Computational diversity in the hippocampus: a matter of components

Multiple contextual elements compose our personal memories: the when, where and what of an event wrapped in how we felt at the time. These episodic memories depend heavily on the hippocampus, a structure critical for both memory encoding and recall. The last few decades of research have begun to elucidate the complexity of memory processing in the hippocampus, and hint that the hippocampus does not uniformly process information along its long axis (dorsal to ventral in rodents). For example, anatomical work shows that the inputs from and outputs to extrinsic brain structures are topographically graded along the dorsal to ventral hippocampal axis. In addition, experimental work points to the dorsal hippocampus as more important in processing spatial information, with ventral hippocampus crucial to emotional processing (for a recent review, see Strange et al. 2014). Moreover, whole-cell patch clamp recordings in hippocampal slices have shown that the biophysics of pyramidal cells differ along the dorsal–ventral axis (Dougherty et al. 2013). Combined, these results have begun to paint a picture of a computationally diverse hippocampus and raise the possibility that the dorsal versus ventral poles potentially process different components of episodic memories. In this issue of The Journal of Physiology, Hönigsperger et al. (2015) provide a new layer to this emerging picture of graded hippocampal processing by identifying a novel molecular component of the dorsal–ventral gradient in the biophysical properties of hippocampal principal cells.

Hönigsperger et al. use whole-cell patch clamp recordings to show that a dorsal–ventral gradient in the M-current, a potassium current conducted by the M-channel, contributes to a dorsal–ventral gradient in hippocampal pyramidal cell biophysics. First, they report a gradient in the basic physiological properties of CA1 pyramidal neurons, such as resting potential and input resistance, which is reflective of a higher M-current conductance in dorsal compared to ventral hippocampus. Second, and perhaps more intriguingly, they report gradients in several temporal and integrative properties that depend, at least in part, on the M-current. Resonant frequency, which measures the input frequency that results in the largest degree of membrane depolarization, is higher in dorsal compared to ventral. This suggests that dorsal and ventral cells may be differentially tuned to respond strongly to inputs arriving at different frequencies. In addition, the presence of a gradient in resonance implicates the M-current as a potential ionic player in temporal codes such as phase precession, which is also graded along the dorsal–ventral axis (Jung et al. 1994). Next, under control conditions, the authors find that the M-current normalizes synaptic integration across the dorsal–ventral axis. Once they blocked the M-current, however, summation increased selectively in dorsal cells. This result raises the interesting possibility that cholinergic suppression of the M-current could differentially impact the integration of inputs in dorsal versus ventral hippocampal pyramidal neurons. Finally, ventral cells show higher intrinsic excitability due to lower M-current conductance, with steeper firing rate versus input curves, lower spiking thresholds and less adaptation in response to excitatory inputs. This finding hints at a potential ionic mechanism for the higher susceptibility of ventral hippocampus to epileptic seizure activity (Bragdon et al. 1986).

How might the ionic diversity along the long axis of the hippocampus shape neural coding and memory processing? An intriguing piece to this puzzle is the finding by Hönigsperger et al. of a gradient specifically in the M-current, which is strongly modulated by acetylcholine. In the hippocampus, the presence of acetylcholine suppresses the M-current (Brown & Adams, 1980) and is known to play a critical role in memory processing. For example, experimental work has shown that cholinergic antagonists infused into the hippocampus interfere with the encoding of episodic memories (for a review, see Hasselmo, 2006). The known effect of cholinergic modulation on the M-current and the functional diversity of the hippocampus raise an interesting question; does acetylcholine play a different role in dorsal versus ventral memory processes? Cholinergic modulation has been proposed to enhance inputs during the encoding of new memories, potentially reducing overlap with items already stored in the hippocampus (Hasselmo, 2006). In dorsal, orthogonal neural codes might be critical for accurately storing spatial locations. For example, reducing overlap between memories for different spatial locations that share many overlapping features, such as the location of several different restaurants. On the other hand, in ventral hippocampus, generalization could be evolutionarily strategic when encoding emotional elements of a memory. Such overlap could, for example, reduce the chance of an animal experiencing a potentially dangerous stimulus in the future.

An additional intriguing, but unanswered, question raised by the Hönigsperger et al. study is whether the gradient in M-current is highly discretized between the dorsal and ventral poles or is more continuous in nature. Recent evidence suggests that the hippocampus is composed of multiple discretized subdivisions of gene expression, which reflect the combined overlap of many gene expression domains (for a review, see Strange et al. 2014). Given the recent report of a dorsal–ventral hippocampal gradient in the HCN channel, it would be interesting for future work to investigate if M- and H-channels significantly contribute to the discretization of hippocampal processing (Dougherty et al. 2013).

In conclusion, Hönigsperger et al. add important insight to our knowledge of the differential organization of hippocampal regions. This contribution is of particular importance, as it contributes to our understanding of the potential molecular substrates of hippocampal memory processes.

Additional information

Competing interests

None declared.

Funding

None declared.