Basal forebrain GABAergic innervation of olfactory bulb periglomerular interneurons

Abstract

Key points: Basal forebrain long-range projections to the olfactory bulb are important for olfactory sensitivity and odour discrimination. Using optogenetics, it was confirmed that basal forebrain afferents mediate IPSCs on granule and deep short axon cells. It was also shown that they selectively innervate specific subtypes of periglomerular (PG) cells. Three different subtypes of type 2 PG cells receive GABAergic IPSCs from the basal forebrain but not from other PG cells. Type 1 PG cells, in contrast, do not receive inputs from the basal forebrain but do receive inhibition from other PG cells. These results shed new light on the complexity and specificity of glomerular inhibitory circuits, as well as on their modulation by the basal forebrain.

Abstract: Olfactory bulb circuits are dominated by multiple inhibitory pathways that finely tune the activity of mitral and tufted cells, the principal neurons, and regulate odour discrimination. Granule cells mediate interglomerular lateral inhibition between mitral and tufted cells' lateral dendrites whereas diverse subtypes of periglomerular (PG) cells mediate intraglomerular lateral inhibition between their apical dendrites. Deep short axon cells form broad intrabulbar inhibitory circuits that regulate both populations of interneurons. Little is known about the extrabulbar GABAergic circuits that control the activity of these various interneurons. We examined this question using patch-clamp recordings and optogenetics in olfactory bulb slices from transgenic mice. We showed that axonal projections emanating from diverse basal forebrain GABAergic neurons densely project in all layers of the olfactory bulb. These long-range GABAergic projections provide a prominent synaptic input on granule and short axon cells in deep layers as well as on selective subtypes of PG cells. Specifically, three different subclasses of type 2 PG cells receive robust and target-specific basal forebrain inputs but have little local interactions with other PG cells. In contrast, type 1 PG cells are not innervated by basal forebrain fibres but do interact with other PG cells. Thus, attention-regulated basal forebrain inputs regulate inhibition in all layers of the olfactory bulb with a previously overlooked synaptic complexity that further defines interneuron subclasses.

Keywords: basal forebrain; interneurone; olfactory bulb.

Figure 1. Conditional expression of ChR2 in various GABAergic neurons of the basal forebrain

A, stereotaxic injection of a viral construct in the basal forebrain of dlx5/6‐Cre mice was used for conditional expression of ChR2–EYFP in GABAergic neurons of the basal forebrain. Bottom left, sagittal section of a brain 4 weeks after the injection. The site of injection (BF, boxed area 1) and the olfactory bulb (OB, boxed area 2) are enlarged in the two other images. EPL, external plexiform layer; GL, glomerular layer; ONL, olfactory nerve layer. B, EYFP expression in the basal forebrain 4 weeks after injection of the same viral construct encoding cytosolic EYFP. Co‐staining of EYFP (green) together with ChAT, CB, PV or CR (red). Some neurons (arrowheads), but not all, co‐express the two markers. Scale bars 50 μm. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 2. Synaptic connections of basal forebrain axons with various inhibitory interneurons of the olfactory bulb

A, light‐evoked consecutive ISPCs evoked in a deep SA cell recorded at V _h = −35 mV in the presence of AMPA and NMDA receptor blockers (10 μM NBQX + 50 μM d‐AP5). The black trace is the average response. Light stimulation (1 ms) at blue arrow. The morphology of this neurobiotin‐filled cell is shown. Continuous lines delimit the glomerular layer (top) and the mitral cell layer (bottom). B, same in a granule cell recorded at V _h = 0 mV. C, light‐evoked IPSCs elicited in a PG cell (V _h = 0 mV) in control condition, in the presence of NBQX (10 μM), d‐AP5 (50 μM), mecamylamine (MECA, 20 μM) and atropine (ATRO, 10 μM) and after the addition of gabazine (5 μM, GBZ) to the cocktail. Several traces are superimposed for each condition. The graph shows the amplitude of consecutive responses in these different conditions. The image on the right is a snapshot of this PG cell during the recording, superimposed on the DIC image of the glomerulus delimited by a dashed line. Scale bar 10 μm. D, average current–voltage relationship of the light‐evoked IPSC in PG cells (n = 12 cells, each plot is an average ± SEM). Inset, IPSCs from a representative PG cell recorded at different holding potentials. E, distribution histograms of the onset latency (left) and onset jitter (right) of light‐evoked IPSCs recorded in PG cells. F, decay time constant distribution of light‐evoked IPSCs recorded in PG, dSA and granule cells. PG and granule cells were recorded at V _h = 0 mV, dSA cells at V _h between −35 mV and −15 mV. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 3. Type 1 PG cells are not contacted by basal forebrain projections but are inhibited by other PG cells

A–C, membrane properties (A), ON‐evoked synaptic responses (B) and responses to basal forebrain axons photo stimulation (C) of a type 1 PG cell recorded in a dlx5/6;ChR2‐EYFP mouse. This cell responded to the stimulation of OSN axons with an inward fast monosynaptic EPSC at V _h = −75 mV and with outward plurisynaptic IPSCs at V _h = 0 mV (B). In contrast, light stimulation of basal forebrain fibres did not evoke any response (C). Several traces are superimposed in each case. D, summary plots for ON‐evoked synaptic response amplitudes of all type 1 PG cells tested at two holding potentials. E, classification of the recorded PG cells into type 1 and type 2 in dlx5/6;ChR2‐EYFP mice. Top, early phase of ON‐evoked EPSCs in a type 1 PG cell (red) and in a type 2 PG cell (black). The type 1 PG cell responded with a short latency EPSC, starting <2 ms after the beginning of the stimulation artifact and with little onset variability between trials. In contrast, responses from the type 2 PG cell had a delayed and more variable onset latency. The graph shows the onset latency vs. onset jitter of ON‐evoked responses in PG cells classified as type 1 PG cells (red circles) and in PG cells classified as type 2 PG cells (black circles) in this study. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 4. CR‐expressing type 2 PG cells receive fast basal forebrain GABAergic inputs

A and B, membrane properties (A) and ON‐evoked synaptic responses (B) of a typical CR‐expressing type 2 PG cell recorded in a dlx5/6;ChR2‐EYFP mouse. This cell typically responded to a suprathreshold depolarization with a single spikelet followed by a passive membrane depolarization (A) and had small ON‐evoked responses at V _h = −75 mV and at V _h = 0 mV (B). C, summary plots for ON‐evoked synaptic responses recorded at V _h = −75 mV and at V _h = 0 mV for cells classified as CR‐expressing PG cells. D, cells classified as CR‐expressing type 2 PG cells also had larger electrical membrane resistances compared to PG cells classified in the other groups (horizontal bars show the averages, P < 0.001 for comparison with each of the 3 other subgroups, Wilcoxon test). E, light stimulation of basal forebrain fibres evoked large and fast GBZ‐sensitive IPSCs in CR‐expressing PG cells. Traces are from the same cell as in A and B. The average IPSC (black trace) is superimposed on several consecutive responses recorded at V _h = 0 mV in ACSF. Bottom, distribution histogram of the decay time constants of light‐evoked IPSCs in cells classified as CR‐expressing PG cells (filled bars) superimposed on the distribution histogram for all the recorded PG cells (open bars). [Color figure can be viewed at wileyonlinelibrary.com]

Figure 5. Type 2 PG cells with short ON‐evoked excitatory response do not interact with other PG cells but receive inputs from the basal forebrain

A and B, membrane properties (A) and ON‐evoked synaptic responses (B) of a typical type 2 PG cell that responded to the stimulation of the OSN axons with a short burst of EPSCs in a dlx5/6;ChR2‐EYFP mouse. The short ON‐evoked excitatory response was the main criterion for classifying PG cells in this subclass. C, summary plots (right) for ON‐evoked synaptic responses recorded at V _h = −75 mV and at V _h = 0 mV indicate that cells in this group also had no ON‐evoked IPSC. D, IPSCs evoked by a light stimulation of the basal forebrain fibres. The average IPSC (black trace) is superimposed on several consecutive responses recorded at V _h = 0 mV in the presence of NBQX and d‐AP5. GBZ abolished the response. Traces are from the same cell as in A and B. Bottom, distribution histogram of the decay time constants of light‐evoked IPSCs in PG cells classified in this subclass (filled bars) superimposed on the distribution histogram for all the recorded PG cells (open bars). E, paired recording of two PG cells projecting in the same glomerulus in the Kv3.1‐EYFP transgenic mouse. The ‘test’ EYFP⁺ PG cell was recorded in the whole‐cell configuration at V _h = 0 mV to monitor IPSCs and at V _h = −75 mV to monitor EPSCs. The ‘control’ PG cell was recorded in the loose cell‐attached configuration to monitor its firing. OSNs were stimulated at an intensity inducing the firing of the control PG cell. The summary graph on the right shows that OSN stimulations did not evoke any outward IPSC in the test cell even when the stimulation was increased by a factor of 10. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 6. Type 2 PG cells with long‐lasting ON‐evoked excitatory response receive slow GABAergic inputs from the basal forebrain

A and B, membrane properties (A) and ON‐evoked synaptic responses (B) of a typical type 2 PG cell that responded to the stimulation of the OSN axons with a long‐lasting burst of EPSCs in a dlx5/6;ChR2‐EYFP mouse. A, like all the cells included in this group, this PG cell responded to a step depolarization with a regular firing. B, ON‐evoked responses recorded at two holding potentials and PSTH of the excitatory response (bottom). C, summary plots for ON‐evoked synaptic responses recorded at V _h = −75 mV and at V _h = 0 mV for cells included in this group. D, IPSCs evoked by a light stimulation of the basal forebrain fibres. The average IPSC (black trace) is superimposed on several consecutive responses recorded at V _h = 0 mV in the presence of NBQX and d‐AP5. GBZ abolished the response. Same cell as in A and B. Bottom, distribution histogram of the decay time constants of light‐evoked IPSCs in PG cells classified in this subclass (filled bars) superimposed on the distribution histogram for all the recorded PG cells (open bars). Cells included in this group had slow IPSCs. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 7. Basal forebrain GABAergic inputs have different presynaptic properties depending on the postsynaptic PG cell subtype

A–C, top row, light‐evoked IPSCs in three PG cells representative of the three subclasses of type 2 PG cells recorded in dlx5/6;ChR2‐EYFP mice (A: CR‐expressing PG cells; B: PG cells with short ON‐evoked excitatory responses; C: regularly firing PG cells with long‐lasting ON‐evoked responses). Each cell was stimulated with 5 flashes of light at 20 Hz. Ten to twelve consecutive responses are superimposed for each cell; the black trace is the average response. Middle row, amplitudes of the nth light‐evoked IPSC relative to the normalized amplitude of the first IPSC recorded in PG cells classified in the 3 subgroups of the corresponding column. Lines connect plots from the same cells. Bottom row, failure rate for each of the five consecutive stimuli and for each cell over multiple trials. Failures were never observed in the third group. [Color figure can be viewed at wileyonlinelibrary.com]