Memory: neurochemical mechanisms of memory

Although the molecular mechanisms of the functioning of a single nerve cell have been studied in many of their manifestations and the principles of organizing interneuronal connections have been formulated, it is still unclear how the molecular properties of neurons ensure the storage, reproduction and analysis of information - memory.

The fact that acquired knowledge (like moral principles) is not passed on by inheritance, and new generations have to learn them anew, allows us to consider that learning is a process of creating new interneuronal connections, and memorization of information is ensured by the brain's ability to reproduce these connections (activate them) when necessary. However, modern neurochemistry is not yet able to present a consistent theory describing how the analysis of factors of the external world is carried out in the living brain. We can only outline the problems that scientists in various fields of neurobiology are working on intensively.

Almost all animal species are capable of analyzing changes in the external environment to some extent and responding to them adequately. At the same time, the repeated reaction of the organism to external influence is often different from the first encounter. This observation shows that living systems have the ability to learn. They have a memory that preserves the animal's personal experience, which forms behavioral reactions and may differ from the experience of other individuals.

Biological memory is diverse. It is inherent not only to brain cells. The memory of the immune system, for example, stores information about a foreign antigen that once entered the body for a long time (often for a lifetime). When encountered again, the immune system triggers a reaction to form antibodies, allowing the infection to be quickly and effectively defeated. However, the immune system “knows” how to react to a known factor, and when encountering an unknown agent, it must develop a behavior strategy anew. The nervous system, unlike the immune system, can learn to create a behavior strategy in new circumstances, based on “life experience,” which allows it to develop an effective response to an unknown irritant.

The main questions that need to be answered when studying the molecular mechanisms of memory are the following: what metabolic changes occur in neurons when they encounter an external stimulus, allowing the information received to be stored for a certain (sometimes long) period of time; in what form is the information received stored; how is it analyzed?

During the process of active learning that occurs at an early age, changes in the structure of neurons are observed, the density of synaptic contacts increases, and the ratio of glial and nerve cells increases. It is difficult to distinguish between the process of brain maturation and structural changes that are molecular carriers of memory. However, it is clear that for the full development of intelligence it is necessary to solve problems presented by the external environment (remember the Mowgli phenomenon or the problems of adaptation to life in nature of animals raised in captivity).

In the last quarter of the 20th century, attempts were made to study in detail the morphological features of A. Einstein's brain. However, the result was rather disappointing - no features distinguishing it from the average brain of a modern person were revealed. The only exception was a slight (insignificant) excess of the ratio of glial and nerve cells. Does this mean that molecular memory processes do not leave visible traces in nerve cells?

On the other hand, it has long been established that inhibitors of DNA synthesis do not affect memory, while inhibitors of transcription and translation worsen memorization processes. Does this mean that certain proteins in brain neurons are memory carriers?

The organization of the brain is such that the main functions associated with the perception of external signals and reactions to them (for example, with a motor reaction) are localized in certain parts of the cerebral cortex. Then the development of acquired reactions (conditioned reflexes) should represent a "closure of connections" between the corresponding centers of the cortex. Experimental damage to this center should destroy the memory of this reflex.

However, experimental neurophysiology has accumulated a great deal of evidence that the memory of acquired skills is distributed across different parts of the brain, and is not concentrated only in the area responsible for the function in question. Experiments with partial damage to the cortex in rats trained to navigate a maze have shown that the time required to restore the damaged skill is proportional to the extent of the damage and does not depend on its localization.

Probably, the development of behavior in the maze includes the analysis of a whole set of factors (olfactory, gustatory, visual), and the areas of the brain responsible for this analysis can be located in different areas of the brain. Thus, although a certain area of the brain is responsible for each component of the behavioral reaction, the overall reaction is carried out through their interaction. Nevertheless, areas have been discovered in the brain whose function is directly related to memory processes. These are the hippocampus and the amygdala, as well as the nuclei of the midline of the thalamus.

Neurobiologists call the set of changes in the central nervous system associated with the recording of information (image, type of behavior, etc.) an engram. Modern ideas about the molecular mechanisms of memory indicate that the participation of individual brain structures in the process of memorizing and storing information does not consist in storing specific engrams, but in regulating the creation and functioning of neural networks that imprint, record, and reproduce information.

In general, the data accumulated in the study of behavioral reflexes and electrical activity of the brain indicate that both behavioral and emotional manifestations of life are not localized in a specific group of neurons in the brain, but are expressed in changes in the interactions of a large number of nerve cells, reflecting the functioning of the entire brain as an integral system.

The terms short-term memory and long-term memory are often used to describe the process of memorizing new information over time. In short-term memory, information can be stored for fractions of a second to tens of minutes, while in long-term memory, information can sometimes be stored for a lifetime. To transform the first type of memory into the second, the so-called consolidation process is necessary. Sometimes it is singled out as a separate stage of intermediate memory. However, all these terms, probably reflecting obvious processes, have not yet been filled with real biochemical data.

Types of memory and their modulation (based on: Ashmarin, 1999)

Types of memory	Inhibitors, effects
Short-term memory	Electroshock, anticholinergics (atropine, scopolamine), galanin, US1 (injection into specific parts of the brain)
Intermediate memory (consolidation)	Energy metabolism inhibitors, ouabain, hypoxia, inhibitors of RNA and protein synthesis (anisomycin, cycloheximide, puromycin, actinomycin O, RNase), antibodies to neurospecific proteins (vasopressin, protein B-100), 2-amino-5-phosphornovaleric acid (6-ARU)
Long-term (lifelong) memory	Inhibitors that irreversibly disrupt it are unknown. Partially suppressed by atropine, diisopropyl fluorophosphate, scopolamine

[ 1 ], [ 2 ], [ 3 ], [ 4 ]

Short-term memory

Short-term memory, which analyzes information coming from various sense organs and processes it, is realized with the participation of synaptic contacts. This seems obvious, since the time during which these processes are carried out is incommensurate with the time of synthesis of new macromolecules. This is confirmed by the possibility of inhibiting short-term memory by synaptic inhibitors, and its insensitivity to inhibitors of protein and RNA synthesis.

The consolidation process takes longer and does not fit into a strictly defined interval (lasting from several minutes to several days). Probably, the duration of this period is affected by both the quality of information and the state of the brain. Information that the brain considers unimportant is not subject to consolidation and disappears from memory. It remains a mystery how the question of the value of information is decided and what are the real neurochemical mechanisms of the consolidation process. The very duration of the consolidation process allows us to consider that it is a constant state of the brain, continuously implementing the "thought process". The diverse nature of the information entering the brain for analysis and the wide range of inhibitors of the consolidation process, different in their mechanism of action, allow us to assume that at this stage various neurochemical mechanisms are involved in the interaction.

The use of compounds listed in the table as inhibitors of the consolidation process causes amnesia (memory loss) in experimental animals - the inability to reproduce the acquired behavioral skill or to present the received information for use.

It is interesting that some inhibitors show their effect after the presentation of the information to be remembered (retrograde amnesia), while others - when used in the period preceding this (anterograde amnesia). Experiments on teaching chickens to distinguish grain from inedible but similar-sized objects are widely known. Introduction of the protein synthesis inhibitor cycloheximide into the chickens' brain did not interfere with the learning process, but completely prevented the skill from being consolidated. On the contrary, introduction of the Na-pump (Na/K-ATPase) inhibitor ouabain completely inhibited the learning process, without affecting the skills that had already been formed. This means that the Na-pump is involved in the formation of short-term memory, but does not participate in consolidation processes. Moreover, the results of experiments with cycloheximide indicate that the synthesis of new protein molecules is necessary for the consolidation processes, but is not needed for the formation of short-term memory.

Therefore, learning during the formation of short-term memory involves the activation of certain neurons, and consolidation involves the creation of long-term interneuronal networks, in which the synthesis of special proteins is necessary for the consolidation of interactions. It should not be expected that these proteins will be carriers of specific information; their formation may be "merely" a stimulating factor for the activation of interneuronal connections. How consolidation leads to the formation of long-term memory, which cannot be disrupted but can be reproduced on demand, remains unclear.

At the same time, it is clear that behind the creation of a stable skill there is the ability of a population of neurons to form a network in which signal transmission becomes most probable, and this ability of the brain can be preserved for a long time. The presence of one such interneuronal network does not prevent neurons from being involved in similar other networks. Therefore, it is clear that the analytical abilities of the brain are very large, if not unlimited. It is also clear that the implementation of these abilities depends on the intensity of learning, especially during the period of brain maturation in ontogenesis. With age, the ability to learn decreases.

Learning ability is closely related to the ability to plasticity - the ability of synaptic contacts to undergo functional reorganizations that occur during functioning, aimed at synchronizing neuronal activity and creating interneuronal networks. The manifestation of plasticity is accompanied by the synthesis of specific proteins that perform known (for example, receptor) or unknown functions. One of the participants in the implementation of this program is the S-100 protein, which belongs to annexins and is found in the brain in especially large quantities (it got its name from the ability to remain soluble at 100% saturation with ammonium sulfate at neutral pH values). Its content in the brain is several orders of magnitude greater than in other tissues. It accumulates mainly in glial cells and is found near synaptic contacts. The content of S-100 protein in the brain begins to increase 1 hour after learning and reaches a maximum in 3-6 hours, remaining at a high level for several days. Injection of antibodies to this protein into the ventricles of rats' brains disrupts the animals' learning ability. All this allows us to consider the S-100 protein as a participant in the creation of interneuronal networks.

Molecular mechanisms of plasticity of the nervous system

The plasticity of the nervous system is defined as the ability of neurons to perceive signals from the external environment that change the rigid determinism of the genome. Plasticity implies the ability to change the functional program of neuronal interaction in response to changes in the external environment.

Molecular mechanisms of plasticity are diverse. Let us consider the main ones using the glutamatergic system as an example. In the glutamatergic synapse, receptors with different properties are simultaneously found - both ionotropic and metabotropic. The release of glutamate into the synaptic cleft during excitation leads to the activation of kainate and AMPA-activated ionotropic receptors, causing depolarization of the postsynaptic membrane. When the transmembrane potential value corresponds to the resting potential value, NMDA receptors are not activated by glutamate because their ion channels are blocked. For this reason, NMDA receptors do not have a chance for primary activation. However, when depolarization of the synaptic membrane begins, magnesium ions are removed from the binding site, which sharply increases the affinity of the receptor to glutamate.

Activation of NMDA receptors causes calcium entry into the postsynaptic zone through the ion channel belonging to the NMDA receptor molecule. Calcium entry is also observed through potential-dependent Ca channels activated by the work of kainate and AMPA glutamate receptors. As a result of these processes, the calcium ion content in the perimembrane regions of the postsynaptic zone increases. This signal is too weak to change the activity of numerous enzymes sensitive to calcium ions, but is significant enough to activate perimembrane phospholipase C, whose substrate is phosphoinositol, and to cause accumulation of inositol phosphates and activation of inositol-3-phosphate-dependent calcium release from the endoplasmic reticulum.

Thus, activation of ionotropic receptors not only causes membrane depolarization in the postsynaptic zone, but also creates conditions for a significant increase in the concentration of ionized calcium. Meanwhile, glutamate activates metabotropic receptors in the synaptic region. As a result, it becomes possible to activate the corresponding G proteins "tied" to various effector systems. Kinases can be activated that phosphorylate various targets, including ionotropic receptors, which modifies the activity of the channel structures of these formations.

Moreover, glutamate receptors are also localized on the presynaptic membrane, which also have a chance to interact with glutamate. Metabotropic receptors of this area of the synapse are associated with the activation of the system for removing glutamate from the synaptic cleft, which works on the principle of glutamate reuptake. This process depends on the activity of the Na-pump, since it is a secondary active transport.

Activation of NMDA receptors present on the presynaptic membrane also causes an increase in the level of ionized calcium in the presynaptic region of the synaptic terminal. The accumulation of calcium ions synchronizes the fusion of synaptic vesicles with the membrane, accelerating the release of the mediator into the synaptic cleft.

When a series of excitatory impulses arrives at the synapse and the total concentration of free calcium ions is persistently elevated, activation of the Ca-dependent proteinase calpain can be observed, which breaks down one of the structural proteins fodrin, which masks glutamate receptors and prevents their interaction with glutamate. Thus, the release of a mediator into the synaptic cleft during excitation provides a variety of possibilities, the implementation of which can lead to amplification or inhibition of the signal, or to its rejection: the synapse operates on a multivariate principle, and the path realized at any moment depends on a variety of factors.

Among these possibilities is the self-tuning of the synapse for the best transmission of the signal that was amplified. This process is called long-term potentiation (LTP). It consists in the fact that with prolonged high-frequency stimulation, the responses of the nerve cell to incoming impulses are amplified. This phenomenon is one of the aspects of plasticity, which is based on the molecular memory of the neuronal cell. The period of long-term potentiation is accompanied by increased phosphorylation of certain neuronal proteins by specific protein kinases. One of the results of the increase in the level of calcium ions in the cell is the activation of Ca-dependent enzymes (calpain, phospholipases, Ca-calmodulin-dependent protein kinases). Some of these enzymes are related to the formation of active forms of oxygen and nitrogen (NADPH oxidase, NO synthase, etc.). As a result, the accumulation of free radicals, which are considered secondary mediators of metabolism regulation, can be registered in the activated neuron.

An important, but not the only result of free radical accumulation in a neuronal cell is the activation of the so-called early response genes. This process is the earliest and most transient response of the cell nucleus to a free radical signal; activation of these genes occurs within 5-10 minutes and continues for several hours. These genes include the groups c-fos, c-jun, c-junB, zif/268, etc. They encode several large families of specific transcription regulator proteins.

Activation of immediate response genes occurs with the participation of the nuclear factor NF-kB, which must penetrate the nucleus through the nuclear membrane to implement its action. Its penetration is prevented by the fact that this factor, which is a dimer of two proteins (p50 and p65), is in a complex with a protein inhibitor in the cytoplasm and is unable to penetrate the nucleus. The inhibitory protein is a substrate for phosphorylation by a specific protein kinase, after which it dissociates from the complex, which opens the way for NF-kB into the nucleus. The activating cofactor of protein kinase is hydrogen peroxide, therefore, a wave of free radicals, capturing the cell, causes a number of the processes described above, leading to the activation of early response genes. Activation of c-fos can also cause the synthesis of neurotrophins and the formation of neurites and new synapses. Long-term potentiation induced by high-frequency stimulation of the hippocampus results in activation of zif/268, encoding a Zn-sensitive DNA-binding protein. NMDA receptor antagonists block long-term potentiation and activation of zif/268.

One of the first to attempt to understand the mechanism of information analysis in the brain and develop a behavioral strategy in 1949 was S. O. Hebb. He suggested that in order to perform these tasks, a functional association of neurons - a local interneuronal network - should be formed in the brain. M. Rosenblatt (1961) refined and deepened these ideas by formulating the hypothesis of "Unsupervised correlation base learning." According to the ideas he developed, in the case of generating a series of discharges, neurons can synchronize due to the association of certain (often morphologically distant from each other) cells through self-tuning.

Modern neurochemistry confirms the possibility of such self-tuning of neurons to a common frequency, explaining the functional significance of series of excitatory "discharges" for the creation of interneuronal circuits. Using a glutamate analogue with a fluorescent label and armed with modern technology, it was possible to show that even when stimulating one synapse, excitation can spread to fairly remote synaptic structures due to the formation of the so-called glutamate wave. The condition for the formation of such a wave is the repeatability of signals in a certain frequency mode. Inhibition of the glutamate transporter increases the involvement of neurons in the synchronization process.

In addition to the glutamatergic system, which is directly related to learning (memorization) processes, other brain systems also participate in memory formation. It is known that the ability to learn shows a positive correlation with the activity of choline acetyl transferase and a negative correlation with the enzyme that hydrolyzes this mediator - acetylcholinesterase. Choline acetyltransferase inhibitors disrupt the learning process, and cholinesterase inhibitors promote the development of defensive reflexes.

Biogenic amines, norepinephrine and serotonin, also participate in the formation of memory. When developing conditioned reflexes with negative (electrical pain) reinforcement, the noradrenergic system is activated, and with positive (food) reinforcement, the rate of norepinephrine metabolism decreases. Serotonin, on the contrary, facilitates the development of skills under conditions of positive reinforcement and negatively affects the formation of a defensive reaction. Thus, in the process of memory consolidation, the serotonergic and norepinephrine systems are a kind of antagonists, and the disorders caused by excessive accumulation of serotonin can apparently be compensated by activation of the noradrenergic system.

The participation of dopamine in the regulation of memory processes has a multifactorial nature. On the one hand, it has been found that it can stimulate the development of conditioned reflexes with negative reinforcement. On the other hand, it reduces the phosphorylation of neuronal proteins (for example, protein B-50) and induces the exchange of phosphoinositides. It can be assumed that the dopaminergic system is involved in memory consolidation.

Neuropeptides released in the synapse during excitation are also involved in memory formation processes. Vasoactive intestinal peptide increases the affinity of cholinergic receptors to the mediator several thousand times, facilitating the functioning of the cholinergic system. The hormone vasopressin, released from the posterior pituitary gland, synthesized in the supraoptic nuclei of the hypothalamus, is transferred by axonal current to the posterior pituitary gland, where it is stored in synaptic vesicles, and from there is released into the blood. This hormone, as well as the pituitary adrenocorticotropic hormone (ACTH), constantly function in the brain as regulators of memory processes. It should be emphasized that this effect differs from their hormonal activity - fragments of these compounds, devoid of this activity, have the same effect on the learning process as whole molecules.

Non-peptide memory stimulants are virtually unknown. The exceptions are orotate and piracetam, which is widely used in clinical practice. The latter is a chemical analogue of gamma-aminobutyric acid and belongs to the group of so-called nootropic drugs, one of the effects of which is increased cerebral blood flow.

The study of the role of orotate in the mechanisms of memory consolidation is associated with an intrigue that excited the minds of neurochemists in the second half of the 20th century. The story began with J. McConnell's experiments on developing a conditioned reflex to light in primitive flatworms, planaria. After creating a stable reflex, he cut the planaria crosswise into two parts and tested the ability to learn the same reflex in animals regenerated from both halves. The surprise was that not only did the individuals obtained from the head part have increased learning ability, but also those regenerated from the tail learned much faster than the control individuals. It took 3 times less time to learn both than for individuals regenerated from the control animals. McConnell concluded that the acquired reaction is encoded by a substance that accumulates in both the head and tail parts of the planaria.

Reproducing McConnell's results on other objects encountered a number of difficulties, as a result of which the scientist was declared a charlatan, and his articles were no longer accepted for publication in all scientific journals. The angry author founded his own journal, where he published not only the results of subsequent experiments, but also caricatures of his reviewers and lengthy descriptions of the experiments he conducted in response to critical comments. Thanks to McConnell's confidence in his own rightness, modern science has the opportunity to return to the analysis of these original scientific data.

It is noteworthy that the tissues of "trained" planarians contain an increased content of orotic acid, which is a metabolite necessary for RNA synthesis. The results obtained by McConnell can be interpreted as follows: conditions for faster learning are created by an increased content of orotate in "trained" planarians. When studying the learning ability of regenerated planarians, we encounter not the transfer of memory, but the transfer of the skill to its formation.

On the other hand, it turned out that when planarian regeneration occurs in the presence of RNase, only individuals obtained from the head fragment demonstrate increased learning ability. Independent experiments conducted at the end of the 20th century by G. Ungar made it possible to isolate from the brain of animals with a reflex of avoiding darkness, a 15-membered peptide called scotophobin (an inducer of fear of darkness). Apparently, both RNA and some specific proteins are capable of creating conditions for the launch of functional connections (interneuronal networks) similar to those that were activated in the original individual.

In 2005, it was 80 years since the birth of McConnell, whose experiments laid the foundation for the study of molecular memory carriers. At the turn of the 20th and 21st centuries, new methods of genomics and proteomics appeared, the use of which made it possible to identify the involvement of low-molecular fragments of transfer RNA in consolidation processes.

New facts make it possible to reconsider the concept of DNA non-involvement in long-term memory mechanisms. The discovery of RNA-dependent DNA polymerase in brain tissue and the presence of a positive correlation between its activity and learning ability indicate the possibility of DNA participation in memory formation processes. It was found that the development of food conditioned reflexes sharply activates certain areas (genes responsible for the synthesis of specific proteins) of DNA in the neocortex. It is noted that DNA activation affects mainly areas that are rarely repeated in the genome and is observed not only in nuclear but also in mitochondrial DNA, and in the latter to a greater extent. Factors that suppress memory simultaneously suppress these synthetic processes.

Some memory stimulants (based on: Ashmarin, Stukalov, 1996)

Specificity of action	Stimulants
	Connection classes	Examples of substances
Relatively specific agents	Regulatory peptides	Vasopressin and its analogues, dipeptide pEOA, ACTH and its analogues
	Non-peptide compounds	Piracetam, gangliosides
	Regulators of RNA metabolism	Orotate, low molecular weight RNA
Broad-spectrum agents	Neurostimulators	Phenylalkylamines (phenamine), phenylalkyloidnonimines (sydnocarb)
	Antidepressants	2-(4-methyl-1-piperazinyl)-10-methyl-3,4-diazaphenoxazine dihydrochloride (azaphen)
	Cholinergic system modulators	Cholinomimetics, acetylcholinesterase inhibitors

The table shows examples of compounds that stimulate memory.

It is possible that the study of DNA's involvement in memory formation processes will provide a well-founded answer to the question of whether there are conditions under which formed skills or impressions can be inherited. It is possible that genetic memory of ancient events experienced by ancestors underlies some as yet unexplained mental phenomena.

According to a witty, though unproven, opinion, the flights in dreams that accompany the final formation of the mature brain, experienced by each of us in our youth, reflect the sensation of flight experienced by our distant ancestors at the time when they spent the night in trees. It is not without reason that flights in dreams never end in a fall - after all, those distant ancestors who did not have time to grab onto the branches when falling, although they experienced this sensation before death, did not give birth to offspring...