Distinct spatiotemporal activity in principal neurons of the mouse olfactory bulb in anesthetized and awake states

Abstract

The acquisition of olfactory information and its early processing in mammals are modulated by brain states through sniffing behavior and neural feedback. We imaged the spatiotemporal pattern of odor-evoked activity in a population of output neurons (mitral/tufted cells, MTCs) in the olfactory bulb (OB) of head-restrained mice expressing a genetically-encoded calcium indicator. The temporal dynamics of MTC population activity were relatively simple in anesthetized animals, but were highly variable in awake animals. However, the apparently irregular activity in awake animals could be predicted well using sniff timing measured externally, or inferred through fluctuations in the global responses of MTC population even without explicit knowledge of sniff times. The overall spatial pattern of activity was conserved across states, but odor responses had a diffuse spatial component in anesthetized mice that was less prominent during wakefulness. Multi-photon microscopy indicated that MTC lateral dendrites were the likely source of spatially disperse responses in the anesthetized animal. Our data demonstrate that the temporal and spatial dynamics of MTCs can be significantly modulated by behavioral state, and that the ensemble activity of MTCs can provide information about sniff timing to downstream circuits to help decode odor responses.

Keywords: anesthesia; awake mouse; calcium imaging; inhibition; sniff dynamics.

Figure 1

Awake GCaMP2 mice display dynamic sniffing behavior compared to anesthetized mice. (A) Average-intensity image of the olfactory bulb in the absence of odor stimulation. A, anterior; L, lateral; M, medial; P, posterior. (B) Schematic showing the expression of GCaMP2 in M/T cells. (C) Fluorescence time course of a single glomerulus in an anesthetized mouse with the corresponding nasal air pressure tracing. Gray shaded area represents the period of odor stimulation. (D) Fluorescence time course and sniff tracing for an awake mouse. Arrows indicate sniff bursts immediately after odor onset and offset. (E) Distribution of sniff frequencies in anesthetized (blue) and awake (red) mice. Inset depicts the same histogram normalized to the peak frequencies. (F) Normalized distribution of sniff frequencies in awake mice during the air period (black), odor onset period (green), and odor offset period (yellow).

Figure 2

Sniff behavior has an effect on odor responses. (A) Time course of fluorescence in the OB in an anesthetized mouse with corresponding sniff trace, showing individual fluorescence fluctuations in response to a sniff. (B) Awake mouse time course. Individual sniff triggered responses are less easily identified. (C) Expanded version of the boxed region in (B) with the original fluorescence values (red) and a filtered time course (black). Fluorescence signal was filtered to remove high frequency noise. (D) Left, ΔF image from stimulation with 1% ethyl valeraldehyde. Responding glomerulus is circled in red. Right, time course plots from the encircled glomerulus (red) and the whole bulb (green) with corresponding sniff traces. Sniff bursts are correlated with response spikes across the whole bulb, including both responding and non-responding areas, even during clean air delivery. (E) Time courses and sniff traces from awake mice depicting the effect of sniff bursts on glomerular activity.

Figure 3

Sniffing behavior can be used to predict the glomerular response. (A) Top, normalized sniff-triggered average representing the generalized shape of the glomerular response to a single sniff in an anesthetized (blue) and awake (red) mouse. Bottom, average sniff traces showing the general shape of a single sniff in the two conditions. (B) Generation of the prediction time course. At each peak in inhalation (gray lines), the sniff-triggered average (green) is added to a baseline value. The sum of all the sniff-triggered averages is the prediction time course (pink). (C) Sniff triggered response amplitude as function of sniff number. Values are relative to the amplitude of the first sniff-triggered response and normalized to the peak sniff-triggered average. These values were used as a lookup table adjust the sniff-triggered averages (D,E). Examples of prediction time courses for anesthetized (D) and awake (E) animals. The images on the left are ΔF projections with white arrows indicating the glomerulus used to derive the time courses.

Figure 4

Global glomerular patterns are conserved across behavioral states. (A) Different odors show different response patterns. The images shown are ΔF images of the response of a single awake mouse to 12 different odors, averaged over three trials. IT, Isopropyl Tiglate; EV, Ethyl Valerate; V, Valeraldehyde; EB, Ethyl Butyrate; 4-H, 4-Heptanone; I, Isoamylamine; MT, Methyl Tiglate; H, Heptanal; T, Thiazole; EP, Ethyl Propionate; ET, Ethyl Tiglate; IA, Isoamyl Acetate; M, medial, A, anterior; Scalebar, 250 μm. (B) Glomerular activation pattern in an anesthetized (left) mouse and awake (right) after presentation of 10% ethyl tiglate. To more clearly distinguish individual hotspots in the anesthetized animal, we filtered the anesthetized image with a spatial high-pass filter (middle). Color bar indicates ΔF. (C) Second example of the similarity between anesthetized and awake activation patterns. These images were taken from a different mouse exposed to 1% isopropyl tiglate. (D) Distribution of correlation coefficients obtained from pair-wise comparison of spatial pattern of responses in different conditions.

Figure 5

Multi-photon optical sectioning confirms the consistency of glomerular activation across states. (A) Schematic from Figure 1A depicting optical sectioning. Multi-photon microscopy excludes light from other layers, resulting in an image consisting only of a thin layer (blue dashed line). (B) Multi-photon image showing the resting activity of the glomerular layer. (C) Fluorescence time courses and sniff plots for an anesthetized (top) and an awake (bottom) mouse. Consistent with findings from wide-field microscopy, anesthetized animals show clear, sniff-locked responses, while awake animals display a noisier time course. (D) ΔF images from the OB of an anesthetized (left) and an awake (right) mouse exposed to 1% ethyl 3-hydroxybutyrate. Color bar indicates ΔF. (E) Response matrix of 13 glomeruli and 20 odors in an anesthetized (top) and an awake (bottom) mouse. Color bar indicates average ΔF in arbitrary fluorescence units. Note that while awake animals had larger glomerular responses, the activation patterns are similar.

Figure 6

Anesthetized, but not awake animals, have diffuse activation patterns. (A) Widefield glomerular activation pattern in an anesthetized (left) and awake (right) mouse after presentation of 10% ethyl tiglate. Note that it is easier to distinguish individual activated glomeruli in the awake mouse. (B) ΔF as a function of distance from the glomerulus in anesthetized (n = 6 mice, 256 glomeruli) and awake (n = 6 mice, 196 glomeruli) mice. Inset, ΔF values are normalized to the ΔF of the glomerulus. Shaded areas denote standard error. (C) Response images (ΔF) obtained with multi-photon imaging in the glomerular layer. Note that neither anesthetized nor awake animals had significant diffuse responses. (D) ΔF values as a function of distance from the glomerulus. Blue, anesthetized (n = 6 mice, 77 glomeruli). Red, awake (n = 6 mice, 99 glomeruli). (E) Odor responses in OB of OMP-synaptopHluorin mice did not exhibit diffuse patterns. Shown are example global activation patterns in the same mouse in anesthetized (left) and awake (right) states. (F) Plot shows the ΔF values for the 18 labeled glomeruli in anesthetized and awake conditions.

Figure 7

Awake mice had highly variable responses. (A) Glomerular activation pattern in an awake mouse evoked by 10% ethyl tiglate. Images represent two trials in the same mouse with the same odor. (B) Top, ΔF responses of six responding glomeruli over 10 different trials. Trials of the same odor were spaced approximately 9 min apart. Note that while the absolute ΔF values varied significantly, the relative values of the six glomeruli remained similar. Bottom, average whole-bulb response over the same 10 trials. While the average response decreases over the 10 trials, the fluctuation cannot be explained solely by habituation. (C) Scatterplot comparing the ΔF of a glomerulus in a single trial to the ΔF values of the same glomerulus in each other trial. (D) Scatterplot similar to c but comparing normalized ΔF values. ΔF was normalized by subtracting the average ΔF value of the entire dorsal bulb. (E) Scatterplot comparing ΔF values normalized by dividing ΔF values by the whole dorsal bulb ΔF.

Figure 8

Consistent response profiles in awake mice can be derived by subtracting whole-bulb responses. (A) Glomerular activation profiles in an awake mouse during two different trials (5 and 9, respectively) of 1% valeradehyde presentation. (B) Response profiles of the glomerulus indicated by the arrows in (A). (C) Glomerular response profiles plotted with the whole dorsal-bulb background response. (D) Corrected glomerular response time courses obtained by subtracting the background response from the glomerular response. (E) Left, glomerular responses of 20 different trials of the same odor. Right, time courses of the same 20 trials after background subtraction. (F) Bar graph depicting the average standard deviation of each point in the time course across all trials of the same odor (n = 6 mice, 79 total odors). Instead of subtracting the background response from the glomerular response, the control algorithm added the two time courses together. Errorbars indicate standard deviation. ^***p < 0.001. (G) Illustration of the utility of removing global responses to better identify specific odor responses. Odors come on at random times, and sniffing frequency varies continuously. Each sniff produces a small response in all glomeruli when there is no odor; when a particular odor is present, a stronger response occurs in a glomerulus that is innervated by OSNs responding to that odor. Some odor responses can be detected by simple threshold crossing (the second and fourth odor response), but others become clearer when responses from background glomeruli are subtracted.