Starved flies were put on wet Kimwipes for 24 hr prior to experimentation. For the temporal consumption assay, flies were starved for 24 hr on wet Kimwipes and then mounted on glass slides using nail polish. After 2 hr of recovery in a humidified chamber, the time spent consuming 1 M sucrose was measured for each fly. Flies were considered nonresponsive if they failed to consume sucrose upon ten consecutive stimulations. For channelrhodopsin-2 experiments, flies were

prepared as previously described (Gordon and Scott, 2009), except that flies were not starved prior to experimentation. Flies were prepared such that all six tarsi remained intact, and the stimulating laser was positioned underneath the fly such that the tarsi and ventral side of the thorax could be simultaneously stimulated. For stimulation, 10 ms light Palbociclib pulses were applied at 30 Hz for a total of 3 s using a 50 mW 473 nm diode pumped solid state laser (Shanghai Dream Lasers). Genetic mosaics Fluorouracil were generated as previously described

(Gordon and Scott, 2009), except that flies were of the genotype tub > Gal80 > ; E564-Gal4,UAS-mCD8::GFP/UAS-Kir2.1; MKRS, hs-FLP. Flies were heat-shocked at 37.5°C for 55 min during late larval to pupal stages. Antibody staining and imaging was carried out as previously described (Wang et al., 2004). The following antibodies were used: rabbit anti-GFP (Invitrogen, 1:1,000), mouse anti-GFP (Invitrogen, 1:1,000), mouse anti-nc82 (Hybridoma bank, 1:500), and rabbit anti-dsRed (Biovision, 1:1,000). Brightness or contrast of single channels was adjusted for the entire image using ImageJ

software. Experiments were performed as previously described (Marella et al., 2012), except that flies were immobilized ventral side up, with cover glass separating the Oxymatrine front tarsi and head of the fly from the recording chamber. E564 neurons were labeled with GFP and PERin neurons identified for recordings based on their fluorescence and anatomical position. For taste stimulations, tastants were delivered to the ipsilateral tarsus using a glass capillary. A stimulus artifact in the recording indicated when stimulation occurred. Data were band-passed filtered between 10 and 300 Hz using a Butterworth-type filter. Prestimulus spike rates were calculated using 15 s of recording prior to stimulation; stimulus spike rates were calculated using 1 s of recording after stimulation. Whole nervous systems (brain and ventral nerve cord) were carefully dissected in cold adult hemolymph-like solution (AHL) lacking calcium and magnesium, then transferred to a room temperature dish with AHL containing calcium and magnesium and gently pinned with the dorsal surface facing up (Wang et al., 2003). Nerves were then individually inserted into a stimulating suction electrode (∼100 kΩ). Stimulus was 10 V, 300 μs delivered at 100 Hz for 100 ms (ten stimulations). G-CaMP3 responses were monitored as previously described (Marella et al.