A mechanosensory receptor required for food texture detection in Drosophila

Abstract

Textural properties provide information on the ingestibility, digestibility and state of ripeness or decay of sources of nutrition. Compared with our understanding of the chemosensory assessment of food, little is known about the mechanisms of texture detection. Here we show that Drosophila melanogaster can discriminate food texture, avoiding substrates that are either too hard or too soft. Manipulations of food substrate properties and flies' chemosensory inputs indicate that texture preferences are revealed only in the presence of an appetitive stimulus, but are not because of changes in nutrient accessibility, suggesting that animals discriminate the substrates' mechanical characteristics. We show that texture preference requires NOMPC, a TRP-family mechanosensory channel. NOMPC localizes to the sensory dendrites of neurons housed within gustatory sensilla, and is essential for their mechanosensory-evoked responses. Our results identify a sensory pathway for texture detection and reveal the behavioural integration of chemical and physical qualities of food.

Figure 1. Flies display feeding and positional preference for substrate texture.

(a) Left: schematic of the two-choice colorant feeding assay. Flies can choose to feed from 5 mM sucrose in substrates composed of 0.5 or 2% agarose coloured with different edible dyes (which are switched in different trials) in alternate wells of a microtitre plate. Feeding preference is determined by abdominal coloration of flies after 90 min. Right: PI of wild-type (w¹¹¹⁸) flies for feeding from 0.5% agarose substrates, calculated as described in the Methods (n=20 experiments). ***P<0.001 (Wilcoxon signed rank test (H₀=0)). (b) Graph of substrate stiffness (target force in milliNewtons measured using Semmes-Weinstein Monofilaments; see Methods) of substrates of different agarose concentrations (red dots). Overlaid are the ranges of similar measurements made from the flesh or skin of the fruit samples shown in the photos (Fujifilm XT1 camera; 18–55 mm objective); compared with agarose, natural food substrates have heterogeneous properties. (c) Left: schematic of two-choice positional preference arena assay. Flies can choose to feed from 5 mM sucrose in substrates composed of different agarose concentrations (without dyes) in alternate chambers of a 90 mm diameter four-quadrant plate. Right: representative images of flies in the assay arena at the start and end of an experiment. Fly position was quantified automatically and used to calculate a Preference Index (PI) as indicated below the images. (d) Time course of PI of wild-type flies for 0.5% agarose over 2% agarose (n=12 arenas). (e) Preference of wild-type flies for 0.5% agarose in the arena assay (at time=90 min, in this and subsequent assays unless otherwise stated) with different alternative substrate agarose concentrations; all substrates contain 5 mM sucrose (n=15 for 0.02% agarose, n=18 for 0.1% agarose, n=14 for 0.25% agarose, n=15 for 0.5% agarose, n=15 for 1% agarose, n=12 for 2% agarose). ns: not significant, ***P<0.001, **P<0.01, *P<0.05 (Wilcoxon signed rank test (H₀=0)).

Figure 2. Appetitive gustatory signals are necessary and sufficient to reveal texture preference.

(a) Preference of starved (grey) or non-starved (red) wild-type flies for 0.5% agarose in a 0.5% versus 2% assay arena in the presence of sucrose (n=12 for both starved and non-starved conditions). ns: not significant, ***P<0.001 (Wilcoxon rank sum test). (b) Left: Preference of wild-type flies for 0.5% agarose in a 0.5% versus 2% arena assay in the presence (grey) or absence (red) of 5 mM sucrose (n=13 with sucrose, n=14 without sucrose). Right: Percentage of flies feeding in the arena assay in the presence (grey) or absence (red) of 5 mM sucrose in parallel assays containing blue food dye in all quadrants, as determined by abdominal coloration of animals after 90 min (n=13 with sucrose, n=14 without sucrose). ***P<0.001 (Wilcoxon rank sum test). (c) Preference of wild-type flies for 0.5% agarose in a 0.5% versus 2% arena assay in the presence of 5 mM sucrose (grey) or 5 mM sorbitol (red) (n=12 for sucrose, n=13 for sorbitol). ***P<0.001 (Wilcoxon rank sum test). (d) Preference for 0.5% agarose in a 0.5% versus 2% arena assay (with 5 mM sucrose) of flies in which Tetanus toxin (TNT) is expressed under the control of Gr64f, Ppk28 or Gr66a promoters. Comparisons are made with control animals expressing an impaired version of this toxin (TNT^IMP). The PI at time=120 min is shown, because animals bearing the Gr64f-Gal4 transgene display delayed decisions; the temporal evolution of PI is shown in Supplementary Fig. 1c. Genotypes are: w;Gr64-Gal4f/UAS-TNT (n=14), w;Gr64f-Gal4/UAS-TNT^IMP (n=12), w;Ppk28-Gal4/UAS-TNT (n=8), w;Ppk28-Gal4/UAS-TNT^IMP (n=8), w;Gr66a-Gal4/UAS-TNT (n=13), w;Gr66a-Gal4/UAS-TNT^IMP (n=12),. ns: not significant, **P<0.01, *P<0.05 (Wilcoxon rank sum test). (e) Preference of wild-type flies for 0.5% agarose in a 0.5% versus 2% arena assay in which 100 μl of 1 M sucrose solution is applied uniformly on the surface of each quadrant and allowed to dry before introducing the flies (n=12). ***P<0.001 (Wilcoxon signed rank test (H₀=0)). (f) Preference of flies for 0.5% agarose in a 0.5% versus 2% arena assay (without sucrose), in which Gr64f sweet-sensing neurons are uniformly activated across all quadrants through optogenetic stimulation. Genotypes: w;Gr64f-Gal4/Gr64f-Gal4 (n=14); w;UAS-CsChrimson/UAS-CsChrimson (n=16); w;Gr64f-Gal4/UAS-CsChrimson (n=19). ***P<0.001 (Wilcoxon rank sum test).

Figure 3. The mechanosensory channel NOMPC is necessary for texture discrimination and expressed in gustatory sensilla neurons.

(a) Preference of the indicated control and nompC mutant flies for 0.5% agarose in a 0.5% versus 2% arena assay (with 5 mM sucrose). Wild-type (n=17), nompC^f00642/+ (n=18), nompC¹/+ (n=15), nompC³/+ (n=15), nompC^f00642/nompC^f00642 (n=17), nompC^f00642/nompC¹ (n=10), nompC^f00642/nompC³ (n=12). ns: not significant, **P<0.01, *P<0.05. Comparisons were made against the wild-type control (Wilcoxon rank sum test with Bonferroni correction for multiple comparisons). (b) Immunofluorescence with anti-NOMPC (green, on a bright-field background) on whole-mount labella. In wild-type animals, NOMPC concentrates in the distal tip of sensory neurons that terminate at the base of each taste sensillum (arrowheads); nompC mutants lack this expression (arrowheads). Scale bar, 10 μm. (c) Immunofluorescence with anti-NOMPC (red) on whole-mount labella of animals in which different gustatory sensory neurons (GSNs) are transgenically labelled (green). Genotypes: w;Gr5a-LexA,LexAop-rCD2:GFP;TM2/TM6B, w;Gr66a-LexA,LexAop-rCD2:GFP;TM2/TM6B. Scale bar, 50 μm. Inset: NOMPC does not colocalise with GSNs (arrowheads). Scale bar, 25 μm. Green and red colocalisation is sometimes apparent in the full-projection because of vertical superposition of the NOMPC-labelled and GSN sensory processes. (d) Immunofluorescence with anti-NOMPC (red) and anti-GFP (green) on whole-mount labella of animals of the indicated genotypes (w;UAS-CD4:tdGFP;nompC-Gal4 and w;LexAop-CD8:GFP-2A-CD8:GFP;nompC-LexA). Scale bar, 50 μm. Inset: NOMPC expression is adjacent to the GFP (arrowheads). The nompC-Gal4 line is not expressed in the labellum (arrowheads). Scale bar, 25 μm. (e) Immunofluorescence with anti-GFP (green) and nc82 (magenta) on whole-mount brains of animals (genotypes as in d). The weak green signal in the SEZ in nompC-Gal4 animals does not originate from labellar neurons. SEZ: Subesophageal Zone; AMMC: Antennal Mechanosensory and Motor Center. Scale bar, 50 μm. (f) Preference for 0.5% agarose in a 0.5% versus 2% arena assay (with 5 mM sucrose) of animals in which different nompC neuron populations are silenced, together with control lines. Left: genotypes: w;nompC-Gal4/UAS-TNT^IMP (n=10) and w;nompC-Gal4/UAS-TNT (n=10). ns: not significant (Wilcoxon rank sum test). Right: genotypes: w;nompC-LexA/LexAop-TNT (n=12), w;nompC-LexA/+ (n=16), w;LexAop-TNT/+(n=12). ***P<0.001, *P<0.05 (Wilcoxon rank sum test with Bonferroni correction for multiple comparisons).

Figure 4. Gustatory sensilla display NOMPC-dependent mechanosensory responses.

(a) Left: schematic of the Drosophila labellum showing the distribution and diverse orientations of different morphological classes of sensilla, including the ‘long' L2 and L3 classes examined in this study. Right: schematic of the cellular organization of L2 and L3 sensilla, which house four GSNs and a single presumed mechanosensory neuron (red) whose dendrites terminate at the base of the cuticular hair. For mechanosensory stimulation, the hair is displaced 20 μm by Piezo-controlled movement of a glass recording pipette placed over the end of the sensillum (see Methods). (b) Representative electrophysiological trace of an L3 sensillum before, during and after mechanosensory stimulation. (i) Before stimulation, basal neuronal spikes of varying amplitudes are visible; large spikes are likely to correspond to the sweet-sensing neuron, and smaller spikes to the other neurons in this sensillum. (ii) during movement of the hair, there is a negative deflection of the base-line and a fast train of larger ‘spike-like' electrical oscillations is detected, which lasts ∼100 ms. (iii) This initial electrical response eventually resolves into a series of small amplitude spikes. For quantifications, see Methods. (c) Representative traces of responses of L3 sensilla to mechanical stimulation and 100 mM sucrose stimulation in the indicated control and nompC mutant genotypes. The horizontal bar indicates the period of hair bending; the arrow indicates the time of contact of the sucrose-containing pipette with the sensillum. (d and e) Neuronal responses of L2 and L3 sensilla to mechanical stimulation (d) and to 100 mM sucrose (e) in the indicated genotypes. Wild-type (n_L2=31, n_L3=33), nompC^f00642/+ (n_L2=17, n_L3=19), nompC^f00642/nompC^f00642 (n_L2=17, n_L3=17), nompC^f00642/nompC¹ (n_L2=17, n_L3=17), nompC^f00642/nompC³ (n_L2=17, n_L3=16), nompC¹/nompC³ (n_L2=14, n_L3=16). All comparisons were made against the wild-type control. ns: not significant, ***P<0.001, **P<0.01, *P<0.05 (Wilcoxon rank sum test with Bonferroni correction for multiple testing). Although the weak hypomorphic nompC^f00642 allele shows decreased responses, these are not statistically different from controls; the behavioural phenotype of this mutant (Fig. 3a) may reflect the consequence of the combination of sensory defects across multiple sensilla.