As any graduate student in neuroscience can attest, the hippocampus, given its importance in memory formation and its popularity as a model system for studying synaptic transmission and plasticity, has been one of the most well-studied brain regions. It has been thought to have a prescribed feedforward architecture, whereby excitatory axons originating in its CA3 subfield run along its transverse axis to form synaptic connections with CA1 pyramidal neurons, which in turn contact other excitatory neurons in the subiculum. This circuitry forces information to flow unidirectionally towards the cortex. On page 1362 of this issue, Jackson et al. challenge this notion of unidirectionality by demonstrating that hippocampal activity can flow 'in reverse', from the subiculum to CA3.

Jackson et al. studied theta oscillations, which are thought to be important for the acquisition and retrieval of memories and for the organization of internal representations of the environment. Such oscillations propagate through the hippocampal network in the form of travelling waves. Using an ex vivo preparation that preserves the entire intra-hippocampal excitatory-inhibitory network, Jackson et al. first showed that there were two distinct theta oscillators in the hippocampus. One of these oscillators was in CA3 and exerted its influence in the canonical direction (toward CA1). However, they also found a second, independent and slightly faster theta oscillator in the subiculum. Surprisingly, this oscillator was sending waves toward CA1, in the opposite direction of what was expected from the canonical circuitry of the hippocampus. Interestingly, the two oscillators showed some phase coupling suggesting some form of interaction. The figure illustrates this phenomenon as it plots the phase coherence between CA3 (x axis) and the subiculum (y axis), with warmer colors indicating stronger phase covariations, and, in this case, a 1:2 phase locking ratio. Using an information theory–based approach, the authors showed that activity in the subiculum preceded CA3 spikes and, in a separate set of experiments, found that subicular inactivation with a sodium channel blocker changed the rhythmicity of CA3. The authors also noted that the subicular oscillator was absent in acute slices of hippocampus, which may explain why this phenomenon had not been previously reported.

What mechanisms may mediate this reverse flow of information? The subicular inactivation led to an increase in the strength of oscillatory activity in CA3, suggesting that inhibition was possibly involved. Consistent with this hypothesis, blocking GABAA receptors pharmacologically led to a decrease in the spread of activity from the subiculum to CA3 and, conversely, optogenetic activation of GABAergic neurons in the subiculum entrained CA3. The role of fast inhibition in this process suggests that long-range interneurons are involved in the backward flow of information from subiculum to CA3.

Jackson et al. went on to confirm, using in vivo recordings, that this reverse flow of information could occur in intact animals that were either exploring their environment or sleeping. Interestingly, a phase coupling similar to the one observed ex vivo was most prominent during REM sleep, suggesting that the reverse flow of information was a state-dependent phenomenon.

This study shows that CA3 and subiculum interact bidirectionally and that, counter to conventional wisdom, the hippocampus is not a 'one-way street'. Future studies will hopefully shed some light on the role that this reversed pattern of activity has in hippocampal function.