Dentate granule cells encode auditory decisions after reinforcement learning in rats

Abstract

Auditory-cued goal-oriented behaviors requires the participation of cortical and subcortical brain areas, but how neural circuits associate sensory-based decisions with goal locations through learning remains poorly understood. The hippocampus is critical for spatial coding, suggesting its possible involvement in transforming sensory inputs to the goal-oriented decisions. Here, we developed an auditory discrimination task in which rats learned to navigate to goal locations based on the frequencies of auditory stimuli. Using in vivo calcium imaging in freely behaving rats over the course of learning, we found that dentate granule cells became more active, spatially tuned, and responsive to task-related variables as learning progressed. Furthermore, only after task learning, the activity of dentate granule cell ensembles represented the navigation path and predicts auditory decisions as early as when rats began to approach the goals. Finally, chemogenetic silencing of dentate gyrus suppressed task learning. Our results demonstrate that dentate granule cells gain task-relevant firing pattern through reinforcement learning and could be a potential link of sensory decisions to spatial navigation.

Nigrostriatal dopamine pathway regulates auditory discrimination behavior

“…Microendscopic procedures were performed as previously described 8 , 53 . Briefly, GRIN lenses 7.3 mm (Inscopix Inc., Palo Alto CA, 1050-002179) or 6.1 mm in length (Inscopix Inc., 1050-002182) were used for both auditory striatal and SNc imaging.…”

Route-dependent spatial engram tagging in mouse dentate gyrus

“…Studies of dynamic hippocampal activity shed light on the proportion of engram cells reactivated across days. One recent study used calcium imaging to examine the dynamics of DG granule cells during learning of a sensory-guided T-maze task, and found that place field stability correlated with behavioral performance (Shen et al, 2021). Alongside other DG imaging studies showing high cross-day similarity of spatial representations in well-trained animals (Hainmueller & Bartos, 2018; Cholvin et al, 2021), we can speculate about the mechanisms underlying the observed 7.8% reactivation in the Full-T group relative to Day 1.…”

“…Splitter cells can emerge early in learning and may be modulated by the turn direction, task phase, or both on the DNMP T-maze task (Levy et al, 2021). DG place fields remap based on task engagement, hinting at the flexibility of DG cells within the same physical location (Shen et al, 2021). Interestingly, CA1 splitter cells appear more stable than classic place fields across days, which might help to explain increased overlap observed in the Full-T group relative to the Half-T group which did not experience context-modulation on Day 2 (Fig 3B; Kinsky et al, 2020).…”

“…Studies of dynamic hippocampal activity shed light on the proportion of engram cells reactivated across days. One recent study used calcium imaging to examine the dynamics of DG granule cells during learning of a sensory-guided T-maze task, and found that place field stability correlated with behavioral performance (Shen et al, 2021).…”

“…Splitter cells can emerge early in learning and may be modulated by the turn direction, task phase, or both on the DNMP T-maze task (Levy et al, 2021). DG place fields remap based on task engagement, hinting at the flexibility of DG cells within the same physical location (Shen et al, 2021).…”