Respiration organizes gamma synchrony in the prefronto-thalamic network

Abstract

Multiple cognitive operations are associated with the emergence of gamma oscillations in the medial prefrontal cortex (mPFC) although little is known about the mechanisms that control this rhythm. Using local field potential recordings from cats, we show that periodic bursts of gamma recur with 1 Hz regularity in the wake mPFC and are locked to the exhalation phase of the respiratory cycle. Respiration organizes long-range coherence in the gamma band between the mPFC and the nucleus reuniens the thalamus (Reu), linking the prefrontal cortex and the hippocampus. In vivo intracellular recordings of the mouse thalamus reveal that respiration timing is propagated by synaptic activity in Reu and likely underlies the emergence of gamma bursts in the prefrontal cortex. Our findings highlight breathing as an important substrate for long-range neuronal synchronization across the prefrontal circuit, a key network for cognitive operations.

Figure 1

Respiration modulates gamma synchrony in the prefronto-thalamic network. (a) mPFC local field potential and its wavelet spectrogram showing transient gamma oscillations nested in the respiration-related potential (RR). Dotted box denotes the burst shown in this figure (d). (b) DAPI-labelled, coronal sections showing electrode tracks in the deep layers of the mPFC, the midline thalamus and the CA1 region of the hippocampus. Cru–sulcus cruciatus, Prs–sulcus praesylvius, CI–capsula interna, Reu–nucleus reuniens. (c) Prefrontal (left) and Reu (right) gamma power versus phase of respiration. (d) An expanded view of a single gamma burst from this figure (a). (e) Polar plot showing the phase preference of gamma bursts to the trough of the respiration-related potential (RR) for all detected gamma bursts. The black line is the mean resultant vector (p_Rayleigh < 0.001). (f) Power spectra from multiple cortical areas during wake episodes. (g) Gamma power was significantly higher in the mPFC than in other cortical sites (one-way Tukey-Kramer post-hoc ANOVA, p < 0.0001). (h) Example comodulograms, showing the strength of phase-amplitude coupling in various recorded sites during wake. (i) Population data of modulation indices calculated from all recordings. Gamma-RR PAC was higher in the mPFC than in all other recorded areas (except RE, post-hoc ANOVA, p <0.001). Gamma-RR PAC in Reu was significantly higher than in the hippocampus, somatosensory, visual, association and auditory cortex but not motor or mPFC (post-hoc ANOVA, p < 0.001). (j) Example cross-correlograms computed from bandpassed (30-80 Hz) signals of the mPFC, Reu and hippocampus in wake, NREM and REM states. Cross-correlogram were computed for mPFC-Reu pairs s (red) and hippocampus-Reu pairs (blue). (k) Signal coherence between the mPFC and Reu and between Reu and the hippocampus was highest in gamma and RR range. (l) Reu lagged the mPFC by a significantly longer latency in wake epochs than in NREM and REM. (m) Reu led the hippocampus by a significantly longer latency in wake epochs than in NREM and REM. (n) An example recording segment showing relatively high synchrony in the mPFC-thalamo-hippocampal network during periods of gamma activity. (o) The mean wavelet coherence between the mPFC and Reu, centered to the onset of exhalation (time = 0 s) showing the modulation of prefronto-thalamic gamma coherence by respiration phase. (p) The mean wavelet coherence between Reu and the hippocampus, centered to the onset of exhalation (time = 0 s) showing the modulation of thalamo-hippocampal gamma coherence by respiration phase.

Figure 2

Respiration-gamma coupling is selective to wakefulness. (a) LFP recording of the mPFC during a wake-to-NREM transition (top) and its wavelet transform (bottom) showing the attenuation of gamma bursts during the transition. (b) NREM-to-REM transition. Sigma and delta power attenuated as the mPFC transitioned to REM. (c) REM-to-wake transition. The 1-Hz, RR oscillation in the mPFC signal emerged in wake although overall gamma power remained comparatively similar to REM levels. (d) The Fourier transform of the mPFC signal from one recording session, separated according to states of vigilance. Wakefulness was characterized by a peak in the 30–40 Hz gamma range, NREM by high delta and sigma power and REM by a low, broad peak in gamma. (e) mPFC gamma power was highest in wakefulness (one-way Tukey-Kramer post-hoc ANOVA, p < 0.0001). (f) mPFC firing rates were highest in wake and REM and lowest in NREM (ANOVA, p <0.001). (g) Variability of gamma power in 5-second windows across states of vigilance, measured as standard deviations. (h) Gamma power versus firing rate across states of vigilance, plotted as measurements in 5-second windows. (i) Delta power decayed exponentially with the emergence of gamma in REM and wake (nonlinear least-square regression, r = 0.92). Dots are 5-second means of gamma or delta-band power in the mPFC signal. (j) Sigma power decayed exponentially with the emergence of gamma in REM and wake (nonlinear least-squares regression, r = 0.97). (k) Normalized gamma power measurements, centered to the onset of spindles, showing increase gamma power during spindles (top). Normalized gamma power measurements, centered to the start of UP states (down). Data are mean ± s.d. Inset: gamma power was significantly higher during UP states compared to down (ANOVA, p <0.001). (l) Phase-amplitude coupling of the mPFC in wake, NREM and REM states. White arrow indicates RR-gamma coupling and red arrow indicates slow-oscillation-spindle coupling. (m) PAC modulation indices in different states, calculated for gamma amplitude (30–60 Hz) and phase of the respiration-related oscillation (0.2–2 Hz). RR-gamma modulation was higher in wakefulness than in NREM and REM (post-hoc ANOVA, p <0.001).

Figure 3

Respiration modulates firing in the nucleus reuniens. (a) Example recording segment showing the modulation of Reu multi-unit activity by the phase of respiration-related potential. (b) The respiration-related potential modulates the firing rate of Reu but not mPFC single-units. From top to bottom: Peri-event firing rate dynamics of a Reu single-unit centered to the peak of the LFP respiration-related oscillation. Dots are single-unit discharges organized from bottom to top as sweeps around the onset of the LFP peak. The shaded area is the firing rate mean ± s.d. of 23 Reu and 21 mPFC single-units, normalized to the mean firing rate of each unit. Black dots indicate bins that were significantly different (t-test, p < 0.01) in comparison to equivalent bins obtained from shuffled peri-event spike histograms. Right: firing rate dynamics of each Reu and mPFC single-units, referenced to the respective LFP peak and normalized to the mean firing rate of each unit. (c) Expanded view of a single gamma burst recorded from the cat mPFC, showing the LFP gamma oscillation and the associated single-unit activity, phase-locked to gamma cycles. Red trace is the mPFC signal and its bandpass (300-8000 Hz, below). The spectrogram (top) shows the wavelet transform of the trace. (d) The interspike interval distribution of a prefrontal single-unit within gamma bursts. Note the ~25 ms interspike intervals indicative of gamma periodicity. Inset shows several traces of the detected spike. Horizontal bar is 1 ms, vertical bar is 20 µV. (e) Example traces of mPFC and reuniens spike/LFP recordings showing the phase preference of prefrontal and reuniens single units for the trough of gamma cycles (left). Polar plot shows the firing probability of a reuniens single-unit on the phase of the gamma cycle (right).

Figure 4

Respiration and the slow oscillation comodulate synaptic activity in the nucleus reuniens. (a) Coronal section of the ventral midline thalamus showing a reuniens neuron labelled with neurobiotin, revealed immunohistochemically by DAB horseradish peroxidase. Reu–nucleus reuniens; mtt–mamillothalamic tract; 3V–ventral third ventricle. (b) In vivo intracellular recording of a reuniens cell labelled in A. (c) Reuniens activity during exhalation in the DOWN state of the slow oscillation at basal and hyperpolarized levels (−0.6 nA and −1.5 nA current injection). (d) Reuniens intracellular events show preference to exhalation phase. Membrane depolarization, EPSPs and action potential discharges were non-uniformly distributed around respiration phase (Rayleigh test, p < 0.001 for EPSP and p <0.001 for action potentials, n = 12). (e) Tri-modal distribution of the membrane potential, indicating hyperpolarizing and depolarizing events typical of the slow oscillation (left and right arrows) and respiration-related EPSPs (middle arrow). (f) Reuniens activity during the cortical UP state. Under hyperpolarizing current injection, UP states induced low-threshold spikes that were not observed during respiration cycles. (g) The Reu membrane potentials was modulated by the slow oscillation. (h) Hyperpolarizing pulses given in DOWN states, during inhalation and exhalation. The membrane voltage response to the same current injection was smaller during exhalation. Input resistance (Ri) was higher during inhalation than exhalation. (i) The membrane potential was significantly higher during the exhalation phase than during inhalation (t-test, p < 0.001). (j) A layer V pyramidal neuron of the mPFC (prelimbic cortex) labelled with neurobiotin, revealed immunohistochemically by streptavidin-Texas Red^TM conjugate. PL – prelimbic cortex; I, II/III, V, VI - cortical laminae. (k) In vivo intracellular recording of a prefrontal cortical cell labelled in J. (l) mPFC intracellular activity is locked to UP states but not to respiration From left to right: uniform membrane depolarization and action potential discharge around respiratory phase (p_Rayleigh= 0.494, n = 12). Non-uniform membrane depolarization and action potential discharge around slow oscillation phase (p_Rayleigh<0.0001, n = 12). (m) Bi-modal distribution of the membrane potential of the mPFC cell labelled in J, indicating hyperpolarizing and depolarizing events typical of the slow oscillation. (n) Spectral content of the reuniens LFP signal showing comodulation of the local field by the slow oscillation and the respiratory cycle.