Temporal responses of bumblebee gustatory neurons to sugars

Abstract

The sense of taste permits the recognition of valuable nutrients and the avoidance of potential toxins. Previously, we found that bumblebees have a specialized mechanism for sensing sugars whereby two gustatory receptor neurons (GRNs) within the galeal sensilla of the bees' mouthparts exhibit bursts of spikes. Here, we show that the temporal firing patterns of these GRNs separate sugars into four distinct groups that correlate with sugar nutritional value and palatability. We also identified a third GRN that responded to stimulation with relatively high concentrations of fructose, sucrose, and maltose. Sugars that were nonmetabolizable or toxic suppressed the responses of bursting GRNs to sucrose. These abilities to encode information about sugar value are a refinement to the bumblebee's sense of sweet taste that could be an adaptation that enables precise calculations of the nature and nutritional value of floral nectar.

Keywords: biological sciences; neuroscience; sensory neuroscience.

Graphical abstract

Figure 1

Stimulation with sugars elicits responses from three gustatory receptor neurons per galeal sensillum (A) Sugar structural relationships and representative extracellular tip-recordings of galeal GRNs stimulated with sugars at 100 mM. (B) Rasters of GRN spikes over 1 s stimulations with sugars of increasing concentrations from 0 to 1,000 mM (blue lettering is used to indicate sugar stimuli that were 500 mM at their maximum concentration in this experiment due to solubility).| (C) Spike sorting exploited inter-spike intervals (ISI) to differentiate GRNs: bursts of spikes with 5–10 ms ISIs (GRN1 spikes) were terminated by a single GRN2 spike marked by an end of burst doublet (EBD) with a very short ISI (<2.5 ms) (Miriyala et al., 2018). A third neuron (GRN3) fired occasionally within bursts as a within burst doublet (WBD) or end of burst triplet (EBT) with ISIs <2.5 ms. Bursts were followed by longer (>15 ms) inter-burst intervals. (D–F) The average firing rates of GRNs 1–3 over 1 s stimulations varied by sugar type and increased as a function of concentration (n = 18 to 24 sensilla from n = 6 to 8 bees per sugar). Points represent mean (±s.e.m) responses across animals. Concentration response curves for GRNs 1 and 2 varied significantly by sugar treatment (GRN1: F_30,807 = 45.55, p < 0.0001; GRN2: F_30,837 = 23.86, p < 0.0001). (G) Activation thresholds for GRNs 1–3 for all sugars. Box colors denote significant activation of a single sugar at a given concentration versus water (linear mixed effects, concentration: GRN1: F_4,257 = 213.21, p < 0.0001; GRN2: F_4,257 = 77.98, p < 0.0001; GRN3: F_4,257 = 20.37, p < 0.0001, sugar: GRN1: F_10,66 = 23.33, p < 0.0001; GRN2: F_10,66 = 11.57, p < 0.0001; GRN3: F_10,66 = 7.58 p < 0.0001; conc:sug: GRN1: F_40,257 = 13.22, p < 0.0001; GRN2: F_40,257 = 7.43, p < 0.0001; GRN3: F_40,257 = 5.48, p < 0.0001) with estimated marginal means (EMM) post hoc multiple comparisons across concentrations for each sugar (colored box denotes p < 0.05 for a sugar versus water at a given concentration). EMM post hoc comparisons between sugars in Figure S1. (H–J) Firing rates (spikes per 100 ms bin) of GRN1 versus GRN2 over 1 s stimulations with sugars that significantly activated GRN2 at 10 (H), 100 (I), and 1,000 mM (J, 500 mM for maltose and melezitose). Points represent mean spikes/s in each bin across all trials for a given sugar, connected with lines over time (over 10 bins). The symbol for the first bin for each sugar (i.e., time = 100 ms) is an open marker with a black outline. GRN2 significantly predicts the trajectory of GRN1 (GLM, GRN2: F_1,194 = 1723.0, p < 0.0001; sugar: F_6,194 = 67.8, p < 0.0001; concentration F_2,194 = 138.6, p < 0.0001). (K) Cluster assignment for each sugar and concentration using the PCs of GRN1-3 responses in 100 ms bins over 1 s of stimulation. Stimuli were assigned to the cluster to which the largest proportion of individual sensilla responses matched (mean cluster consensus across stimuli: 0.79, CI 0.07). Boxes surround stimuli assigned to a single cluster. Sugars are abbreviated throughout the manuscript as follows: sucr = sucrose, fruc = fructose, malt = maltose, mele = melezitose, gluc = glucose, sbtl = sorbitol, srbs = sorbose, treh = trehalose, lact = lactose, xylo = xylose, mann = mannose.

Figure 2

Nutritional value of sugars corresponds with palatability (A and B) Feeding behavior over 2 min varies by sugar type. Bees consumed more of the sugar solutions at the highest concentrations (top, 100 mM, (A), Kruskal-Wallis, ${\text{Χ}}^2$₁₁ = 48.5, p < 0.0001, n = 10 to 13; 1000 mM, (B) ${\text{Χ}}^2$₁₁ = 69.4, p < 0.0001, n = 8 to 13 per sugar) and spent more time in contact with the most phagostimulatory sugars over the first bout (middle, 100 mM, ${\text{Χ}}^2$₁₁ = 25.8, p < 0.05, n = 10 to 13; 1000 mM, ${\text{Χ}}^2$₁₁ = 42.7, p < 0.0001, n = 8 to 13 per sugar) and entire 2-min period (bottom, 100 mM, ${\text{Χ}}^2$₁₁ = 38.4, p < 0.0001, n = 10 to 13; 1000 mM, ${\text{Χ}}^2$₁₁ = 69.5, p < 0.0001, n = 8 to 13 per sugar). Blue letters denote 500 mM concentrations. Boxplots show median and interquartile range. Asterisks represent significant differences in sugar consumption versus water (Wilcoxon rank-sum multiple comparisons). (C) The probability of eliciting the proboscis extension reflex depended on sugar type (Kruskal-Wallis, ${\text{Χ}}^2$₁₀ = 36, p < 0.0001, n = 5 per sugar). (D) Survival of bumblebees fed with 1000 mM (or 500 mM for lactose, maltose, and melezitose) solutions depended on sugar type (Kaplan-Meier Log-Rank statistic = 1576, df = 11, p < 0.0001). Holm-Sidak post hoc analysis showed significant differences in survival between the following groups of sugars (from highest to lowest survival): (1) maltose; (2) fructose, sucrose, melezitose; (3) glucose, sorbitol, trehalose; and (4) sorbose, lactose, xylose, mannose, and water. (n = 33 to 42 bees per treatment). (E) Total consumption of sugar solutions per 24 h period over the 10-day longevity assay in (D).

Figure 3

GRN3 responses are not affected by the addition of secondary compounds in fructose (A) Representative tip-recordings from galeal A-type sensilla stimulated with 100 mM fructose or 100 mM fructose plus 1 mM bitter compound (quinine, caffeine), salt (KCl, NaCl), or amino acid (glutamate, proline, phenylalanine). Individual sensilla were tested with all compounds (in total 24 sensilla, n = 8 bees). (B) Average GRN3 spike rates (over 1 s stimulation) to fructose are not affected by the addition of secondary compounds (linear mixed effects, F_6,267 = 0.452, p > 0.05). Responses to quinine are not included; quinine shuts down all GRN responses. Boxplots indicate median and interquartile range, and violin plots denote the density distribution. (C) Firing rates (spikes per 100 ms bin) of GRN1 versus GRN2 over 1 s of stimulation with fructose and fructose mixtures. Points represent mean spikes/s in each bin across all trials for a given sugar, connected with lines over time (over 10 bins). GRN2 significantly predicts the trajectory of GRN1 over time (GLM, GRN2: F_1,65 = 786.1, p < 0.0001; sugar F_6,65 = 36.0, p < 0.0001). (D) Responses to fructose and fructose mixtures cluster in two groups (with or without quinine) using a consensus clustering analysis of the binned GRN responses (Figure S3).

Figure 4

GRN responses to sucrose are attenuated by the addition of nonnutritive secondary compounds (A) Bees fed with sucrose mixed with a nonnutritive sugar (500 mM equimolar solutions) over 10 days did not survive if the mixture included mannose and lactose but did not have a different rate of mortality if mixed with sorbose. Those fed with sucrose mixed with xylose were more likely to survive than the control. Letters denote significant differences in survival (Kaplan-Meier log-rank statistic = 370.9, df = 4, p < 0.001, n = 20 to 23 bees per sugar). (B) Feeding behavior over 2 min did not vary in the total volume consumed with the addition of a nonnutritive sugar (ANOVA: F_5,88 = 1.66, p = 0.15, n = 15 to 17 per sugar solution). Boxplots denote median and interquartile range. (C) Over 24 h, the quantity of food consumed per bumblebee was significantly less if the bees had been fed with the sucrose-mannose mixture (ANOVA with Tukey multiple comparisons, F_5,18 = 10.53, p < 0.001, n = 4 cages, 20 bees per treatment). Asterisks represent significance versus sucrose alone. Boxplots denote median and interquartile range. (D) Average GRN1 (left) and GRN2 (right) firing rates following stimulation with sucrose (500 mM) and equimolar (500 mM) sugar mixtures, including a sucrose-fructose mixture (Aligned Rank Transform, GRN1: F_5,263 = 26.8, p < 0.0001; GRN2: F_5,258 = 19.7, p < 0.0001). Violin plots show the distribution of responses of individual sensilla, with boxplots overlaid denoting median and interquartile range. Asterisks denote significant differences of GRN responses to the mixture versus sucrose (red boxes). (E) Firing rates (spikes per 100 ms bin) of GRN1 versus GRN2 over 1s of stimulation with sucrose and sucrose mixtures. Points represent mean spikes/s in each bin across all trials for a given sugar, connected with lines over time (over 10 bins). GRN2 significantly predicts the trajectory of GRN1 over time; slopes do not vary significantly between sugars (GLM, GRN2: F_1,48 = 195.7, p < 0.0001; sugar F_5,48 = 36.8, p < 0.0001, GRN2∗sugar F_5,48 = 214.9, p > 0.05). (F) The consensus clustering analysis of the binned GRN responses to sucrose and sucrose mixtures did not partition the stimuli into separate groups.