driver (Fig. 5 B). All behavioral phases were typically present in the experimental animals, although phase II was extremely brief. Most animals persisted in walking and digging through most of the 15 min temperature pulse, but then transitioned almost immediately into the expansion phase after perching. Consistent with our observations under other conditions, phase III was relatively uniform in duration and lasted on average ∼15 min. All animals remained sedentary during wing expansion, indicating that execution of the expansion program potently inhibits perch selection behaviors. In addition to performing the expansion motor programs, animals expressing 1× UAS-TRPM8 in the high-perturbation condition displayed normal cuticle tanning after NCCAP activation (Fig. 5 C, compare left, right). NCCAP activation thus appears to induce bursicon release into the hemolymph. We confirmed this directly by measuring hormone titers after the temperature shift by Western blot (Fig. 5 D).
Thus, under conditions that normally bias the animal to persist in perch selection behaviors, mild activation of NCCAP initiates hormone release and induces both the behavioral and physiological hallmarks of bursicon action. These results indicate that the neural pathways mediating perch selection and/or environmental input regulate the wing expansion program at, or above, the level of NCCAP, and that neurons downstream of NCCAP are not targets of modulation. In addition, our results demonstrate that NCCAP neurons are not only necessary for wing expansion but also sufficient to drive this entire process.
The introduction of genetically encoded effectors, which permit the targeted manipulation of neuronal activity in vivo, is increasingly facilitating neurobiological studies of behavioral choice in vivo, as exemplified by the work presented here. Using a new method for stimulating neurons in live animals, we demonstrate that the NCCAP network, first implicated in governing ecdysis, effects the fly's decision to expand its wings after eclosion. This decision is normally coupled to the decision to perch, which we demonstrate is based on evaluation of environmental variables. The postponement of wing expansion by the fly under adverse circumstances is thus a consequence of a value-based choice to prolong search behavior and the delayed activation of a command network responsible for the execution of the wing expansion decision.
Perch selection is environmentally regulated
As indicated in Figure 6, our results support a model in which the motor programs underlying perch selection respond primarily to environmental input. Although the environmental features monitored by the fly and the sensory channels that process them remain to be determined, it is clear from our experiments that flies assess conditions during phase I and assign longer search periods to more adverse environments. The observation that expansion is not deferred indefinitely, even under the high-perturbation condition, suggests that the benefits of continued searching are weighed against the risks of further delay. How these risks, which may include predation and desiccation, are represented physiologically to encode the “value” of different environments is still unclear, but the work described here should facilitate their investigation.
Elucidating which neurons mediate the decision to perch will also require further investigation. Although perching can be induced by the activation of NCCAP using UAS-TRPM8, NCCAP is clearly not normally required for terminating phase I, since its suppression does not affect perching. Also, the perching that does occur when NCCAP is stimulated by UAS-TRPM8 is more tightly linked temporally with the initiation of expansion (i.e., phase III) than with the onset of NCCAP stimulation (i.e., the shift to 18°C), suggesting that neurons downstream of NCCAP mediate phase I termination, most likely at the level of the motor networks underlying perch selection and expansion, as shown in Figure 6. Inhibition of one motor system by another has been demonstrated to explain such behavioral hierarchies in other cases, as in the dominance of feeding over withdrawal in mollusks (Davis, 1979).
NCCAP regulates the expansion program
In contrast to perch selection, the expansion program shows no obvious environmental dependence, but it is strongly dependent on levels of NCCAP activity for its initiation. This is consistent with the known effects of bursicon, which is released by a subset of NCCAP (Luan et al., 2006) and is required for both wing extensibility (Reynolds, 1976, 1977; Dai et al., 2008) and phase III behaviors (Baker and Truman, 2002). The probability of expansion occurring at all is likely to depend on whether the bursicon released into the CNS reaches a critical threshold, while the initiation of expansion most likely depends on the timing and/or rate of bursicon release. The rate of release should decline with NCCAP suppression, which will reduce the excitability of the bursicon-expressing neurons, and may account for both the prolongation of phase II under this condition and the appearance of graded wing expansion deficits. It is also possible that the timing of bursicon release was modulated by our manipulations of NCCAP if delay circuits intrinsic to NCCAP are responsible for initiating its secretion (see below).
Although it remains to be determined whether NCCAP participates in the decision to expand, it is clear that its activation drives all the behavioral and somatic processes necessary for wing expansion (see Fig. 6), placing NCCAP neurons high in the execution pathway of the decision and indicating that their function in wing expansion is command-like. Because this command-like function relies on bursicon and not direct synaptic activation, it differs from familiar command systems that mediate fast behavioral switches, such as those involved in defensive escape (Edwards et al., 1999), and more closely resembles systems in which hormones or neuromodulators elicit profound behavioral transitions such as those responsible for egg-laying in mollusks (Kupferman, 1967), stomatogastric ganglion modulation in crustacea (Hooper and Marder, 1984), and ecdysis in insects (Truman and Riddiford, 1970). Because of the technical difficulty involved, the ability of such systems to drive behavioral programs upon activation has rarely been demonstrated. However, as illustrated by the work presented here, the introduction of genetic techniques for in vivo neuronal activation should make such demonstrations increasingly possible.
Coupling perch selection and expansion: insights from ecdysis
The work presented here argues that the motor programs for perch selection and wing expansion have distinct regulatory mechanisms, but leaves unanswered the question of how the decisions to perch and expand are coupled: Specifically, the mechanism by which NCCAP is activated following perching remains to be clarified. As indicated in Figure 6, NCCAP could, in principle, receive input from either the perch selection motor network or from sensory processing pathways. Such input is likely to be indirect since perching and expansion are separated by a delay (i.e., phase II), which varies considerably in duration between individual animals and different conditions (see Figs. 2, 3). This delay period, during which bursicon secretion is initiated, may act, at least in part, to permit bursicon to plasticize the cuticle of the wing and body before expansion.
Behavioral transitions during ecdysis in both Drosophila and the hawkmoth Manduca sexta, are known to be regulated by inhibitory delay circuits (Zitnan and Adams, 2000; Fuse and Truman, 2002). In the case of eclosion hormone, the delay circuit temporally separates the somatic and behavioral aspects of the hormone's action and has been proposed to act as a control point for environmental modulation by light, which accelerates eclosion, perhaps by suppressing the inhibitory delay circuit (Baker et al., 1999; Truman, 2005). Interestingly, the modulation of eclosion by light is eliminated in animals lacking NCCAP, supporting a role for these neurons in the inhibitory pathway (Park et al., 2003). Because ablation of the EH-expressing neurons causes wing expansion deficits, it is possible that in addition to modulating eclosion, EH may similarly modulate wing expansion by simultaneously upregulating excitability in both the bursicon-expressing neurons and in a delay circuit that inhibits them (McNabb et al., 1997; McNabb and Truman, 2008). By analogy to the action of EH on ecdysis, the latter circuit, which could include non-bursicon-expressing NCCAP neurons, would inhibit bursicon release until inhibition is alleviated by an environmentally mediated signal. Such a model might explain not only the wing expansion deficits observed in animals lacking EH-expressing neurons, but also the eventual wing expansion of animals kept in the high-perturbation condition. In these animals, run down in the delay circuit may trigger bursicon release before the decision to perch, as happens when NCCAP is activated by UAS-TRPM8, and thus also account for their abbreviated phase II (see Fig. 2).
NCCAP function and behavioral adaptation
The merits of this model, and others that invoke delay circuits, remain to be tested. It is interesting to note, however, that the mechanism(s) coupling environmental input to NCCAP is likely to be added during metamorphosis, because the function of this network changes from an intrinsically regulated mediator of pupal ecdysis to an extrinsically modulated mediator of wing expansion (Park et al., 2003; Kim et al., 2006; Luan et al., 2006). Despite its broad phylogenetic conservation, this network also appears to be differentially regulated in different species, with CCAP-expressing neurons being direct targets of EH in many insects, but not Drosophila (Ewer and Truman, 1996). Such differential regulation may underlie, at least in part, species-specific differences in the coordination of wing expansion with adult emergence: In some insects, expansion is initiated at emergence without benefit of environmental input, and in others, like honeybees, it is initiated and completed before emergence. Investigation of the basis of these differences, and of the neural mechanisms that adapt NCCAP function to the individual developmental needs and life histories of different insects, should provide insight into general mechanisms of behavioral adaptation (Katz and Harris-Warrick, 1999).
In general, the work presented here demonstrates that the behavioral programs used by Drosophila to achieve wing expansion can serve as a simple and fruitful model for investigating networks underlying behavioral choice. Elucidation of the neuronal pathways that mediate the decisions to perch and expand the wings will shed light on the development and evolution of behavioral networks as well as their architecture and function. Indeed, the decision-making architecture outlined here for wing expansion may prove relevant to the type of instinctive behaviors, first described by Craig (1918), in which an environmentally sensitive “appetitive” phase is coupled to a “consummatory” phase consisting of a stereotyped, motor pattern. Finally, tools that permit one to manipulate decision-making in living, behaving animals, such as UAS-TRPM8, which we introduce here, clearly will play an essential role in mapping the neural networks underlying behavior in Drosophila and other animals.
This work was supported by the Intramural Research Program of the National Institute of Mental Health. We thank David Julius for providing the TRPM8 cDNA and John Ewer, Paul Taghert, and the Bloomington stock center for fly stocks. We appreciate early help with behavioral assays from Hailyn Nielsen and William Lemon. Special thanks to Grace Gray for editorial assistance and to Howard Nash for insightful comments and persistent insistence on clarity of thought.
- Correspondence should be addressed to Benjamin H. White, National Institute of Mental Health, National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20892.