Animals experience a wide range of behavioral events at the same spatial location. Variations in experience at specific locations are expressed in firing patterns of hippocampal place cells by an orthogonal space and rate coding scheme in which distinct sets of neurons are activated at different spatial locations, whereas experience and behavioral context modulate the distribution of activity across the active place cell population in each place (Leutgeb et al., 2005). While integration of space and experience is thought to be part of the neural basis for episodic memory, the wider circuits supporting such integration remain to be characterized. A potential source of contextual input to the hippocampus is the medial prefrontal cortex (mPFC) (e.g. Jones and Wilson, 2005; Navawongse and Eichenbaum, 2013). These prefrontal neurons, however, do not directly project to the hippocampus. Here we asked if a thalamic midline nucleus, the nucleus reuniens (Re), could serve as a potential relay between mPFC and the hippocampus. The Re has reciprocal anatomical connections with mPFC and gives rise to a major excitatory input to CA1 (Vertes et al., 2007).

We recorded the activity of neurons in Re as well as CA1 when animals performed a continuous alternation task in a figure-8 maze. In this task, the majority of CA1 cells with a place field on the stem of the maze showed a trajectory-dependent change of their peak firing rate (Wood et al., 2000). Our findings support the idea that Re is a source of trajectory-dependent firing, because 1) many Re cells showed trajectory-dependent rate change, 2) CA3 cells, which do not receive Re inputs, showed significantly smaller rate change compared to that of CA1 cells, and 3) lesions of Re caused significant reduction of rate change in CA1 cells. We further validated the lesion studies by optogenetic manipulation of Re activity. As many thalamic nuclei have relay functions, we also explored possible sources of contextual information to the Re. We recorded the activity of mPFC cells in the figure-8 alternation task and found that a significant proportion of mPFC cells exhibited trajectory-dependent rate change similar to the one observed in Re and CA1 cells. Finally, we asked whether we can predict animals’ future trajectories from the firing rate of neurons in mPFC, Re and CA1 on the stem of the alternation task, and our analysis of correct and incorrect trials suggests that Re plays a key role in relaying decision-related information from mPFC to CA1. Taken together, our results identify a functional circuit consisting of mPFC, Re and CA1, that is required for conjunctive representation of current position and future trajectory in hippocampal neurons.