Labellar slices. Young adult flies (1 to 4 days after eclosion) were immobilized on ice. The labella were isolated, and each labellar lobe was cut into transverse slices in Drosophila saline. To record bristle MSNs, the cut was made parallel and close to the inner labellar surface; to record peg MSNs, the cut was also made parallel to the inner labellar surface, and the inner labellar part was kept intact. The labellar slice was stabilized in the recording chamber and continuously perfused with 95% O2/5% CO2 (v/v)–bubbled Drosophila saline. The saline contained the following: 178 mM NaCl, 3 mM KCl, 4 mM MgCl2, 1.5 mM CaCl2, 26 mM NaHCO3, 1 mM NaH2PO4, 5 mM N-tris(hydroxymethyl) methyl-2-aminoethanesulfonic acid (TES), and 5 mM trehalose, bubbled with 95% O2/5% CO2 (pH 7.4). The dissection solution was made by replacing NaHCO3, NaH2PO4, and TES in Drosophila saline with 5 mM Hepes and 27 mM NaCl (pH 7.4, adjusted with NaOH), bubbled with oxygen. All chemicals were obtained from Sigma-Aldrich and were freshly dissolved in Drosophila saline daily.

Neurons were visualized on an upright microscope (Scientifica) with infrared (IR)–differential interference contrast (Olympus). The image was captured with an IR charge coupled device (DAGE-MTI) and displayed on a television monitor (Sony).

Patch-clamp recordings were made with MultiClamp 700B (Molecular Devices). The patch electrodes were made from borosilicate glass (WPI) with a P-1000 or P-97 puller (Sutter). The cell bodies of sensory neurons in labellar slices were small (diameter, 3 to 4 μm), thus requiring a recording pipette tip of ~ 0.2 μm and a resistance of ~15 to 20 megohms filled with intracellular saline [185 mM K-gluconate, 5 mM NaCl, 2 mM MgCl2, 0.1 mM CaCl2, 1 mM EGTA, and 10 mM Hepes (pH 7.4); ~390 mOsm]. For perforated patch-clamp recordings, amphotericin B was dissolved in dimethyl sulfoxide, diluted with intracellular saline to a final concentration of 200 μg/ml, and backfilled into the recording pipette. For whole-cell patch-clamp recordings, guanosine 5′-triphosphate (GTP)–tris (0.5 mM) and Mg–adenosine 5′-triphosphate (ATP) (4 mM) were added to the intracellular saline. For suction-pipette recordings, a recording pipette with a tip diameter of ~ 2 μm and a resistance of ~2 megohms filled with dissection solution was used. Typically, a loose seal (~15 megohms) between the recording pipette and cell body was obtained. To access the cell bodies of MSNs, a suction-recording pipette filled with protease XIV (2 mg/ml; Sigma) was used to locally digest the sheath cells that wrap a neural cluster of either a peg or bristle.

To measure the I-V relationship, voltage-sensitive Na+ channels and K+ channels were blocked by a mixture of TTX (50 nM), tetraethylammonium chloride (10 mM), and sometimes also 4-AP (10 mM). Current and voltage signals were digitized and recorded with Digidata 1440A and pClamp 10.2 (Molecular Devices), filtered at 2 kHz, and sampled at 5 kHz. Recorded currents were low-pass–filtered at 200 Hz (unless stated otherwise) for display. The voltage was clamped at −80 mV unless stated otherwise. Measured voltages were corrected for a liquid junction potential.

Labella-brain preparation. Young adult flies (1 to 4 days after eclosion) were immobilized on ice. The head was dissected and transferred to a recording chamber. The antenna, compound eyes, and brain cuticle were removed by fine forceps. The labella-brain preparation was then stabilized in the chamber with the anterior side up, continuously perfused with a saline solution bubbled with 95% O2/5% CO2 (pH 7.4) at room temperature. The saline was composed of the following: 103 mM NaCl, 3 mM KCl, 4 mM MgCl2, 1.5 mM CaCl2, 26 mM NaHCO3, 1 mM NaH2PO4, 5 mM TES, 20 mM d-glucose, 17 mM sucrose, and 5 mM trehalose. For whole-cell patch-clamp recordings, the recording pipette was filled with internal solution consisting of the following: 140 mM K-gluconate, 6 mM NaCl, 2 mM MgCl2, 0.1 mM CaCl2, 1 mM EGTA, and 10 mM Hepes (pH 7.2), with an osmolarity of 270 mOsm. For perforated patch-clamp recordings, the pipette was backfilled with the internal solution that contains amphotericin B (200 μg/ml) and then filled with regular internal solution. For cell-attached recordings, the pipette was filled with the saline solution with NaHCO3 replaced by 10 mM Hepes (pH 7.2).

Mechanical and chemical stimulation. Sensillar deflection was achieved by pushing the sensillar bristle with a glass pipette, which was similar to the suction-recording pipette and has a “7-shaped” tip. The pipette was attached to a piezo actuator (Physik Instrumente), which was mounted on a micromanipulator (Scientifica). Mechanical deflection was quantified by the movement of the glass pipette. Under the 60× water lens, the pipette tip was positioned against the mid-point of a targeted sensillar bristle. The piezo actuator was controlled by voltage signals from the analog output of Digidata 1440 (Molecular Devices).

For chemical stimulation of the sensory neurons in the labellar slice, rapid solution changes were used. The rapid solution change was produced by translating the interface between the two following streams across the recorded labellar sensory neurons with an electronic stepper (Warner Instruments). Different solutions ran through a three-barrel tube (Warner Instruments), whose tips were positioned ~100 μm away from the labellar slice. The solution flow was driven by gravity and was controlled by solenoid valves (The Lee Company) and a valve controller (AutoMates Scientific). The inner width of each square barrel of the perfusion tubing was ~600 μm, emitting a solution readily covering the labellar slice.

Optogenetic stimulation in electrophysiological recordings. Flies expressing CsChrimson were raised on standard food. Labellar slices or labella-brain preparations were first incubated in the Hepes-buffered saline with 100 μM all trans-Retinal (Sigma-Aldrich) for 20 to 25 min and then washed and perfused with regular saline bubbled with 95% O2/5% CO2 (pH 7.4). The labella were stimulated with a red light-emitting diode (617 nm; M617F1, Thorlabs) of 1.75 mW/cm2 through an optic fiber (inner diameter, 200 μm) that was positioned approximately 50 μm away. Light intensity was measured by a power meter (Model 1936-R, 918D-ST-UV, Newport).

Single-cell labeling by neurobiotin. Neurobiotin (Vector Lab) was dissolved in a modified internal solution [70 mM K-gluconate, 6 mM NaCl, 2 mM MgCl2, 0.1 mM CaCl2, 1 mM EGTA, 4 mM Mg-ATP, 0.5 mM GTP-tris, and 10 mM Hepes (pH 7.4)] for osmolarity balance of the 2% neurobiotin. The recording pipette was filled with neurobiotin-containing internal solution. After breaking into whole-cell mode, depolarizing currents (200 ms, 2 Hz) were injected for 20 min, which facilitated diffusion of positively charged neurobiotin into the recorded neuron. The recording pipette was gently detached from the cell after another 20-min wait of neurobiotin diffusion within the cell. The labellar preparation was transferred into 4% paraformaldehyde, fixed for 4 hours on ice, and then washed by phosphate-buffered saline (PBS) for three times at a 20-min interval, blocked in 5% bovine serum albumin in PBST (1% Triton X-100 in PBS) for 2 hours at room temperature. The preparations were incubated with Streptavidin-568 (1:500; Invitrogen, catalog number S11226) overnight at 4°C, washed by PBST (1% Triton X-100 in PBS) for three times at an interval of 20 min, and then mounted in the glycerol. Images were acquired on a Nikon A1R+ confocal microscope with a 25× water immersion objective.

Synaptic labeling by GRASP and immunostaining. The fly brains were dissected in PBS, transferred to 4% paraformaldehyde (on ice) for 1 hour, and then washed by PBS and blocked in 5% normal goat serum (1% Triton in PBS) for 2 hours at 23°C. After incubation with mouse anti-nc82 [1:100; DSHB (Developmental Studies Hybridoma Bank) catalog number nc82] overnight at 4°C, the brains were washed three times with 0.5% PBST, followed by incubation of secondary antibodies, Alexa Fluor 568–congugated anti-mouse and Alexa Fluor 647–conjugated anti-mouse (1:200 each; Invitrogen, catalog numbers, A-11004 and A-32728, respectively), for 6 hours at room temperature. The brains were washed three times and then mounted in glycerol. Images were acquired in 0.5-μm sections on a Nikon A1R+ confocal microscope with a 25× water immersion objective.

Behavioral assays. Flies (2 to 3 days after eclosion) were food-deprived on a wet Kimwipe for 24 hours. The fly was anesthetized on ice and inserted into a 1-ml pipette tip with the fly head and proboscis exposed. To conduct proboscis extension reflex and labellar spreading experiments, a 1-ml syringe was used to apply a small drop of sucrose solution (500 mM) to the labella or forelegs. To conduct proboscis retraction experiments, a small drop of sucrose solution (500 mM) was used to stimulate the labella and engage the fly to feeding. The behaviors were recorded under a dissection microscope (M205, Leica) with a color camera (DFC450 C, Leica).

For optogenetic behavioral experiments, flies were raised on standard food mixed with 100 μM all trans-Retinal. Newly enclosed flies were collected and starved for 24 hours. The fly proboscis was stimulated with red light (617 nm; M617F1, Thorlabs). The light intensity is of 1.75 mW/cm2 unless stated otherwise.

For natural feeding experiments, a drop of food was placed in a small cavity of a sylgard-coated plate and covered by a coverslip, thus facilitating the video capture by limiting the range of fly walking. When food was omitted, this setup could be also used to provide physical touch of the coverslips for the extended proboscis induced by optogenetical activation of sweet-sensing GRNs. The behaviors were recorded with a high-speed camera (MV-A5031MU815, Dahua Technology).

Scanning electron microscopy. Samples were collected and fixed in 0.1 M phosphate buffer (pH 7.4) that contains freshly prepared 2.5% glutaraldehyde for 4 hours at room temperature. Samples were washed by 0.1 M phosphate buffer three times at a 10-min interval and then post-fixed in 1% OsO4 for 1 hour at room temperature. Subsequently, the samples were washed with 0.1 M phosphate buffer three times and then treated with increasing concentrations of acetone (30, 50, 70, 85, 90, 95, and 100%) for approximately 10 min in each solution, followed by three washes in 0.1 M phosphate buffer. Samples were treated with critical point drying, mounted on a scanning electron microscope (SEM) stub with a copper tap, and then sputter-coated with gold for 1.5 min. Images were collected with a scanning electron microscope (FEG QUANTA 450) at 20 kV.

Focused ion beam SEM. Samples were collected and fixed in 0.1 M phosphate buffer (pH 7.4), containing freshly prepared 4% (w/v) paraformaldehyde and 2.5% glutaraldehyde for 2 hours at room temperature and then overnight at 4°C. Specimens were washed and post-fixed in 1% OsO4 with 2% potassium ferrocyanide for 1 hour at room temperature. After rinsing several times in phosphate buffer, samples were dehydrated in a graded ethanol series and embedded in Spurr’s resin (SPI Supplies, PA, USA). Focused ion beam SEM imaging was performed with a Helios Nanolab G3 UC scanning electron microscope (Thermo Fisher Scientific), and the automated data collection was guided by the Auto Slice and View G3 1.7.2 software (Thermo Fisher Scientific). The samples were imaged in the backscattered electron mode with through-the-lens and in-column detectors. The ion beam milling was performed at 30 kV and 2.5 nA, and images were recorded with an electron beam at 2 kV and 0.2 nA and a working distance of 2 mm. The image resolution was 6144 × 4096 with a horizontal filed width of 16.8 μm, and the z-step size was 20 nm.

Quantification of the time of proboscis retraction. We used two independent analysis methods to quantify the time of proboscis retraction. One is to manually count the video frames one by one; the other is computer-assisted image processing. For manual analysis, the frames were counted from the frame with the start of optogenetic activation until the video frame with a full proboscis retraction. For computer-aided analysis, the image processing software was written in MATLAB. In the videos, the experimental fly appears bright in a black ground under IR illumination. The fly was stabilized in a pipette tube, which only allows the proboscis to move. Thus, the change of brightness area in the video frames reflects the movement of proboscis. The maximal reduction in the brightness area indicates a final proboscis retraction. The time of proboscis retraction is calculated as the total frames from optogenetic activation to a full proboscis retraction time the duration a single frame. These two methods yield similar results.

Statistics. Data are presented as mean values accompanied by SEM. Statistical parameters including the exact value of n, precision measures (means ± SEM), and statistical significance were reported in the figure legends.