Activity-dependent lateral-enhancement of spike times

a) Left: STA map with an excitatory region near the electrode location and presumably several surrounding inhibitory spots. Right: the significance map (P < 0.01 relative to shuffled data, see Methods).

b) Schematic illustration of the experimental setups. Photo-stimulation of each spot alone (hotspot or lateral spot conditions marked by orange and green text, respectively) or paired stimulation (marked in blue) using four different light intensities.

c) Raster plots and smoothed PSTHs of the response to light stimulation of the hotspot (top, orange) and paired stimulation (middle, blue). Note, the increase in spike time accuracy within and across trials when both spots are activated (middle panel). Light stimulation did not affect the average firing rate (lower panel). The effect of light stimulation of the lateral spot alone is shown in green.

d-e) Examples of MTCs time-frequency wavelet analysis from two different mice. Example pair #1 is the pair displayed in c. Both examples show a strong gamma rhythm following paired stimulation. In pair #1, gamma power peaked at ~58Hz, and pair #2 at ~48Hz.

f) Paired stimulation increased the spikes’ temporal precision. Mean ± SEM of the change in spikes entrainment at the population level (N = 319/511 values from all pairs and light intensities that significantly responded to light stimulation, P = 0.13 and P < 0.001 for shuffled (green) and real (purple) data, respectively; two-tailed paired t-test). In brown are values that exceeded the 95% confidence interval of the shuffled data distribution values of increased and decreased spike entrainment, respectively; confidence interval is marked by dashed black lines).

g) Lateral entrainment is activity-dependent. The moving average of the data shown in f is plotted as a function of the firing rate of the postsynaptic MTC (N = 50 values of increased entrainment). The increase in entrainment was largest when the neuron fired at ~40Hz. The color code is the same as in f. The shuffled data is shown in a dashed green line.

h) Spike entrainment does not depend on the distance between the MTC pair. No significant correlation was found between the increase in spike-entrainment and the distance from the hotspot (r = 0.05, P = 0.73, Spearman correlation; N = 50 values with significant increase in spike entrainment, brown dots in g).

Spike entrainment between and within trials

a) Light stimulation protocol. An image containing N randomly distributed light patches (N = 5 in this example) is projected on the dorsal bulb in each trial. The light patterns are shown in the bottom panel. The yellow rectangle marks the region around the recording electrode. The spiking activity and the respiratory signal are shown above. Multiplying each pattern by the firing rate it evoked and averaging across all trials gives the STA activity map (see Methods). Scale bars: 0.5 second; 110 µm.

b) Paired activation enhances the spike precision at the gamma range within each trial (P = 0.04, two-tailed paired t-test, see Methods). Delta entrainment was computed per trial (see Methods). The mean is marked with a dashed green line.

c) Power spectral density (PSD) during the light stimulation duration (100 ms) of the example shown in Figure 1c-d. Dashed lines represent the 95% confidence interval constructed using bootstrapping.

Activity-Dependent Lateral Suppression of MTCs is Confined in Space

a) An example of an MTC receptive field (STA map). The white rectangles mark the spots exposed to light stimulations. The hotspot location is marked with an electrode drawing.

b) Mean ± SEM firing rates of light stimulating pair #1 from a for each of the three conditions across all four light intensities. Activation of pair #1 (blue) caused a reduction in the recorded MTC firing rates only when the recorded MTC fired above ~25 spikes/sec. Lateral stimulation alone did not inhibit the recorded MTC (green). Zero denotes the baseline firing rate. *P < 0.05, **P < 0.01, two-tailed paired t-test.

c) PSTHs of pair #1 and #2 responses to light stimulations at four different light intensities. Paired stimulation of pair #1 evoked activity-dependent suppression. Paired stimulation of more distant neurons in pair #2 did not affect the recorded MTC firing rates across any tested light intensities. Light stimulation is marked with a blue bar (0.1 sec). PSTHs without a p-value showed no significant change between the firing rates of paired stimulation and hotspot stimulation alone.

d) Summary analysis of the effect of paired activation on MTCs firing rate. Each point marks the percentage change in firing rate (Y-axis) relative to the firing rate elicited by light stimulating the hotspot alone (X-axis). Suppressive effects (i.e., negative activity change) occurred mainly when the MTC fired in the gamma range (~30-80 Hz). Color-code denotes the distance between the light-activated MTC pair. Filled circles mark significant activity change (P < 0.05, two-tailed unpaired t-test, N = 51/319 data points). The red line shows the moving average. Only light intensities that elicited a significant light response were analyzed (319/511; P < 0.05, two-tailed paired t-test).

e) Lateral suppression degrades with distance. Spearman correlation between the change in firing rate for all significant inhibitory pairs (the filled circles in d, N = 51) and their distance to the hotspot (r = 0.45, P = 0.001).

f) Mean firing rate of the pixels located on each of the four diagonals in the Z-scored STA maps, centered relative to the ‘hotspot’. The centered map and the diagonals are shown in Supplementary Figure 2b. Gray lines show the response of the four sections taken from the origin towards the four corners (the four dashed lines in Supplementary Figure 2b). The blue line represents the average across all four sections. Zero denotes the hotspot location.

MTC lateral suppression is activity- and spatially-dependent

a) Light stimulating a lateral spot without stimulating the hotspot has no effect on the recorded neuron’s baseline firing rate, regardless of its firing rates. Color code as in Figure 2d.

b) Lateral inhibition is confined in space. All STA maps were centered at the hotspot location (N = 27 Z-scored maps, see Methods).

c) MTC lateral suppression is effective only when the target MTC is activated. Upper panels: an example of an MTC STA map and the corresponding significance map. Lower panels: the same maps recomputed by excluding all light patterns that stimulated the region around the hotspot (All pixels with significant excitation P < 0.01, relative to shuffled data). No inhibitory regions are detected after exclusion.

d) Population analysis across all MTC STA maps (N = 27) of the percent of inhibitory pixels in the original STA map and when we excluded the light patterns that hit the hotspot area. A pixel is defined as inhibitory if its value is below two standard deviations from the shuffled distribution (see Methods). The percent of inhibitory pixels drops considerably (P = 1.8e-6, two-tailed paired t-test), in the excitation-excluded map.

Spike entrainment and suppression are mediated by two different circuits

a) Light-stimulation of two different MTC pairings sharing the same postsynaptic MTC. Light-activating pair #1 (left) caused potent entrainment (P < 0.05, two-sample bootstrap) without affecting the light-evoked firing rate (P = 0.59, two-tailed paired t-test), whereas light-activating pair #2 (right) suppressed the MTC firing rate (P = 0.036, two-tailed paired t-test), without affecting the spikes precision (P > 0.05, two-sample bootstrap).

b) Two different light intensities were applied to pair #1, which had differential effects on suppression and entrainment. High light intensity increased spike entrainment without affecting the firing rate (left panel in a). In contrast, lower intensity reduced the light-evoked firing rate (P = 0.007, two-tailed paired t-test), with no effect on the spikes’ gamma entrainment (P > 0.05, two-sample bootstrap).

GC activation increases MTC synchrony in an activity-dependent and location-independent manner

a) Cre-dependent AAV injected into the GC layer (GCL) of Gad2-Cre mice. A representative example showing restricted ChR2 expression in the GCL and the external plexiform layer (EPL), into which GCs extend dendrites (red, mCherry-ChR2; blue, DAPI). MCL, mitral cell layer; GL, glomerular layer. Scale bar, 0.1 mm.

b) Schematic illustrations of the experimental setup. Left: Three weeks post injection, MTCs were recorded while light-activating subsets of GCs. Right: MTC activity was recorded in response to odor stimulation alone (purple) or combined with light-activation of GC columns near the recording electrode or distant from it (blue). Scale bar, 330µm.

c) MTC spike synchrony to the gamma oscillation (ΔPPC1) significantly increases when we light-stimulated columns of GCs compared to odor-only stimulation (N = 54, P = 0.0016, two-tailed paired t-test). Only cell-odor pairs that were significantly odor-excited were analyzed (N = 18/31 cell-odor pairs; three spots were stimulated per cell odor-pair).

d) MTC spike entrainment does not depend on the GC location. The relation between the change in PPC1 caused by odor and GC stimulation as a function of the distance of the light-stimulated spot from the recording electrode. No significant correlation was found (N = 54 values from 18 cell-odor pairs, r = −0.03, P = 0.84, Spearman correlation). Zero denotes the spot above the recording electrode.

e) MTC spike entrainment is activity-dependent. The change in synchrony peaked when MTCs fired at ~25Hz.

f) Odor-evoked spike reference analysis. Two spike raster plots are shown, for odor only (left, purple), and odor with light-activation of a GC column (right, blue). In each raster plot, spikes are plotted relative to a randomly chosen spike during the odor presentation period (N = 400 spikes references, see Methods). Note, the potent spike entrainment when GCs are activated. This analysis was performed on a cell that had a sufficiently high firing rate. This cell is likely a tufted cell due to its potent entrainment at the high gamma range, as shown in (Burton and Urban, 2021; Fukunaga et al., 2014).

g) The power spectral densities (PSD) for the two conditions in f. A multi-taper analysis of the circular convolution of each spike raster plot was used to compute the PSD (see Methods).

MTC-to-MTC firing rate suppression is not mediated by GCs

a) MTC-to-MTC lateral suppression is not mediated by GCs. Similar to Figure 2d, the change in odor-evoked activity following GC activation is plotted as a function of the recorded MTC odor-evoked firing rates. Filled blue circles denote significant activity change (P < 0.05, two-tailed unpaired t-test). A moving average is shown in red.

b) Odor-evoked firing rates are not suppressed when a GC column is activated, irrespective of the odor-evoked MTC firing rate. Two plots are shown for activation of proximal (i.e., in the electrode vicinity, N = 18, blue) and distal GCs (all other locations, N = 36, green). The slopes of a linear fit and their P-values are plotted in each panel.