Activation of raphe nuclei triggers rapid and distinct effects on parallel olfactory bulb output channels

Abstract

The serotonergic raphe nuclei are involved in regulating brain states over timescales of minutes and hours. We examined more rapid effects of raphe activation on two classes of principal neurons in the mouse olfactory bulb, mitral and tufted cells, which send olfactory information to distinct targets. Brief stimulation of the raphe nuclei led to excitation of tufted cells at rest and potentiation of their odor responses. While mitral cells at rest were also excited by raphe activation, their odor responses were bidirectionally modulated, leading to improved pattern separation of odors. In vitro whole-cell recordings revealed that specific optogenetic activation of raphe axons affected bulbar neurons through dual release of serotonin and glutamate. Therefore, the raphe nuclei, in addition to their role in neuromodulation of brain states, are also involved in fast, sub-second top-down modulation similar to cortical feedback. This modulation can selectively and differentially sensitize or decorrelate distinct output channels.

Figure 1. Raphe stimulation excites mitral and tufted cells at rest

a) Histological section showing an electrolytic lesion in the raphe nucleus introduced with the stimulating electrode at the end of an experiment. b) Three exemplar trials showing breathing traces measured with a chest strap around the time of raphe stimulation (three 1ms pulses at 10Hz; orange vertical bars). Upward deflections correspond to inhalation. c) Peristimulus time histogram (PSTH) of breathing rate across 5 animals. Orange bar denotes time of raphe stimulation and black lines are SEM. d) Resting fluorescence image of TCs in an exemplar experiment shown in (e). e) Average trace showing TC response to raphe activation (purple). Time course of fractional fluorescence intensity for all cells in the field of view in (d) in response to raphe activation (orange bars). Cell 15 is plotted at top. f) Bar plot of all imaged TCs showing fluorescence change in trials with raphe stimulation when compared to blank trials with no raphe stimulation (288 cells, 12 mice, p = 9.80×10⁻²⁹, Wilcoxon signed-rank). g) Resting fluorescence image of MCs in one experiment, analyzed in (h),(i). h) Average trace showing an example MC responding to activation of raphe (bluish-green). Time course of fractional fluorescence intensity for all cells in the field of view in (g) in response to raphe activation (orange bar). Cell 7 is plotted at top. i) Bar plot of fluorescence changes in all imaged MCs for trials with raphe stimulation compared to blank trials with no stimulation (238 cells, 13 mice, p= 4.68×10⁻¹⁰, Wilcoxon signed-rank).

Figure 2. TC odor responses are sensitized by raphe inputs

a) 2-photon image showing resting fluorescence of GCaMP6s in TCs. b) Time course of florescence responses of two TCs to raphe stimulation (black), odor stimulation (light green) and raphe and odor stimulation (purple). Bars denote odor timing (light green) and raphe stimulation timing (orange). Error bars are SEM. c) Time course of responses to 6 odors in 62 TCs in the imaged region shown in (a). Light green and orange bars at bottom indicate the timing of the odor and raphe stimulation, respectively. d) Heat map representing the double rank ordered odor responses without (light green) and with (purple) raphe activation. Each row indicates the odor response amplitudes of single cells, with the rows themselves rank ordered in increasing amplitude of responses (summed across all odors). The order of cells was determined in the left panel and was maintained for the right panel (with raphe stimulation) for direct comparison. Bar plots on the right show the normalized summed activity of single cells and bar plots at the bottom show normalized summed activity for odors. e) Bar plot (top) showing the change in the number of odors that elicited responses without and with raphe stimulation for cells shown in c (372 cell-odor pairs, p = 7.50×10⁻¹³, Wilcoxon signed-rank). Bar plot (bottom) showing the change in the number of odors that elicited responses without and with raphe stimulation for all cell odor pairs from 6 animals (1087 cell-odor pairs, p = 1.68×10⁻¹⁰, Wilcoxon signed-rank). f) Cumulative distribution of the change in florescence for TCs without raphe stimulation (light green) or with raphe stimulation (purple) for all the cell odor pairs shown in c and d (top). Same for TCs without (black) or with raphe stimulation (purple) for all the cell odor pairs (n=1087) across 6 animals (bottom). g) Scatter plots showing change in odor response for each TC when the raphe is stimulated (compared to odor response on its own), as a function of raphe response (left) and odor response (right) for that TC. Each dot is a single cell odor pair. h) Three dimensional scatter plot showing the responses for TCs (372 odor cell pairs) for odor and raphe stimulation. The prediction surface, shown as a grid, is from the interaction based regression model (see Methods). i) Scatter plot showing the predicted based on interactive model vs observed responses for odor and raphe stimulation (colors depict different odors, R2 = 0.88). j)Scatter plot showing the predicted vs observed responses for odor and raphe stimulation (colors depict different animals, R2 = 0.86 for n=6 animals, 1087 cell-odor pairs, p =1.29× 10⁻²¹, F test).

Figure 3. MC odor responses are bidirectionally modulated by raphe inputs

a) 2-photon image showing resting fluorescence of GCaMP6s in MCs. b) Time courses of florescence of two MCs in response to raphe stimulation (black), odor stimulation (light green) and raphe and odor stimulation (bluish-green). Bars are odor timing (light green) and raphe stimulation timing (orange). Error bars are SEM. c) Time course of responses to 6 odors in 47 MCs in the imaged region. Green and red bars at bottom indicate the timing of the odor and raphe stimulation, respectively. d) Heat map representing the double rank ordered odor responses without (light green) and with (bluish-green) raphe activation. Each row indicates the odor response amplitudes of single cells, with the rows themselves rank ordered in increasing amplitude of responses (summed across all odors). The order of cells was determined in the left panel and was maintained for the right panel (with raphe stimulation) for direct comparison. Bar plots on the right show the normalized summed activity of single cells and bar plots at the bottom show normalized summed activity for odors. e) Bar plot (top) showing the change in the number of odors that elicited responses without and with raphe stimulation for cells shown in c (p = 0.61 for n=47 cells for Wilcoxon signed-rank). (Bottom) Bar plot showing the change in the number of odors that elicited responses without and with raphe stimulation for all cell odor pairs (p = 0.85 for 1050 cells from 7 animals, Wilcoxon signed-rank). f) Cumulative distribution of the change in florescence for MCs without (light green) or with (bluish-green) raphe stimulation for all the cell odor pairs shown in c and d (top). Cumulative distribution of the change in florescence for MCs without raphe stimulation (black) or with raphe stimulation (bluish-green) for all the cell-odor pairs (n=1050) across 7 animals (bottom). g) Scatter plots showing change in odor response for each MC when the raphe is stimulated (compared to odor response on its own), as a function of raphe response (left) and odor response (right) for that MC. Each dot is a single cell odor pair. h) Three dimensional scatter plot showing the responses for MCs (282 odor cell pairs) for odor and raphe stimulation (R² = 0.29). Grid shows the prediction manifold based on interaction based regression model (see methods).

Figure 4. Optogenetic activation of raphe leads to excitation of principal cells

a) Confocal image of the raphe fibers innervating different layers of the OB in a TPH2-ChR2:YFP mouse. b) Extracellular spikes from an isolated unit (820 spikes superimposed at top) recorded in the OB are represented in a raster plot (bottom). Twenty trials of dorsal surface stimulation with blue light (timing indicated by blue bar). c) PSTH of activity of a single putative M/TC (left) in the TPH2-ChR2:YFP mouse exhibiting excitation from rest when raphe fibers are activated via blue light (100 ms bins, same for all PSTHs). Population PSTH of all putative M/TCs (right) from extracellular recordings in the TPH2-ChR2:YFP mouse (n=17 cells). Gray shading indicates SEM. d) PSTH of activity in a single putative M/TC in the TPH2-ChR2:YFP mouse reveals excitatory responses to blue light stimulation (top left, timing indicated by the blue bar) and electrical stimulation of the raphe (bottom left, timing indicated by orange bar). Population PSTH of all putative M/TCs both with electrical stimulation (bottom right) and ChR2 stimulation (top right) in the TPH2-ChR2:YFP mouse (n=9 cells). Firing rate in the resting period for each cell was normalized before averaging, and the final average was normalized to the peak for clarity of display. e) Time course of responses in 24 TCs in 3 animals for i) light stimulation (gray), ii) odor stimulation (light green, methyl tiglate) and iii) odor plus light stimulation (purple). Light green and blue bars at bottom indicate the timing of the odor and ChR2 stimulation, respectively. f) Time courses of fluorescence changes in a single TC in response to odor stimulation alone (black) and odor stimulation along with light stimulation of raphe (purple) for 2 different odors. Shading indicates SEM. g) Scatter plot showing change in odor responses caused by concurrent raphe stimulation, as a function of odor response (left). Bar graph (right) showing distribution of changes in odor responses in TCs caused by optical stimulation of the raphe for 144 odor-cell pairs from 3 animals (p = 1.2× 10⁻²³ for Wilcoxon signed-rank test). h) Time course of responses in 7 MCs from 2 animals for i) light stimulation (gray), ii) odor stimulation (light green, ethyl valerate) and iii) odor plus light stimulation (bluish-green). Light green and blue bars at bottom indicate the timing of the odor and ChR2 stimulation, respectively. i) Time courses of fluorescence changes in a single MC in response to odor stimulation alone (black) and odor stimulation along with light stimulation of raphe (bluish-green) for 2 different odors. Shading indicates SEM. j) Scatter plot showing change in odor responses in MCs caused by concurrent raphe stimulation, as a function of odor response (left). Bar graph showing distribution of changes in odor responses in MCs caused by optical stimulation of the raphe for 64 odor-cell pairs from 2 animals (right).

Figure 5. Modulation of TCs by optogenetic activation of raphe projections in main olfactory bulb slices

a) Schematic diagram depicting TC recordings from OB slices. b) (Left) Whole cell voltage clamp recordings from a TC at 0mV and −70 mV showing IPSCs (top) and EPSCs (bottom) with (purple) and without (black) raphe fiber activation. (Right) Average time series for 6 TCs showing EPSCs elicited by brief optogenetic activation of raphe fibers (black arrow). c) (Bottom) Integrated charge transfer in a TC at −70 mV with (purple) optogenetic activation of raphe fibers. Blue bars depict the timing of raphe fiber activation. (Top) change in integrated charge transfer at 0 mV with (purple) optogenetic activation of raphe fibers (Black dotted reference line). d) Summary of all 6 recorded TCs, showing increase in both inhibitory charge transfer (top, p = 3.1×10⁻²⁷ for 6 cells from 4 animals, Wilcoxon signed-rank test) and excitatory charge transfer (bottom, p = 3.3×10⁻³ for Wilcoxon signed-rank test, n = 6 cells from 4 animals) by light stimulation. e) Normalized percentage change in the charge transfer for EPSCs (left, −70mV) and IPSCs (right, 0mV) in presence of different antagonists: 5-HT receptor antagonist (50 μM methysergide, p = 5.01×10⁻⁴ and p = 3.92×10⁻³ for excitatory and inhibitory charge transfer respectively, Wilcoxon rank-sum test, n = 8 cells from 5 animals), glutamate receptor antagonists (20 μM CNQX and 50 μM AP5, p = 7.61×10⁻²⁹ and 2.10×10⁻⁷ for excitatory and inhibitory charge transfer respectively, Wilcoxon rank-sum test, n = 7 cells from 5 animals), and GABA receptor antagonist (20 μM gabazine,, p = 2.0×10⁻³ and p = 1.92×10⁻²¹ for excitatory and inhibitory charge transfer respectively, Wilcoxon rank-sum test, n = 6 cells from 6 animals). Data for some conditions are not visible since they are close to 0 and significance is shown with *. * signifies p<0.05, ** signifies p<0.01 and *** signifies p<0.001. f) Example trace of a TC recorded under current clamp showing spikes in response to raphe fibers activation. Blue bars indicate blue light timing. g) Scatter plots comparing the firing rate of TCs for different step current injection with and without serotonergic fiber activation in presence of different antagonists: no drug (purple, p = 1.21×10⁻³, Wilcoxon rank-sum test, n = 5 cells from 5 animals), glutamate receptor antagonists (20 μM CNQX and 50 μM AP5, black solid circles, p= 1.43×10⁻⁸, Wilcoxon rank-sum test, n=5 cells from 5 animals), 5-HT receptor antagonist (50 μM methysergide, open circles, p= 0.072, n=5 cells from 3 animals, Wilcoxon rank-sum test), and GABA receptor antagonist (20 μM gabazine, squares, p =0.068, n=5 cells from 4 animals, Wilcoxon rank-sum test). Firing rates calculated for an 800 ms period starting at the time of first light stimulation. ** signifies p<0.01. h) Light-triggered changes in firing rates of TCs for no current injection (left, p= 9.21×10⁻³, Wilcoxon rank-sum test, n = 5 cells from 4 animals) and averaged across different levels of current injections (right). The changes for different pharmacological manipulations were normalized to the condition with no drugs applied (ACSF). * signifies p<0.05, ** signifies p<0.01 and *** signifies p<0.001.

Figure 6. Modulation of MC activity by optogenetic activation of raphe fibers

a) Schematic diagram depicting MC recordings from OB slices. b) Plot of the relationship between injected current and firing rate for 2 different MCs showing both inhibitory and excitatory effects of raphe fiber activation. c) Scatter plots comparing the firing rate of MCs for different step current injection with and without raphe fiber activation. Arrow points to trials where light stimulation induced firing in MCs that were quiescent (p= 2.0×10⁻³, Wilcoxon rank-sum test, n = 7 cells from 4 animals). ** signifies p<0.01. d) Whole cell voltage clamp recordings from a MC at 0mV and −70 mV showing IPSCs (top) and EPSCs (bottom) with raphe fiber activation. Blue bars depict the timing of ChR2 fiber stimulation. e) Ratio of integrated excitatory to inhibitory charge transfer in TCs and MCs (n=6 each) following optogenetic stimulation of raphe fibers. f) Scatter plots comparing the firing rate of MCs for different step current injection with and without raphe fiber activation in presence of different antagonists: 5-HT receptor antagonist (50 μM methysergide, open circles, p = 1.1×10⁻¹⁹, Wilcoxon rank-sum test, n = 5 cells from 3 animals), glutamate receptor antagonists (20 μM CNQX and 50 μM AP5, solid circles, p = 7.2×10⁻³¹, Wilcoxon rank-sum test, n = 5 cells from 4 animals), and GABA antagonist (20 μM gabazine, squares, p = 8.79×10⁻³, Wilcoxon rank-sum test, n=5 cells from 4 animals). g) Average light-triggered changes in firing rates of all MCs averaged across different levels of current injections. The changes for different pharmacological manipulations were normalized to the condition with no drugs applied. * signifies p<0.05, ** signifies p<0.01 and *** signifies p<0.001.

Figure 7. ETCs receive direct excitatory inputs from raphe fibers

a) Schematic diagram depicting ETC recordings from OB slices. b) Whole cell voltage clamp recordings from at −70 mV showing short latency EPSCs in ETC (top) and long latencies in TC (middle) and MC (bottom). Timing of light stimulus is shown in blue. – c) Four successive trials from an ETC, showing short latency EPSCs at −70 mV for raphe fiber activation. d) Histograms showing the latencies of EPSCs in ETCs (gray, n=10 cells), TCs (purple, n= 8) and MCs (bluish-green, n=6) in response to optogenetic activation of raphe fibers (p=3.2×10⁻¹⁷ and p = 2.4×10⁻²⁰ for TC and MC respectively, Wilcoxon rank-sum test).

Figure 8. Raphe activation leads to sensitization of TC odor responses and decorrelation of MC odor responses

a) Pseudo-color plot showing the pattern of odor responses in a region of interest for in vivo experiment with TCs for two odors without (left) and with (right) raphe activation. Ellipses denote individual cell bodies. b) Similar plot for MC population from a different experiment. c) Principal component projections of TC responses from one experiment in the space of the first 3 components, for 4 different odors without and with raphe activation. d) Principal component projections for MC responses in a different experiment. e) Matrix showing the correlation between odor responses of TCs (6 odors) without (left) and with (right) raphe stimulation (p = 2.16× 10⁻⁷ for Wilcoxon rank-sum test, n= 6 odors). Note that diagonal elements have a value of 1 by definition. f) Analogous matrix for MCs (p = 7.8×10⁻⁴ for Wilcoxon rank-sum test, n= 6 odors). g) Cumulative density plots comparing the inter-odor distances measured in PCA space (left) and through direct correlation between different odors (right) for the TC experiment shown in panels a and c. h) Analogous density plots for MC experiment illustrated in b and d. i) Bar graphs depicting change in the inter-odor separation for TC population when raphe is stimulated for 8 separate experiments (p<0.05, Wilcoxon rank sum test, n= 8 animals, see Methods for the criterion). The change was calculated as the Kolmogorov distance between the cumulative density plots in PCA space (left) and as change in the correlation (right). j) Bar graphs depicting change in the inter-odor separation for MC population when raphe is stimulated for 7 separate experiments (p<0.05, Wilcoxon rank sum test, n=7 animals, see Methods for details).