Two GABAergic intraglomerular circuits differentially regulate tonic and phasic presynaptic inhibition of olfactory nerve terminals

Abstract

Olfactory nerve axons terminate in olfactory bulb glomeruli forming excitatory synapses onto the dendrites of mitral/tufted (M/T) and juxtaglomerular cells, including external tufted (ET) and periglomerular (PG) cells. PG cells are heterogeneous in neurochemical expression and synaptic organization. We used a line of mice expressing green fluorescent protein under the control of the glutamic acid decarboxylase 65-kDa gene (GAD65+) promoter to characterize a neurochemically identified subpopulation of PG cells by whole cell recording and subsequent morphological reconstruction. GAD65+ GABAergic PG cells form two functionally distinct populations: 33% are driven by monosynaptic olfactory nerve (ON) input (ON-driven PG cells), the remaining 67% receive their strongest drive from an ON-->ET-->PG circuit with no or weak monosynaptic ON input (ET-driven PG cells). In response to ON stimulation, ON-driven PG cells exhibit paired-pulse depression (PPD), which is partially reversed by GABA(B) receptor antagonists. The ON-->ET-->PG circuit exhibits phasic GABA(B)-R-independent PPD. ON input to both circuits is under tonic GABA(B)-R-dependent inhibition. We hypothesize that this tonic GABA(B)R-dependent presynaptic inhibition of olfactory nerve terminals is due to autonomous bursting of ET cells in the ON-->ET-->PG circuit, which drives tonic spontaneous GABA release from ET-driven PG cells. Both circuits likely produce tonic and phasic postsynaptic inhibition of other intraglomerular targets. Thus olfactory bulb glomeruli contain at least two functionally distinct GABAergic circuits that may play different roles in olfactory coding.

FIG. 1.

GAD65+ cells in the glomerular layer are periglomerular (PG) cells. A: neurons expressing high levels of green fluorescent protein (GFP) are easily visualized in the glomerular layer using an epifluorescent microscope. B: differential interference contrast (DIC) optics, allowing whole cell patch-clamp recording and biocytin filling of identified neurons (P, patch pipette attached to GFP+ cell, → in A and B). C: photograph of biocytin-labeled processes and spines (→). D: reconstructions of 2 GAD65+ biocytin-labeled cells showing a cell that received bursts of spontaneous excitatory postsynaptic currents [sEPSCs; external tufted (ET) driven] on the right and one receiving single sEPSCs [olfactory nerve (ON) driven] on the left. These cells exhibit a wide range of dendritic structures consistent with PG cells. Scale bar = 50 μm. E: the intrinsic properties of GAD65+ PG cells are similar to those described in rat. F: not all PG cells express GFP. The negative PG cells were physiologically similar to GFP-positive PG cells. G: Scholl analysis, which measures the number of intersections (ordinate) with spheres of increasing radius from the soma (abscissa), shows the majority of GAD65+ dendrites are within 50–60 μm of the soma.

FIG. 2.

GAD65+ cells differ in their sEPSC inputs. A: in voltage clamp, ∼33% of GAD65+ PG cells exhibit spontaneous single EPSCs. Bottom trace: an expansion of the EPSCs in the boxed region. B: the remaining 67% of cells exhibit spontaneous bursts of EPSCs. Bottom trace: an expansion of the boxed region showing a burst of 4 sEPSCs. C: burst-EPSC cells were identified on the basis of observed EPSC bursts exceeding chance by ≥2.5 SD (- - -; see methods). The histogram shows the number of SDs (abscissa) observed sEPSC bursts exceed chance for 114 GAD65+ PG cells (burst-EPSC cells, ▪, n = 56; single-EPSC cells, , n = 58). Single-EPSC cells cluster around 0 SD, which indicates no significant differences between observed bursts and chance.

FIG. 3.

GAD65+ PG cells receive either mono- or polysynaptic inputs. A: in voltage clamp, single-EPSC cells respond to ON stimulation with a single, short-latency EPSC. B: burst-EPSC cells responded to ON stimulation with a burst of EPSCs at longer more variable latency. ON-evoked responses show 6–10 superimposed EPSCs. C: histogram showing the number of EPSCs elicited by ON stimulation (abscissa) in 114 GAD65+ PG cells. Single-EPSC cells (green bars, n = 58) exhibited averages of only 1 or 2 EPSCs per stimulation, whereas the majority of burst-EPSC cells (red bars, n = 56) exhibited an average of 4–10 EPSC/stimulation. Regions of single- and burst-EPSC cell overlap in the histogram are shown in yellow. D: Hectograph showing latency to ON-evoked EPSCs (abscissa) in burst-EPSC cells (red bars, n = 58) is significantly greater than single-EPSC cells (green bars, n = 56). However, while burst-EPSC cells generally show longer latency than single-EPSC cells, there is considerable overlap in the distribution of latencies across the population (yellow). Latency was measured from the artifact generated by ON stimulation to the onset of the 1st evoked current evoked by minimum effective stimulation intensity. E: histogram showing the jitter in response to ON-evoked EPSCs (abscissa) in burst-EPSC cells (red bars, n = 58) has no overlap with response jitter in single-EPSC cells (green bars, n = 56). Jitter >309 μs only occurred in burst-EPSC cells, whereas jitter <298 μs only occurred in single-EPSC cells. F: single-EPSC cells respond to ON stimulation with a single, short-latency EPSC after a lateral cut through the EPL (severing mitral cell apical dendrites). G: burst-EPSC cells continue to respond to ON stimulation with a burst of EPSCs after a lateral cut through the EPL. H: in single-EPSC cells, increasing stimulus strength above minimal effective stimulation (MES) never produced EPSC bursts (top trace, MES; bottom trace, 2× MES). I: burst-EPSC cells respond to increased stimulus intensity with reduced latency and jitter (top trace, MES; bottom trace, 2× MES). J: graph showing latency in ON-driven (Ond) cells (green) is unchanged with increased stimulus strength (green), whereas ET-driven (Etd) cells (red) exhibit reduced latency. K: graph showing negligible changes in jitter (green) with increased stimulus strength in ONd cells, whereas ETd cells (red) exhibit reducing jitter as stimulus strength increases.

FIG. 4.

Spontaneous EPSCs, ON-evoked latency, jitter, and pattern of response are correlated. A–C: scatter plots of these parameters showing single-EPSC cells (green, n = 56) form a separate cluster from burst-EPSC cells (red, n = 58). D: single-EPSC (green bars) and burst-EPSC cells (red bars) are unambiguously assigned to 2 clusters using a K-means test (dashed line and arrows). K-means separates cells into clusters but does not indicate the probability of a cell belonging to 1 or the other cluster. Thus as a 2nd pass, we applied a Fuzzy clustering test (Euclidean distances) which calculates the probability of each cell being part of clusters 1 or 2. The histogram shows this test resulted in the same separation into 2 clusters consisting of the 58 single-EPSC in 1 cluster and the 56 burst-EPSC cells in a 2nd cluster. The probability distribution indicates that most cells clearly separated into 1 or the other cluster with only 2 cells exhibiting somewhat ambiguous classification (probabilities between 0.2 and 0.8. E: schematic showing ON→PG and ON→ET→PG glomerular circuits.

FIG. 5.

GABA_B-dependent ON-evoked paired-pulse depression (PPD) in ONd GAD65+ PG cells. A: schematic showing the experimental design with ONd PG cells whole cell patch clamped and a stimulating electrode placed in the olfactory nerve. B: PPD in response to ON stimulation occurs in ONd-PG cells (top trace). The figure shows 5 individual traces superimposed with interstimulus intervals between the conditioning and test pulses of 25, 50, 75, 100, 200, and 400 ms. In the presence of the GABA_B receptor antagonist CGP55845 PPD is attenuated (bottom traces). C: mean PPD plotted as the ratio of EPSC amplitude evoked by the test pulse to conditioning pulse (a ratio of 1.0 indicates no PPD) vs. interstimulus interval. The black line shows PPD under normal conditions (black asterisks indicates significant PPD at 1 asterisk, P < 0.01; 2 asterisks, P < 0.001). The red line shows the ratio of EPSCs evoked by the test pulse to conditioning pulse in the presence of CGP55845 (green asterisks show significantly difference from normal PPD at 1 asterisk, P < 0.05). PPD is not completely eliminated in the presence of CGP55845 (significant PPD remains at interstimulus intervals of 200–1,000 ms, red asterisks indicates significance at 1 asterisk, P < 0.01; 2 asterisks, P < 0.001). All statistical comparisons were performed using multifactorial ANOVA. D: step-depolarization (prepulse) of an ONd PG cell (from holding membrane potential –70 to 0 mV) was used to evoke GABA release prior to stimulation of ON. The magnitude of ON-evoked EPSCs was reduced by prior step depolarization (center left trace) compared with control EPSCs (left trace). This reduction was blocked, and enhanced over control, by application of the GABA_B receptor antagonist, CGP55845 (center right and right traces). E: overlay of ON-evoked EPSCs in ONd PG cells (black trace), when preceded by step depolarization (green trace) and when preceded by step depolarization in the presence of CGP55845 (red traces). F: quantification of step depolarization evoked depression in ONd PG cells (n = 8 cells). ON-evoked EPSCs were significantly depressed by a prior step-depolarization (green bar, 2 asterisks, P < 0.001, ANOVA). Addition of CGP55845 increased the response magnitude and blocked step depolarization evoked depression (3 asterisks, P < 0.0001). Washout of the CGP55845 restored EPSC amplitude close to control (no statistical difference between control and wash). G: ONd PG cells are under tonic GABA_B dependent inhibition. ON-evoked EPSC amplitude is increased in the presence of 10 μM CGP55845 (black, control EPSC, red CGP55845). H: quantification of tonic inhibition in ONd PG cells (n = 5 cells). The addition of CGP55845 increased the amplitude of ON-evoked EPSCs by 27.6%.

FIG. 6.

ETd GAD65+ PG cells do not show GABA_B dependent ON-evoked PPD. A: schematic showing the experimental design with ETd PG cells whole cell patch clamped and a stimulating electrode placed in the olfactory nerve. B: PPD in response to ON stimulation in ETd GAD65+ cells is not modulated by GABA_B receptors (top trace shows normal conditions; lower red trace is in the presence of CGP55845). C: mean PPD plotted as the ratio of EPSCs evoked by the test pulse to conditioning pulse (a ratio of 1.0 indicates no PPD) vs. interstimulus interval. The black line shows ETd PG cells do exhibit PPD under normal conditions (black asterisks indicates significant PPD at 1 asterisk, P < 0.05; 2 asterisks, P < 0.01), the red line shows the ratio of EPSCs evoked by the test pulse to conditioning pulse in the presence of CGP55845. There is no significant effect of CGP55945 on the ETd PG cell PPD at any interstimulus interval. D: overlay of ON-evoked EPSCs in an ETd PG cell. Left: 8 superimposed control EPSCs (gray) and mean EPSC (black). Right: 8 superimposed EPSCs (red) and mean EPSC (black trace) when preceded by step depolarization. E: quantification of EPSCs in ETd PG cells (n = 7 cells) show ON-evoked EPSCs are unaffected by a prior step-depolarization. F: ETd PG cells are under tonic GABA_B dependent inhibition. ON-evoked EPSC amplitude is increased in the presence of 10 μM CGP55845. Left: 10 superimposed control EPSCs (gray) and mean EPSC (black). Right: 10 superimposed EPSCs (red) and mean EPSC (black) in the presence of CGP55845. G: quantification of tonic inhibition in ETd PG cells (n = 5 cells). The addition of CGP55845 increased the integrated area of ON-evoked EPSCs by 25%.

FIG. 7.

ET cells show PPD that is independent of GABA_B activation. A: schematic showing the experimental design with ET cells whole cell patch clamped and a stimulating electrode placed in the ON. B: PPD in response to ON stimulation occurs in ET cells (top trace). The figure shows 5 individual traces superimposed with interstimulus intervals between the conditioning and test pulses of 50, 100, 200, and 400 ms. In the presence of the GABA_B receptor antagonist CGP55845, PPD is still present. C: mean PPD plotted as the ratio of EPSCs evoked by the test pulse to conditioning pulse (a ratio of 1.0 indicates no PPD) vs. interstimulus interval. The black line shows PPD under normal conditions (black asterisks indicates significant PPD at 4 asterisks, P < 0.0001; 2 asterisks, P < 0.001), the red line shows the ratio of EPSCs evoked by the test pulse to conditioning pulse in the presence of CGP55845 (red lines; 10 μM, n = 5 cells). There is no significant effect of CGP55945 at any interstimulus interval. D: ET cells are also under tonic GABA_B-dependent inhibition. ON-evoked EPSC amplitude is increased in the presence of 10 μM CGP55845 (black, control EPSC, red CGP55845). E: quantification of tonic inhibition in ET cells (n = 5 cells). The addition of CGP55845 increased the amplitude of ON-evoked EPSCs by 26% (red bar). F: spontaneous EPSCs are also under tonic inhibition. sEPSCs in an ET cell (black) increase amplitude in the presence of 10 μM CGP55845 (red). G: the cumulative probability plot for sEPSC frequency, which calculates the percentage (ordinate) of total EPSCs with frequencies less than the values shown on the abscissa, shows an increase in frequency of sEPSCs (n = 5 cells, P < 0.01). Inset: the normalized population means. H: cumulative probability graph of sEPSC amplitude. The addition of CGP55845 increased sEPSC amplitude by 22%.