Types of Memory Storage

It has been well established that experience can leave memory traces by inserting neurotransmitter receptors into postsynaptic membranes. This type of memory traces are usually short (minutes to hours), because neurotransmitter receptors are constantly cycled into and out from postsynaptic membranes due to a constitutive pathway and busy activities at the hippocampus (review 1; review 2). Since microtubules are essential for transporting neurotransmitter receptors and other molecules for synaptic strength and dendritic growth, Chapter 9 proposed another type of memory traces: modification of the microtubular network. Such memory traces require not only calcium ions, but also microtubule-associated proteins (MAPs). They are more difficult to create and thus harder to be replaced by new events. Its memory capacity, however, is limited by the number of microtubules. The hippocampus, which is the first region to acquire memory, has a small size. Old memory could be replaced by new memory within days. For long term storage, the memory stored in the hippocampus must be transferred to other brain regions such as the neocortex (review). There is strong evidence suggesting that the hippocampus and neocortex communicate with each other for memory consolidation (reference 1; reference 2).

Memory Consolidation During Sleep

Electroencephalogram (EEG) exhibits several types of brain waves. The one with the lowest frequency (~1 Hz) is called the delta wave, which is predominant in babies and also observed in children and adults during slow wave sleep - the sleep characterized by the slow delta wave. Slow waves have been demonstrated to play a critical role in memory consolidation (review). They becomes less likely to occur as one ages (reference 1; reference 2).

A slow wave oscillates between a depolarization phase and a hyperpolarization phase, each lasting for several hundred milliseconds. The hyperpolarization phase is known as the "DOWN state", during which nearly all cortical neurons are silent. The depolarization phase is called the "UP state" with intense firings by cortical neurons. The slow wave can originate from anywhere in the cortex spontaneously (reference), and propagate to other regions as a traveling wave (reference). It has been well established that the slow wave is coupled to "sharp waves" which originate in the hippocampus. UP states have been found to precede hippocampal sharp waves (review). These results indicate that, during slow wave sleep, the neocortex sends message to the hippocampus, looking for new memory traces. The hippocampus replies by generating sharp waves, which then travel to the neocortex and modify its memory storage.

A related feature of brain waves is the K-complex, which consists of a large amplitude biphasic fluctuation, either from "down" to "up" or from "up" to "down" (reference). The duration of the K-complex is about the same as a single cycle (~1 second) of the slow wave. K-complexes are also less likely to occur as one gets older.

Mechanisms of Slow Waves

The spontaneous generation of UP states could arise from excitation by glutamates (reference), but the mechanism for DOWN states is less clear. It has been proposed that the hyperpolarization was caused by outward K⁺ currents, possibly through ATP-sensitive K⁺ channels (reference), because the intense firing during the UP state may increase the intracellular Na⁺ level, which then activates Na⁺/K⁺-ATPase and consumes ATP. The ATP-sensitive K⁺ channels open when the ATP level decreases. This model sounds reasonable. However, it depends on preceding intense firings while the K-complex may not have a preceding up phase (reference). It is also unclear how the model can explain the age-dependence of both slow waves and K-complexes.

Here enters the microtubule. As mentioned in Chapter 9, the average length of microtubules in dendrites is age-dependent. Shorter microtubules may represent those which are not fixed in a microtubular track toward specific destination. If somehow these free microtubules can cause the DOWN state, then the age-dependence of both slow waves and K-complexes can be explained. The question is: how?

Microtubules are highly negatively charged, about 50 electrons per tubulin dimer (reference). If they can adsorb on the membrane surface, they will have significant effects on membrane properties. Strictly speaking, the opening probability of voltage-gated ion channels depends not only on the membrane potential, but also on the local electric field from nearby surface charges. For instance, high extracellular Ca²⁺ concentration, which neutralizes the negative surface charges of a cell membrane, can shift the voltage dependence of Na⁺ and K⁺ channels (those involved in the generation of action potentials) by as much as 30 mV (reference). Since surface charges are usually fixed, their effects can be ignored in most cases. However, in our case, the surface charges play a crucial role.

Cellular membrane surfaces are negatively charged. To adsorb on the membrane surface, microtubules must rely on counterions, which can mediate attraction between like-charges (reference 1; reference 2). Multivalent counterions are usually more effective than monovalent counterions. In the following, we shall show that the DOWN state can be explained by the assumption that Ca²⁺ ions, but not monovalent cations, can mediate the adsorption of microtubules on the intracellular membrane surface.

Synaptic stimulation may cause Ca²⁺ accumulation near the postsynaptic site. Part of them may loosely bind to the negative charges on the intracellular membrane surface, forming a layer of counterions. These Ca²⁺ ions make the nearby ion channels feel more depolarized than the membrane potential. Thus, binding of Ca²⁺ ions on the intracellular membrane surface makes a neuron more excitable. On the other hand, these Ca²⁺ ions can mediate the adsorption of free microtubules if they are around. As a result, the Ca²⁺ ions serve as the counterions for both microtubules and membrane surface charges. The adsorption of highly negatively charged microtubules should make nearby ion channels feel more hyperpolarized, which counteracts the excitatory effects of Ca²⁺ ions. If sufficient number of free microtubules can adsorb on the membrane surface, neuronal firings will be suppressed, resulting in the DOWN state. Since memory traces generally fix microtubules in a specific track toward its destination, the number of free microtubules should decrease with age. This explains the age-dependence of the DOWN state.

This model is supported by the observed duration of Ca²⁺ accumulation. In the absence of additional synaptic stimulation, Ca²⁺ ions eventually diffuse away. The adsorbed microtubules should also leave the membrane surface because they cannot bind to the membrane surface without divalent counterions. In basal dendrites, upon stimulation by glutamate microiontophoresis, the duration of Ca²⁺ accumulation was found to be about 1 second (reference, figure 1) - the same as a single cycle of the slow wave. This experiment also simultaneously measured the membrane potential at the soma. The duration of the somatic plateau depolarization was about 300 milliseconds. This means that Ca²⁺ accumulation can last much longer than the somatic plateau depolarization. According to the present model, the period during the somatic plateau depolarization may be identified as the UP state, and the subsequent period until substantial decrease in Ca²⁺ accumulation could be the DOWN state, during which sufficient Ca²⁺ accumulation can mediate the adsorption of free microtubules to suppress neuronal firings.

The present model requires Ca²⁺ accumulation to induce the DOWN state, but it does not require preceding intense firings. As mentioned above, some K-complexes do not have a preceding up phase. These K-complexes may originate from synaptic stimulation which causes Ca²⁺ accumulation, but without generating action potentials.

Author: Frank Lee
First Published: September 22, 2009