Twitches during N2 and N3 are coupled with sleep spindles but not delta oscillations in 6-month-old infants

Abstract

Twitches are discrete movements that characterize REM sleep. However, recent work showed that twitches also occur during NREM sleep in human infants beginning around 3 months of age, a time when sleep spindles and the cortical delta rhythm are also emerging. Further, NREM twitches co-occur with sleep spindles, suggesting a unique functional role. Given that NREM sleep is composed of distinct substages, we investigated whether twitching and twitch-spindle coupling are differentially expressed during N2 and N3 sleep. In 6-month-old human infants (n=21; 7 females), we recorded EEG, respiration, and video during daytime sleep. We found that the high intensity of twitching during N2 resembled REM more than N3. In contrast, sleep spindles exhibited similar temporal characteristics during N2 and N3. Also, despite differences in the intensity of twitching during N2 and N3, significant twitch-spindle coupling occurred in both stages. Finally, twitch rate was inversely related to delta power across NREM periods. These findings suggest that although twitching occurs during REM, N2, and N3 sleep at this age, its expression is compatible with some sleep components (e.g., rapid eye movements, sleep spindles) but not others (e.g., cortical delta), highlighting the continuing need to better understand the dynamic organization of sleep and its individual components in early development.

Keywords: delta oscillations; development; infancy; myoclonic twitches; sensorimotor system; sleep spindles.

Figure 1.. Representative behavioral and EEG data in a 6-month-old human infant.

(A) Continuous 37.5-min record of N1 (pink), N2 (purple), N3 (green), wake (W, grey) and REM (red). EEG data are from the C3 electrode site and are represented as a sonogram (0–30 Hz) and delta power (0.5–4 Hz, black trace). The onset of spindles (blue ticks), limb twitches (black ticks), and REMs (red ticks) are also shown. (B) Same infant and data presentation as in (A) but for 2-minute periods of N2, N3, and REM sleep.

Figure 2.. Characteristics of twitching during REM and NREM sleep.

(A₁) Mean rate of twitching (twitches/min), (A₂) mean duration of twitches (s), and (A₃) mean (±SEM) proportion of twitching across various body parts during NREM (blue) and REM (red) sleep. Grey lines show data for individual infants. (B_1–3) Same as in (A_1–3) but for N2 (green) and N3 (purple) sleep. (C) Log-survivor plot of inter-twitch intervals during periods of REM, N2, and N3 sleep. ITIs were pooled across sleep sessions: REM = 1795 twitches; N2 = 1239 twitches; N3 = 2060 twitches. (D₁) Mean (±SEM) twitch-burst rate (bursts/min) and (D₂) mean (±SEM) proportion of twitches occurring in a burst during REM, N2, and N3 sleep. Grey dots show data for individual infants. * p < .05

Figure 3.. Characteristics of spindles do not differ between N2 and N3 sleep.

(A) Mean spindle rate (spindles/min), (B) mean spindle frequency (Hz), and (C) mean spindle amplitude (μV) during N2 (green) and N3 (purple) sleep. Grey lines show data for individual infants. (D) Log-survivor plot of inter-spindle intervals (ISIs) during N2 (green) and N3 (purple) sleep. ISIs were pooled across sleep sessions: N2 = 286 spindles; N3 = 1104 spindles.

Figure 4.. Twitches are inversely related to delta power.

(A) Representative linear regressions from two infants showing the number of twitches in relation to mean delta power (μV²) across 1-min windows of artifact-free NREM sleep. (B) Multilevel linear mixed-effects analysis of the effect of within-infant changes in delta power on the expected number of twitches within each 1-min window.

Figure 5.. Spindles and twitches are coupled in N2 and N3 sleep.

(A₁) Illustration of our method for assessing the probability of a sleep spindle given a twitch: P(Spindle | Twitch). For every twitch, we determined whether a spindle (from onset to offset) overlapped with that twitch. (A₂) Perievent histograms showing the P(Spindle | Twitch) during N2 (top; purple) and N3 (bottom; green) sleep. Each plot shows mean probabilities with SEM (shaded values). For each plot, 95% confidence intervals based on shuffled data are shown in grey. (A₃) Mean (±SEM) P(Spindle | Twitch) during N2 (purple solid circle) and N3 (green solid circle). Grey lines show observed data for individual infants. Expected mean probabilities (unfilled circles) based on shuffled data are also shown (expected probabilities for individual infants are not shown). (B_1–3) Same as in (A_1–3) but for the probability of a twitch given a sleep spindle: P(Twitch | Spindle). Black * significant main effect (p<.05) between observed and expected values. Colored * significant observed and expected probability differences in N2(purple) and N3(green) (p<0.05). † Significant interaction (p<.05) between sleep state and observed-expected probability differences.

Figure 6.. Organization of sleep components of REM, N2, and N3 at 6 months of age.

Venn diagram to illustrate the sleep-stage distribution of twitching (high and low intensity), sleep spindles, rapid eye movements, and cortical activity (delta, desynchronized). The red double arrows indicate the coupling of twitches and spindles that occurs in N2 and N3.