Augmenting hippocampal-prefrontal neuronal synchrony during sleep enhances memory consolidation in humans

Memory consolidation during sleep is a crucial process for long-term memory storage. The prevailing systems-level memory consolidation theory proposes that the hippocampus initially encodes declarative memories, which then gradually transfer to the neocortex. This process relies on offline reactivation of hippocampal information, particularly during slow-wave sleep (SWS), often manifested as hippocampal ripples (80-120 Hz oscillations). These ripples are believed to interact with neocortical slow waves (<4 Hz) and thalamocortical sleep spindles (9-16 Hz) to facilitate the neocortical encoding of new memories. While numerous correlative studies in humans and rodents support this theory, direct causal evidence in humans linking hippocampal-neocortical interactions during sleep to memory consolidation has been lacking. This study aimed to address this gap by implementing a novel closed-loop deep brain stimulation (DBS) paradigm. The researchers hypothesized that enhancing the temporal coupling between MTL ripples, neocortical slow waves, and thalamocortical spindles during NREM sleep would directly improve overnight declarative memory consolidation. This approach allowed for a causal manipulation of neural activity during sleep, providing a more robust test of the systems-level consolidation theory than previous correlative studies.

Decades of research have highlighted sleep's role in memory consolidation. Early work by Jenkins and Dallenbach demonstrated reduced forgetting during sleep compared to wakefulness. Systems-level consolidation theories propose a two-stage model: initial hippocampal encoding followed by a gradual transfer to the neocortex. This transfer involves offline reactivation of hippocampal memories during SWS, largely linked to hippocampal ripple events. Rodent studies have causally linked hippocampal ripples to memory consolidation and demonstrated their widespread impact on neocortical activity during sleep. In humans, studies using non-invasive methods or rodent neuronal recordings have shown extensive hippocampal-neocortical interactions during sleep, but lacked direct causal evidence linking these interactions to memory consolidation. Previous studies in humans have shown memory enhancement through open-loop and closed-loop stimulation during encoding, but the present study focused on the offline consolidation phase during sleep, taking advantage of the unique opportunity presented by intracranial recordings and stimulation in neurosurgical patients.

Eighteen neurosurgical patients with pharmacoresistant epilepsy, implanted with intracranial depth electrodes for clinical reasons, participated. The study employed a within-participant design, comparing an intervention night (with real-time closed-loop stimulation) and an undisturbed night. A visual paired-association task was used to assess memory, with testing before and after sleep. Real-time closed-loop stimulation was delivered intermittently in 5-min blocks for ~90 min during early NREM sleep. One MTL electrode served as a synchronization probe to trigger brief (50 ms), high-frequency (100 Hz) stimulation in a neocortical electrode (typically orbitofrontal cortex white matter) based on the active phase of MTL slow waves. Two stimulation modes were used: synchronized stimulation (sync-stimulation), time-locked to MTL slow-wave active phases, and mixed-phase stimulation, with identical stimulation but without regard to slow-wave phase. Electrophysiological data (iEEG and single-neuron activity) were recorded simultaneously. Memory was assessed using recognition accuracy (hit rate – false-alarm rate) and association accuracy. Electrophysiological analyses included spectral analysis (spindle power) and time-domain analysis (event detection probabilities and rates for slow waves, spindles, and ripples) and analysis of phase-locking of neuronal spiking activity to MTL slow waves. Statistical analyses included binomial tests, Wilcoxon signed-rank tests, Wilcoxon rank-sum tests, and Spearman correlation.

Synchronized stimulation significantly improved recognition memory accuracy compared to undisturbed sleep in all six participants receiving prefrontal cortex white matter sync-stimulation (P=0.01, binomial test). Mixed results were seen for sync-stimulation in other cortical regions, and a trend towards degraded performance was observed with mixed-phase stimulation. Sync-stimulation immediately increased spindle power and detection probability relative to sham stimulation (P<10⁻³⁰ and P<10⁻¹⁰ respectively, Wilcoxon signed-rank tests). Mixed-phase stimulation did not show this increase and actually decreased spindle detection probability (P<10⁻¹⁰). Individual changes in recognition memory accuracy correlated highly with the degree to which stimulation affected immediate spindle occurrence (Spearman correlation, p=0.69, P=0.013). Sync-stimulation also resulted in a prolonged enhancement of spindle rate and slow-wave phase-locking of neuronal spiking activity, particularly in distant regions (Wilcoxon signed-rank test, P=0.007 and P<10⁻⁹). Importantly, the enhancement of spindle rates correlated with memory improvement (Wilcoxon rank-sum test, P<10⁻¹⁰). Sync-stimulation also increased the temporal coupling between MTL ripples and neocortical slow waves, as well as the triple co-occurrence of ripples, spindles and slow waves, and these effects correlated with memory improvement (Spearman correlation, p=0.8, P=0.007).

The study provided strong causal evidence supporting the role of hippocampo-thalamocortical synchronization during sleep in human memory consolidation. The precise temporal coupling of stimulation to MTL slow-wave active phases was crucial for the observed memory benefits, highlighting the importance of these specific time windows for effective inter-regional communication. The widespread electrophysiological effects observed, even in contralateral hemispheres, are likely due to the unique state of brain activity during sleep, enabling local low-amplitude stimulation to propagate widely. Although the study was performed on patients with epilepsy, the within-subject design and consistent findings across various clinical profiles minimize the influence of these factors. While previous studies have shown memory enhancement via closed-loop stimulation during encoding or using acoustic stimulation, this study uniquely manipulated hippocampal-neocortical coupling during sleep, revealing the direct causal link between synchronized oscillations and memory consolidation. The observed improvement in recognition memory, but not necessarily association, highlights a potential mechanism for sleep's role in reducing false memories.

This study provides compelling evidence for the causal role of hippocampo-thalamocortical synchronization during sleep in human memory consolidation. Real-time closed-loop DBS synchronized with MTL slow waves enhanced sleep electrophysiology and improved recognition memory. The results strongly support existing models of systems-level memory consolidation and suggest a novel therapeutic approach for memory disorders.

The study's limitations include the use of a patient population with epilepsy, potential medication effects on sleep, and the specific stimulation parameters. The generalizability of the findings to healthy individuals requires further investigation. The relatively small sample size also necessitates replication studies with larger cohorts. Future research should optimize stimulation parameters and explore other brain regions and stimulation targets to further refine this promising therapeutic strategy.