Author response: Slow oscillation-spindle coupling predicts enhanced memory formation from childhood to adolescence

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Precise temporal coordination of slow oscillations (SO) and sleep spindles is a fundamental mechanism of sleep-dependent memory consolidation. SO and spindle morphology changes considerably throughout development. Critically, it remains unknown how the precise temporal coordination of these two sleep oscillations develops during brain maturation and whether their synchronization indexes the development of memory networks. Here, we use a longitudinal study design spanning from childhood to adolescence, where participants underwent polysomnography and performed a declarative word-pair learning task. Performance on the memory task was better during adolescence. After disentangling oscillatory components from 1/f activity, we found frequency shifts within SO and spindle frequency bands. Consequently, we devised an individualized cross-frequency coupling approach, which demonstrates that SO-spindle coupling strength increases during maturation. Critically, this increase indicated enhanced memory formation from childhood to adolescence. Our results provide evidence that improved coordination between SOs and spindles indexes the development of sleep-dependent memory networks. eLife digest Sleep is essential for consolidating the memories that we made during the day. As we lie asleep, unconscious, our brain is busy processing the day’s memories, which travel through three parts of the brain before they are filed away. First, the hippocampus, the part of the brain that stores memories temporarily, replays the memories of the day. Then the reactivated memories pass through the thalamus, a central crossroads in the brain, so they can be embedded in the neocortex for long-term storage. Neuroscientists can eavesdrop on the brain at work, day or night, using a technique called EEG. Short for electroencephalogram, an EEG detects brain waves like the bursts of electrical activity known as sleep spindles and slower sleep waves called slow oscillations. These two brain wave patterns represent how the brain processes memories as people sleep – and it is all about timing. If the two patterns are running in sync, then the brain’s memory systems are thought to be communicating well and memories are more likely to be stored. But patterns of sleep spindles and slow oscillations change dramatically between childhood and adolescence. Memory consolidation also improves in those formative years. Still, it is not yet known if better synchronization between sleep spindles and slow oscillations explains how memory formation improves during this period; that is the going theory. To test it out, Hahn et al. completed a unique study examining how well a group of 33 children could store memories, and then again when the same group were teenagers. Both times, the group was asked to memorise and then recall a set of words before and after a full night’s sleep. Hahn et al. measured how much their memory recall improved and whether their brain wave patterns were in sync, looking for any changes between childhood and adolescence. This showed that children whose sleep spindles stacked better with their slow oscillations had improved memory formation once they became teenagers. This work highlights how communication between memory systems in the brain improves as children age, and so does memory. Moreover, it suggests that if disturbances were to be detected in patterns of sleep spindles and slow oscillations, there might be some problem with memory storage. It also points to brain stimulation as a possible treatment option for such problems in the future. Introduction Active system memory consolidation theory proposes that sleep-dependent memory consolidation is orchestrated by three cardinal sleep oscillations (Diekelmann and Born, 2010; Helfrich et al., 2019; Klinzing et al., 2019; Mölle et al., 2011; Piantoni et al., 2013; Rasch and Born, 2013; Staresina et al., 2015): (1) Hippocampal sharp-wave ripples represent the neuronal substrate of memory reactivation (Vaz et al., 2019; Wilson and McNaughton, 1994; Zhang et al., 2018), (2) thalamo-cortical sleep spindles are thought to promote long-term potentiation (Antony et al., 2018; De Gennaro and Ferrara, 2003; Rosanova and Ulrich, 2005; Schönauer, 2018; Schönauer and Pöhlchen, 2018), while (3) neocortical SOs provide temporal reference frames where memory can be replayed, potentiated and eventually transferred from the short-term storage in the hippocampus to the long-term storage in the neocortex, rendering memories increasingly more stable (Chauvette et al., 2012; Diekelmann and Born, 2010; Frankland and Bontempi, 2005; Rasch and Born, 2013). Importantly, these three oscillations form a temporal hierarchy, where ripples and spindles are nested in SO peaks, with ripples also being locked to spindle troughs. This hierarchy likely constitutes an endogenous timing mechanism to ensure that the neocortical system is in an optimal state to consolidate new hippocampus-dependent memories (Chauvette et al., 2012; Clemens et al., 2011; Helfrich et al., 2019; Klinzing et al., 2016; Klinzing et al., 2019; Latchoumane et al., 2017; Niethard et al., 2018; Piantoni et al., 2013; Staresina et al., 2015). Recent findings indicate that the precise temporal coordination of SO-spindle coupling is deteriorating over the lifespan, which contributes to age-related memory decline (Helfrich et al., 2018b; Muehlroth et al., 2019; Winer et al., 2019). It is currently unclear if similar principles apply to brain maturation and how the dynamic interplay of SOs and spindles is initiated. Critically, the transition from childhood to adolescence is marked by considerable changes in sleep architecture and cognitive abilities similar to the transition from young adulthood to old age (Carskadon et al., 2004; Huber and Born, 2014; Iglowstein et al., 2003; Ohayon et al., 2004; Shaw, 2007; Shaw et al., 2006). Previous research mainly focused on the individual development of SOs and sleep spindles across brain maturation, showing that these cardinal sleep oscillations undergo a substantial evolution in their defining features such as amplitude, frequency, distribution and occurrence (Campbell and Feinberg, 2009; Campbell and Feinberg, 2016; Goldstone et al., 2019; Hahn et al., 2019; Kurth et al., 2010; Nicolas et al., 2001; Purcell et al., 2017; Shinomiya et al., 1999; Tarokh and Carskadon, 2010). Currently, two major obstacles hamper our understanding of how the precise temporal interplay between SOs and spindles predicts brain development and memory formation. First, pronounced changes in sleep oscillatory activity pose major methodological challenges for assessing and comparing SOs and sleep spindles across the age spectrum (Muehlroth and Werkle-Bergner, 2020). Second, memory performance was rarely tested in developmental sleep studies, thus, impeding our understanding of the functional significance of temporal SO-spindle coupling for memory formation. Here, we leverage a unique longitudinal study design from childhood to adolescence to investigate how SO-spindle coupling emerges during development and infer its functional significance for developing memory networks. To account for the substantial morphological alterations of SO and spindle morphology across brain maturation, we developed a principled methodological approach to assess SO-spindle coupling. We utilized individualized cross-frequency coupling analyses, which enable a clear demonstration of SO-sleep spindles coupling during both developmental stages. Critically, over the course of brain maturation from childhood to adolescence, more spindles are tightly coupled to SOs, which directly predicts improved memory formation. Results Using a longitudinal study design, we tested 33 healthy participants during childhood (age: 9.5 ± 0.8 years; mean ± SD) and during adolescence (age: 16 ± 0.9 years). At both time points participants underwent full-night ambulatory polysomnography at their home during two nights (adaptation and experimental night; Figure 1A) and performed a declarative memory task during the experimental night (Figure 1B). After encoding, participants recalled word pairs before and after a full night of sleep. As previously shown (Hahn et al., 2019), memory recall improved from childhood to adolescence (Figure 1C; F1,32 = 38.071, p<0.001, η2 = 0.54) and immediate recall was better than delayed recall (F1,32 = 6.408, p=0.016, η2 = 0.17; Maturation*Recall Time interaction: F1,32 = 2.059, p=0.161, η2 = 0.06). Next, we assessed the relationship of sleep-dependent memory consolidation (delayed recall – immediate recall) between childhood and adolescence and found no correlation between the two maturational stages (Figure 1—figure supplement 1A; for a direct comparison of sleep-dependent memory consolidation see Figure 1—figure supplement 1B). During adolescence, memory consolidation was superior after a sleep retention interval compared to a wake retention interval (Figure 1—figure supplement 1C), indicating a beneficial effect of sleep on memory. Figure 1 with 1 supplement see all Download asset Open asset Study design and behavioral results. (A) Longitudinal study design. Participants were recorded during childhood (blue) and adolescence (red). Recording periods were separated by 7 years. Participants underwent two full-night ambulatory polysomnographies in their habitual sleep environment at both time points respectively. The first night served adaptation purposes. At the following experimental night, participants performed a declarative word pair learning task during which they encoded and recalled semantically non-associated word pairs before sleep. The post-sleep recall was separated by a 10 hr (childhood) and 8 hr sleep during the retention interval (adolescence). (B) Word pair task design. Participants encoded 50 word pairs during childhood (blue) and 80 word pairs during adolescence (red). Every word pair presentation was followed by a fixation cross. Participants were instructed to imagine a visual connection between the two words. Timing parameters are indicated in the respective colors for childhood (blue) and adolescence (red). During the recall trial, only the first word of the word pair was presented and participants had to recall the corresponding word. Participants received no performance feedback. (C) Behavioral results for the word pair task. Performance was measured as percentage of correctly recalled word pairs. Participants showed a higher performance during adolescence. Black circles indicate individual recall scores. Oscillatory signatures of NREM sleep during childhood and adolescence To investigate whether SO-spindle coupling accounts for enhanced memory formation from childhood to adolescence, we first assessed the oscillatory signatures of NREM (2 and 3) sleep. We compared spectral estimates during childhood and adolescence using cluster-based permutation tests (Maris and Oostenveld, 2007) across frequencies from 0.1 to 20 Hz (Figure 2A; at electrode Cz). We found that EEG power significantly decreased from childhood to adolescence between 0.1 to 13.6 Hz (cluster test: p<0.001, d = −2.74) and 14.6 to 20 Hz (cluster test: p<0.001, d = −1.60; Figure 2A). However, inspection of the underlying spectra revealed that this effect was driven by (I) an overall offset of the 1/f component of the power spectrum on the y-axis and (II) by a shift of the peak frequency in the spindle band. In order to mitigate the prominent power difference, we first z-normalized the signal in the time domain, which alleviated the differences above ~15 Hz (Figure 2B). This analysis showed increased spectral power during childhood from 0.3 to 8.4 Hz (cluster test: p=0.002, d = −084), which was broadband and not oscillatory in nature. In addition power differences between 10.6 to 12.8 Hz (cluster test: p=0.040, d = −1.07) and 13.4 and 14.8 Hz (cluster test: p=0.046, d = 1.12) directly reflected the spindle peak frequency shift from childhood to adolescence. To account for the differences in broadband 1/f and oscillatory components, we disentangled the 1/f fractal component (Figure 2C) from the oscillatory residual (Figure 2D) by means of irregular-resampling auto-spectral analysis (IRASA Helfrich et al., 2018b; Wen and Liu, 2016). We found that a significant decrease in the fractal component between 0.3 and 10.8 Hz from childhood to adolescence (Figure 2C; cluster test: p<0.013, d = −0.90), accounted for the prominent broadband power differences as observed in Figure 2A. To assess true oscillatory brain activity, we subtracted the fractal component (Figure 2C) from the normalized power spectrum (Figure 2B), to isolate SO and spindle oscillations in the frequency domain (Figure 2D). Figure 2 with 2 supplements see all Download asset Open asset Oscillatory signatures of NREM sleep. (A) Uncorrected EEG power spectra (mean ± standard error of the mean [SEM]) during NREM (NREM2 and NREM3) sleep at Cz during childhood (blue) and adolescence (red). Grey overlays indicate significant differences (cluster-corrected). Note the overall power decrease from childhood to adolescence. (B) Z-normalized EEG power spectra. Same conventions as in (A). Significant differences indicate a change in the fractal component of power spectra (0.3–8.4 Hz) and a spindle frequency peak shift (10.6–14.8 Hz) from childhood to adolescence. (C) Extracted 1/f fractal component. Same conventions as in (A). Decrease of the fractal component (0.3–10.8 Hz) from childhood to adolescence. (D) Oscillatory residual of the NREM power spectra obtained by subtracting the fractal component (C) from the z-normalized power spectrum (B). Oscillatory residual shows clear dissociable peaks in the SO and sleep spindle frequency range (dashed boxes) during both time points, indicating true oscillatory activity. (E) Spindle amplitude development. Spindle amplitude (exemplary depiction at Cz, left, mean ± SEM) as extracted from the oscillatory residuals (D) indicating an increase in 1/f corrected amplitude within a centro-partial cluster (right) from childhood to adolescence. Grey dots represent individual values. Asterisks denote cluster-corrected two-sided p<0.05. T-scores are transformed to z-scores to indicate the difference between childhood and adolescence. (F) Spindle frequency peak development. Spindle frequency peak (mean ± SEM) as extracted from the oscillatory residual (D). Same conventions as in (E). Spindle peak frequency increases at all electrodes from childhood to adolescence. Based on the oscillatory residuals, we then extracted the individual peak frequency and the corresponding amplitude in the SO and sleep spindle range for each electrode in every participant during childhood and adolescence. After discounting 1/f effects, we found that spindle amplitude (Figure 2E) increased in a centro-parietal cluster (cluster test: p=0.005, d = 0.63), whereas spindle peak frequency (Figure 2F) accelerated at all channels from childhood to adolescence (cluster test: p<0.001, d = 1.57). SO amplitude and frequency decreased from childhood to adolescence (Figure 2—figure supplement 1A,B). Both, SO and spindle features have been previously related to memory formation (Gais et al., 2002; Huber et al., 2004; Lustenberger et al., 2017; Schabus et al., 2004; Schabus et al., 2006). However, neither spindle nor SO amplitude or peak frequency changes explained the behavioral differences (Figure 2—figure supplement 1C,D). Note, we also observed a peak in the theta band, which was unrelated to behavior (for theta peak frequency and amplitude correlations with behavior see Figure 2—figure supplement 1E). Individual features of discrete SO and sleep spindle events After having established the cardinal features of SO and spindle oscillations during childhood and adolescence, we then individually adjusted previously used SO and spindle detection algorithms (Helfrich et al., 2018b; Mölle et al., 2011; Staresina et al., 2015) according to the individual peak frequencies (Bódizs et al., 2009; Ujma et al., 2015). We considered the possibility that two distinct spindle frequency peaks exist (Anderer et al., 2001; Werth et al., 1997), but inspecting the oscillatory residuals did not indicate two clearly discernable peaks in individual electrodes of the majority of participants (for exemplary oscillatory residuals see Figure 2—figure supplement 2A). Because we observed the typical antero-posterior spindle frequency gradient (Cox et al., 2017; De Gennaro and Ferrara, 2003; Zeitlhofer et al., 1997) with slower frontal and faster posterior spindles (Figure 2—figure supplement 2B), we used the highest peak in the spindle range at every electrode as the most representative oscillatory event for the detection algorithm. Importantly, individualized SO and spindle event detections closely followed spectral sleep patterns during childhood and adolescence (Figure 3A,B; event detections are superimposed in white). Figure 3 with 1 supplement see all Download asset Open asset Individual features of discrete SO and sleep spindle events. (A) Hypnogram (top) and full-night spectrogram (bottom) at electrode Cz of an exemplary subject during childhood. White lines in the spectrogram indicate the amount of events detected by the individually adjusted detection algorithms for sleep spindles (upper trace) and SO (lower trace). (B) Same conventions as in (A) but for the same individual during adolescence. (C) Spindle-SO co-occurrence expressed as the percentage of SO detections that coincide ±2.5 s with a sleep spindle detection at electrode Fz during NREM2 and NREM3 sleep during childhood (blue) and adolescence (red). Note the high co-occurrence of spindles and SOs during NREM3 at both recording time points. (D) Grand average of z-normalized sleep spindle events (mean ± SEM) during childhood (blue) and adolescence (red) at electrode Fz with the corresponding SO-low-pass filtered (<2 Hz) EEG-trace (inset). Note that there is no baseline difference between −2.5 s and −2 s (dashed box). The significant difference in the −1.5 to -0.5 s interval (grey shaded area, SO-filtered inset) indicates an increased amount of coupled SO-sleep spindle events during adolescence. Further note, no amplitude differences in the SO-filtered signal around the spindle peak at 0 s (i.e. time point of the phase readout). Grand average spindle frequency is distorted by the individually adjusted event detection criteria. (E) SO-spindle coupling features. Data are shown for electrode Fz during NREM3. Left: Exemplary spindle-locked average for a single subject during childhood with the corresponding SO-filtered signal in black. Note that the spindle amplitude peak coincides with the maximum peak in the SO-component. Right: Normalized phase histograms of spindle events relative to SO-phase of an exemplary subject during childhood. 0° denotes the positive peak whereas ±π denotes the negative peak of the SO. (F) Same conventions as in (E). Left: Exemplary spindle-locked average of the same single subject as in (E) during adolescence. Note the clearer outline of a SO-component compared to during childhood indicating a stronger SO-spindle coupling. Right: Normalized phase histograms of spindle events relative to SO-phase of same exemplary subject as in (E) during adolescence. Note the reduced spread in SO-phase. (G) Left: Grand average baseline-corrected (−2 to −1.5 s) SO-trough-locked time frequency representation (TFR). Schematic SO-component is superimposed in black. Note the alternating pattern within the spindle frequency range indicating a modulation of spindle activity by SO-phase. Right: Circular plot of preferred phase (SO phase at spindle amplitude maximum) per subject during childhood. Dots indicate the preferred phase per subject. The line direction shows the grand average preferred direction. The line length denotes the mean resultant vector (i.e. sample variance of preferred phase and therefore does not represent coupling strength). Note that most subjects show spindles coupled to or just after the positive SO-peak at 0°. Data are shown for electrode Fz during NREM3. (H) Same conventions as in (G). Left: SO-trough-locked TFR indicating a modulation in spindle activity depending on SO-phase. Right: Circular plot of preferred phase per subject during adolescence. Note that there are no preferred phase changes but an overall reduced spread in preferred phase on the group level during adolescence as indicated by a longer mean resultant vector (red line). Next, we quantified how many separate SO and spindle event detections co-occurred within a 2.5 s time window (reflecting ±2 SO cycles around the spindle peak; Helfrich et al., 2019). Note that the co-occurrence rate does not actually indicate coupled SO-spindle events but directly reflects the percentage of detected spindle events that are concomitant with detected SO events. Co-occurrence rate was higher in NREM3 than NREM2 sleep during childhood and adolescence (Figure 3C; F1,32 = 2334.19, p<0.001, η2 = 0.99). Subsequently we restricted our analyses to NREM3 sleep to avoid spurious cross-frequency coupling estimates caused by the lack of simultaneous detections during NREM2 sleep (Aru et al., 2015; for circular plots including NREM2 sleep see Figure 4—figure supplement 1D). To ensure reliable coupling estimates, we further Z-normalized individual spindle-locked data epochs in the time domain (Figure 3D) for all subsequent analyses to avoid possible confounding amplitude differences (Aru et al., 2015; Cole and Voytek, 2017; Helfrich et al., 2018b). in the grand average spindle directly the enhanced SO-spindle which in the time domain when more events are coupled to the SO This effect can also be in single subject spindle-locked data (Figure Note that we found no differences in the underlying SO-component around the spindle peak time point of phase thus, that the alleviated possible amplitude differences (Figure for spindle-locked data see Figure supplement To further the between SO phase and spindle activity, we also assessed this effect in the domain by SO-trough-locked (Figure The alternating pattern (i.e. spindle power during the and increases during the within the spindle frequency range during childhood and adolescence indicated an of SO phase on spindle activity. To the interplay of SO and spindle oscillations, we cross-frequency analyses and 2014; Helfrich et al., 2018b; Staresina et al., 2015). this is mainly to used methods to assess cross-frequency coupling (Helfrich et al., it can be by their (Aru et al., 2015). we a approach by first power differences (Figure and Figure 3D) and the of oscillations in the signal (Figure and Figure Next, we extracted the SO phase during every spindle peak at every electrode and for every subject. Then we the preferred phase mean and coupling strength during childhood and adolescence for all events at a electrode Figure for exemplary phase We that both of SO-spindle coupling were not by differences in the of detected events using a (Figure 4—figure supplement maturation SO-spindle coupling First, we assessed in which phase of the SO spindles and how maturation the preferred coupling direction. We found that coupling direction did not change at frontal electrodes (Figure where spindles were locked to the SO peak during childhood circular mean ± SD) and adolescence circular mean ± However, we detected differences in p<0.001, d = p<0.001, d = Figure 4—figure supplement which did not with behavior (Figure 4—figure supplement 1B). After showing that sleep spindles are locked to the SO peaks, we quantified how spindles are embedded in the preferred SO phase by assessing the respective coupling strength – circular strength increased from childhood to adolescence across all electrodes (cluster test: p<0.001, d = Figure indicating that frontal sleep spindles more tightly locked to their respective preferred SO phase as the phase on frontal stable across maturation. To further the of the coupling strength increase on we the percentage of spindle events that at the preferred phase as compared to all detected events (Figure The is directly related to the coupling strength and was to further and the observed with our coupling strength analyses, percentage of spindles within the preferred phase increased in a cluster Figure decreased in an cluster from childhood to adolescence d = These results an overall increase in SO-spindle coupling from childhood to adolescence. Figure with 1 supplement see all Download asset Open asset strength development and correlations with memory formation. (A) strength development. strength mean ± SEM) increases from childhood to adolescence (exemplary data at left, dots indicate individual at all electrodes indicating that more spindles within the preferred phase during adolescence than during childhood. Asterisks indicate cluster-corrected two-sided p<0.05. T-scores are transformed to z-scores to indicate the difference between childhood and adolescence. (B) in preferred Same conventions as in (A). plot the percentage of sleep spindles (mean ± SEM) that in a ± around the individual preferred coupling spindles in preferred phase increase from childhood to adolescence in a cluster but decrease in an (C) Left: correlation between the individual coupling strength increases from childhood to adolescence adolescence – and recall performance (delayed – delayed Asterisks denote significant with higher coupling strength increases showed stronger recall performance from childhood to adolescence. This effect was at electrode = plot with line). (D) between coupling strength increase and sleep-dependent memory consolidation – immediate – immediate from childhood to adolescence at electrode = This indicates that subjects with a higher developmental increase in coupling strength show higher sleep on memory consolidation. SO-spindle coupling development predicts memory formation After having established SO-spindle coupling and their from childhood to adolescence, we tested the that these changes also maturational differences in recall performance and sleep-dependent memory consolidation. We utilized cluster-based correlation analyses to differences in coupling strength to differences in recall performance (delayed – delayed We observed a significant frontal cluster mean = Figure showing that a stronger increase in SO-spindle coupling strength from childhood to adolescence related to improved recall performance from childhood to adolescence. This relationship was most pronounced at electrode = Figure using a approach with a frequency Hz) the test = Figure 4—figure supplement 1E). Next, we only considered spindle events that with detected to (1) ensure more precise coupling by the of oscillations and (2) to the that the temporal of these events is for sleep-dependent memory consolidation (Diekelmann and Born, 2010; Helfrich et al., 2018b; Muehlroth et al., 2019; Rasch and Born, 2013). Participants with a stronger coupling strength increase also showed enhanced sleep-dependent memory consolidation (delayed recall immediate recall) from childhood to adolescence = Figure electrode To whether the SO-spindle coupling strength predicts sleep memory we coupling strength during adolescence with the difference in memory