Article Figures & Data
Figures
- Figure 1.
Experimental paradigm and behavioral results. A, An experimental schedule. In the pre-training (PRE) and post-training (POST) phases, two odors were applied in pseudo-random order, at least five times each. The training phase (TRAIN) started with the sucrose-rewarded odor (CS+), alternating with the unrewarded odor (CS-). Ca2+ imaging was performed during PRE and POST phases. Intervals between odor stimuli and between training repetition cycles were both 90 s. B, The numbers of animals trained with the specified odors for CS+ and CS-. Different odors were equally used as rewarded and unrewarded stimulus. C, Behavioral data for all bees (N = 18) during TRAIN (trials 1–10) and POST (trials 11–20) phases. The probability of CRs by a population [p(CR)] rapidly increased and saturated after two pairings of odor and sugar. A small fraction of bees initially generalized toward CS-.
- Figure 2.
Bouton responses and their temporal properties. A, Frontal view of the honeybee brain with schematic overlay: uniglomerular PNs (yellow) connect the AL with the MB calyx (Ca) via the l-ALT (arrowhead). d: dorsal, l: lateral. B, A confocal image of stained brain viewed frontally. PN somata were stained with a mixture of the fura-dextran and the rhodamine dextran. Arrow, a site of injection. Scale bars, 100 μm. C, A confocal image of rhodamine fluorescence in the MB calyx. Individual boutons are visualized. Scale bar, 20 μm. D, An example of identified locations of synaptic boutons based on their Ca2+ responses (bee #11, also in Figs. 3, 4). E, Color-coded Ca2+ responses superimposed on raw fluorescence images of PN boutons for two trials during PRE (left) and POST (right) phases in response to rewarded (upper row) and unrewarded (lower row) odors. The same animal as in D. 8on: 2-octanone, 8al: octanal. F, An example of temporal dynamics of Ca2+ activity for individual boutons (heat map) and their average traces (black curve) for CS+ (upper) and CS- (lower) odors in a representative bee #11. Changes between PRE (left) and POST (right) phases were also calculated per boutons (middle). Data were transformed by the hyperbolic tangent function (Extended Data Fig. 2-1; see also Materials and Methods). Individual bouton traces show the trial-averaged activities based on signal quality check (Extended Data Figs. 2-2, 2-3).
- Figure 3.
CS+ evoked responses in all recorded boutons. Color-coded CS+ odor responses of all 1652 PN boutons recorded from 18 bees. Ca2+ activity of individual boutons in PRE (left) and POST (right) phases as well as their differences (ΔNR, middle) are shown. Black horizontal lines separate boutons of different animals, whose IDs are given as ordinate labels. Odors were applied between 0 and 3 s. Measures of neural plasticity and SNRs were calculated on the basis of all time steps between 0 and 4 s (enclosed by red horizontal lines). 8ol: octanol, 6ol: hexanol.
- Figure 4.
CS- evoked responses in all recorded boutons. Color-coded CS- odor responses of all 1652 PN boutons recorded from 18 bees. Same as in Figure 3.
- Figure 5.
Temporal profiles of odor-evoked PN bouton activity. A, Time courses of the mean excitatory responses (CS+ and CS- pooled) of PRE (black) and POST (gray) phases with previously published Ca2+ imaging data of PNs (red; Peele et al., 2006; n = 112). B, Time courses of the mean inhibitory response (black) and the difference in excitatory responses (ΔNR/PRE-POST, gray) with previously published Ca2+ imaging data of MB feedback neurons (red; Haehnel and Menzel, 2010; n = 30). C, Mean latencies of excitatory and inhibitory responses relative to valve opening of the odor stimulation device.
- Figure 6.
Positive correlation between a conditioned behavior and bouton plasticity. A, B, Scatter plots of behavioral performance during training phase (ΔCR) against neural plasticity (ΔNR) for CS+ (A) and CS- (B) odors. C, Jackknife resampling of Spearman's rank order correlations (ρ) on high SNR (n = 8 for CS+, n = 10 for CS-) and full data (n = 18) to estimate robustness and dependence on outliers. D, Corresponding p values of C. Thick dashed line indicates a level of significance of 0.05.
Tables
Types of excitatory odor responses Types of response changes Stimulus Only ON Only OFF ON and OFF None Only up Only down Both None CS+ 39.5 3.3 9.9 47.2 14.2 33.0 7.8 45.0 CS- 40.9 2.7 12.3 43.6 18.7 31.4 5.7 44.2 Both 15.8 0.0 0.8 16.6 2.8 11.0 0.6 20.7 Types of excitatory odor responses of untrained animals: all 1652 boutons were categorized by their response profiles of (1) only ON responses, (2) only OFF responses, (3) both ON and OFF responses, or (4) no detected responses to CS+ and/or CS- odors. Types of excitatory bouton response changes by learning: all 1652 boutons were categorized by their response profiles of (1) only up-regulated, (2) only down-regulated, (3) both up- and down-regulated, or (4) no detected change to CS+ and/or CS- odors.
Extended Data Figure 2-1
Transformation of Ca2+ responses. A, Tangens hyperbolicus transformation of ΔF/F signal. Black curve represents transformation across the total range of ΔF/F values encountered in all boutons of bee #9. B, Transformation of exemplified single bouton activities. Black shades indicate odor stimulation for 3 s. Large excitatory response values are compressed by the transformation while smaller excitatory and inhibitory response values around zero are enhanced. C, Color code visualization of all 107-bouton response in bee #9 in response to the CS+ odor before (PRE) and after (POST) training and the difference (POST-PRE). The top row shows the untransformed ΔF/F response signals. The color map spans the complete data range across all boutons as shown in A, top. The bottom row shows the transformed tanh (ΔF/F) response signals. The color map spans the complete range of transformed data across all boutons as shown in A, right. This leads to a compression of strong excitatory responses (reddish colors) and increases the dynamic color range for weaker excitatory and inhibitory responses. In effect, we observe clear excitatory and inhibitory responses across many boutons. The same classical “jet” 64-bit color map (as shown in A) is used for upper and lower row. Download Figure 2-1, TIF file.
Extended Data Figure 2-2
Response statistics within and across experimental phases. A, Distributions of pairwise Euclidean distances of odor responses of different trials within and between different experimental phases. More similar bouton activity patterns in the same experimental phase than between different phases. Gray areas show smoothed distribution densities of Euclidean distances. Boxplots show median, lower to upper quartiles with whiskers extending to the most extreme data points. Two-sided pairwise KS statistics applied. B, Distributions of pairwise Euclidean distances of odor responses within the same animal and between different animals. More similar boutons activity pattern within the same animal than between different animals. C, SNRs sorted according to odor type. Histogram of SNRs by odors. A Gaussian kernel density estimation of the underlying distribution was added. The orange line indicates the median of all SNRs that was used as a threshold to separate high and low SNRs. Download Figure 2-2, TIF file.
Extended Data Figure 2-3
Decorrelation of spatial bouton response pattern. A, Linear correlation coefficient R between the bouton response pattern to the CS+ odor and the CS- odor during the first second after stimulus onset PRE and POST training for all 18 bees. B, The average correlation coefficient is smaller (p = 0.049, Wilcoxon signed rank test) after training (POST). C, Average correlation coefficients before (PRE) and after (POST) training during spontaneous activity are close to zero and do not change from PRE to POST. Download Figure 2-3, TIF file.
In this issue
