Dissecting the signaling mechanisms underlying recognition and preference of food odors

Abstract

Food is critical for survival. Many animals, including the nematode Caenorhabditis elegans, use sensorimotor systems to detect and locate preferred food sources. However, the signaling mechanisms underlying food-choice behaviors are poorly understood. Here, we characterize the molecular signaling that regulates recognition and preference between different food odors in C. elegans. We show that the major olfactory sensory neurons, AWB and AWC, play essential roles in this behavior. A canonical Gα-protein, together with guanylate cyclases and cGMP-gated channels, is needed for the recognition of food odors. The food-odor-evoked signal is transmitted via glutamatergic neurotransmission from AWC and through AMPA and kainate-like glutamate receptor subunits. In contrast, peptidergic signaling is required to generate preference between different food odors while being dispensable for the recognition of the odors. We show that this regulation is achieved by the neuropeptide NLP-9 produced in AWB, which acts with its putative receptor NPR-18, and by the neuropeptide NLP-1 produced in AWC. In addition, another set of sensory neurons inhibits food-odor preference. These mechanistic logics, together with a previously mapped neural circuit underlying food-odor preference, provide a functional network linking sensory response, transduction, and downstream receptors to process complex olfactory information and generate the appropriate behavioral decision essential for survival.

Keywords: glutamatergic transmission; neuropeptide signaling; olfactory sensory neurons; olfactory sensory signaling; preference of food odors.

Figure 1.

An automated microdroplet assay for sensorimotor responses to food odors. A, The protocol for measuring food-odor preference. B, Schematic of the microdroplet assay for olfactory preference between two bacterial strains, E. coli OP50 and P. aeruginosa PA14. The sapphire window holds 12 droplets (2 μl each) of NGM buffer that contain individual young adults. The animals are subjected to two air streams that are odorized with the smell of OP50 or the smell of PA14 and alternate every 30 s. The swimming behavior is recorded. The frequency of Ω bends is analyzed by a customized software and the PA14 preference index is calculated as indicated (See Materials and Methods). C, Sample turning frequency and PA14 preference index generated by multiple wild-type animal. Mean ± SEM.

Figure 2.

The cGMP-gated TAX-2/TAX-4 channel acts in the AWB and AWC sensory neurons to facilitate food-odor preference of PA14 to OP50. A, B, The tax-2(ks10) and tax-4(ks28) loss-of-function mutants are defective in preference of the smell of PA14 to the smell of OP50. C, D, Expressing the tax-4 genomic DNA rescues the preference defect in the tax-4(ks28) mutants and selectively expressing a wild-type cDNA of tax-4 in the AWB and AWC neurons also rescues the preference defect in the tax-4(ky791) mutant animals. E–H, The cng-1(ok3292) and cng-3(jh113) mutants (E, F), as well as the osm-9(ky10) mutants (G, H) are normal in olfactory preference between OP50 and PA14. I–L, The tax-2(p691);tax-4(p678) double mutants (I, J) and each of the single mutants (K, L) are severely defective in the olfactory preference of PA14 over OP50. In J, error bars are smaller than the size of the circles. For all, the transgenic animals and their nontransgenic siblings, as well as mutants, were compared with the wild-type control N2 measured in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≥ 3 assays, mean ± SEM.

Figure 3.

A subset of the tax-2-expressing neurons negatively regulates food-odor preference. A–D, The tax-2(p694) mutants that have lost the activity of TAX-2 in only a subset of the tax-2-expressing sensory neurons (BAG, AQR, AFD, and ASE) exhibit enhanced preference toward the smell of PA14 in comparison with the smell of OP50 (A, B) and expressing a full-length tax-2 genomic transgene rescues the enhanced PA14 preference in tax-2(p694) mutants (C, D). E–N, Expressing the wild-type tax-2 cDNA in the ASE (Pflp-6), AFD (Pgcy-8), BAG (Pflp-17), and AQR (Pgcy-32) sensory neurons (E, F) or in BAG (G, H) or AQR (I, J) alone rescues enhanced PA14 preference in the tax-2(p694) mutants, but expressing the tax-2 cDNA in either AFD (K, L) or ASE (M, N) alone does not rescue. In N, error bars are smaller than the size of the circles. O, The animals that express the egl-1 cDNA in the BAG sensory neuron exhibit enhanced PA14 preference. For all, transgenic and nontransgenic siblings, as well as mutants, were compared with wild-type N2 animals that were examined in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≥ 4 assays, mean ± SEM.

Figure 4.

A G-protein signaling pathway regulates food-odor recognition. A, B, Two different mutations in the Gα-encoding gene odr-3 disrupt olfactory preference of PA14 to OP50, but mutating several other Gα-encoding genes does not alter olfactory preference for PA14. C, D, Mutations in two genes that encode guanylyl cyclases odr-1 and daf-11 disrupt the food-odor preference between PA14 and OP50. E–H, Selective knockdown of the odr-1 activity or the daf-11 activity in the AWB and AWC sensory neurons in wild-type animals abolishes the preference of PA14 smell to OP50 smell. I–L, Animals containing null mutations in eat-16 are defective in the olfactory preference of PA14 in comparison with OP50, but deleting rgs-3 does not alter PA14 preference. For all, transgenic animals and their nontransgenic siblings, as well as mutants, were compared with the wild-type N2 tested in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≥ 3 assays, mean ± SEM.

Figure 5.

The AWC sensory neuron regulates food-odor recognition through glutamatergic neurotransmission. A–F, The loss of EAT-4 activity in the eat-4(ky5) mutants results in a loss of olfactory preference of PA14 to OP50 (A, B) and this defect is rescued by the expression of a genomic DNA of eat-4 (C, D) or selective expression of a wild-type eat-4 cDNA in the AWB and AWC sensory neurons (E, F, Podr-1::eat-4), but expressing the wild-type eat-4 cDNA in AWB alone does not rescue (E, F, Pstr-1::eat-4). G, H, Selective knockdown of eat-4 activity in AWC (Pnlp-1::eat-4RNAi) abolishes olfactory preference of PA14 to OP50. I–L, Animals lacking tdc-1 or tbh-1 are wild-type in the olfactory preference of PA14 in comparison with OP50 (I, J), and application of exogenous tyramine or octopamine does not change the preference of PA14 over OP50 (K, L). For all, transgenic animals and their nontransgenic siblings, as well as mutants, were compared with the wild-type N2 animals tested in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≥ 3 assays, mean ± SEM.

Figure 6.

The AWB sensory neuron mediates PA14 odor preference via neuropeptidergic signaling. A, B, Mutations in genes that encode peptide-processing enzymes, egl-3 and egl-21, disrupt food-odor preference of PA14 to OP50. C–F, Selective knockdown of egl-3 activity in AWB and AWC (C, D, Podr-1::egl-3RNAi) or in AWB alone (E, F, Pstr-1::egl-3RNAi) significantly reduces PA14 preference in wild-type animals. G, H, Deleting the AWB-expressing neuropeptide-encoding gene nlp-9 significantly reduces the olfactory preference of PA14 to OP50. I, J, Selective knockdown of the nlp-9 activity in AWB alone reduces the preference of the PA14 smell to OP50 smell. K, L, Deleting the AWC-expressing neuropeptide-encoding gene nlp-1 significantly reduces the food-odor preference of PA14 to OP50. M, N, Expressing the wild-type nlp-1 activity in AWC olfactory sensory neuron rescues the defect of the olfactory preference of PA14 in comparison with OP50 in nlp-1(ok1470) mutants. O, P, Loss of npr-18 or knocking down npr-18 activity decreases the food-odor preference of PA14 over OP50. Q, Mutations in nlp-9, npr-18, and egl-3 significantly disrupt the aversive response to the repulsive odorant 2-nonanone, a response mediated by AWB sensory neurons. For all, transgenic animals and their nontransgenic siblings, as well as mutants, were compared with the wild-type N2 animals tested in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≤ 3 assays (A–P) or n ≥ 2 separate days (Q), mean ± SEM.

Figure 7.

Combinatorial effects of multiple glutamate receptors on food-odor response to PA14 and OP50. A, B, Single mutations in a series of genes that encode glutamate receptors do not alter food-odor preference for PA14. C, D, The double mutants glr-1;glr-2 and glr-1;nmr-1 reduce PA14 preference. E, F, The double mutant glr-1;glc-3 exhibits wild-type PA14 preference. For all, transgenic animals and their nontransgenic siblings, as well as mutants, were compared with wild-type N2 animals tested in parallel, two-tailed Student's t test. ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05; no asterisk denotes no statistical difference (p > 0.05), n ≤ 3 assays, mean ± SEM.

Figure 8.

The signaling pathways underlying the olfactory preference of PA14 to OP50. A, Diagram represents the neuronal network for naive and learned olfactory preference for PA14 over OP50 (Ha et al., 2010). Anatomically, the network consists of the major olfactory sensory neurons AWB and AWC, as well as the downstream interneurons and motor neurons that mediate head bending. Functionally, our analysis has shown that neurons highlighted in blue regulate olfactory preference of PA14 in naive animals (i.e., animals cultivated with OP50 as food) and neurons highlighted in red mediate learned preference of PA14 after training with PA14. Arrows and lines denote chemical and electrical synapses, respectively. B, The AWC/AWB signal transduction and neurotransmission pathways that regulate the recognition and preference of the odors generated by E. coli OP50 and P. aeruginosa PA14. C. elegans genes implicated in food-odor recognition and preference and their mammalian homologues are shown on the right. In the schematic, the genes in black font are required for food-odor recognition and the genes in red font are specifically required for only food-odor preference.