Neural activity suppression in the mediodorsal thalamus precedes the occurrence of hippocampal ripples

Forming a persistentmemory trace requires hippocampal-cortical interaction. Population bursts in the hippocampal network occurring during awake immobility or NREMsleep propagate throughout the entire hippocampal formation and generate transient dynamic interactions locally, but also between the hippocampus (HPC) and cortex. This synchronized population activity is revealed in the local field potentials as brief, high-frequency (about 200 Hz) oscillations, or ripples, which are thought to mediate the hippocampal-cortical communication underlying memory consolidation [1]. The medial prefrontal cortex (mPFC) receives direct input from the HPC andmanymnemonic processes depend on these two brain regions [2]. The HPC-mPFC pathway is considered critical for consolidation of declarative memory and is currently one of the most studied memory-related pathways. A memory-supporting network is, however, not limited by the HPC and the mPFC. The thalamic mediodorsal (MD) nucleus is likely a part of an extended memory network. The MD is reciprocally connected with the mPFC and has long been implicated in different mnemonic functions [3]. Our fMRI-based mapping of the whole brain activity associated with ripples occurrence suggested that silencing of a subset of subcortical regions, including thalamus, may reduce interference for hippocampal-cortical communication [4]. We characterized neural activity in the MD around times of the hippocampal ripples in spontaneously behaving rats. Generally, the MD population activity was strongly suppressed around ripples. A substantial reduction of MD firing occurred 0.4–2.4 sec (mean: 1.1±0.1 sec) before the ripple peak and lasted for 2.1±0.2 sec. Moreover, the degree of MD activity suppression correlated with the ripple amplitude. The ripple-associated decrease of the MD firing rate was the strongest and the most consistent during awake immobility. In contrast, during NREM sleep bidirectional modulation of the MD activity was observed: the MD firing was actually enhanced around ripples that were temporally coupled with sleep spindles, while it was decreased around spindle-uncoupled ripples. Our results suggest possible competitive interaction between the hippocampal-cortical and thalamo-cortical networks supporting ‘off-line’ and ‘on-line’ information processing, respectively