(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Value-related learning in the olfactory bulb occurs through pathway-dependent perisomatic inhibition of mitral cells [1]

['Sander Lindeman', 'Sensory', 'Behavioural Neuroscience Unit', 'Okinawa Institute Of Science', 'Technology Graduate University', 'Okinawa', 'Xiaochen Fu', 'Janine Kristin Reinert', 'Izumi Fukunaga']

Date: 2024-03

Associating values to environmental cues is a critical aspect of learning from experiences, allowing animals to predict and maximise future rewards. Value-related signals in the brain were once considered a property of higher sensory regions, but their wide distribution across many brain regions is increasingly recognised. Here, we investigate how reward-related signals begin to be incorporated, mechanistically, at the earliest stage of olfactory processing, namely, in the olfactory bulb. In head-fixed mice performing Go/No-Go discrimination of closely related olfactory mixtures, rewarded odours evoke widespread inhibition in one class of output neurons, that is, in mitral cells but not tufted cells. The temporal characteristics of this reward-related inhibition suggest it is odour-driven, but it is also context-dependent since it is absent during pseudo-conditioning and pharmacological silencing of the piriform cortex. Further, the reward-related modulation is present in the somata but not in the apical dendritic tuft of mitral cells, suggesting an involvement of circuit components located deep in the olfactory bulb. Depth-resolved imaging from granule cell dendritic gemmules suggests that granule cells that target mitral cells receive a reward-related extrinsic drive. Thus, our study supports the notion that value-related modulation of olfactory signals is a characteristic of olfactory processing in the primary olfactory area and narrows down the possible underlying mechanisms to deeper circuit components that contact mitral cells perisomatically.

Funding: This work was supported by Grant-in-Aid for Scientific Research (C) 22K06490 ( https://www.jsps.go.jp/english/e-grants/ to SL & IF) and OIST Graduate University ( https://www.oist.jp/ to SL, XF, JKR, & IF). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2024 Lindeman et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Here, we show that the olfactory bulb exhibits robust and consistent reward-related signals during a trace olfactory conditioning paradigm, where mice discriminate between closely related olfactory mixtures. This phenomenon is characterised by widespread inhibitory responses following the rewarded odour presentation, in mitral cells but not tufted cells. This divergence is not explained by the odour identity or sampling strategy and reflects the congruence of sensory drive and contextual signals. By imaging from specific subcellular compartments of mitral cells, we demonstrate that the divergent responses first become evident perisomatically. Depth-resolved imaging from the dendrites of adult-born granule cells suggests that the cell type-specific modulation may involve an extrinsic drive to putative mitral cell-targeting granule cells.

In general, the effects of such perturbations depend on the output neuron type—mitral cells versus tufted cells. Mitral and tufted cells of the olfactory bulb differ physiologically and morphologically [ 23 – 27 ] and project to different downstream areas [ 26 ]. As a result, they are regarded as the starting points of parallel olfactory processing. Possible origins of the cell type-dependent modulation may include different sources of modulatory signals [ 17 ], receptor types expressed [ 28 ], or connectivity with distinct sets of local interneurons [ 25 , 29 – 31 ]. For example, a recent study in naïve mice demonstrated that feedback modulations of mitral versus tufted cells preferentially involve the piriform cortex versus the anterior olfactory cortex, respectively [ 17 ]. Within the olfactory bulb, it is yet unknown how diverse modulatory signals originating from different brain regions reach the output neurons in a cell type-specific manner. Resolving the nature and mechanisms underlying the cell type-specific modulation is crucial for understanding how long-range inputs from multiple brain regions couple into the intricate local circuitry to alter their computations.

The nature of this apparent reward-related modulation in the olfactory bulb remains unresolved. For example, one study observed that evoked responses to rewarded versus unrewarded odours in the olfactory bulb diverge only transiently during learning [ 12 ]. Such a transient modulation could be explained by dynamic changes in the level of animal’s engagement [ 14 ], where the learning-related modulation corresponds mainly to the changes in the inputs from the sensory periphery arising from sniff pattern changes. Rodents indeed adjust the odour sampling patterns exquisitely according to the behavioural contexts [ 15 , 16 ]. However, given that the olfactory bulb is a major target of feedback and neuromodulatory projections from many brain regions, value-related information could affect how the olfactory bulb represents odours. For example, electrical and optogenetic stimulations and pharmacological manipulations of neuromodulatory and feedback inputs to the olfactory bulb change the gain of odour responses in the principal neurons of this region [ 17 – 22 ].

Decision and value-related modulations of sensory responses are featured prominently in higher sensory areas [ 1 , 8 ]. However, recent studies indicate that even early stages of sensory processing, especially in rodents, participate in value-like representations [ 9 – 11 ]. The olfactory system is an extreme case in this regard, where apparent reward-related modulation is readily observed as peripherally as in the olfactory bulb [ 12 , 13 ], the primary olfactory region situated just one synapse away from the site of sensory transduction. This peripheral location, along with the saliency of olfactory cues for rodents, makes the olfactory bulb an attractive structure to study the mechanisms that generate value-like signals in the brain [ 12 ].

Sensory systems of the brain play crucial roles in guiding animals’ choices. One such role played by the systems is in reward-driven learning, where the internal representations of sensory cues are adjusted as a result of past reward encounters, to influence their future behavioural choices. In addition to long-term adjustments, decades of studies across brain areas have demonstrated that reward expectations are potent and dynamic modulators of sensory activity. For example, stimulus evoked responses in many sensory regions of the brain scale with the quantity of expected reward [ 1 – 5 ], which is often interpreted as representations of the subjective value [ 6 , 7 ]. Such a system where sensory processing is fine-tuned flexibly may be crucial for maximising returns in a dynamic and uncertain world [ 7 ].

Results

The olfactory bulb integrates both feedforward sensory stimuli, as well as long-range projections from other brain areas (Fig 1A). The latter input is thought to convey behavioural contextual signals to the olfactory bulb and tune activity patterns flexibly. To study how the behavioural context modulates the olfactory bulb output in olfactory decision-making, we trained head-fixed mice to perform an olfactory discrimination task (Fig 1). Water-restricted mice were trained to associate a rewarded odour (S+ odour) with a water reward, and an unrewarded odour (S- odour) with no water delivery (Fig 1B). Note that this paradigm includes a trace period, as we reasoned that an early cessation in the feedforward signal may maximise the chance of observing context-related activity patterns.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Widespread inhibition is observed in mitral cells in response to rewarded odour. (A) Schematic showing 2 major sources of inputs to the olfactory bulb. Left arrow represents olfactory nerve inputs. Right arrow represents long-range inputs from other brain areas. (B) Odour was presented for 1 s and did not overlap with the reward that was delivered 2 s after the odour offset. (C) Odours used in the Go/NoGo olfactory discrimination tasks. Blue font corresponds to the rewarded (S+) odour. Mice were head-fixed and had a cranial window implanted. (D) Time course of task acquisition for easy and difficult discriminations defined in C. Imaging sessions took place in proficient mice (n = 6 mice). (E) Left, imaging configuration. Mitral cell (MC) and tufted cell (TC) somata were distinguished by depth. Right, example fields of view for TC somata and MC somata. (Fi–iii) Responses to S+ and S- odours measured in TC somata. (Fi) Colormap representation of fluorescence change over time for all ROIs. (Fii) Average responses to S+ (blue) and S- (black) odours from all ROIs. (Fiii) Scatter plot comparing S- vs. S+ response amplitudes for the period shown in Fii. Each point represents 1 ROI, and the data shown are from all sessions and mice. Dotted line represents unity (S- amplitude = S+ amplitude). Individual points correspond to ROIs. Black dots indicate S+ and S- responses that were significantly different. (Gi–iii) Same as Fi–iii, but for MC somata. N = 150 and 428 ROIs, and 3 and 6 mice for TC somata and MC somata, respectively. Source data can be found in Fig 1 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g001

The mice were first trained to discriminate between easily distinguishable odour mixtures, which comprised ethyl butyrate (EB) and methyl butyrate (MB), mixed at 80%/20% ratio versus a 20%/80% ratio for the S+ versus S- odours, respectively (Fig 1C). When the mice reached a criterion of 80% accuracy (3 ± 0.9 days, n = 6 mice, Fig 1D), they were trained to discriminate between more similar odour mixtures (“Difficult discrimination task”; 60%/40% mixture of EB and MB versus a 40%/60% mixture). This is a task known to engage many components of the olfactory bulb circuitry [32]. Well-trained mice discriminated between these similar mixtures in 1.63 ± 0.53 s (S1 Fig), with comparable sniffing patterns for the S+ versus S- odours (S2 Fig), consistent with previous reports where similar odours and reward timing were used [33,34].

In mice proficiently performing the difficult olfactory discrimination task, we studied the responses of olfactory bulb output to the S+ versus S- odours. The calcium indicator GCaMP6f was expressed in mitral and tufted cells using Tbx21-Cre mice crossed with Ai95D mice [35,36] and was imaged using a two-photon microscope (n = 428 regions of interest (ROIs) in 6 mice, and n = 150 ROIs in 3 mice, respectively; Figs 1E–1G, and S3). Mitral and tufted cells were distinguished by depth (Fig 1E). Tufted cells responded largely similarly to both odours (mean ΔF/F during odour = 0.628 ± 0.135 and 0.655 ± 0.148 for S+ and S-, respectively; p = 0.777, Wilcoxon rank-sum test; mean ΔF/F post-odour = 0.203 ± 0.264 and 0.237 ± 0.264 for S+ and S-, respectively; p = 0.149, Wilcoxon rank-sum test; Fig 1F). Peculiarly, responses of the mitral cell somata to the S+ odour were characterised by widespread inhibitory responses (mean ΔF/F S+ = −0.048 ± 0.058; S- = −0.022 ± 0.054; p < 0.001, Wilcoxon rank-sum test; Fig 1G). We also observed that the S+ odour evoked less inhibition on trials where mice did not generate anticipatory licks (S4 Fig). This dominance of inhibition for the S+ odour was present soon after the odour onset but was particularly pronounced during the post-odour period (mean ΔF/F S+ = −0.048 ± 0.095; S- = 0.034 ± 0.102; p < 0.001, Wilcoxon rank-sum test; Fig 1G). The earliest time when the S+ and S- responses diverge significantly was 818 ± 540 ms after the odour onset (mean ± standard deviation; n = 16 fields of view, 6 mice).

The late onset of the reward-associated inhibition in mitral cells raises the question regarding the underlying drive: Is the inhibitory component locked to the anticipatory motor output, or to the odour? To analyse this, we divided the rewarded trials into 2 sets based on the animals’ reaction times (“early onset” versus “late onset”) and reverse-correlated the GCaMP6f signals to the onsets of anticipatory signals (“lick-aligned average”; Fig 2). If the peak of inhibition in the averages occur at the same time for the early lick sets and late lick sets, it would imply that the majority of the inhibition is locked more to the behavioural output (Fig 2B). This analysis revealed, in contrast, that the time of peak inhibition is shifted depending on the reaction time (Pearson’s correlation coefficient = −0.555, p = 0.026; n = 16 fields of view, 6 mice; Fig 2C–2E), indicating that the inhibition is, on average, locked to the odour.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Reward-related inhibition is locked to the odour presentation. (A) Fluorescence change from mitral cell somata aligned to the time of first anticipatory lick after odour onset (S+ trials). (B) Predictions for 2 alternative hypotheses for late-lick vs. early-lick trials; if inhibition is generally odour-locked, a shift in the peak inhibition is observed in the lick-aligned average. If the reward-related inhibition is locked to generation of licks, the trough times will be the same for late vs. early lick trials relative to the onset time of anticipatory lick. (C) Lick raster plots from 2 example sessions. Late (early) vs. early (black) lick trials were defined as trials where the first anticipatory lick occurred later or earlier than the median lick onset time for each session. The second dotted line represents the timing of water delivery. (D) Lick-aligned averages for early vs. late lick for all sessions (black and red traces, respectively). Each trace is an average for 1 session. Top row is for 5 sessions with the smallest range in the reaction times (see panel E). Bottom row is for sessions where reaction times ranged more widely. (E) Shift in the peak trough time in the odour-aligned (left) and lick-aligned (right) averages compared against mean difference in the lick onsets for the late vs. early trials. Each point corresponds to 1 imaging session. Pearson’s correlation coefficient = −0.555, p = 0.026 (n = 16 sessions, 6 mice). Source data can be found in Fig 2 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g002

The prevalence of inhibitory responses in mitral cells following the rewarded odour presentation is striking, but this level of inhibitory dominance has not been reported previously, even though several studies already studied how mitral cells respond to odours during difficult odour discrimination paradigms [37–39]. The difference here may be the short duration of odour pulse used, followed by a 2-s long trace period. It is possible that, with a longer odour presentation, the feed-forward component may dominate over any intrinsic or modulatory influences in the olfactory bulb that underly the reward-related inhibition (Fig 3A). To test this possibility, in well-trained mice, we presented the odours for a longer period (4 s), making the task a delay task (Fig 3B). In this condition, mitral cells responded to the rewarded and unrewarded odours similarly (Fig 3C–3E). Notably, both responses were characterised by widespread inhibitory component (% of ROIs showing significant inhibition = 18.1 for S+ and 11.9 for S-, and 35.2 for S+ and 23.9 for S- in early and late time windows, respectively, n = 5 mice). While the divergent response is still present, the magnitude of this divergence is significantly reduced during long odour presentation (S5 Fig). This result indicates that the response divergence in the post-odour period may be uncovered when olfactory responses are allowed to evolve in the absence of feed-forward inputs.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Longer odour presentation masks the appearance of divergent responses. (A) Schematic showing dominance of sensory drive. (B) A 4-s odour pulse overlapped temporally with reward delivery. (C) Colour map display showing GCaMP6f fluorescence change from mitral cell somata evoked by S+ and S- odours. (D) Average fluorescence change from all ROIs in response to S+ (blue) and S- (black) odours. Mean and SEM shown. (E) Scatter plots of average fluorescence change for S+ vs. S- odours for the time period indicated in D. Black dots indicate S+ and S- significantly divergent responses (N = 210 ROIs, 5 mice). Source data can be found in Fig 3 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g003

To test if the behavioural state of the animal is crucial for the response divergence in mitral cells, we used 2 pseudo-conditioning paradigms using the same odours (Fig 4A and 4B). In the first case (“Disengaged”), the water was delivered every trial, approximately 15 s before the odour presentation (Fig 4B). In the second case (“Random association”), we delivered the water on randomly selected trials, so that both 60/40 and 40/60 odour mixtures were followed by water 50% of the time (Fig 4B). These 2 paradigms decouple the odour-reward association, while inducing different levels of engagement in the head-fixed mice [40]. Imaging sessions took place after the mice, previously trained on the difficult discrimination task, were switched to, and experienced at least 1 session of the new paradigm (Fig 4C).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Occurrence of reward-related inhibition depends on the behavioural context, not odour identity. (A) Hypotheses on the source of signals underlying differential S+ and S- responses in mitral cell somata; it could derive from sensory stimuli (top) or from long-range inputs to the olfactory bulb (bottom). (B) Behavioural paradigms to decouple reward association while disengaging mice (middle) or engaging mice (“Random association”). In disengagement sessions, reward was delivered every trial, preceding odour presentations. In random association sessions, reward followed both mixtures of EB and MB 50% of the time. (C) Timeline of experiments. Mice first performed difficult olfactory discrimination, then went through either disengagement or random association sessions. Imaging took place from day 2 in both cases. (D) Number of anticipatory licks (licks within a 3-s window from odour onset) for the 2 odours for 3 behavioural paradigms. Individual points correspond to each imaging session analysed. *** Corresponds to p < 0.001 (post hoc Tukey–Kramer multiple comparisons after 1-way ANOVA). (E) Average fluorescence change of all ROIs (mitral cell somata) for the odours indicated. (F) Comparison of fluorescence change in response to the 2 odours for the odour period for disengagement sessions (left) and random association sessions (right). Individual points correspond to ROIs. Darker points represent significantly divergent responses. (G) Same as F, but for post-odour period. N = 125 ROIs, 3 mice for disengagement and 301 ROIs, 5 mice for random association. Source data can be found in Fig 4 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g004

In both control paradigms, the head-fixed mice showed no preferential licking for the 60/40 mixture (average anticipatory licks for disengagement paradigm = 1.5 ± 1.6 and 0.7 ± 0.8 on 60/40 and 40/60,respectively; p = 0.999, 1-way ANOVA with post hoc Tukey–Kramer multiple-comparisons; average anticipatory licks for random association paradigm = 6.9 ± 6.1 and 5.5 ± 4.3 on 60/40 and 40/60, respectively; p = 0.890, 1-way ANOVA with post hoc Tukey–Kramer multiple comparisons, Fig 4D). Importantly, disengagement and randomised paradigms differed in the general levels of anticipatory licks (average anticipatory licks for all trials = 1.1 ± 1.3 and 6.2 ± 5.2 for disengagement and random association paradigms, respectively; p = 0.0021, 1-way ANOVA with post hoc Tukey–Kramer multiple comparisons), indicating that different levels of behavioural engagement were indeed achieved by these paradigms. A difference between the 2 behavioural states included a general reduction in the mitral cell inhibition when the mice were disengaged, which is consistent with a previous observation [41]. Importantly, in both cases, the mitral cell somata responded similarly to the 2 odour mixtures (mean ΔF/F for disengagement = −0.02 ± 0.05 and −0.01 ± 0.06 during odour for 60/40 and 40/60, respectively; p = 0.465; for post-odour = 0.05 ± 0.09 and 0.05 ± 0.09; p = 0.553; mean ΔF/F for random association = −0.03 ± 0.09 and −0.04 ± 0.10 during odour; p = 0.259; post-odour = −0.01 ± 0.14 and 7.7 × 10−5 ± 0.14; p = 0.617, Wilcoxon rank-sum test; Fig 4E–4G). Note that the inhibition during the post-odour, anticipatory period that is normally present in discriminating mice was generally reduced in the 2 control paradigms. Curiously, the reward-related inhibition was reduced in mice performing an easy olfactory discrimination (S6 Fig). Together, these results indicate that the observed divergent responses in mitral cell somata are state dependent, and not explained by the odour identities.

What is the origin of the widespread inhibition associated with the rewarded odour? Previous studies showed that a variety of feedback and neuromodulatory projections to the olfactory bulb modulate the physiology of olfactory bulb neurons [18–20,42–44]. Further, several studies showed that such modulations manifest differently for mitral cells and tufted cells [17,18,38,45]. Recent works indicate that mitral cells receive more potent feedback modulation from the piriform cortex [17,18]. Further, anterior piriform cortex has been reported to contain neurons that show value-like signals [46]. Thus, even though it is beyond the scope of the current work to systematically investigate all sources, the anterior piriform cortex is a reasonable candidate for the source of the contextual signal resulting in the mitral cell-specific, reward-related inhibition we observe.

To test the involvement of the piriform cortex, we pharmacologically inactivated the ipsilateral anterior piriform cortex while the head-fixed mice performed the difficult olfactory discrimination task (Fig 5A). This was achieved by infusing the GABA A agonist, muscimol, unilaterally through an implanted canula. Muscimol and control sessions were carried out on alternate days, but the same fields of view were sampled for the 2 conditions, so that the responses of the same ROIs could be compared directly. The infusion of muscimol disrupted the behavioural performance significantly (behavioural accuracy = 64.0 ± 14.5% during muscimol sessions; 92.8 ± 7.8% during control sessions; p = 0.004, Wilcoxon rank-sum test; n = 6 control sessions and 6 muscimol sessions, 3 mice; Fig 5F). When responses of mitral cells were imaged in this condition, divergence in the rewarded versus unrewarded odour responses was significantly reduced (mean ΔF/F during odour = −0.007 ± 0.073 and −0.023 ± 0.091 for S+ and S-, respectively; p = 0.174, Wilcoxon rank-sum test; mean ΔF/F post-odour = 0.067 ± 0.124 and 0.079 ± 0.171 for S+ and S-, respectively; p = 0.252, Wilcoxon rank-sum test, Fig 5G–5I). This was characterised by a reduction in the inhibitory responses evoked by the rewarded stimulus (normalised S+—S- difference = −0.007 ± 0.156 and 0.043 ± 0.142 in control and muscimol sessions respectively; p = 0.008, Wilcoxon rank-sum test; Fig 5J), and during the post-odour phase (normalised S+—S- difference = −0.263 ± 0.175 and −0.040 ± 0.168 in control and muscimol sessions, respectively; p = 3.53 × 10−18, Wilcoxon rank-sum test; Fig 5J). Together, these data indicate that an intact piriform cortex and/or accurate behavioural performance are required to observe the widespread inhibitory responses associated with the rewarded odour.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Intact piriform cortex is needed to observe the reward-related inhibition. (A) Hypothesis tested; anterior piriform cortex is necessary for mitral cell divergence during odour discrimination. (B) Muscimol solution (2 mM; 500 nL) was infused via an implanted cannula targeted to anterior piriform cortex. (C) Timeline of experiments. After mice were trained on the discrimination task, control and muscimol sessions alternated. One imaging session occurred per day. (D) Example of DiI location infused via an implanted cannula. Scale bar = 1 mm. (E) Summary of cannula tip locations. (F) Accuracy in performance (% of trials with correct lick response) for control vs. muscimol sessions. Individual points correspond to each imaging session. P = 0.004. (G) Example fields of view matched across 2 conditions. (H) Time course of fluorescence change for control (top) and muscimol (bottom) sessions in matched ROIs. (I) Average fluorescence change for each ROI for S+ and S- odours during odour and post-odour periods. Darker points represent significantly divergent responses. (J) Cumulative fraction of ROIs for normalised difference in fluorescence changes evoked by S+ and S-. N = 123 ROIs, 3 mice. Source data can be found in Fig 5 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g005

The results so far indicate that the widespread inhibitory responses associated with the rewarded odours come from sources extrinsic to the olfactory bulb. One of the major targets of such long-range projections within the olfactory bulb is the granule cells. These cells contact mitral cells on their lateral dendrites at a deeper portion of the external plexiform layer, although other inhibitory neurons contact mitral cells in the deeper subcellular compartments as well [47,48]. If the granule cells convey the contextual signals to mitral cells, the divergent responses may be observable perisomatically, but not in the superficial compartment (Fig 6A). To test this, we compared the GCaMP6f signals from the apical dendrites of mitral cells in the glomeruli versus signals from the somata, which reflect signals derived from all subcellular compartments. Since tufted cells and mitral cells both send their apical dendrites to the glomeruli, to study signals from mitral cells in isolation, we used Lbhd2-CreERT2::Ai95D mice, where GCaMP6f is expressed predominantly in mitral cells [33] (Fig 6A and 6B).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Reward-related inhibition originates perisomatically. (A) Schematic showing possible subcellular compartments where divergent signals may arrive, namely, apical dendrites in the glomerulus (upper arrows), and deeper, lateral dendrites (lower arrows). (B) Left, confocal image showing GCaMP6f preferentially in mitral cells (MCs) in Lbhd2-CreERT2::Ai95D mice. Right, illustration of imaging planes to obtain signals from the MC apical dendrites and somata. Scale bar = 100 μm. (C) Analysis of GCaMP6f signals from MC apical dendrites. (Ci) Example field of view from the glomerular layer. (Cii) Average fluorescence change from all ROIs (glomeruli) in response to S+ and S- odours. (Ciii) Comparison of average GCaMP6f fluorescence change for individual ROIs evoked by S+ vs. S- odours for the periods indicated. Darker points represent significantly divergent responses. (Di–iii) Same as Ci–iii, but for MC somata. Scale bar = 50 μm. N = 140 ROIs, 4 mice for apical dendrites and 321 ROIs, 7 mice for somata. Source data can be found in Fig 6 data, Dryad. https://doi.org/10.1371/journal.pbio.3002536.g006

Imaging from the superficial plane, the apical dendrites showed no significant differences between responses to S+ and S- odours (mean ΔF/F during odour = 0.307 ± 0.439 and 0.328 ± 0.466 for S+ and S-, respectively; p = 0.687; post-odour = 0.338 ± 0.512 and 0.382 ± 0.549 for S+ and S-, respectively; p = 0.423, Wilcoxon rank-sum test, Fig 6C). As before, signals from the mitral cell somata imaged in the Lbhd2-CreERT2::Ai95D mice were characterised by the widespread inhibitory component (mean ΔF/F during odour = −0.058 ± 0.077 and −0.034 ± 0.061 for S+ and S-, respectively; p = 1.86 × 10−4; post-odour = −0.024 ± 0.130 and 0.038 ± 0.131 for S+ and S-, respectively; p = 1.08 × 10−5, Wilcoxon rank-sum test; Fig 6D). Divergent responses were also observed in the lateral dendrites in the vicinity of mitral cell somata (S7 Fig). Together, these data suggest that the reward-related inhibition in mitral cells originates perisomatically.

If the inhibition in response to the rewarded cue in mitral cells is mediated via granule cells, we should observe a greater GCaMP6f signal change to S+ odours specifically in the granule cells that target mitral cells (Fig 7A). The granule cells whose dendrites ramify in the deeper portion of the external plexiform layer are thought to synapse with mitral cells [25,49,50], where mitral cell lateral dendrites are found. These mitral cell-targeting granule cells are, however, present intermixed with tufted cell-targeting granule cells.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. An experimental approach to image from superficial and deep adult-born granule cells. (A) Schematic of local circuitry and hypothesis; mitral cells (MCs) synapse with granule cells (GCs) whose dendrites ramify in the lower portion of the external plexiform layer. The deep-ramifying granule cells may receive the contextual signal that leads to the divergent responses in MC somata. (B) Lower and upper portions of the external plexiform layer can be distinguished by the density of MC dendrites. tdTomato is preferentially expressed in MCs in Lbhd2-CreERT2::Ai14 mice. Scale bars = 30 μm for top and bottom images. (C) Depth-dependent pixel intensity histogram from an example z-stack obtained with a two-photon microscope in a Lbhd2-CreERT2::Ai14 mouse. (D) Adult-born granule cells (abGCs) are made to express GCaMP6f by injecting AAVs in the subventricular zone (SVZ). Adult-born granule cells are imaged 4 weeks after injection. RMS = rostral migratory stream. (E) An example confocal image showing the site of AAV injection targeted to the lateral wall of the subventricular zone. Scale bar = 1 mm. (F) Example confocal images showing amplified GCaMP6f signal from deep abGC (green), shown with tdTomato signals (red) from MCs and DAPI signals (blue). Scale bars = 100 μm and 50 μm for left and right images, respectively. (G) Another example confocal image from a separate animal showing a mixture of adult-born GCs with deep and superficial dendrites. Scale bar = 100 μm. https://doi.org/10.1371/journal.pbio.3002536.g007

To distinguish the putative mitral cell-targeting granule cells from those that target tufted cells, the depth of the external plexiform layer needs to be distinguished accurately in vivo. Towards this end, we crossed Lbhd2-CreERT2 mice with Ai14 mice to express tdTomato preferentially in mitral cells. We reasoned that despite the tissue curvature or non-uniform thickness of the external plexiform layer, this method would allow us to accurately separate the deeper portion from the superficial portion based on the density and distribution of the tdTomato expression. Indeed, the deep portion of the external plexiform layer showed higher density of thin red fluorescent processes (Fig 7B and 7C), while at more superficial depths, we observed occasional fluorescence from thick processes, likely corresponding to the primary dendrites of mitral cells.

To study if the divergent odour responses in mitral cells can be explained by the evoked activity of putative mitral cell-targeting granule cells, we turned to adult-born granule cells that develop their dendrites in the deep external plexiform layer (Fig 7D and 7E). Adult-born granule cells are thought to be critical for refining odour responses in mitral and tufted cells when mice need to discriminate between similar odours [51–55]. Further, since the mature adult-born granule cells form dendro-dendritic synapses with mitral cell lateral dendrites, where GABA release can occur locally [56], we sought to image directly from dendritic gemmules. Due to their small size, we were cautious to exclude images from sessions that showed motion artefact, which was determined by correlating the structural fluorescence pattern to the baseline period and discarded those that showed low correlation (S8 Fig). As a result, 70% (1,343/1,917 trials) of the acquired data was discarded.

We first characterised how the deep versus superficial dendritic gemmules respond to the rewarded versus unrewarded odours as mice performed difficult discrimination (Fig 8A and 8B). Inhibitory responses were generally more prevalent than excitatory responses in both cases, for both odours. In the deep gemmules, in the early phase, we observed slightly more inhibitory responses for the rewarded odour (Fig 8A and 8B). However, in the late phase, the S+ and S- response distributions almost completely overlapped. We wished to understand how well these granule cell dendritic responses could be explained by the local presynaptic counterparts, that is, against the distribution of evoked responses in mitral cells and tufted cells. The S+ versus S- tuning showed a close overlap between tufted cells and superficial gemmules of adult-born granule cells (Fig 8C). On the other hand, S+ versus S- tuning distribution of deep gemmules during the late phase could not be explained by the mitral cell tuning distribution (Fig 8C). There was a tendency for these gemmules to respond more positively to the rewarded odours than would be predicted from mitral cell activity alone. In other words, our data suggests that mitral cell-targeting granule cells may receive an additional excitatory drive associated with the rewarded odour.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002536

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.