Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Mar 2;12:48.
doi: 10.3389/fncel.2018.00048. eCollection 2018.

Precision of Classification of Odorant Value by the Power of Olfactory Bulb Oscillations Is Altered by Optogenetic Silencing of Local Adrenergic Innervation

Affiliations
Free PMC article

Precision of Classification of Odorant Value by the Power of Olfactory Bulb Oscillations Is Altered by Optogenetic Silencing of Local Adrenergic Innervation

Daniel Ramirez-Gordillo et al. Front Cell Neurosci. .
Free PMC article

Abstract

Neuromodulators such as noradrenaline appear to play a crucial role in learning and memory. The goal of this study was to determine the role of norepinephrine in representation of odorant identity and value by olfactory bulb oscillations in an olfactory learning task. We wanted to determine whether the different bandwidths of olfactory bulb oscillations encode information involved in associating the odor with the value, and whether norepinephrine is involved in modulating this association. To this end mice expressing halorhodopsin under the dopamine-beta-hydrolase (DBH) promoter received an optetrode implant targeted to the olfactory bulb. Mice learned to differentiate odorants in a go-no-go task. A receiver operating characteristic (ROC) analysis showed that there was development of a broadband differential rewarded vs. unrewarded odorant-induced change in the power of local field potential oscillations as the mice became proficient in discriminating between two odorants. In addition, the change in power reflected the value of the odorant rather than the identity. Furthermore, optogenetic silencing of local noradrenergic axons in the olfactory bulb altered the differential oscillatory power response to the odorants for the theta, beta, and gamma bandwidths.

Keywords: associative learning; local field potential; noradrenaline; olfaction; optogenetics.

Figures

Figure 1
Figure 1
Behavioral performance of the mice in the go-no go task. (A1–A3) Example of the percent correct performance in the go-no go odorant discrimination task for a mouse learning to differentiate odorants in the APEB odorant pair. Percent correct was calculated in a window of 20 trials. Blue points are percent correct below 65% and red points are above 80%. (B–D) Mean and 95% confidence interval for the percent correct performance in the first thirty trials of the first training session (blue, naïve) and the last thirty trials of the last trainings session (red, proficient). Data for individual mice are shown in light grey. The difference in percent correct performance between naïve and proficient was statistically significant for all odorant pairs (ranksum test, p < 0.001, pFDR = 0.05). (E–G) Time course of the mean and 95% confidence intervals for the number of licks per second calculated for the last thirty trials of the last session for the same mice. The red line denotes the interval when the odorant was applied for 2.5 s. The odorant pairs and number of mice are: (B,E) IAMO, 5, (C,F) APEB, 8, (D,G) EAPA, 5.
Figure 2
Figure 2
The rewarded (S+) and unrewarded (S–) odorants elicit broadband increases and decreases in LFP power in the olfactory bulb in mice proficient in the go-no go odorant discrimination task. Panels (A1,A2) show representative examples of the raw traces of the olfactory bulb LFP in response to the S+ (A1, top) and S– (A2, bottom) odorants for the APEB odorant pair. The mouse was exposed to the odorant for 2.5 s (black bar). In A to F the S+ odorant was 1% acetophenone diluted in mineral oil (AC) and the S stimulus was 1% ethylbenzoate (EB) (APEB odorant pair). Panels (B1,B2) shows an example of the time course for the average Δ power spectrogram (in decibels) for the response to the S+ (B1, top) and S– (B2, bottom) odorants for the last 30 trials in the last training session where this animal reached >80% percent correct responses (“proficient”). Δ power used for panels (C–G) was calculated as the average power during the first 2 s of odorant application, computed in a logarithmic scale, in a sliding 1 s window during odorant application minus the average power in the interval from 2.1 to 0.6 s before exposure to the odorant. Panel (C) shows the average Δ power spectrum calculated during odorant application for frequencies spanning theta to high gamma calculated from LFP measurements in 16 electrodes in the last 30 trials of the last training session for six mice. The shadow displays a 95% confidence interval. Panels (D1–D4) shows for four different bandwidths ranging from theta to high gamma the histogram for the number of LFP recordings per electrode at each average Δ power and scatter plots for the average Δ power calculated in the last 30 trials for each LFP recorded from each electrode in the six mice discriminating odorants in the APEB odorant pair. In the plot on the right side of the histogram the solid line shows the average Δ power. Panels (E1–E4) shows examples of the receiver operating characteristic (ROC) graphs (Fawcett, 2006) estimated from the S+ and S– Δ power distributions for the last 30 trials of the last go-no go session for training a mouse to differentiate between AC (S+) and EB (S–). The blue points are the true positive and false positive rates for each trial and the red line is a best fit of the ROC curve. The area under the ROC (auROC, defined here to fall between −0.5 and 0.5) was significantly different from zero (the diagonal) using a z-test (Cardillo, 2008) (p < 0.05). Panels (F1–F4) shows the histogram of the auROCs calculated for the last 30 trials in the last go-no go training session for each electrode LFP in each bandwidth for all electrodes recorded from in the six mice (all S+ and S– trials were included in the ROC calculation). Significant auROCs are shown in light brown, and auROCs that were not statistically significant are shown in light blue. auROC significance was tested using the z-test and the p-values were corrected for multiple comparisons by calculating the significance p-value corrected for the false discovery rate (Curran-Everett, 2000) (pFDR = 0.04). Panels (G1–G4) show the percent of single electrode LFP auROCs significantly different from zero for the different odorant pairs used in this study. The ROCs for the blue bars were calculated using all S+ and S– trials, the ROCs for the red bars were calculated using only Hit and CR trials. S+/S– odorants: IAMO: 1% isoamyl acetate/mineral oil, APEB: 1% acetophenone/1% ethylbenzoate and EAPA: 0.1% ethyl acetate/0.05% ethyl acetate + 0.05% propyl acetate. *The p-value for a Chi-Squared testing for the difference in the number of significant LFPs is smaller than the pFDR = 0.037). Number of mice: IAMO: 5, APEB: 8, EAPA: 5, 16 electrodes per mouse). The behavioral performance is shown in Figure 1.
Figure 3
Figure 3
The difference between the rewarded and unrewarded odorant-elicited change in LFP power develops as the mice become proficient in learning to discriminate between the odorants. Panels (A1–A4) show an example of the increase in the odorant-induced change in power (Δ power) elicited by the reinforced odorant as a mouse became proficient in discriminating between the rewarded (IA: 1% isoamyl acetate) and unrewarded (mineral oil: MO) odorants. Panels (A1,A2) are the spectrograms for the average Δ power in decibels for first 30 trials in the first go-no go session where mice learned to discriminate between IA (S+, A1) and MO (S–, MO)(naïve period). Panels (A3,A4) are the average Δ power spectrograms for the last 30 trials in the last session for S+ (A3) and MO (A4) (proficient period). Black bar: duration of odorant application. Panels (B1–B4) are histograms (left) and scatter plots (right) for the auROCs for the average Δ power elicited by S+ vs. the average Δ power elicited by S– for the different bandwidths for the naïve period for IA vs. MO (blue) and proficient period (red) for LFPs measured in 16 electrodes in five mice. The difference in Δ power auROCs between learning and proficient is statistically significant for all bandwidths when tested using a non-parametric permutation based ANOVA (Delorme and Makeig, 2004) (p < 0.001, pFDR = 0.05). Panels (C1–C4) are auROCs for the average Δ power during naïve and proficient periods for the different bandwidths for 0.1% ethyl acetate/0.05% ethyl acetate + 0.05% propyl acetate (EAPA odorant pair). The difference in Δ power auROCs between naïve (blue) and proficient (red) is statistically significant for all bandwidths when tested using a non-parametric permutation based ANOVA (Delorme and Makeig, 2004) (p < 0.001, pFDR = 0.05). Panels (D1–D4) show the percent of single electrode LFP auROCs significantly different from zero for the different odorant pairs used in this study. S+/S– odorants: IAMO: 1% isoamyl acetate/mineral oil, APEB: 1% acetophenone/1% ethylbenzoate and EAPA: 0.1% ethyl acetate/0.05% ethyl acetate + 0.05% propyl acetate. *The p-value for a Chi-Squared testing for the difference in the number of significant LFPs is smaller than the pFDR (pFDR = 0.037). Number of mice: IAMO: 5, APEB: 8, EAPA: 5, 16 electrodes per mouse. The auROC histograms are not shown for APEB. The behavioral performance is shown in Figure 1.
Figure 4
Figure 4
ROC analysis of the odorant-elicited change in LFP power (Δ power) for trials when the animal responds by licking to the unrewarded odorant (false alarm, FA). ROC analysis of the per trial LFP Δ power calculated when the animals were performing >80% correct. ROC was calculated for Δ power calculated in FA trials vs. Δ power calculated in correct rejection (CR) trials where the animal refrained for licking to the unrewarded odorant (FA/CR) or for Δ power calculated in Hit trials where the animal was licking for the rewarded odorant vs. Δ power calculated in CR (Hit/CR). (A,B) IAMO odorant pair, (C,D) EAPA odorant pair. (A,C) Histograms for LFP Δ power in decibels for Hits (red), CRs (blue) or FAs (green). (B,D) ROC analysis for Hit/CR (red) or FA/CR (green). All auROCs were significantly different from the diagonal (p < 0.0001, pFDR = 0.05). auROC significance was tested using the z-test and the p-values were corrected for multiple comparisons by calculating the significance p-value corrected for the false discovery rate (Curran-Everett, 2000).
Figure 5
Figure 5
The odorant-elicited change in LFP power (Δ power) reverses polarity when the rewarded odorant is reversed. (A1) Percent correct discrimination as a function of trial number for a mouse becoming proficient in differentiating 0.1% ethyl acetate (EA) as the rewarded odorant from a mixture of 0.05% ethyl acetate and 0.05% propyl acetate (EA+PA) as the unrewarded odorant (forward session). (A2) Percent correct discrimination as a function of trial number for the same mouse learning to differentiate these two odorants after the reward was shifted to EA+PA (reversed session). Percent correct was computed in a window of 20 trials. The points are blue when percent correct <65% and red when percent correct >80%. (B) Percent correct for the last 30 trials for the forward and reversed go-no go sessions for five different mice. A ranksum test indicates that the percent correct was different between the forward and reversed sessions (p = 0.04). (C,D) Histograms and point plots of the mean LFP Δ power computed in the last 30 trials of the forward and reversed go-no go sessions for EA (C) and EA+PA (D) in all bandwidths (LFP was recorded from 16 electrodes in 5 mice). A permuted ANOVA test of the difference in LFP Δ power between the forward and reversed sessions yielded a p < 0.0001 for all bandwidths and for both odorants (pFDR = 0.05).
Figure 6
Figure 6
Optogenetic silencing of norepinephrine axons in the olfactory bulb does not alter the broadband odorant-elicited change in LFP power (Δ power). (A1–A4) Example of the Δ power spectrograms recorded in one go-no go session with a DBH-Cre eNpHR3.0 mouse for S+ (0.1% ethyl acetate, A1, A3) and S– (0.05% ethyl acetate + 0.05% propyl acetate, A2,A4) for 20 trials before (A1,A2) and 20 trials during (A3,A4) local optogenetic activation in the OB of eNpHR3.0 expressed in noradrenergic axons. The laser was activated for 3.5 s starting at the time the mouse entered the odorant port, 1–1.5 s before odorant application. Black bar: Duration of odorant stimulation. (B1–B4) Examples of auROCs for Δ power for S+ (0.1% ethyl acetate) and S– (0.05% ethyl acetate + 0.05% propyl acetate) (EAPA odorant pair) for 20 trials before (B1,B3) and 20 trials during (B2,B4) local optogenetic activation of eNpHR3.0 expressed in noradrenergic axons in the OB. (B1) auROC for beta LFP Δ power before laser stimulation, (B2) auROC for beta LFP Δ power during laser stimulation. (B3) auROC for high gamma LFP Δ power before laser stimulation and (B4) auROC for high gamma LFP Δ power after during laser stimulation. (C1,C2) Histograms (left) and scatter plots (right) for Δ power auROCs for theta bandwidth before (Pre L, blue) and during (Laser, red) laser stimulation for the APEB odorant pair. (C1,C2, left) DBH-Cre mice (n = 4, 16 LFP electrodes each), (C1,C2, right) DBH-Cre eNpHR3.0 mice (n = 6, 16 LFP electrodes each). The auROC in (C1) was calculated using LFP Δ power recorded in S+ vs. S– trials. The auROC in (C2) was calculated using LFP Δ power recorded in Hit vs. CR trials. The interaction term of an N-way ANOVA with mouse genotype and laser stimulation as the two factors was significant for the S+/S– auROC (C1) (p = 0.016 < pFDR = 0.031), but was not significant for the Hit/CR auROC (C2) (p = 0.19 >pFDR = 0.031). (D) Expression of EYFP in the locus coreuleus of DBH-Cre eNpHR3.0 mice. (E) Performance of mice in the go-no go odorant discrimination task for the last 20 trials in the last training session (blue, Pre L), and in the first 20 trials in the subsequent training session where the laser was turned on for the duration of odorant delivery (red, Laser). Results are shown for the three odorant pairs for DBH-Cre mice (E1), and DBH-Cre eNpHR3.0 mice (E2). A paired t-test yields a significant difference in performance between the two sessions for DBH-Cre eNpHR3.0 mice discriminating between the IAMO odorants (p = 0.002, pFDR = 0.008, n = 4). The number of DBH-Cre mice tested was 5 for IAMO, 4 for APEB, and 8 for EAPA, and the number of DBH-Cre eNpHR3.0 mice was 4 for IAMO. Six for APEB and 4 for EAPA.
Figure 7
Figure 7
Optogenetic silencing of norepinephrine axons in the olfactory bulb elicits statistically significant changes in the auROC for the odorant-elicited change in power of the lick-related LFP (Δ power LR-LFP). (A) Examples of traces showing phase locking of the theta LFP filtered at 6–12 Hz with licks for three trials when the animal was proficient in differentiating odorants for the APEB odorant pair in the go-no go task. Licks were detected as an increase in voltage elicited when the tongue touched the lick spout. The black bar shows the duration of odorant application (2.5s). (B) Mean LFP and 95% confidence intervals, computed by bootstrapping, recorded when the LFP was triggered by the onset of the lick, for all licks occurring for 2 s after odorant application for 30 trials for the mouse whose raw traces are shown in (A) (B1, S+, B2 S–). The lick-locked LFP is shown for the time interval from −0.5 to 0.5 s centered at lick onset (lick-related LFP, LR-LFP). (C) Spectrogram for the odorant-induced change in LR-LFP power for S+ (C1) and S– (C2) for these 30 trials. (D) Theta LFP phase of the lick onset for S+ (D1) and S– (D2) for these 30 trials. (E1–E4) The auROC for Δ power LR-LFP computed for 30 S+ and S– trials increases for all bandwidths between the first 30 trials of the first go-no go training session (naïve, blue) and the last 30 trials of the last training session (proficient, red) for animals learning to discriminate the APEB odorant pair. (A1) theta, (A2) beta, (A3) low gamma, (A4) high gamma. For each panel a histogram of auROC values is shown on the left, and a scatter plot is shown on the right. For all bandwidths the p-value for a permuted N-way ANOVA testing for the difference in auROC between naïve and proficient was <0.0001, pFDR = 0.05 (9 mice, 16 electrodes each). (F1–F3) The percent of significant auROCs for Δ power LR-LFP computed for 30 S+ and S–trials increases for all bandwidths between the first 30 trials of the first go-no go training session (naïve, blue) and the last 30 trials of the last training session (proficient, red) for animals learning to discriminate the IAMO (F1), APEB (F2), and EAPA (F3) odorant pairs (percent significant auROC is shown for all bandwidths). The number of significant auROCs differed between naïve and proficient trials for all odorant pairs and all bandwidths when tested with a Chi Squared test (p < pFDR = 0.05, number of mice 7 for IAMO, 9 for APEB, and 9 for EAPA, 16 electrodes per mouse). (G) Examples of the effect of optogenetic silencing of the noradrenergic fibers in the OB on the Δ power LR-LFP auROCs calculated with Hits and CRs for the last 30 trials in the session where the animal was proficient in differentiating odorants for the APEB odorant pair (Pre L, blue) and the first 30 trials in a subsequent session where light was applied for 3.5 s starting when the animal entered the port (Laser, red). Shown are the histograms (left) and scatter plots (right) for Δ power LR-LFP auROCs for theta (G1) and beta (G2) bandwidths before (Pre L, blue) and during (Laser, red) laser stimulation for the APEB odorant pair. Data are shown for DBH-Cre mice (C1,C2, left, n = 4, 16 LFP electrodes each) and DBH-Cre eNpHR3.0 mice (C1,C2, right, n = 6, 16 LFP electrodes each). The auROC in panels (C1,C2) was calculated using LR-LFP Δ power recorded in Hit and CR trials 0.3 s after the onset of the lick. The interaction term of an N-way ANOVA with mouse genotype and laser stimulation as the two factors was significant for the Hit/CR auROC for both (C1,C2) (p ≤ 0.0001, <pFDR = 0.022). H. Difference between genotypes of the change in auROC elicited by optogenetic silencing of local noradrenergic fibers (ΔauROC = change in auROC for DBH-Cre eNpHR3.0 mice—change in auROC for DBH-Cre mice). Shown are the mean and the estimate of the 95% confidence intervals for the ΔauROC for all bandwidths for IAMO (H1), APEB (H2), and EAPA (H3). Data for the different bandwidths are shown in different colors: theta (blue), beta (red), low gamma (purple), and high gamma (black). The LR-LFP Hit/CR auROC was computed as shown for examples in panel (G). In H2 arrows points to ΔauROC data points corresponding to the data shown in panels (G1) (theta, APEB) and (G2) (beta, APEB). Asterisks denote ΔauROCs found to be significant for the interaction term of an N-way ANOVA with mouse genotype and laser stimulation as the two factors (p ≤ pFDR, pFDR = 0.015 for IAMO, 0.022 for APEB and 0.001 for EAPA, the number of DBH-Cre mice was 5 for IAMO, 4 for APEB, and 8 for EAPA and the number of DBH-Cre eNpHR3.0 mice was 4 for IAMO, 6 for APEB, and 4 for EAPA, 16 electrodes per mouse).

Similar articles

See all similar articles

Cited by 3 articles

References

    1. Amarante L. M., Caetano M. S., Laubach M. (2017). Medial frontal theta is entrained to rewarded actions. J. Neurosci. 37, 10757–10769. 10.1523/JNEUROSCI.1965-17.2017 - DOI - PMC - PubMed
    1. Aston-Jones G., Waterhouse B. (2016). Locus coeruleus: From global projection system to adaptive regulation of behavior. Brain Res. 1645, 75–78. 10.1016/j.brainres.2016.03.001 - DOI - PMC - PubMed
    1. Beshel J., Kopell N., Kay L. M. (2007). Olfactory bulb gamma oscillations are enhanced with task demands. J. Neurosci. 27, 8358–8365. 10.1523/JNEUROSCI.1199-07.2007 - DOI - PMC - PubMed
    1. Bouret S., Sara S. J. (2005). Network reset: a simplified overarching theory of locus coeruleus noradrenaline function. Trends Neurosci. 28, 574–582. 10.1016/j.tins.2005.09.002 - DOI - PubMed
    1. Buzsáki G. (2010). Neural syntax: cell assemblies, synapsembles, and readers. Neuron 68, 362–385. 10.1016/j.neuron.2010.09.023 - DOI - PMC - PubMed

LinkOut - more resources

Feedback