Memory  >   The Single Dendritic Branch as a Fundamental Memory Unit

A "memory unit" is the elementary physical component of an enduring memory. Accumulating evidence suggests that the memory unit is not an individual synapse or spine. Rather, it may contain a cluster of synapses within the same dendritic branch (Poirazi and Mel, 2001; Polsky et al., 2004; Larkum and Nevian, 2008; Branco and Häusser, 2010; Govindarajan et al., 2011; Winnubst and Lohmann, 2012; Fu et al., 2012; Yang et al., 2014; Gökçe et al., 2016). These clustered synapses act synergistically to produce NMDA spikes and plateaus (Chapter 8).

The evidence for synaptic clustering first came from theoretical studies indicating that clustered synapses can function as a separately thresholded neuron-like summing unit. The non-linear summation of clustered synaptic inputs may increase information storage capacities (Poirazi and Mel, 2001). Direct experimental proof has long been lacking until recently. Fu et al. (2012) showed that, during the acquisition phase of repetitive motor learning, a third of new dendritic spines emerge in clusters, and that most such clusters are neighboring spine pairs. Yang et al. (2014) reported that after motor learning new spines are formed on different sets of dendritic branches in response to different learning tasks. Gökçe et al. (2016) observed clusters of synaptic inputs on the basal dendrites of layer 5 pyramidal cells in mouse visual cortex.

As mentioned in the previous chapter, clustered synaptic input can produce NMDA spikes; repetitive clustered input can further produce the NMDA plateau. During repetitive glutamatergic stimulation, glutamate is accumulated at the basal dendrites of engram cells. The buildup of glutamate is called a "glutamate pond" (Figure 9-1), which has been demonstrated to play a critical role in producing NMDA plateaus. It is the level of the glutamate pond, rather than excitatory postsynaptic potentials (EPSPs), that controls NMDA plateaus and engram cell activation (Oikonomou et al., 2014).


Figure 9-1. Illustration for the glutamate pond. Repetitive glutamatergic input can build up a "glutamate pond" to produce the NMDA plateau and activate engram cells. [Adapted from: Oikonomou et al., 2014]

The NMDA plateau is accompanied by Ca2+ influx through NMDARs, which engulfs the entire length of the respective dendritic branch (Oikonomou et al., 2014). Since Ca2+ plays a central role in the formation of enduring memory, the individual spines in a dendritic branch cannot act separately. They must work as a whole. Therefore, an entire dendritic branch may serve as a fundamental memory unit. The strength of a memory unit then depends on the number of synapses in the memory unit which is likely to change dynamically. In response to learning processes, new synapses could be added while existing synapses might be eliminated from a memory unit. A neuron may consist of multiple memory units, encoding different objects and events. In the medial temporal lobe, each neuron was estimated to encode 50 - 150 distinct objects (Rey et al., 2015).


Author: Frank Lee
First published: April, 2018