Circadian regulation of human cortical excitability

Abstract

Prolonged wakefulness alters cortical excitability, which is essential for proper brain function and cognition. However, besides prior wakefulness, brain function and cognition are also affected by circadian rhythmicity. Whether the regulation of cognition involves a circadian impact on cortical excitability is unknown. Here, we assessed cortical excitability from scalp electroencephalography (EEG) responses to transcranial magnetic stimulation in 22 participants during 29 h of wakefulness under constant conditions. Data reveal robust circadian dynamics of cortical excitability that are strongest in those individuals with highest endocrine markers of circadian amplitude. In addition, the time course of cortical excitability correlates with changes in EEG synchronization and cognitive performance. These results demonstrate that the crucial factor for cortical excitability, and basic brain function in general, is the balance between circadian rhythmicity and sleep need, rather than sleep homoeostasis alone. These findings have implications for clinical applications such as non-invasive brain stimulation in neurorehabilitation.

Figure 1. Experimental protocol.

Participants underwent a 29 h sustained wakefulness protocol under constant routine conditions (no time-of-day information, constant dim light (<5 lux), external temperature and semi-recumbent posture, regular liquid isocaloric intake, sound proofed rooms). TMS-evoked EEG potential (TEP) were recorded eight times (>250 trials per session; violet triangles ) and test batteries including the psychomotor vigilance task (PVT; turquoise circle ) were completed 12 times. TMS/EEG sessions were scheduled throughout the 29-h period with higher frequency around the wake-maintenance (WMZ) and sleep-promoting zones (SPZ), the timing of which was predicted based on habitual sleep times (data realigned a posteriori). During TMS/EEG sessions, participants performed a visual vigilance task consisting in maintaining a constantly moving cursor in the centre of a computer screen to assess simultaneous performance and exclude vigilance lapses. Saliva samples were collected hourly for melatonin and cortisol assays, together with subjective sleepiness and affect measures. Relative clock time displayed is for a participant with a 23:00–07:00 sleep–wake schedule. Prep: 5 preparatory hours, including test battery task practice (<5 lux). Baseline night: 8 h night of sleep in darkness at habitual sleep times and under EEG recording.

Figure 2. Non-linear changes in cortical excitability with wakefulness extension.

(a) TMS-evoked potentials (TEP; 0–30 ms post TMS) measured at the electrode closest to the hotspot, averaged in each of the eight TMS/EEG sessions, in a representative participant (habitual sleep time: 23:00–07:00). Hotspot location was provided by the neuronavigation system. Time course of TEP amplitude (b) and slope (c) with respect to the circadian cycle. Data were averaged (mean±s.d.) after standardization (z-score) and realignment to individual circadian phase (n=22; melatonin secretion onset=0°). Mean z-scored melatonin profile is displayed in grey with respect to circadian phase (bottom X axis). The top x axis indicates relative clock time for a participant with a 23:00–07:00 sleep–wake schedule. Both TEP amplitude and slope significantly changed across the 29 h of sustained wakefulness (PROC MIXED; n=22; main effect of circadian phase: amplitude F_(7,128)=8.17, P<0.0001; slope: F_(7,/129)=5.91, P<0.0001). Post hoc analysis revealed (1) a significant increase from the first to the last session (n=22; S1 versus S8: amplitude: P_corr=0.0025; slope: P_corr =0.0635], (2) a local decrease from the second afternoon session (S2) to the third evening session (S3) in the hypothetical WMZ (n=22; S2 versus S3: amplitude: P_corr=0.037; slope: P_corr=0.058), (3) a sharp increase during the biological night (n=22; S3 versus S7: amplitude and slope: P_corr<0.0001], (4) ceasing after the seventh session, at the end of the theoretical SPZ (n=22; S7 versus S8: amplitude and slope: P_corr>0.8).

Figure 3. The circadian system modulates cortical excitability.

(a) Individual cortical excitability measured by TEP amplitude (dashed line represents average z-scored TEP amplitude) was fitted with linear (red) and 24 h period sine-wave (blue) functions to mimic sleep pressure build-up and the circadian signal respectively. Error sum of squares (ESS) was <10 for both indices (amplitude: ESS linear fit=4.9, P<0.0001; ESS sine fit= 4.1, P<0.0001; slope: ESS linear fit=5.19, P<0.0001; ESS sine fit=4.24, P<0.0001). (b) Slow wave activity across the first four cycles of sleep baseline night was fitted to compute individual dissipation rate (schematically shown by red arrow). Each dot represents SWA of an individual sleep cycle (four identical symbols per participant). (c) Regression analysis showed that individual dissipation rate was positively correlated with the increase in cortical excitability from first to last session, recorded 24 h apart, at the same circadian phase, following a normal night of sleep and after sleep deprivation (n=18; amplitude: P=0.044; r²=.23; slope: P=0.036, r²=0.25). (d) Cortisol (yellow) and subjective stress (red) levels. Salivary cortisol concentration was not significantly different between the first and the last protocol samples, collected 24 h apart, at the same circadian phase, following a normal night of sleep and after sleep deprivation (n=22; F_(28,482)=13.44; P<0.0001). Dashed line: shape of TEP z-scored amplitude dynamics. (e) Regression analysis revealed that individual fitted amplitude of cortisol secretion over the protocol was positively associated with the decrease in cortical excitability measured around the wake-maintenance zone (n=20; amplitude: P=0.017; r²=0.24; slope: P=0.023, r²=0.21).

Figure 4. Cortical excitability dynamics is associated with changes in system-level brain function measures and in behaviour.

(a) Time course of relative theta (4.5–7.5 Hz) power (%) in spontaneous waking EEG (blue) and subjective sleepiness (black) (mean±s.d.). Both variables showed significant variation over the sleep deprivation protocol (PROC MIXED; n=22; main effect of circadian phase: P<0.001; Supplementary Fig. 1 for details). Dashed line: shape of TEP z-scored amplitude dynamics. (b,c) ANCOVAs showed that relative theta power (b) (n=22; amplitude: r²=0.19, P=0.004) and subjective sleepiness (c) (n=22; amplitude: r²=0.69, P<0.0001) were significantly and positively associated with both indices of cortical excitability. Amplitude × circadian phase interactions was not significant (P>0.28). (d) Time course of performance to the vigilance task performed simultaneously to TMS/EEG recordings (mean±s.d.). The task consisted of maintaining a constantly moving cursor in the centre of a computer screen. Small inset depicts a representative well-rested and sleep-deprived (SD) session. Task performance (average distance kept from the screen center) significantly changed with time awake (PROC MIXED; n=22; main effect of circadian phase: F_(7,122)=13.78; P<0.0001). (e) An ANCOVA revealed that vigilance task performance impairment was associated to TEP amplitude/slope increase (n=22; amplitude: r²=0.44, P<0.0001; slope: r²=0.43, P<0.0001). Amplitude/slope × circadian phase interaction was not significant (P>0.69).