Indirect feedback inhibition from hilar mossy cells enhances the sparse activation of dentate granule cells and frequency-dependent pattern separation

The discrimination of similar episodes and places and their representation as distinct memories depend on a process called pattern separation that is performed by the circuitry of the hippocampal dentate gyrus (hDG). Excitatory hilar mossy cells (MCs) support pattern separation through their connections with different inhibitory interneuron (INT) populations that provide feedback inhibition to granule cells (GCs). In this study, we investigated how MCs and their synaptic connections with cholecystokinin-expressing interneurons (CCKIs) and parvalbumin-expressing interneurons (PVIs) influence the dynamics of hDG circuitry by using pharmacological agents to reduce neurotransmission from MCs or different INTs, genetically modified (GM) mice to selectively remove MCs from hDG circuitry and optogenetics to manipulate INT activity during GC whole cell/field recordings (age 8-10 weeks, n=5-18 animals/group). In addition, we used computational models to simulate hDG circuit dynamics. Our results showed that the net influence of MCs on GC activity is frequency-dependent and inhibitory, likely through the recruitment of INTs, since (i) GC responses to 20Hz and 50Hz stimulation of the medial perforant pathway (MPP) were less depressed in GM mice (that lacked MCs) compared with WT, but not at 5Hz; (ii) GC responses to 20Hz and 50Hz stimulations of the MPP were less depressed in the presence of GABAA receptor antagonist picrotoxin (PTX) in WT animals but not to 5Hz; (iii) GC responses to 20Hz and 50Hz stimulation of the MPP were less depressed, compared to controls, by pharmacologically reducing (40.9%±2.2%) MC-GC synaptic transmission with the type 1 cannabinoid receptor agonist WIN 55,212-2 (WIN). Secondly, our preliminary results confirmed that CCKIs and PVIs provide feedback inhibition of GCs at different stimulation frequencies: (i) GC responses to 10Hz and 20Hz stimulation of MPP were increased by blocking CCKI-GC synaptic transmission using the N-type voltage-gated calcium channel inhibitor ω-Conotoxin GVIA in WT hippocampal slices; (ii) GC responses to 5Hz and 10Hz stimulation of MPP were restored by activating ChR2-expressing PVIs in GM animals. Finally, our computational model showed that the removal of MCs from DG circuitry led to the disinhibition of GC activity through their indirect MC-basket cell-GC connections. Together, these findings suggest that the net inhibitory influence of MCs on GC activity is frequency-dependent due to the recruitment of CCKIs and PVIs at different stimulation frequencies.