Behavioral timing of interictal spikes, but not rate, correlates with impaired working memory performance.

In temporal lobe epilepsy, interictal spikes (IS) - hyper-synchronous bursts of network activity - occur at high rates in between seizures. We sought to understand the influence of IS on working memory by recording hippocampal local field potentials from male epileptic mice while they performed a delayed alternation task. Interestingly, the rate of IS during behavior did not correlate with performance. Instead, we found that IS were correlated with worse performance when they were spatially non-restricted and occurred during running. In contrast, when IS were clustered at reward locations, animals tended to perform well. A machine learning decoding approach revealed that IS at reward sites were larger than IS elsewhere on the maze, and could be classified as occurring at specific reward locations. Finally, a spiking neural network model revealed that spatially clustered IS preserved hippocampal replay, while spatially dispersed IS disrupted replay by causing over-generalization. Together, these results show that the spatial specificity of IS on the maze, but not rate, correlates with working memory deficits.Significance Statement In people with epilepsy, the hippocampus can generate large electrical discharges in the period between seizures called interictal spikes. Previous studies have proposed that interictal spikes cause memory impairments. We use a mouse model of epilepsy and computer simulations to study how interictal spikes impact navigation to remembered rewards. We find that when interictal spikes occur uncontrollably throughout the maze memory performance is worse, and in contrast, when they are sequestered to reward locations memory performance is better. Together our results show that interictal spikes are correlated with corrupted memory depending on when and where they occur during learning.