In addition of detecting thalamic sleep spindles in human subjects, we revealed their association with thalamic ripples. SP(ripple) made up around 20% of overall ANT and MD sleep spindles, respectively. These spindles were of longer duration and characterized by different time-frequency activity profiles than sleep spindles in the absence of ripples: SP(ripple) were associated with an excess of thalamic ripple and lower (5-15 Hz) frequency activity compared to SP(pure). This 5-15 Hz activity occurred later, roughly 500 ms after sleep spindle onset, which is assumed to reflect the elongation of ripple-associated sleep spindles. The increased ripple-band activity was the basis of ripples-associated sleep spindle definition; however, no signs of power difference were observed in the same time window at scalp derivations.
The coordinated interactions between hippocampal ripples and cortical sleep spindles plays crucial role in memory formation (Latchoumane et al. 2017; Siapas and Wilson 1998; Staresina et al. 2015). It was proposed that NREM sleep-based memory formation depends on the hierarchical nesting of slow waves, sleep spindles and hippocampal ripples, as these oscillations are instrumental in transferring information from the hippocampus to the neocortex (Diekelmann and Born 2010; Staresina et al. 2015). In the past few years, accumulating evidence were found for the existence of ripples outside of the hippocampus (for a review, see McKenzie, Nitzan, and English 2020). Ripples were detected in several cortical areas such as somatosensory and motor cortices (Averkin et al. 2016), olfactory cortices (Manabe et al. 2011), parahippocampal regions (Axmacher, Elger, and Fell 2008) and higher order associational cortices as well (Khodagholy, Gelinas, and Buzsáki 2017). These cortical ripples were frequently coupled with slower oscillations including sleep spindles. In the prefrontal cortex, mesio-temporal and neocortical structures, ripples were detected both before sleep spindles and locked to the spindle troughs (Bruder et al. 2021; Peyrache, Battaglia, and Destexhe 2011). Similar results were observed confirming that cortical ripples were embedded into spindle troughs during natural sleep (Averkin et al. 2016). Although it is not entirely clear, whether ripples in the thalamus are connected with pathological processes related to epilepsy. A close relationship between epilepsy and high-frequency oscillations (80–600 Hz, HFOs) occurrence is repeatedly found in animal and human studies (for a review, see Jiruska et al. 2017). In a recent study, Rektor et al. (2016) found HFOs during wakefulness in the human ANT up to 240 Hz frequency, and 500 Hz frequency in one case, which was the first report of HFOs in the human thalamus. Deutschová et al (2021) reported a relationship between prestimulation HFO power and ANT-BDS treatment reponse, with reduced awake resting 65–500 Hz activity forecasting better outcome. In the current study, ripples (100–200 Hz) were revealed in both ANT and MD nuclei during NREM sleep. Ripple events were in part coupled with sleep spindles, but also detected outside spindles during NREM sleep. We found a strong phase-amplitude coupling measured by the modulation index between sleep spindle phase and ripple amplitude. However, thalamic 100–200 Hz ripples emerging during sleep spindles were not associated with pathological processes, rather, SP(ripple) were shown to indicate preserved cognitive ability. Here we found a positive correlation between the density of SP(ripple) in the ANT and general intelligence (overall cognitive functions) of our patients. This latter result indicates that 100–200 Hz ripples detected in the human ANT during NREM sleep spindles may contribute to the physiological expression of thalamocortical oscillations. Furthermore, these results suggest that thalamic ripples could indicate physiological forms of neural activity.
The temporal dynamics of sleep spindles in the thalamus and the scalp show that cortical spindle preceded thalamic spindles, especially those associated with ripples. Sleep spindles are generated by the reticular thalamic nucleus and propagate to the cortex through thalamo-cortical network (Steriade 2005; Steriade et al. 1987). The temporal advantage of the scalp sleep spindles suggest that ANT and MD sleep spindles are propagated through cortico-thalamic networks, and not directly through thalamo-thalamic projections. It was suggested that cortico-thalamic feedback projections from the cortical sites to the reticular thalamus is responsible for the large-scale synchronization of sleep spindles (Destexhe, Contreras, and Steriade 1998). The ANT has a bidirectional connection with the anterior cingulate cortex, retrosplenial cortex, and subiculum, whereas the MD is interconnected with the medial-prefrontal cortex (Aggleton et al. 2010; Mitchell and Chakraborty 2013; Pergola et al. 2018). The recent results suggest that sleep spindles propagate through these cortical sites to the ANT and MD. Thus, the thalamus seems to play an important interface between the propagation of these neural oscillations in the hippocampal-prefrontal network.
Former studies suggested that the human ANT contribute to the epileptic network, and the ANT became an important target for DBS in epilepsy treatment (Hodaie et al. 2002; Salanova 2018; Sweeney-Reed et al. 2016). Furthermore, interictal discharges were also detected in the MD (Sweeney-Reed et al. 2016). Although, there is no direct anatomical connection between the ANT and MD, this report indicates that the MD might be involved in the epileptic circuitry as well. Our current results further support this assumption. Besides interictal discharges, the presence of different sleep spindle (slow vs fast) could be also indicative for specific network dysfunctions. The density of slow SP(pure) in the MD showed a negative correlation with the year since epilepsy onset. In contrast to slow spindles, fast SP(pure) density showed a positive correlation with seizure prevalence. These results suggest that the ANT and MD are involved in different epileptogenic circuitries. The occurrence of fast sleep spindles may indicate the altered functioning of the MD due to network-based changes through epilepsy propagation, whereas the occurrence of fast spindles seems physiological in the ANT. This was also indicated by the lower overall fast spindle density in the MD. Sleep spindles in the mediodorsal thalamus were also indicative in schizophrenia patients, where a negative association were found between the volume of the MD and scalp-recorded sleep spindle density (Ferrarelli and Tononi 2017). To our best knowledge, this is the first report where sleep spindles were separated to fast and slow spindles, showing that the appearance of fast sleep spindles may indicate pathological mechanisms in the mediodorsal thalamic functions.