Olfactory neuromodulation of motion vision circuitry in Drosophila

Abstract

It is well established that perception is largely multisensory; often served by modalities such as touch, vision, and hearing that detect stimuli emanating from a common point in space; and processed by brain tissue maps that are spatially aligned. However, the neural interactions among modalities that share no spatial stimulus domain yet are essential for robust perception within noisy environments remain uncharacterized. Drosophila melanogaster makes its living navigating food odor plumes. Odor acts to increase the strength of gaze-stabilizing optomotor reflexes to keep the animal aligned within an invisible plume, facilitating odor localization in free flight. Here, we investigate the cellular mechanism for cross-modal behavioral interactions. We characterize a wide-field motion-selective interneuron of the lobula plate that shares anatomical and physiological similarities with the "Hx" neuron identified in larger flies. Drosophila Hx exhibits cross-modal enhancement of visual responses by paired odor, and presynaptic inputs to the lobula plate are required for behavioral odor tracking but are not themselves the target of odor modulation, nor is the neighboring wide-field "HSE" neuron. Octopaminergic neurons mediating increased visual responses upon flight initiation also show odor-evoked calcium modulations and form connections with Hx dendrites. Finally, restoring synaptic vesicle trafficking within the octopaminergic neurons of animals carrying a null mutation for all aminergic signaling is sufficient to restore odor-tracking behavior. These results are the first to demonstrate cellular mechanisms underlying visual-olfactory integration required for odor localization in fruit flies, which may be representative of adaptive multisensory interactions across taxa.

Figure 1

Visual and Olfactory Information Are Integrated to Generate More Robust Behavioral Outputs (A) The electronic visual flight simulator records wing kinematics from a fixed fly in response to sensory stimuli. The difference in wing beat amplitude (ΔWBA) across the two wings is proportional to yaw torque. Steering torque is activated by movement of the panoramic grating projected on the circular display of light-emitting diodes (LEDs) [20]. The arena is equipped with a laminar flow olfactometer. (B) Average modulation of ΔWBA optomotor response to a velocity impulse in the yaw axis with and without paired odor presentation. The sum of two exponential functions is fitted to the impulse responses (smooth line). Asterisk indicates two-way paired t test, p < 0.05 comparing peak amplitude values of fits to responses by individual flies. n = 15. (C) Magnetic-tether flight simulator records body orientation in response to a spatially restricted odor plume. A video image tracks the fly’s angular heading changes on a magnetic tether allowing free movement in the yaw plane. A narrow plume of odor is delivered from one side of the arena. (D) Exemplar flight orientation responses to an odor plume located at 180° (as in C) shown for T4T5-blocked flies (purple trace) and parental controls (black and gray traces). (E) Inactivation of the T4T5 local motion-detecting neurons (T4T5-Gal4/UAS-Kir, n = 25) inhibits stabilization of odor plume tracking. Time in plume, for each category acquisition and tracking, is total time spent within ±10° of the odor nozzle over the time period defined in (D). T4T5-Gal4/+, n = 20; UAS-Kir/+, n = 19). Mean ± SEM are shown. Asterisk denotes significant difference (two-way paired t test, p < 0.05).

Figure 2

Two-Photon Calcium Imaging and Characterization of Hx Tuning Properties (A) Perspective-matched LED arena display within the imaging apparatus, equipped with an olfactometer. (Bi) Posterior view confocal image of Hx dendrites within the right lobula plate via Odd-skipped-Gal4/GFP. The dashed white rectangle indicates the imaging ROI. Neuropil is indicated in purple (nc82 staining). (Bii) Dorsal view of Hx dendrites within the lobula plate via Odd-skipped-Gal4/GFP (same as Bi). White dashed box within the lobula plate indicates enlarged region shown in (Biii). Scale bar, 25μm. (Biii) Enlarged cross-section of lobula plate demonstrates the four layers of the lobula plate, with Hx innervation restricted to layer 2. Arrowheads indicate layer-specific directional tuning of T4T5 innervation [23]. (C) Average ΔF/F from Hx in response to a vertical bar revolving in each of two horizontal directions, either back to front (red) or front to back (blue), across the full 216° display. Black line represents superimposed ipsilateral azimuthal receptive-field fit. n = 7 animals. (D) Mean responses ± SEM to a vertical bar of varying width revolving in each of two horizontal directions across the display. n = 7 animals. (E) Directional tuning of Hx. A square-wave grating (27° wavelength) was moved in each direction as indicated on the x axis, and maximum ΔF/F was normalized to the largest response observed. Points indicate mean responses ± SEM. Red point and arrowhead indicate the stimulus direction giving maximum response, used in (F). n = 7 animals. (F) Temporal frequency tuning of Hx. A square-wave grating was moved at constant velocity from back to front. Points indicate mean responses ± SEM. n = 7 animals.

Figure 3

Odor-Induced Modulation of Hx Activity (A) Intracellular calcium response to visual motion by Hx neurons expressing GCaMP6m. Mean ΔF/F ± 1 SEM is shown. ^∗p < 0.05, rank-sum test on peak response amplitude compared between epoch one (prior to odor stimulation) and each following epoch, n = 13 animals. (B) Mean maximum ΔF/F from each individual fly compared between epochs one (prior to odor stimulation, black circle) and two (paired odor, orange circle) from (A). ^∗p < 0.05, rank-sum test, n = 13 animals. (C) Mean maximum ΔF/F from each individual fly compared between epochs one (prior to odor stimulation, black circle) and two (water vapor control, gray circle) from (A). Epochs one and two were not statistically different via rank-sum test, n = 6 animals. (D) T4T5-Gal4 expression pattern within the visual ganglia. ROIs shown in (E) are from the lobula plate. (E) Mean ΔF/F ± 1 SEM for T4T5 terminals in the lobula plate. n = 6 animals. (F) Mean maximum ΔF/F for T4T5 terminals for individual animal in epochs one and two. n = 6 animals. (G) R27B03-Gal4 expression pattern includes HSE neurons within the lobula plate, imaged within the primary HSE dendrite (white dashed box). (H) Mean ΔF/F ± 1 SEM for HSE. n = 7 animals. (I) Mean maximum ΔF/F for HSE for each individual animal in epochs one and two. n = 6 animals.

Figure 4

Octopaminergic Neurons Innervating the Lobula Plate Are Activated by Odor, Make Close Contact with Hx, and Are Required for Behavioral Plume Tracking (A) Intracellular calcium dynamics (ΔF/F, GCaMP6s) of octopaminergic terminals innervating the lobula plate in response to olfactory stimulation. Asterisks indicates significance between odor off (black line) and odor on (orange line) shown above the mean ΔF/F response (two-way paired t test, p < 0.005). n = 6 animals. (B) Mean maximum ΔF/F for each individual animal during a period preceding the odor pulse (black) and during the odor pulse (orange). Horizontal bars over the ΔF/F response in (A) indicate the measurement epochs. n = 6 animals. (C) GFP expression by GRASP indicates octopaminergic (Tdc2-Gal4) connections with Hx (Odd-Gal4). Inset shows Hx arborization pattern to highlight similarity in GFP profile between GRASP and the lobula plate tangential cell. (D) Mean time ± SEM spent in odor plume during the duration of the experiment (olfactory flight simulator; Figure 1C) for flies carrying a null mutation in the Drosophila vesicular monoamine transporter dVMAT rescued with either a wild-type DVMAT transgene in octopaminergic neurons (Tdc2-Gal4/ VMAT) or a DVMAT trafficking mutant (Tdc2-Gal4/ Δ3VMAT). Asterisk indicates significant difference (two-way paired t test, p < 0.05) between VMAT (n = 32 animals) and Δ3VMAT (n = 21 animals). Also shown is mean time in plume for Tdc2-Gal4/VMAT-rescued flies exposed to water rather than vinegar (n = 32 animals, ^∗p < 0.05 by two-way paired t test).