Sleep in the human hippocampus: a stereo-EEG study

Abstract

Background: There is compelling evidence indicating that sleep plays a crucial role in the consolidation of new declarative, hippocampus-dependent memories. Given the increasing interest in the spatiotemporal relationships between cortical and hippocampal activity during sleep, this study aimed to shed more light on the basic features of human sleep in the hippocampus.

Methodology/principal findings: We recorded intracerebral stereo-EEG directly from the hippocampus and neocortical sites in five epileptic patients undergoing presurgical evaluations. The time course of classical EEG frequency bands during the first three NREM-REM sleep cycles of the night was evaluated. We found that delta power shows, also in the hippocampus, the progressive decrease across sleep cycles, indicating that a form of homeostatic regulation of delta activity is present also in this subcortical structure. Hippocampal sleep was also characterized by: i) a lower relative power in the slow oscillation range during NREM sleep compared to the scalp EEG; ii) a flattening of the time course of the very low frequencies (up to 1 Hz) across sleep cycles, with relatively high levels of power even during REM sleep; iii) a decrease of power in the beta band during REM sleep, at odds with the typical increase of power in the cortical recordings.

Conclusions/significance: Our data imply that cortical slow oscillation is attenuated in the hippocampal structures during NREM sleep. The most peculiar feature of hippocampal sleep is the increased synchronization of the EEG rhythms during REM periods. This state of resonance may have a supportive role for the processing/consolidation of memory.

Figure 1. Magnetic resonance imaging (MRI) scan of intracranial electrodes implanted into the medial temporal lobe (a and b panels, coronal and sagittal views, respectively).

Red line cross indicates the location of the two electrode contacts within the hippocampus. C and d panels show the MRI scan of the same locations before the electrode implant.

Figure 2. Time course of EEG power in the 0.5–30.0 Hz frequency range across the first three NREN-REM sleep cycles.

REM periods are indicated by the black rectangles on top of the figure. Data are expressed in a logarithmic scale. Scalp recordings have been obtained by means of a bipolar (Cz-Fz) montage. Intracranial (cortex and hippocampus) stereo-EEG recordings have been obtained by a bipolar montage between two contiguous contacts of a platinum-iridium microwire. Very low frequency range (VLF, 0.5–1.0 Hz); delta frequency range (1.1–5.0 Hz); theta frequency range (5.1–8.0 Hz); alpha frequency range (8.1–12.0 Hz); sigma frequency range (12.1–15.0 Hz); beta frequency range (15.1–30.0 Hz).

Figure 3. Examples of EEG patterns recorded from intracranial (cortex and hippocampus) and scalp electrodes during slow wave sleep (SWS) and rapid eye movement sleep (REM).

Electrooculogram (EOG) and chin electromiogram (EMG) are also reported. See text for other methodological details about recordings. A clear slow rhythm is present in the hippocampal recording also during REM sleep.

Figure 4. Time course of the very low frequency/delta power ratio (0.5–1.0/1.1–5.0 Hz) across the first three NREM-REM cycles, separately for scalp and hippocampal recordings.

REM periods are indicated by the black bars on top of the figure.

Figure 5. NREM/REM EEG power ratio in the entire (Hz by Hz) 0.5–30 Hz frequency range of the whole night, separately for scalp and hippocampal recordings.

The horizontal line indicate the mean REM sleep EEG power (value = 1). Mean values close to this line indicate that NREM and REM sleep power are equivalent in that frequency bin.

Figure 6. Linear decrease of delta power (1.1–5.0 Hz) across the first three sleep cycles in the cortical, hippocampal and scalp recordings.

Values are expressed in a logarithmic scale.