Contribution of parvalbumin and somatostatin-expressing GABAergic neurons to slow oscillations and the balance in beta-gamma oscillations across cortical layers

Abstract

Cortical interneurons are classified into several subtypes that contribute to cortical oscillatory activity. Parvalbumin (PV)-expressing cells, a type of inhibitory interneuron, are involved in the gamma oscillations of local field potentials (LFPs). Under ketamine-xylazine anesthesia or sleep, mammalian cortical circuits exhibit slow oscillations in which the active-up state and silent-down state alternate at ~1 Hz. The up state is composed of various high-frequency oscillations, including gamma oscillations. However, it is unclear how PV cells and somatostatin (SOM) cells contribute to the slow oscillations and the high-frequency oscillations nested in the up state. To address these questions, we used mice lacking glutamate decarboxylase 67, primarily in PV cells (PV-GAD67 mice) or in SOM cells (SOM-GAD67 mice). We then compared LFPs between PV-GAD67 mice and SOM-GAD67 mice. PV cells target the proximal regions of pyramidal cells, whereas SOM cells are dendrite-preferring interneurons. We found that the up state was shortened in duration in the PV-GAD67 mice, but tended to be longer in SOM-GAD67 mice. Firing rate tended to increase in PV-GAD67 mice, but tended to decrease in SOM-GAD67 mice. We also found that delta oscillations tended to increase in SOM-GAD67 mice, but tended to decrease in PV-GAD67 mice. Current source density and wavelet analyses were performed to determine the depth profiles of various high-frequency oscillations. High gamma and ripple (60-200 Hz) power decreased in the neocortical upper layers specifically in PV-GAD67 mice, but not in SOM-GAD67. In addition, beta power (15-30 Hz) increased in the deep layers, specifically in PV-GAD67 mice. These results suggest that PV cells play important roles in persistence of the up state and in the balance between gamma and beta bands across cortical layers, whereas SOM and PV cells may make an asymmetric contribution to regulate up-state and delta oscillations.

Keywords: CSD; PV cells; beta oscillation; gamma oscillation; mouse; neocortex; slow oscillation; wavelet analysis.

Figure 1

Multi-channel probe positions and LFP recordings. (A) The multi-channel probe was positioned across the cortico-hippocampal area (NC, neocortex; upper, upper layers; deep, deep layers; CA1, hippocampal area around the Cornu Ammonis areas; DG, hippocampal area around the dentate gyrus). The hole on the sagittal section (red square) shows the position of the probe. (B) The temporal profile example of the LFP recorded from the probe in (A).

Figure 2

LFP and MUA temporal profile and the up state detected from the WPS of the LFP. (A) The LFP temporal profile example from channel 14 (top), the MUA temporal profile from all channels (middle), and the WPS (bottom) of a control mouse (semi-transparent red, up state detected from the WPS; colored bar, normalized amplitude of WPS). (B) PV-GAD67 mouse under the same conditions described in (A). The up state tended to be shorter and the down state longer in duration compared with the control mouse. (C) SOM-GAD67 mouse under the same conditions described in (A). The down state tended to be longer in duration compared with that in the control mouse.

Figure 3

A comparison of the state duration, delta oscillation, and firing rate of MUA of PV-GAD67 and SOM-GAD67 with control mice. (A) A comparison of the normalized duration average of the up state between PV-GAD67 or SOM-GAD67 mice and control mice (red bar, the normalized duration of the up state of PV-GAD67 mice; green bar, the normalized duration of the up state of SOM-GAD67 mice; black bar, the normalized duration of the control mice). (B) A comparison of the normalized down state average under the same conditions described in (A). (C) A comparison of the normalized firing rate average of MUA under the same conditions described in (A). (D) A comparison of the normalized delta oscillation power average under the same conditions described in (A). Welch's two-sample t-test together with Benjamini-Hochberg correction, N = 5 in PV-GAD67 mice, N = 4 in SOM-GAD67 mice, N = 9 in control mice; ^**P < 0.01, error bars indicate SEM.

Figure 4

CSD depth and temporal profile in the up state. (A) LFPs example from all channels (upper) and the CSDs example calculated from the LFPs in control mice. (B) Up state start-triggered average example of LFPs (upper) and that of CSDs (lower) in control mice. The probe position according to the CSDs is placed on right side of the CSDs. The same calculations as described in (A) and (B) for the (C,D) PV-GAD67 mouse and for the (E,F) SOM-GAD67 mouse (semi-transparent red, up state; NC, neocortex; hippo, area around the hippocampus; colored bar, normalized CSD pattern).

Figure 5

The CSD WPS depth profiles in each high frequency band. (A) CSDs (top) and WPS example calculated from the CSDs sorted according to each high frequency band (from the bottom alpha, 7–14 Hz; beta, 15–30 Hz; low gamma, 30–60 Hz; high gamma, 60–90 Hz; ripple, 100–200 Hz) in control mice. (B) Up state start-triggered average of CSDs (upper) and of WPSs (lower five rows) in control mice. N, neocortex; H, region around the hippocampus; semi-transparent red, up state; color bar, relative WPS. (C) The same as A, but for PV-GAD67 mice. (D) The same as (B), but for PV-GAD67 mice. (E) The same as A, but for SOM-GAD67 mice. (F) The same as (B), but for SOM-GAD67 mice.

Figure 6

Comparison of the CSD wavelet power spectrum depth profile in each high frequency band between PV-GAD67 or SOM-GAD67 and control mice. (A) Comparison of the CSD WPS depth profile average in each high frequency band. Comparisons between the PV-GAD67 and control mice and between the SOM-GAD67 and control mice are shown in the upper and lower panels, respectively. The frequency range in each band from left (alpha) to right (ripple) is the same as that described in Figure 5. Red line, relative WPS difference between PV-GAD67 and control mice; green line, the difference between SOM-GAD67 and control mice; central line, the relative WPS of the control mice was set at zero. (B) Comparison of CSD WPS depth profile average among high-frequency bands as (A), and its statistical results. Welch's two-sample t-test together with Benjamini-Hochberg correction, N = 5 PV-GAD67 mice, N = 4 SOM-GAD67 mice, N = 9 control mice; ^*P < 0.05, **P < 0.01; error bars indicate SEM.