Odor mixtures of opposing valence unveil inter-glomerular crosstalk in the Drosophila antennal lobe

Abstract

Evaluating odor blends in sensory processing is a crucial step for signal recognition and execution of behavioral decisions. Using behavioral assays and 2-photon imaging, we have characterized the neural and behavioral correlates of mixture perception in the olfactory system of Drosophila. Mixtures of odors with opposing valences elicit strong inhibition in certain attractant-responsive input channels. This inhibition correlates with reduced behavioral attraction. We demonstrate that defined subsets of GABAergic interneurons provide the neuronal substrate of this computation at pre- and postsynaptic loci via GABA_B- and GABA_A receptors, respectively. Intriguingly, manipulation of single input channels by silencing and optogenetic activation unveils a glomerulus-specific crosstalk between the attractant- and repellent-responsive circuits. This inhibitory interaction biases the behavioral output. Such a form of selective lateral inhibition represents a crucial neuronal mechanism in the processing of conflicting sensory information.

Fig. 1

Establishing mixture ratio at which repellent odor reduces behavioral attraction. a Schematic drawing of the FlyWalk assay. Individual adult female flies are placed in small glass tubes where pulses of single odors or binary mixtures are presented in a continuous airflow (adapted from refs. ). b, c Behavioral responses to ethyl acetate (ETA, 10⁻², blue-green), benzaldehyde (BEA, 10⁻¹, red), and their binary mixture (MIX( + ), yellow). b Quantified behavior from individual flies (n = 30) stimulated with ethyl acetate, benzaldehyde, and their binary mixture. Line represents mean upwind speed; shadow indicates SEM. Gray bars in b, d represent odor pulse (1 s). c Box plots represent net upwind displacement of 4 s from odor onset. Colored dots and gray lines represent individual flies (Wilcoxon signed rank test). d, e Same as in b, c but for a lower concentration of ethyl acetate (10⁻³). d Quantified behavior from individual flies (n = 30) stimulated with ethyl acetate (bright blue-green), benzaldehyde (red), and their binary mixture (dark orange). e Box plots represent net upwind displacement of 4 s from odor onset. f Schematic drawing of the T-maze assay. g Box plots showing behavioral preference indices in the T-maze assay to benzaldehyde (10⁻¹), ethyl acetate (10⁻²/10⁻³), MIX( + ) and MIX(−) against the solvent control (MOL) (n = 16–17; one-way ANOVA with posthoc Tukey test). Box plots here and in all following figures represent the median value (horizontal line inside the box), the interquartile range (height of the box), and the minimum and maximum value (whiskers, excluding the outliers) of each experimental group

Fig. 2

Glomeruli responding to the attractive odor reveal mixture inhibition. a Schematic of odor delivery system connected to 2-photon microscope. FM flowmeter, cont. continuous, O1/O2 odor 1/odor 2. b, c Representative odor-evoked calcium responses in PNs from three focal planes. Gray-scale images represent AL structure with identified glomeruli. Calcium responses are shown to ethyl acetate (10⁻²/10⁻³), benzaldehyde (10⁻¹), MIX (+) and MIX(−). Scale bar = 20 µM. d Mean PN activity of strongest activated repellent-responsive (DL1, DL5, red) and attractant-responsive glomeruli (DM1, DM2, DM3, DM4, blue-green) during stimulation with ethyl acetate (10⁻², blue-green), benzaldehyde (10⁻¹, red) and their binary mixture (MIX(+), yellow). Odor responses of all annotated glomeruli (in total 34) are shown in Supplementary Fig. 2. Upper panel, averaged time traces of calcium signals with SEM (shadow); gray bar represents 2 s odor stimulation. Lower panel, mean fluorescence signals during odor stimulation; individual flies are given by single dots and lines; mean is indicated by black thick line (n = 9, paired t-test). Pairwise comparisons of mixture responses to the response of the strongest single component (i.e. either ethyl acetate or benzaldehyde) are shown for each animal. e Same as in d for ethyl acetate at 10⁻³ (bright blue-green), benzaldehyde (10⁻¹, red) and MIX(−) (orange) (n = 11, paired t-test). f PCA of six most activated glomeruli during stimulation with the odors shown in b–e. Colored dots represent individual measurements. Shadows represent 95% ellipses for each odor. MIX(−) and benzaldehyde representations are not significantly different (one-way ANOSIM, Rho similarity index). g Box plots represent net upwind displacement in the FlyWalk within 4 s following stimulation with ethyl acetate (10⁻³, bright blue-green), benzaldehyde (10⁻², bright red), and MIX( + ) (yellow). Colored dots and gray lines represent individual flies. (n = 30, Wilcoxon signed rank test). h Mean PN activity of strongest activated repellent- and attractant-responsive glomeruli during stimulation with the odors from g (n = 8, paired t-test). i Same data as in g with ethyl acetate at 10⁻⁴ (turquoise, n = 30, Wilcoxon signed rank test). j Mean PN activity of strongest activated repellent- and attractant-responsive glomeruli during stimulation with the odors from i (n = 8, paired t-test)

Fig. 3

Different binary mixtures evoke glomerulus-specific inhibitions. a, b Upper panel, box plots represent net upwind displacement in the FlyWalk within 4 s following stimulation with ethyl acetate (10⁻² and 10⁻³, blue-green/bright blue-green), methyl salicylate (10⁻³, magenta) and their binary mixtures (MIX(+) and MIX(−), yellow/orange). Colored dots and gray lines represent individual flies (n = 30, Wilcoxon signed rank test). Lower panel, mean PN activity of strongest activated attractant- and repellent-responsive glomeruli during stimulation with the odors from a or b, respectively (n = 6, paired t-test). c Left, box plots represent net upwind displacement in the FlyWalk within 4 s following stimulation with balsamic vinegar (10⁻², blue), benzaldehyde (10⁻¹, red), and their binary mixture (MIX(−), orange) (n = 30, Wilcoxon signed rank test). Right, mean PN activity of strongest activated attractant- and repellent-responsive glomeruli during stimulation with the odors used in the FlyWalk  (n = 6, paired t-test). d Left, box plots represent net upwind displacement in the FlyWalk within 4 s following stimulation with balsamic vinegar (10⁻², blue), geosmin (10⁻³, pink), and their binary mixture (MIX(−), orange; n = 30, Wilcoxon signed rank test). Right, mean PN activity of strongest activated attractant- and repellent-responsive glomeruli during stimulation with the odors used in the FlyWalk (n = 6, paired t-test). Odor responses to the different mixture combinations of all annotated glomeruli (in total 34) are shown in Supplementary Fig. 3

Fig. 4

Selective silencing of input channels reveals glomerulus-specific inhibition. a Schematic of the experimental design: Or10a-expressing ORNs (targeting DL1) are not functional in a Or10a^−/− mutant background. Color code indicates glomerulus-specific activation by the attractive (bright blue-green) or repellent (red) odor. b Representative odor-evoked calcium responses in PNs from three focal planes of Or10a^−/− mutant fly expressing UAS-GCaMP6s in PNs. Gray-scale images represent the AL structure highlighting the attractant- (DM1, DM2, DM3, and DM4) and repellent-responsive glomeruli (DL1 and DL5) with colored circles. Calcium responses are shown to stimulation with ethyl acetate (10⁻³), benzaldehyde (10⁻¹), and their binary mixture (MIX(–)). Scale bar = 20 µM. c Mean PN activity of repellent- and attractant-responsive glomeruli during stimulation with ethyl acetate (10⁻³, bright blue-green), benzaldehyde (10⁻¹, red), and MIX(−) (orange) in Or10a^−/− mutant flies. Individual flies are given by individual dots and lines; mean is indicated by black thick line (n = 10, paired t-test). Pairwise comparisons of mixture responses to the response with the strongest single component (i.e. ethyl acetate or benzaldehyde) are shown for each animal. d Schematic of the experimental design: Or7a-expressing ORNs (targeting DL5) are not functional in a Or7a^−/− mutant background. Color code indicates glomerulus-specific activation by the attractive (bright blue-green) or repellent (red) odor. e Representative gray-scale and pseudcolored images of odor-evoked calcium responses in PNs from three focal planes of a Or7a^−/− mutant fly expressing UAS-GCaMP6s in PNs. Calcium responses are shown to stimulation with ethyl acetate (10⁻³), benzaldehyde (10⁻¹), and their binary mixture (MIX(−)). Scale bar = 20 µM. f Mean PN activity of repellent- and attractant-responsive glomeruli during stimulation with ethyl acetate (10⁻³, bright blue-green), benzaldehyde (10⁻¹, red), and MIX(−) (orange) in Or7a^−/− mutant flies (n = 12, paired t-test). g Schematic of the T-maze assay. h Box plots showing behavioral preference indices in the T-maze of Or10a^−/− mutant (pink), Or7a^−/− mutant (purple), and w1118 flies (gray) to the odors benzaldehyde (10⁻¹), ethyl acetate (10⁻²/10⁻³), MIX( + ), and MIX(−) against the solvent control (MOL). (n = 15–19, one-way ANOVA with posthoc Tukey test, **p < 0.01). Filled boxes are significantly different from zero, empty boxes not (Student’s t-test)

Fig. 5

Optogenetic activation of repellent-responsive glomeruli unveils glomerular crosstalk. a Schematic of the experimental design: artificial activation of CsChrimson by red light in Or10a-expressing ORNs (targeting DL1) during stimulation with ethyl acetate. Color code reflects activation by light (red) or attractive odor (bright blue-green). b Calcium signals (time traces and pseudocolored images) of glomerulus DL1 to photoactivation with increasing intensities of 619 nm light for 2 s. Gray boxes indicate light stimulation. Lines represent averaged response, shadows give SEM (n = 3). Dashed box marks the light intensity used for further experiments. c Mean PN activity of repellent- and attractant-responsive glomeruli during stimulation with either light (red dots), ethyl acetate (10⁻³, bright blue-green dots), or both combined (additional red rectangles) in flies expressing CsChrimson in ORNs of DL1 and GCaMP3 in PNs (n = 19, paired t-test). d Schematic of the experimental design: artificial activation of CsChrimson by red light in Or7a-expressing ORNs (targeting DL5) during stimulation with ethyl acetate. e Same experiment as in b but for CsChrimson expression in ORNs of DL5 (n = 4). f Same experiment as in c but for CsChrimson expression in ORNs of DL5 (n = 21, paired t-test). g Box plots showing preference indices in the T-maze assay of flies with artificial activation of ORNs expressing Or10a (DL1), Or7a (DL5) or both via CsChrimson by red light alone (bulb) or red light combined with an odor (bulb + odor) against the dark arm of the T-maze without (Blank) or with solvent (MOL) (n = 22–24, one-way ANOVA with posthoc Sidak’s multiple comparisons test, *p < 0.05, ***p < 0.001). Treatment and genotypes are indicated by the table below. In our assay, control flies (no all-trans retinal) showed reproducible slight attraction to light. Based on this, we used the control flies as the comparison point for calculating optogenetically driven avoidance

Fig. 6

MIX(−)-induced inhibition can be blocked by GABA- and glutamate antagonists. a Schematic illustrating the experimental design: GABA_B antagonist CGP54626 (50 µM) is applied while calcium responses of PNs are monitored (green). b Mean PN activity of the four attractant-responsive glomeruli showing the effect of CGP54626 (CGP) compared to saline (S) and wash-out (W) during stimulation with ethyl acetate (10⁻³, ETA, bright blue-green) and MIX(−) (orange). Individual flies are given by single dots and lines; mean is indicated by black thick line (n = 10, paired t-test). c Box plots represent normalized peak response differences of odor responses of the glomeruli shown in b. Differences were calculated by subtracting calcium signals to MIX(−) from those to ethyl acetate during different treatments (i.e. 1 represents strongest mixture inhibition, while 0 means no inhibition). Circles show individual animals (n = 10, one-way ANOVA with Bonferroni’s multiple comparisons test). d Schematic illustrating the experimental design: GABA_A and glutamate antagonist picrotoxin (100 µM) is applied while calcium responses of PNs are monitored (green). e, f Same representations as in b, c for picrotoxin (PTX) compared to saline (S) (n = 10, Student’s t-test). g Schematic illustrating the experimental design: mixture of CGP54626 (50 µM) and picrotoxin (100 µM) is applied to block GABA_A, GABA_B, and glutamate receptors while calcium responses of PNs are monitored (green). h, i Same representations as in b, c for the combined application of picrotoxin and CGP54626 (PTX + CGP) compared to saline (S) (n = 9, Student’s t-test)

Fig. 7

Mixture inhibition takes place at pre- and postsynaptic sites. a Schematic illustrating the effect of different RNAi-lines used to block GABA_A, GABA_B, or glutamate receptors selectively in PNs, while odor responses in PNs are monitored with 2-photon imaging (green). UAS-empty-RNAi serves as the control. b Mean PN activity of the four attractant-responsive glomeruli during stimulation with ethyl acetate (10⁻³, bright blue-green) and MIX(−) (orange) in flies expressing the indicated RNAi lines (c) in PNs. Individual flies are given by single dots and lines; mean is indicated by black thick line (n = 10, paired t-test). c Box plots represent normalized peak response differences of odor responses of the glomeruli shown in b. Circles show individual animals (n = 10, one way ANOVA with Bonferroni’s multiple comparisons test). d Schematic illustrating the effect of different RNAi-lines used to block GABA_A, GABA_B, or glutamate receptors selectively in ORNs, while odor responses in PNs are monitored with 2-photon imaging (green). UAS-empty-RNAi serves as the control. e Mean PN activity of the four attractant-responsive glomeruli during stimulation with ethyl acetate (10⁻³, bright blue-green) and MIX(−) (orange) in flies expressing the indicated RNAi lines (f) in ORNs. Individual flies are given by individual dots and lines; mean is indicated by black thick line (n = 8–12, paired t-test). f Box plots represent normalized peak response differences of odor responses of the glomeruli shown in e. Circles show individual animals (n = 8–12, one way ANOVA with Bonferroni’s multiple comparisons test)

Fig. 8

Defined subsets of GABAergic LNs mediate mixture inhibition. a–d Upper panel, immunostaining against GFP (white) and GABA (red) within the AL of four different Gal4 lines that label LN subpopulations with intact (empty-RNAi) or silenced (Gad-RNAi) GABA production. Scale bar = 20 µm. Lower panel, barplots represent mean PN activity of the four attractant-responsive glomeruli during stimulation with MIX(−) in flies with intact or silenced GABA production in four different LN subpopulations (shown in a–d). Individual flies are given by individual dots. Data are represented by mean+SEM (n = 7–15, Student’s t-test, **p < 0.01, ***p < 0.001). Silencing GABA release of LNs labeled by NP3056-Gal4 results in a strong MIX(−) response in PNs of glomerulus DM3, while glomeruli DM1 and DM4 show strong mixture responses when HB4-93-Gal4 LNs were silenced. This response increase to MIX(−) indicates a relief of the mixture-induced inhibition. e, f Left, representative individual patchy LNs, labeled by photoactivating PA-GFP in single somata of NP3056-Gal4 LNs (e), and HB4-93-Gal4 LNs (f). To facilitate glomerular identification, GH146-QF, QUAS-mtdtomato (blue) was expressed. Scale bar 20 µm. Right, neuronal reconstructions of two exemplary single LNs. Glomeruli that are supposed to be connected by these LNs are highlighted. g Boxplots representing the quantification of the florescence signal of the presynaptic marker UAS-Syt::HA (left panel) and the postsynaptic marker UAS-homer-GCaMP3 (right panel) expressed under control of NP3056-Gal4 (n = 10 for pre-, n = 12 for postsynapses, two-sample t-test). h Same as in g for HB4-93-Gal4 (n = 12 for pre-, n = 10 for postsynapses, one-way ANOVA with posthoc Dunnett’s multiple comparisons test). *p < 0.05, **p < 0.01, ***p < 0.001. See Supplementary Fig. 9b, c for immunostaining of these markers

Fig. 9

Circuit model for glomerulus-specific crosstalk in the fly AL. a Stimulation with the attractive odor ethyl acetate activates the attractant-responsive glomeruli DM1, DM2, DM3, and DM4, which results in behavioral attraction. b The repellent odor benzaldehyde is blended with the attractive odor ethyl acetate. Benzaldehyde activates the repellent-responsive glomeruli DL1 and DL5 which induce an inhibition of the attractant-responsive glomeruli via two different subsets of GABAergic LNs: DL1 is mediating the inhibition of glomeruli DM1 and DM4 via HB4-93-type LNs, while glomeruli DM3 and, to some extent, DM2 are inhibited by DL5 via NP3056-type LNs. This inhibitory crosstalk shifts the mixture representation towards the repellent odor and consequently leads to a reduced behavioral attraction