Intraglomerular lateral inhibition promotes spike timing variability in principal neurons of the olfactory bulb

Abstract

The activity of mitral and tufted cells, the principal neurons of the olfactory bulb, is modulated by several classes of interneurons. Among them, diverse periglomerular (PG) cell types interact with the apical dendrites of mitral and tufted cells inside glomeruli at the first stage of olfactory processing. We used paired recording in olfactory bulb slices and two-photon targeted patch-clamp recording in vivo to characterize the properties and connections of a genetically identified population of PG cells expressing enhanced yellow fluorescent protein (EYFP) under the control of the Kv3.1 potassium channel promoter. Kv3.1-EYFP(+) PG cells are axonless and monoglomerular neurons that constitute ∼30% of all PG cells and include calbindin-expressing neurons. They respond to an olfactory nerve stimulation with a short barrage of excitatory inputs mediated by mitral, tufted, and external tufted cells, and, in turn, they indiscriminately release GABA onto principal neurons. They are activated by even the weakest olfactory nerve input or by the discharge of a single principal neuron in slices and at each respiration cycle in anesthetized mice. They participate in a fast-onset intraglomerular lateral inhibition between principal neurons from the same glomerulus, a circuit that reduces the firing rate and promotes spike timing variability in mitral cells. Recordings in other PG cell subtypes suggest that this pathway predominates in generating glomerular inhibition. Intraglomerular lateral inhibition may play a key role in olfactory processing by reducing the similarity of principal cells discharge in response to the same incoming input.

Keywords: glomerulus; inhibition; olfactory bulb; periglomerular cell.

Figure 1.

Characterization of EYFP⁺ PG cells in Kv3.1–EYFP transgenic mice. A, Immunofluorescence staining for CB (top), CR (middle), and TH (bottom) in the glomerular layer of Kv3.1–EYFP transgenic mice. Note that all CB⁺ PG cells express EYFP, but some EYFP⁺ PG cells do not express CB (arrows). Scale bars, 20 μm. B, Intrinsic membrane properties were diverse in EYFP⁺ PG cells. Depolarizing and hyperpolarizing steps (500 ms) were injected in EYFP⁺ PG cells recorded in current-clamp at a membrane potential of approximately −60 mV. The three examples show individual traces from three cells responding differently to a step depolarization. C, Two biocytin-filled EYFP⁺ PG cells. Dashed lines delimit the glomerulus into which each PG cell projected its dendrites. D, Spontaneous EPSCs in an EYFP⁺ PG cell recorded at V_h = −75 mV. Onset, Individual spontaneous EPSCs (black traces) and their average (white trace) are superimposed. E, Responses evoked by stimulation of OSNs (arrow) recorded in an EYFP⁺ PG cell in the cell-attached (top) or whole-cell voltage-clamp mode at two intensities of stimulation indicated on the left (bottom; V_h = −75 mV). Individual traces are superimposed. Insets are zooms on the initial part of the response. Note that the latency and jitter of the responses decreased with the intensity of stimulation. The histogram plots the timing of individual EPSCs in responses evoked by the highest intensity of stimulation (bin width, 20 ms; stimulation at t = 0).

Figure 2.

A large fraction of CR⁺ PG cells is weakly connected to the glomerular network. A, Immunofluorescence staining for CR (red, middle) revealed that only a fraction of CR⁺ PG cells express EGFP (green, left) in CR–EGFP transgenic mice. Scale bar, 20 μm. B, Spontaneous EPSCs recorded in the whole-cell voltage-clamp mode (V_h = −75 mV) in a CR–EGFP⁺ PG cell. Onset, Individual events (black) superimposed with the average EPSC (white). Distribution histograms compare the frequency of spontaneous EPSCs in CR–EGFP⁺ PG cells (left) and Kv3.1–EYFP⁺ PG cells (right). C, Five consecutive ON-evoked whole-cell responses (V_h = −75 mV) in a CR–EGFP⁺ PG cell. This stimulus (200 μA, arrow) produced the largest responses. Distribution histograms compare the maximal amplitudes of ON-evoked responses in CR–EGFP⁺ PG cells (left) and Kv3.1–EYFP⁺ PG cells (right).

Figure 3.

PG cell functional diversity. A, ON stimulation (arrow) evoked different types of responses in EYFP⁻ PG cells. Type 1 PG cells (red traces) responded with a fast-onset monosynaptic EPSC (the inset is a zoom on the initial part of the response), whereas type 2 PG cells responded with plurisynaptic responses of variable amplitudes and durations (3 examples shown with blue traces). V_h = −75 mV. In each case, several traces are superimposed. B, The onset latency versus onset jitter of individual ON-evoked responses is plotted for EYFP⁺ PG cells (blue circles) and EYFP⁻ type 1 PG cells (red circles). Dark symbols and bars show the mean ± SD. Strong intensities of stimulation were used here (110 ± 101 μA for EYFP⁺ PGs and 239 ± 122 μA for EYFP⁻ PGs). C, Distribution histograms of the membrane resistance (top) and maximal amplitude of ON-evoked responses (eEPSC; bottom) in EYFP⁻ type 2 PG cells (n = 33). D, The duration of the ON-evoked response is plotted against the frequency of spontaneous EPSCs for each individual EYFP⁻ type 2 PG cells (white circles). Values for EYFP⁺ PG cells (blue circles) are also plotted for comparison. Durations were estimated from the fit of the latency histogram of each cell. E, Comparison of ON-evoked firing in EYFP⁺ and EYFP⁻ PG cells. Action potentials were monitored using paired loose cell-attached recordings (scheme). Middle, Example of a pair that responded to ON stimulation with a similar threshold but different time course. One representative trace (top) and raster plots of 80 consecutive episodes (bottom) are shown. Each point indicates a spike. OSNs were stimulated (arrow) with increasing intensities. Right, Summary graph of the experiment. Each column shows a pair. White and blue vertical bars show the range of stimulus intensities that evoked the firing of the EYFP⁻ and EYFP⁺ PG cells, respectively. Columns with no white bars (pairs 13–20) indicate pairs in which the EYFP⁻ PG cells did not respond to the stimuli tested.

Figure 4.

Synaptic connections of EYFP⁺ PG cells. A, Paired recordings between mitral (M) and EYFP⁺ PG cells. Left, Single action potentials evoked in the mitral cell produced fast EPSCs in the PG cell recorded at V_h = −70 mV. Six representative EPSCs are superimposed in gray, and the average EPSC across all trials is shown in black. Right, In another pair, a train of 10 action potentials evoked in the PG cell produced a barrage of ISPCs in the mitral cell recorded at V_h = 20 mV. Six sweeps are superimposed, and the average from all trials appears in black. B, Same for connected pairs of tufted (T) and EYFP⁺ PG cells. C, Same for connected pairs of ET and EYFP⁺ PG cells. IPSCs in A and B were recorded in the presence of NBQX (10 μm) and d-AP-5 (50 μm); C in ACSF.

Figure 5.

EYFP⁺ PG cells mediate lateral inhibition between mitral, tufted, and ET cells projecting into the same glomerulus. A, Paired recording between two mitral cells (M1 and M2) projecting into the same glomerulus. A current step (700 pA, 100 ms) was used to evoke a train of action potentials in M1 (top trace) while simultaneously monitoring voltage-clamped responses in M2 (scheme). Inward EPSCs were recorded at V_h = −80 mV (top traces, 5 traces superimposed). Outward IPSCs were recorded at V_h = 5 mV (bottom gray traces) in control conditions and after a transient puff of GBZ (50 μm, 100 ms) within the glomerulus. This protocol was repeated eight times. Traces before (control) and after (GBZ) the puff are superimposed. Average traces are shown in black. B, Same experiment in a pair of mitral (M) cell–tufted (T) cell. C, Same experiment in a pair of mitral (M) cell–ET cell. Experiments in A–C in the presence of d-AP-5 (50 μm) and CPCCOEt (100 μm). D, Summary plot showing the amplitudes of the averaged postsynaptic current obtained at negative potentials (V_h = −70/−80 mV) and at a potential near 0 mV, under control conditions and in the presence of GBZ. Each plot is a mitral (white), tufted (black), or ET (gray) cell. Horizontal lines show the means. E, Paired recording between a mitral cell (M) recorded in the whole-cell mode and an EYFP⁺ PG cell recorded in the loose cell-attached mode (PG). Both neurons projected into the same glomerulus. A train of action potentials induced by a current step (400 pA, 100 ms) in the mitral cell (top trace) produced the firing of the PG cell (middle trace). Bottom, Raster plots of spontaneous and evoked action potentials in the PG cell.

Figure 6.

Activity of EYFP⁺ PG cells in anesthetized mice. A, Picture of an EYFP⁺ PG cell first recorded in the cell-attached mode (top) and then filled with Alexa Fluor 594 (red) during whole-cell recording (bottom). The spontaneous firing from this cell (top trace) was recorded in the cell-attached configuration before breaking the membrane. The corresponding respiration cycle is shown below the trace (upward deflection corresponds to inhalation). Respiration cycles were aligned, and PSTHs per respiration cycle were constructed (right). Bin width, 2 ms. The orange trace is the best Gaussian fit of the histogram. The average respiration cycle waveform is shown below the PSTH. Bottom, Spontaneous EPSCs recorded in the whole-cell voltage-clamp configuration. Right, Thirty consecutive respiration cycles are superimposed, as are their corresponding whole-cell recordings. The white trace is a typical event. B, Competing EPSCs and IPSCs underlie the spontaneous rhythmic activity of tufted cells in vivo. Left, Two-photon image of an EYFP⁺ tufted cell during whole-cell recording in vivo. The patch pipette containing Alexa Fluor 594 (red) comes from the right. Middle traces, Slow outward inhibitory currents locked to the respiration rhythm (resp) dominated the spontaneous activity of this cell at V_h = 0 mV (top), whereas slow inward excitatory currents dominated the activity at V_h = −80 mV (bottom). Right, Fifty consecutive respiratory cycles with their correlated currents are superimposed. Black traces are the average.

Figure 7.

EYFP⁺ PG cells mediate OSN-evoked feedforward inhibition. A, Mitral cell (M) responses evoked by a stimulation of OSNs in control conditions (top) and in the presence of CPCCOEt (100 μm) and d-AP-5 (50 μm; bottom). Individual traces recorded at V_h = −10 mV are superimposed. B, ON-evoked feedforward inhibition in a tufted cell (T, top) and in an ET cell (bottom). Individual traces recorded at V_h = −10 mV in the presence of CPCCOEt (100 μm) and d-AP-5 (50 μm) are superimposed. Right, Summary plots comparing the half-width and latency of the ISPC in mitral, tufted, and ET cells. Horizontal bars are the average. C, Mitral cell responses evoked by minimal stimulation of OSNs that evoked a response (success) in ∼50% of the trials in the presence of CPCCOEt plus d-AP-5 before (top, control) and after (bottom, 4 μm) bath application of GBZ. Individual traces recorded at V_h = −35 mV are superimposed in gray. Average responses are shown in black. D, ON-evoked mitral cell responses in the presence of CPCCOEt plus d-AP-5 (top) and during local perfusion of GBZ (25 μm) inside the glomerulus to which the recorded mitral cell projected (middle trace). Single traces recorded at V_h = 5 mV. Note that intraglomerular perfusion of GBZ did not block spontaneous IPSCs. Bottom, Normalized area of the evoked IPSCs (measured during the first 100 ms) over time during local perfusion of GBZ within the glomerulus. Average from seven cells (5 mitral and 2 tufted cells). E, Same as in D, but GBZ was applied transiently (50 μm, 150 ms). Bottom, Summary plot for one cell for which GBZ was applied within the glomerulus (filled circles) or in the EPL below the glomerulus (open circles). F, Paired recording of a mitral cell and an EYFP⁺ PG cell projecting into the same glomerulus. The mitral cell was recorded in the whole-cell mode (V_h = 15 mV), whereas the PG cell was recorded in loose cell-attached mode. A minimal stimulation (18 μA) was used to stimulate OSNs in the presence of CPCCOEt plus d-AP-5. IPSCs in the mitral cell (top) were correlated with action potentials in the PG cell (middle). Bottom, Raster plot of every successful trial in the PG cell.

Figure 8.

Intraglomerular inhibition reduces output frequency and spike timing redundancy in mitral cells. A, Schematic of the experiment. B, Paired recording of ON-evoked responses in two mitral cells (M1 and M2) projecting into the same glomerulus. M1 (top traces) was recorded in whole-cell mode at V_h = 5 mV. Five responses are superimposed under control conditions (left) and during local perfusion of GBZ (25 μm) within the glomerulus (right). Middle, Raster plots of the firing of M2 recorded simultaneously in loose cell-attached mode. The dashed line indicates the stimulation (25 μA) time. Bottom, Mean firing rate across nine mitral cells recorded in cell-attached mode in control conditions (left, blue) and in the presence of GBZ (right, red). Thick traces are grand averages, and gray traces are averages for each cell. Bin width, 10 ms. C, Zoom on the initial phase of the responses shown in B. Bottom, PSTH for the same M2 cell in control conditions (top) and in the presence of GBZ (bottom). Bin width, 2 ms. D, Top, Cumulative density function (CDF) of the first three ISIs (first 4 spikes) in the 0–200 ms time window for nine mitral cells recorded in cell-attached mode. Blue colors correspond to the control data, and red colors indicate GBZ application. The three ISIs are shown in different columns. Bottom, Same as the top for the 200–400 ms time window. E, Significance testing of the difference between ISI distribution in control conditions and in the presence of GBZ. Two-sampled Kolmogorov–Smirnov test p values are shown in log scale. Null hypothesis is that the ISI distribution in control conditions and in the presence of GBZ are the same. x-Axis refers to the ISI, and the y-axis refers to the cell number. p values are higher within the 0–200 ms than in the 200–400 ms time period. Significantly different ISIs in control and GBZ are marked with a white dot (α values = 0.001, n = 9 mitral cells). F, In the presence of GBZ, the first two ISIs after stimulation become highly reliable across trials. CV_ISI of the first three ISIs. In F and G, each bar shows the mean ± SD of CV_ISI in control conditions (blue) and in the presence of GBZ (red) for the first three ISIs within the 0–200 and 200–400 ms time windows. G, In the presence of GBZ, in each trial, the first two ISIs after stimulation become highly reliable across cells in a simulated ensemble of nine mitral cells (n = 14 random trials).