Block Rockin' Beats - Glutamate Excitation and GABA Inhibition

I'm currently reading Joseph LeDoux's excellent book "Synaptic Self" - I highly recommend it. Chapter 3 of the book - "The Most Unaccountable Machinery" - does a splendid job of covering the basic working mechanisms of neurons, axons, dendrites and synapses, as well as the history behind some of the most important discoveries in neurobiology. The section covering inhibition was particularly enlightening for me, so I'd like to use this post to capture the key points on inhibition and the roles of Glutamate and GABA.
In a previous post (Neurotransmitters - molecular messages), the following definition of GABA was quoted from another excellent (and free!) book: "Discovering the Brain" by Sandra Ackerman: GABA (gamma-aminobutyric acid) often acts as a fast synaptic transmission inhibitor. Unlike dopamine or serotonin, which have diverse roles, GABA consistently acts as an “off” signal; the cerebellum, retina, and spinal cord all use this transmitter to inhibit signals, as do many other parts of the brain and nervous system. GABA's inhibitory effect comes about in the following way: the transmitter opens a channel in the membrane through which negatively charged chloride ions can enter the cell. This influx hyperpolarizes the cell and makes it less likely to be excited by incoming stimuli. GABA receptor sites show some tendency to bind barbiturates and the “minor tranquilizers,” the benzodiazepines. Curiously, the presence of GABA in low concentrations enhances the binding of benzodiazepines to receptor sites. This pattern indicates that GABA and the benzodiazepines cannot be competing for exactly the same sites. Instead, an array of recent studies have yielded the view that the GABA receptor site is in fact a multifunctional set of proteins that contain the chloride ion channel and distinct subsites for binding of benzodiazepines, other tranquilizers such as barbiturates, and GABA itself.
Unlike GABA, Glutamate (the 'G' in MSG) is an excitatory neurotransmitter. Dr. LeDoux refers to Glutamate and GABA as "The Chemical Brothers" (hence this post's title :) )

From Synaptic Self: Inhibition is a very useful device in neural circuits. It adds termendously to the specificity of information processing, filtering out random excitatory inputs, preventing them from triggering activity. Only if the excitatory inputs arrive simultaneously can they overcome the inhibition and elicit activity. And once activity is elicited, inhibition is important for keeping the excitation in check and resetting the circuit.

From Wikipedia: GABA acts by binding to specific receptors in the plasma membrane of both pre- and postsynaptic cells. This binding causes the opening of ion channels to allow either the flow of negatively-charged chloride ions into or positively-charged potassium ions out of the cell. This will typically result in a negative change in the transmembrane potential.

Three general classes of GABA receptor are known. These include GABA_A and GABA_C ionotropic receptors, which are ion channels themselves, and GABA_B metabotropic receptors, which are G protein-coupled receptors that open ion channels via intermediaries

From The Synaptic Self:Glutamate receptors (such as the NMDA receptor) tend to be located out on the dendrites, especially in the spines, whereas GABA receptors tend to be found on the cell body, or on the part of dendrites close to the cell body. In order for glutamate-mediated excitation to reach the cell body to help trigger an action potential, it has to get past the GABA guard. Excitation coming down a dendrite and headed for the cell body can be extinguished by GABA.

Without GABA inhibition, neurons would send out action potentials continuously under the influence of glutamate, and would eventually literally fire themselves to death. ... Overactivity of glutamate, and the resulting injury to neurons, actually plays an important role in stroke and other vascular disorders of the brain, as well as in epilepsy and possibly Alzheimer's disease.

This is Spinal Tap - Dendritic Spines

The picture at right is truly amazing. It overlays three color-coded images of dendritic spines in a living mouse's brain, collected 45 minutes apart. White regions indicate stable dendritic segments. Green shows spines that retracted and red shows spines that sprouted during the observation period.

From A New Window to View How Experiences Rewire the Brain: Howard Hughes Medical Institute researchers have developed sophisticated microscopy techniques that permit them to watch how the brains of live mice are rewired as the mice learn to adapt to new experiences.

Their studies show that rewiring of the brain involves the formation and elimination of synapses, the connections between neurons. The technique offers a new way to examine how learning can spur changes in the organization of neuronal connections in the brain.
...
“Our first observations of the large-scale structure of neurons, their axons and dendrites, revealed that they were remarkably stable over a month.” Dendrites and axons are highly branched structures, where dendrites are the input side of neurons and axons the output side.

“However, when we zoomed in closer, we found that some spines on dendrites appeared and disappeared from day to day,” said Svoboda. These spines stipple the surface of dendrites, like twigs from a branch, and form the receiving ends of synapses, which are the junctions between neurons where neurotransmitters are released.

“This finding was quite unexpected, because the traditional view of neural development has been that when animals mature, the formation of synapses ceases, which is indicated by stable synaptic densities,” said Svoboda. “However, the flaw in this view has been that a stable density only indicates a balanced rate of birth and death of synapses. It doesn’t imply the absence of the formation of new synapses, but it was often interpreted that way.”

In their experiments, Svoboda and his colleagues observed that about twenty percent of spines disappeared from one day to the next, offset by the formation of new spines.

“While we were surprised at the rate of turnover of some spines, we were also surprised at the incredible stability of other spines,” said Svoboda. The spines appeared to fall into different classes. And while there were those that turned over rapidly, other spines, typically the larger ones, persisted for months.

For this post, I've pulled together a lot of different articles on dendritic spines. They are amazingly complex, and are remarkably self-sufficient - they've got local mitochondria for powering their cellular machinery, can not only locally assemble proteins but also control the structure of the proteins that are to be assembled there, and can control their shape to help preserve long term memories...

From ScienceDaily (Dec. 22, 2004): Compared with most cells, neurons are more elongated and complex. Neurons are divided into different sections: the long, thin axon and the branching, tree-like dendrites that receive signals from other neurons via the suction cup-shaped synapses. Synapses, tiny gaps separating neurons, consist of a presynaptic ending at the very tip of an axon that contains neurotransmitters, a postsynaptic ending that contains neurotransmitter receptor sites, and the space between the two endings. Both presynaptic and postsynaptic endings require the energy generated by mitochondria.

Critical functions, such as synapses' transmission of information and ability to change rapidly in response to stimuli, are managed by distant cellular compartments that can become isolated from their nearest mitochondria. Sheng and postdoctoral fellow Zheng Li explored whether having the power source far away, like a too-distant room heater, would affect synaptic function.

The researchers looked at mitochrondria in living hippocampal neurons. The hippocampus, a seahorse-shaped brain region in the temporal lobe, is known to play a critical role in memory.

When synapses were stimulated, the researchers found, mitochrondria in the dendrites changed from being long and thin to more aggregated, collecting in globules near the enlarged dendritic spines, as if the mitochondria were reporting for duty at the active part of the neuron.

"Our studies reveal that mitochrondria dynamically redistribute into dendritic protrusions in response to synaptic excitation … The dendritic distribution of mitochondria appears to be an essential and limiting factor for synaptic density and plasticity," the authors wrote. If mitochondria don't migrate into the distant reaches of the dendrites--where most synapses are present--the synapses become less numerous and lose some of their ability to respond to external input. Enhancing mitochondria with the nutrient creatine also promoted synaptic density, the study showed.


ScienceDaily (Jan. 15, 2006): Harvard University biologists have identified a molecular pathway active in neurons that interacts with RNA to regulate the formation of long-term memory in fruit flies. The same pathway is also found at mammalian synapses.

Changes in the dendrites themselves outside of the synaptic region have also been shown to be critical to memory formation. From ScienceDaily (Nov. 3, 2005): Neurons experience large-scale changes across their dendrites during learning, say neuroscientists at The University of Texas at Austin in a new study that highlights the important role that these cell regions may play in the processes of learning and memory.
...
Dendrites--the thin branch-like extensions of a neuron cell--receive many inputs from other neurons that transmit information through contact points called synapses. Much attention has been focused on the role that changes at synapses play in learning. They change in ways that make it easier for connected neurons to pass information.

Johnston and his colleagues show that learning and memory are likely to not only involve changes at synapses, but also in dendrites. They found that h-channels, which are distributed throughout the dendrite membrane and allow the passage of potassium and sodium ions into and out of the neuron, are altered during learning.

"The h-channels undergo plasticity, not near the synapse but probably throughout the dendritic tree," says Johnston.

To record the changes during learning, cells from the rat hippocampus (an important area of the brain for short-term memory) were electrically stimulated using a high frequency pattern called theta-bursts. Theta-bursts mimic the electrical stimulus that shoots through neurons when animals perform a learning task. The researchers found that when stimulated with theta-bursts, hippocampus neurons showed h-channel plasticity and a rapid increase in the synthesis of h-channel proteins.

The proteins were produced in the rat hippocampal neurons within 10 minutes, which is pretty rapid for cells, says Johnston.

"This really pushes the envelope with respect to how fast a neuron can produce new proteins important for learning," he says.

Learning and memory researchers know that protein synthesis in neurons is related to long-term memory, because protein synthesis inhibitors block long-term memory in animals.

Johnston says it's possible that the new proteins are being used by the neuron to build more h-channels in the dendrite membrane. He has a working hypothesis that h-channels may help buffer receiving neurons from being barraged and over-stimulated by inputs coming from information transmitting neurons.


From the article RNA splicing occurs in nerve cell dendrites: Researchers at the University of Pennsylvania School of Medicine have discovered that nerve-cell dendrites have the capacity to splice messenger RNA (pre-mRNA), a process once believed to only take place in the nucleus of cells.
In the nucleus of a mammalian cell, a gene is copied into mRNA, which possesses both exons (mature mRNA regions destined to code for proteins) and introns (non-coding regions). mRNA splicing works by cutting out introns and merging together the remaining exon pieces, resulting in an mRNA capable of being translated into a specific protein.

The vast array of proteins within the human body arises in part from the many ways that mRNAs can be spliced and reconnected. Specifically, splicing removes pieces of intron and exon regions from the RNA, with the resulting spliced RNA often being made into protein. Should the RNA have different exons spliced in and out of it, then different proteins can be made from this RNA.
...
Dendrites, which branch from the cell body of the neuron, play a key role in the communication between cells of the nervous system, allowing for many neurons to connect with each other to form a network. Dendrites detect the electrical and chemical signals transmitted to the neuron by the axons of other neurons. The synapse is the neuronal structure where this chemical connection is formed, and investigators surmise that the synapse is where learning and memory occur.
...
Protein diversity is a key aspect to the complexity of the central nervous system. Proteins are the workhorses of the cell and are generally responsible for insuring that cells function properly. When proteins interact with one another they can elicit specific physiological responses, including the generation and maintenance of memories. Changing protein identity, as can occur with splicing, can change the ability of the protein to interact with other proteins and therefore potentially change such physiological processes. With the dendrite being the initial site in the neuron where learning is thought to occur, the ability to create a diversity of mRNAs, through local splicing, and subsequent protein translation may permit exquisitely sensitive control of these cellular functions.

"The regulation and timing of the expression of proteins is what makes the central nervous system function," says Eberwine. The diversity and redundancy of the nervous-system proteins may serve to help maintain the system over a lifetime. However, failure in protein regulation or proper expression in neurons may give rise to cognitive dysfunction. "Most neurodegenerative and psychiatric illnesses exhibit dendrite dysfunction, therefore, the inability to properly generate spliced RNAs in dendrites or proteins may underlie aspects of these disease processes."

From ScienceDaily (Sep. 24, 2007): A molecular "recycling plant" permits nerve cells in the brain to carry out two seemingly contradictory functions -- changeable enough to record new experiences, yet permanent enough to maintain these memories over time. Michael Ehlers of Duke University Medical Center has done a study that shows the importance of receptor recycling in the dendritic spines plays in maintaining memories. "This process occurs on a time scale of minutes or hours, so the acquisition of new neurotransmitter receptors and their recycling is an on-going process. Memory loss may result from receptors escaping from the synapse. ... We believe that the existence of this recycling ability explains in part how individual dendritic spines retain their unique identity amidst this constant molecular turnover," Ehlers said. "The system is simultaneously dynamic and stable. ... If the receptors don't get recycled, you see a gradual loss of synaptic function that is associated with reduced cognitive ability," Ehlers said. "These dendritic spines are where learning and memories reside. These are the basic units of memory." (See the post on Clathrin - the super cool cellular transport machine for more information on receptor recycling.)

From Establishing Synaptic Independence: How Neurons Create Diffusional Barriers by Bernardo Sabatini: The optimal diffusional isolation of each spine likely depends on the state of the associated synapse. For example, at certain times the synapse may need to exchange components with the dendrite, requiring a relatively open neck. At other times, it may be beneficial to constrict the neck and favor the accumulation of plasticity-inducing molecules in the spine head. We found that the strength of diffusional coupling across the neck is a dynamically regulated parameter that is adjusted in response to changes in activity levels. Furthermore, the diffusional isolation of each synapse is controlled by a local signaling loop, such that when a synapse and the postsynaptic cell are consistently coactive, the neck of the spine that houses the active synapse is rapidly constricted. Since these patterns of activity are known to induce strengthening of hippocampal synapses, we propose that adjusting the properties of the neck may titrate the threshold for plasticity induction by determining the lifetimes of plasticity-inducing molecules near active synapses.

More...
Synapses in a State: A Molecular Mechanism to Encode Synaptic History and Future Synapse Function by Johanna Montgomery
Dynamin-dependent NMDAR endocytosis during LTD and its
dependence on synaptic state discusses the idea of synaptic state

Actin Lessons Part II: Memorabilia

Recall from the previous post, that when a neuron's axon fires repeatedly the relevant genes (in that neuron) turn on, and the synapses that are holding the short-term memory when the synapse strengthening proteins find them, become, in effect, tattooed (from Making Memories Stick by R. Douglas Fields)

It appears that this 'tattooing' process involves enzymes that cause actin to change the shape of the synapse, broadening it so that more receptors can be brought into play. Much progress has been made in the past 10 years or so to understand the details of what is going on.

From ScienceDaily (Jun. 14, 2004): Neuroscientists at the Picower Center for Learning and Memory at MIT show for the first time that storage of long-term memories depends on the size and shape of synapses among neurons in the outer part of the brain, the cerebral cortex.
...
When an experience or a fact is repeated enough or elicits a powerful emotional response, it shifts from short- to long-term memory. It moves from the hippocampus, in the innermost fold of the temporal lobe, to the brain’s outermost region, the cortex, which controls higher functions like abstract thought and speech. MIT researchers studied how structural and functional alterations of synapses--physical and chemical connections among neurons--in the cortex affect the animal’s ability to store long-term memory.

The MIT research provides the first evidence that synaptic structure and function in the cortex are critical for their long-term storage as memories are transferred from the hippocampus to the cortex.

PAK (p21-activated kinase), a critical regulator of synaptic architecture, was inhibited in the mutant mice. In humans, the gene that encodes this enzyme is tied to mental retardation. "Overall, we believe that one of the underlying mechanisms for mental retardation is synaptic malformation that leads to cognitive dysfunction," Hayashi said. "One of the most significant findings of this study is that the structure of synapses links to the function of synapses, and the size of synapses reflect the strength of the synapse. The bigger the synapse, the more vesicles to carry more neurotransmitters and the more channel proteins to permeate and bind to the neurotransmitters."

From Science Daily: University of California, Irvine researchers, using newly developing microscope techniques, have captured first-time images of the changes in brain cell connections following a common form of learning.

Detailed in the Journal of Neuroscience, the study shows that synaptic connections in a region of rats’ brains critical to learning change shape when the rodents learn to navigate a new, complex environment. In turn, when drugs are administered that block these changes, the rats don’t learn, confirming the essential role the shape change plays in the production of stable memory. “This is the first time anyone has seen the physical substrate, the ‘face,’ of newly encoded memory." said Gary Lynch, professor of psychiatry and human behavior at UC Irvine and leader of one of the two research teams involved in the studies.

The new behavioral experiments followed a series of recent discoveries by the UC Irvine group concerning the synaptic changes responsible for long-term potentiation (LTP), a physiological effect closely related to memory storage. Those studies used brain slices from rats, maintained alive in experimental chambers, to identify chemical markers for synapses that had recently experienced the LTP effect. The new study used live animals.

Working with advanced microscopic techniques called restorative deconvolution microscopy, the UC Irvine team found that the LTP-related markers appear during learning and are associated with expanded synapses in the hippocampus. Because the size of a synapse relates to its effectiveness in transmitting messages between neurons, the new results indicate that learning improves communication between particular groups of brain cells.

Because proteins continually biodegrade, ongoing maintenance operations need to be performed within the synapse or else it will not keep its shape. Prof. Yadin Dudai and his colleagues at the Neurobiology Department of Weizman Institute of Science have found that the process of storing long-term memories involves a miniature molecular machine that must run constantly to keep memories going. They also found that jamming the machine briefly can erase long-term memories. Their findings, which appeared today in the journal Science, may pave the way to future treatments for memory problems.

Dudai and research student Reut Shema, together with Todd Sacktor of the SUNY Downstate Medical Center, trained rats to avoid certain tastes. They then injected a drug to block a specific protein into the taste cortex – an area of the brain associated with taste memory. They hypothesized, on the basis of earlier research by Sacktor, that this protein, an enzyme called PKMzeta, acts as a miniature memory 'machine' that keeps memory up and running. An enzyme causes structural and functional changes in other proteins: PKMzeta, located in the synapses – the functional contact points between nerve cells – changes some facets of the structure of synaptic contacts. It must be persistently active, however, to maintain this change, which is brought about by learning. Silencing PKMzeta, reasoned the scientists, should reverse the change in the synapse. And this is exactly what happened: Regardless of the taste the rats were trained to avoid, they forget their learned aversion after a single application of the drug. The technique worked as successfully a month after the memories were formed (in terms of life span, more or less analogous to years in humans) and all signs so far indicate that the affected unpleasant memories of the taste had indeed disappeared. This is the first time that memories in the brain were shown to be capable of erasure so long after their formation.

'This drug is a molecular version of jamming the operation of the machine,' says Dudai. 'When the machine stops, the memories stop as well.' In other words, long-term memory is not a one-time inscription on the nerve network, but an ongoing process which the brain must continuously fuel and maintain. These findings raise the possibility of developing future, drug-based approaches for boosting and stabilizing memory.

Oster, Eichele and Leitges of The Max-Planck Institue provide more info on PKMzeta here. They've found that PKMzeta is found to be broadly expressed in most of the cortex, the limbic system, and the thalamus.

Another key protein is a-CaMKII. From ScienceDaily (May 21, 2001) A study by scientists from UCLA and Johns Hopkins University reveals the role of a protein that must be present in the cortex for information to be converted from short-term into lifelong memories. It indicates that the a-CaMKII protein triggers changes in cell-to-cell communication needed for establishing permanent memories in the cortex.

This was followed up by research from Yale - ScienceDaily (Mar. 16, 2005): A family of proteins that help build the cytoskeleton, or the bones of the cell, also play an important role in learning and memory, according to a study published in the Journal of Neuroscience 25: 2138–2145 (February–2005)

“If you learn to do something new, your neurons have to adapt and change to create a stronger, more direct pathway between neurons,” Picciotto said. “The protein ß–adducin appears to be important for making those new connections.”

In this study, the mice that did not have the protein were not able to strengthen a synapse in the hippocampus, which is the area of the brain that enables us to remember people, places and things. “If the mice don’t have ß–adducin, they can’t make a new map,” Picciotto said. “It’s not enough to just have the electrical properties, the (cyto)skeleton is very important in making long–lasting changes between nerve cells that result in learning.”

Nature has a review called "Actin in action: the interplay between the actin cytoskeleton and synaptic efficacy" by Cingolani and Goda that ties together many of these discoveries into a coherent picture: Synapse regulation exploits the capacity of actin to function as a stable structural component or as a dynamic filament. Beyond its well-appreciated role in eliciting visible morphological changes at the synapse, the emerging picture points to an active contribution of actin to the modulation of the efficacy of pre- and postsynaptic terminals. Moreover, by engaging distinct pools of actin and divergent signalling pathways, actin-dependent morphological plasticity could be uncoupled from modulation of synaptic strength.

From the corrected version of Box 3 of the article: Synaptic activity regulates actin dynamics and spine morphology through multiple signalling pathways. Key regulators of actin polymerization are the GTPases of the Rho family and serine/threonine kinases, which ultimately target actin-binding proteins (Table 1).

In the figure, actin-binding proteins are shown in red, Rho GTPases are shown in blue, kinases are shown in golden yellow, phosphatases are shown in green and GEFs are shown in dark orange.

In dendritic spines, opening of the Ca2+-permeable NMDA (N-methyl-D-aspartate) receptors by repetitive synaptic stimulation leads to the elevation of intracellular Ca2+ and the subsequent activation and translocation of Ca2+/calmodulin-dependent protein kinase II (CaMKII) to the postsynaptic density. Once activated, CaMKII can phosphorylate and activate multiple downstream signalling targets.

In addition, CaMKII, which is present in dendritic spines at high levels, has been proposed to also function as a structural molecule that can bind to and bundle F-actin through its subunit. Among the intracellular signalling events that are relevant for actin dynamics, of particular interest is the activation of the calmodulin-dependent kinase kinase (CaMKK)–CaMKI signalling pathway. These kinases form a multimolecular complex with the guanine-nucleotide exchange factor (GEF) PIX.

CaMKI-mediated phosphorylation of PIX increases its GEF activity, resulting in the activation of the small GTPase Rac1. PIX also interacts with p21-activated kinase 1 (Pak1), which can be activated by Rac1-triggered autophosphorylation.

In turn, Pak1 phosphorylates several downstream signalling molecules that directly modulate F-actin. For instance, Pak1-mediated phosphorylation activates LIM kinase (LIMK). The only known targets of LIMK are the members of the ADF/cofilin family (actin-depolymerizing proteins), which are inactivated upon LIMK-mediated phosphorylation.

Alternatively, Rac1 can be activated by a second GEF, kalirin 7, which interacts with and is phosphorylated by CaMKII. The activation of LIMK can also be promoted by the small GTPase RhoA, through RhoA-specific kinase (ROCK).

Interestingly, RhoA interacts with NMDA receptors, ROCK and the actin-polymerizing protein profilin II in an activity-dependent manner. Also,alpha-actinin, which is involved in crosslinking and bundling actin filaments, associates with NMDA receptors in an activity-dependent manner. When NMDA receptors open in response to stimuli delivered at low frequency to induce long-term depression, the effects on actin dynamics are opposite to those of CaMKII-initiated pathways, as ADF/cofilin family proteins are turned on following the activation of the phosphatase calcineurin.

More...
ScienceDaily (Nov. 10, 2000): info on NMDA receptors

Will you remember me? I will remember you...

If there was one experience that pushed my interest in neurobiology beyond the activation threshold and kick-started the process that led to the creation of this blog, it was reading Dr. R. Douglas Fields' article "Making Memories Stick" in the Feb. 2005 issue of Scientific American (ref.). I'd long been interested in molecular biology but had been intimidated by the level of jargon and assumed knowledge that filled most articles. Dr. Fields' article explained the inner workings of a neuron so clearly and lucidly that I was able to get a basic understanding of what was happening, and was motivated to try to learn more about molecular biology and neurobiology in particular. This blog is essentially the notes I've been making as I try to learn more about the details of how biology works at a molecular, cellular and neuronal level. So, things have finally come full circle. Let's take a deeper look at what makes memories stick...

From Making Memories Stick by R. Douglas Fields:
As far back as 1949, a psychologist named Donald Hebb ...proposed that, like an orchestra player who cannot keep up, a synapse on a neuron that fires out of sync with the other inputs to the neuron will stand out as odd and should be eliminated, but synapses that fire together - enough so as to make the nueron fire an action potential - should be strengthened. The brain would thus wire itself up in accordance with the flow of impulses through developing neural circuits, refining the original general outline.
...
By stimulating neurons to fire action potentials in different patterns and then measuring the amount of mRNA from genes known to be important in forming neural circuits or in adapting to the environment, we found ...we could turn on or off particular genes simply by dialing up the correct stimulus frequency on our electrophysiological stimulator, just as one tunes into a particular radio station by selecting the correct signal frequency.
...
If [the hippocampus] is dissected from a rat and kept alive in a salt solution, microelectrodes and electronic amplifiers can record the electrical impulses from individual synaptic connections on a neuron. By administering a burst of electrical shocks to a synapse, causing it to fire in a specific pattern, that synaptic connection can be strengthened. That is to say, the synapse produces about twice as much voltage in response to subsequent stimulations after it has received the high-frequency stimulus.
This increased strength, termed long-term potentiation (LTP), can be, despite its name, relatively short-lived. When test pulses are applied at a series of intervals after the high frequency stimulus, the voltage produced by the synapse slowly diminishes back to its original strength within a few hours. Known as early LTP, this temporary synaptic strengthening is a cellualr model of short-term memory.

Remarkably, if the same high-frequency stimulus is applied repeatedly (three times in our experiments), the synapse becomes strengthened permanently, a state called late LTP. But the stimuli cannot be repeated one after the other. Instead, each stimulus burst must be spaced by sufficient intervals of inactivity (10 minutes in our experiments). And adding chemicals that block mRNA or protein sysnthesis to the salt solution bathing the brain slice will cause the synapse to weaken to its original strength within two to three hours. Just as in whole organisms, the cellular model of short-term memory is not dependent on the nucleus, but the long-term formation of memory is.
...
Strong stimulation, either from the repeated firing of a single synapse or from the simultaneous firing of several synpases on a cell, depolarizes the cell membrane, causing the cell to fire action potentials of its own, which in turn causes voltage-sensitive calcium channels to open. The calcium ions interact with enzymes that activate the transcription factor CREB, which activates the genes for manufacturing synapse-strengthening proteins. The cell's nucleus "listens," in effect, to the cell's output - firing action potientials - to determine when to permanently strengthen a synapse and make a memory last. (see diagram at top of post.
More info on the role CREB plays is available here.)
...
One does not always know beforehand what events should be commmited permanently to memory. The moment-to-moment memories necessary for operating in the present are handled well by transient adjustments in the strength of individual synapses. But when an event is important enough or is repeated enough, synapses fire to make the neuron in turn fire neural impulses repeatedly and strongly, declaring "this is an event that should be recorded". The relevant genes turn on, and the synapses that are holding the short-term memory when the synapse strengthening proteins find them, become, in effect, tattooed.

(Some recent research that explains how this 'tattooing' works will be covered in the next post). From another excellent article by Dr. Fields, Erasing Memories (SciAm Mind Dec 2005):
There is a complication, however. The molecules that establish current flow around synapses are proteins, and all proteins in the body degrade and are replaced constantly over a period of hours or days. To strengthen a neural connection for a lifetime, some other process must take place to bolster the physical structure of a synapse or form additional synapses between the neurons involved.
The transition from temporary to permanent memory is called consolidation. Many experiments have deterined that consolidation requires many hours, and it can be enhanced or blocked in various ways. ... If the next 10 seconds of your life could be your last, you will remember them, even if the interval is folowed by another 10 dramatic seconds, and so on. ... The heightened state of attention, stress and novelty stimulates the consolidation phase of memory.
Neuroscientists have discovered how this consolidation happens. An epinephrine (a.k.a. adrenaline) rush releases a flood of stress hormones and neurotransmitters that activate the amygdala, the brain region that processes fear and emotion. The amygdala connects to many other regions where different kinds of memories are stored, and it boosts incoming data that have emotional impact. Consolidation, therefore, possibly can be aided by increasing levels of these neurotransmitters or hormones.
...
A growing body of evidence suggests that memory consolidation continues "off-line" while we sleep, in part because sleep involves periodic surges of some of the same hormones and neurotransmitters that are aroused in stressful and novel situations.


A research group headed by Jan Born at the University of Lübeck has found that sleep not only strengthens the content of a memory, but also the temporal structure of episodic memories, probably by replaying them in the forward direction (ScienceDaily (Apr. 18, 2007))

Returning to Erasing Memories:... [Work by Phillipe Peigneux] illustrates that memory consolidation requires sorting through fresh memories, integrating them with other memories, and shuttling them to different brain regions for permanent storage. Short-term memories deemed dispensible are discarded.
...
[James R. Misanin and others at Rutgers University] found that a consolidated memory could be erased if a lab rat was shocked right after being forced to recall the experience. ... [R]ecalling the memory had somehow made it vulnerable to disruption. This phenomenon has been termed reconsolidation.

In the article, Dr. Fields mentions that research by Joseph E. LeDoux and his colleagues at New York University has shown that microinjecting protein synthesis inhibitors into a rat's amygdala can block memory reconsolidation if applied shortly after the memory is recalled.
From Joseph LeDoux's book The Synaptic Self: The recent discovery , made by Karim Nader and Glenn Schafe in my lab, is that protein synthesis in the amygdala seems necessary for a recently activated memory to be kept as a memory. That is, if you take a memory out of storage you have to make new proteins (you have to restore, or reconsolidate it) in order for the memory to remain a memory. One way of thinking about this is that the brain that does the remembering is not the brain that formed the initial memory. In order for the old memory to make sense in the current brain, the memory has to be updated.

From Cellular and Systems Reconsolidation in the Hippocampus by Jacek Debiec, Joseph E. LeDoux and Karim Nader:
Indeed, reconsolidation and consolidation have been found to share a number of common properties, including: (1) requirement of protein synthesis in order for the memory to persist, (2) time windows during which protein synthesis blockade is effective, and (3) that protein synthesis blockage in the same brain region, the amygdala, disrupts both. Given these similarities, it seemed parsimonious to conclude that a new memory and a reactivated, consolidated memory share a common memory state, as originally proposed by Lewis (1979). Thus, instead of just occurring once, memory storage may instead be a process that is reiterated with each use of the memory.

... and from the abstract:
Cellular theories of memory consolidation posit that new memories require new protein synthesis in order to be stored. Systems consolidation theories posit that the hippocampus has a time-limited role in memory storage, after which the memory is independent of the hippocampus. Here, we show that intra-hippocampal infusions of the protein synthesis inhibitor anisomycin caused amnesia for a consolidated hippocampal-dependent contextual fear memory, but only if the memory was reactivated prior to infusion. The effect occurred even if reactivation was delayed for 45 days after training, a time when contextual memory is independent of the hippocampus. Indeed, reactivation of a hippocampus-independent memory caused the trace to again become hippocampus dependent, but only for 2 days rather than for weeks. Thus, hippocampal memories can undergo reconsolidation at both the cellular and systems levels.

Do you remember this quote, near the start of Dr. Fields' article 'Making Memories Stick': Donald Hebb ...proposed that, like an orchestra player who cannot keep up, a synapse on a neuron that fires out of sync with the other inputs to the neuron will stand out as odd and should be eliminated.?
The official term for the weakening of unhelpful synapses such as these is Long Term Depression (LTD). From the Howard Hughes Medical Institute (2004) LTP and LTD: strengthening and weakening synaptic connections

OUT OF SYNC, LOSE THE LINK
"The prevailing view used to be 'use it or lose it,'" says Bear. But a more accurate conclusion may be "neurons that fire out of sync lose their link.
...
When a neuron receives mismatched signals, synapses lose receptors ... If the loss of receptors is sufficiently prolonged, Bear suspects, the synapse eventually will disappear.
...
Bear's research into LTD has had other, and sometimes quite unexpected, consequences. His lab also has investigated a form of LTD triggered by the activation of another receptor on receiving neurons. One of the proteins affected by this receptor is the one that's missing in fragile X syndrome, the most common cause of inherited mental retardation in humans.

As always, the more I learn about things the more questions I have! Next post, we'll zoom in further to look at the changes the synapse undergoes when it is 'strengthened', and then start to zoom out to look at memory from a higher 'system-level' perspective.

More...
Podcast interview with R. Douglas Fields
Synapse signalling complexes and networks: machines underlying cognition" by Seth G.N. Grant

Actin Lessons - part 1. Cytoskeletal proteins are similar to G-proteins

I happened to stumble upon Martin Rodbell's 1994 Nobel Lecture paper: "Signal Transduction: Evolution of an Idea" again recently. I really enjoy reading these Lecture papers as they are a) written by the scientists that did the breakthrough research, b) contain a lot of insights into the creative process behind their discoveries and c) are intended for a general audience. Rodbell's paper is a good example. The following excerpt from his lecture paper bridges two areas I had no idea were related: G-protein receptors and the cytoskeleton. (The picture at right is from the web page for Andres Lebensohn of the Kirschner Lab at Harvard. It shows the assembly of an actin network.)

G-PROTEINS ARE SIMILAR IN STRUCTURE AND REGULATION TO
CYTOSKELETAL PROTEINS. by Martin Rodbell

During these studies, my attention was drawn to the striking similarities in the properties of G-proteins with those of tubulin and actin, the major cytoskeletal elements in cells. For example, G-proteins, like actin and tubulin, are associated with the inner aspect of the surface membrane, adhering possibly both through intrinsic membrane proteins, such as receptors, and to membrane lipids. Of particular interest is the fact that all three types of multimeric proteins are subject to regulation by either GTP (G-proteins and tubulin) or ATP (actin) and their hydrolytic products (dinucleotides and Pi). Receptors regulate exchange of bound nucleotides (GDP with GTP) and act catalytically in the process. Similarly, the excursion of a single myosin molecule during muscle contraction along the chain of actin multimers is governed by the exchange of bound ADP with ATP and the hydrolysis of ATP to ADP and Pi. As stated previously, GTP-turnover (production of GDP+Pi) is essential for the rapid and sustained actions of hormones; release of bound Pi is the crucial rate-limiting process in the overall dynamics of signaling. The same is true for myosin/ actin interactions.

With these similarities in structure and regulation, G-proteins can be classified as part of the cytoskeletal matrix, with the primary functional difference that G-proteins serve as chemical signaling devices whereas tubulin and actin serve as mechano-signaling devices. The release of monomers from multimers is the basis for chemical signaling by G-proteins. Dynamic changes in the disaggregation-aggregation cycle of actin and tubulin multimers are also regulatory devices designed for regulating the interactions or movement between specialized components of cells. Based on evidence accumulated over the past decade, all three types of cytoskeletal proteins are connected in some manner to a variety of signaling systems that adhere to the cytoskeletal matrix, including heterotrimeric G-proteins, so-called small molecular weight G-proteins, protein kinases and phosphatases, and other proteins or systems that communicate between the surface membrane and the interior of cells. These components form web-like structures that possibly interact in a flickering manner in response to activation of membrane receptors, including those that are growth promoting.

This is particularly fascinating given the recent breakthroughs in understanding the functional basis of memory formation that point to a physical change in the shape of the synapse (i.e. a cytoskeletal response to a neural action potential) as well as some of the molecular machinery that keeps the synapse in this configuration. My next blog will dig into what is currently known about how memories are formed in more depth.

Station to Station: Action Potentials in Neurons

Overview

From Sandra Ackerman's book Discovering the Brain: The actual signals transmitted throughout the brain come in two forms, electrical and chemical. The two forms are interdependent and meet at the synapse, where chemical substances can alter the electrical conditions within and outside the cell membrane.

A nerve cell at rest holds a slight negative charge (about –70 millivolts, or thousandths of a volt, mV) with respect to the exterior; the cell membrane is said to be polarized. The negative charge, the resting potential of the membrane, arises from a very slight excess of negatively charged molecules inside the cell.

A membrane at rest is more or less impermeable to positively charged sodium ions (Na⁺), but when stimulated it is transiently open to their passage. The Na⁺ ions thus flow in, attracted by the negative charge inside, and the membrane temporarily reverses its polarity, with a higher positive charge inside than out. This stage lasts less than a millisecond, and then the sodium channels close again. Potassium channels (K⁺) open, and K⁺ ions move out through the membrane, reversing the flow of positively charged ions. (Both these channels are known as voltage-gated, meaning that they open or close in response to changes in electrical charge occurring across the membrane.)

Over the next 3 milliseconds, the membrane becomes slightly hyperpolarized, with a charge of about –80 mV, and then returns to its resting potential. During this time the sodium channels remain closed; the membrane is in a refractory phase.

An action potential—the very brief pulse of positive membrane voltage—is transmitted forward along the axon; it is prevented from propagating backward as long as the sodium channels remain closed. After the membrane has returned to its resting potential, however, a new impulse may arrive to evoke an action potential, and the cycle can begin again.

Gated channels, and the concomitant movement of ions in and out of the cell membrane, are widespread throughout the nervous system, with sodium, potassium, and chlorine being the most common ions involved. Calcium channels are also important, particularly at the presynaptic boutons of axons. When the membrane is at its resting potential, positively charged calcium ions (Ca²⁺) outside the cell far outnumber those inside. With the advent of an action potential, however, calcium ions rush into the cell. The influx of calcium ions leads to the release of neurotransmitter into the synaptic cleft; this passes the signal to a neighboring nerve cell.

Having taken a close look at the electrical side of the picture, we are in a better position to see where the chemistry comes in. Molecules of neurotransmitter are released into a synaptic cleft and bind to specific receptor sites on the postsynaptic side (the dendrite or dendritic spine), thereby altering the ion channels in the postsynaptic membrane. Some neurotransmitters cause sodium channels to open, allowing the influx of Na⁺ ions and thus a lessening of negative charge inside the cell membrane. If a considerable number of these potentials are received within a short interval, they can depolarize the membrane enough to trigger an action potential; the result is the transmission of a nerve impulse. The substances that can cause this to occur are the excitatory neurotransmitters. By contrast, other chemical compounds cause potassium channels to open, increasing the outflow of K⁺ ions from the cell and making excitation less likely; the neurotransmitters that bring about this state are considered inhibitory.

A given neuron has a great quantity of sites available on its dendrites and cell body and receives signals from many synapses simultaneously, both excitatory and inhibitory. These signals often amount to a rough balance; it is only when the net potential of the membrane in one region shifts significantly up or down from the resting level that a particular neurotransmitter can be said to be exerting an effect. Interestingly, in the membrane's overall balance sheet, the importance of a particular synapse varies with its proximity to where the axon leaves the nerve cell body, so that numerous excitatory potentials out at the ends of the dendrites may be overruled by several inhibitory potentials closer to the soma. Other kinds of synapse regulate the release of neurotransmitters into the synaptic cleft, where they go on to affect the postsynaptic channels as described above.

So, let's dig into this in more detail:

In every living cell there is always a difference in electrostatic potential between the inside and outside of a cell: the cell is polarized. (ref 1,ref 2). This membrane potential is due to ion pumps such as Na-K-ATP-ase, an enzyme. Within cells, the concentration of Na+ ions is lower, and that of K+ ions higher, than in the surrounding fluid. Na+, K+-ATPase and other ion pumps must work all the time in our body. If they were to stop, our cells would swell up, and might even burst, and we would rapidly lose consciousness. A great deal of energy is needed to drive ion pumps - in humans, about 1/3 of the ATP that the body produces.

The Na+, K+-ATPase enzyme exposes binding sites to 3 Na+ ions when opened to the inside of the cell, and binding sites to 2 K+ ions when opened to the outside of the cell, and uses the energy from ATP to change its shape from one configuration to the other. This results in a higher concentration of +ve ions outside of the cell than inside, as 3 Na+ ions are pumped out of the cell for every 2 K+ ions that are pumped into the cell. The resulting relative electric potential (voltage difference) is the driving 'electro motif force' (EMF) that causes ions to flow when ion channels, triggered by neurotransmitters, open.

From Fred Wolf's page at the Max Planck Institute for Dynamics and Self-Organization: Every living cell maintains a voltage difference across its cell membrane. (how this is done). Nerve cells distinguish themselves from other cells in that they use this voltage difference to process and transmit messages. When a nerve cell receives an impulse, the voltage across the cell membrane is reversed. This "action potential" spreads out through the long appendages of the cell with high speed. At the end of the appendages it is transmitted to other cells. In 1952, Alan Lloyd Hodgkin and Andrew Fielding Huxley described in a mathematical model how such an action potential originates on the basis of measurements on neurons of the squid. The Hodgkin-Huxley model, for which the scientists received the 1963 Nobel Prize in Medicine, has since then served to explain the signal processes in all neurons.

According to the Hodgkin-Huxley model, an action potential is initiated when the voltage across the membrane of the nerve cell reaches to a certain threshold value. Voltage gated sodium channels react to this voltage change by opening up and triggering an avalanche-like reaction. Positively charged sodium ions flow through the open channels into the cell, which leads to a further increase of the membrane potential and the opening of additional sodium channels. The threshold and the speed with which the action potential originates vary from cell to cell - for any individual cell however, these parameters are specified for the most part by the characteristics of its sodium channels.

There is always a difference in electrostatic potential between the inside and outside of a cell: the cell is polarized. the result of the distribution of ions across the cell membrane and the permeability of the membrane to these ions. The ion channels themselves are positioned and held in place by the cytoskeleton of the neuron. In fact the Baz/Par-6/aPKC complex, a highly conserved protein cassette that functions during the establishment of polarity in a number of cell types, is also involved in the development of new synaptic boutons. (ref.)

From the University of Bristol: How do neurons work? : Under resting conditions, the potassium channel is more permeable to potassium ions than the sodium channel is to sodium ions. So there is a slow outward leak of potassium ions that is larger than the inward leak of sodium ions. This means that the membrane has a charge on the inside face that is negative relative to the outside, as more positively charged ions flow out of the neuron than flow in. The membrane is said to be polarised.

From reference.com: The voltage of an inactive cell remains close to a resting potential with excess negative charge inside the cell. When the membrane of an excitable cell becomes depolarized beyond a threshold, the cell undergoes an action potential (it "fires), often called a "spike" (see Threshold and initiation).

An action potential is a rapid change of the polarity of the voltage from negative to positive and then vice versa, the entire cycle lasting on the order of milliseconds. Each cycle — and therefore each action potential — has a rising phase, a falling phase, and finally an undershoot (see Phases).

From http://www.du.edu/~kinnamon/3640/actionpotential/ap1.html : Conformational changes in the voltage-dependent Na⁺ channel constitute the basis of the action potential. Time course of an action potential: The opening of a few Na⁺ channels leads to an initial depolarization. If this depolarization exceeds a threshold value then a rapid regenerative opening of many other Na⁺ channels follows, resulting in the depolarizing phase of an action potential. Since the Na⁺ channel spontaneously inactivates, the repolarization of the membrane occurs automatically. (Reichert, Introduction to Neurobiology)

From the commentary by Frank Werblin, UC Berkely to his animated and interactive tutorial's action potential animation at http://mcb.berkeley.edu/courses/mcb64/action_potential.html :

The action potential begins with a small depoloarization that comes from somewhere else. This causes ALL ion gates involved to begin their openning trajectory, but each at a different rate. First the sodium activation gate at the top of the sodium channel begins to open, allowing the sodium to enter the membrane and depolarize it further. Then the sodium inactivation gate at the bottom of the sodium channel closes, terminating the inward flow of sodium.

More slowly, the potassium activation gate begins to open. There's an outward flow of potassium ions, hyperpolarizing the membrane. This outward flow continues until the membrane overshoots the resting level in the hyperpolarizing direction. The potassium gate opens with depolarization, but the current hyperpolarization causes the potassium gate to close, terminating the action potential. http://www.du.edu/~kinnamon/3640/actionpotential/ap1.html

Bucket Brigade

(The diagrams at right are from B. Fleming, U. of Waterloo )

From the University of Bristol: How do neurons work? :

Excess ions are subsequently pumped in/out of the neuron.

In some neurons, after hyperpolarization a secondary, longer-term effect occurs: a post-excitatory restoration of membrane potential, termed after-hyperpolarization (AHP).

If this were all there was to it, then the action potential would propagate in all directions along an axon. But action potentials move in one direction. This is achieved because the sodium channels have a refactory period following activation, during which they cannot open again. This ensures that the action potential is proagated in a specific direction along the axon.

For a neuron to integrate all of the various pulses that it is recieving from various other neurons, the pulses need to 'line up' to a certain degree in order to be 'coincident' and push the neuron over the activation threshold. For this to happen, the pulse width would need to be greater than the propagation delay introduced by the axon. The pulse width is on the order of 2ms. The propagation along an unmyelinated axon depends on a number of factors (e.g. the cross-sectional width of the axon, the type of axon, etc.) but is typically around 50 m/s. (ref.) which would imply that the axon needs to be less than 10 cm in length.

Axons that are used a lot become sheathed in 1 mm long sections of myelin (oligodendrocytes) that promote more efficient and faster conduction. The action potential in myelinated axons jumps from one gap in the myelin (called a 'node of Ranvier') to the next in a process called 'saltatory conduction'. Myelination causes the resistance of the membrane to increase by a factor of 200 and the capacitance per unit area of the axon to decrease by a factor of 200 (ref.)

Recent research by B. Stevens, S. Porta, L. Haak, V. Gallo and R. Douglas Fields has shown that axons release Adenosine in response to action potential firing and that this is used as a chemical signal by oligodendrocyte progenitor cells to initiate sheathing axons that fire frequently.

More information:

MASS ACTION IN THE NERVOUS SYSTEM EXAMINATION OF THE NEUROPHYSIOLOGICAL BASIS OF ADAPTIVE BEHAVIOR THROUGH THE EEG (2004) by Professor Walter J. Freeman University of California, Berkeley, USA (http://sulcus.berkeley.edu/MANSWWW/MANSHTML/MANSChapt3.html) : provides a lot of the equations and details about how neurons operate.

The sea inside your skull - ion homeostasis

Previous posts have covered a number of the low-level building blocks that are used by cells in the brain - things like ion channels, neurotransmitters, receptors, clathrin, vesicles, etc. This post focuses on some important pieces of infrastructure that are needed to enable the brain to do its thing. 

Neurons operate in an aqueous medium - a kind of salt water bath, water that is full of postively charged ions (cations like sodium, potassium and calcium) and negatively charged ions (anions like chlorine).   Water molecules are V shaped and have a non-uniform distribution of charge - i.e. one end of the water molecule is more positively charged than the other end.  Like charges repel and unlike charges attract.  As a result, a sphere of these 'polar' water molecules tends to surround the ions (a 'sphere of hydration').   Complicating the picture further is the fact that charged particles like ions are influenced by both concentration gradients (ions flow to the region with the lower concentration of that ion) and electrical gradients (electro-magnetic fields, charge distribution)

The resting nerve has mostly Na+ on the outside of the membrane and K+ and negatively charged, non-diffusible protein on the inside. Because there is a little leakage of K+ to the outside (K+ is more permeable than Na+), the net membrane charge is positive on the outside and negative on the inside.  (Ref.: PLASMA MEMBRANE AND MEMBRANE POTENTIAL IN THE NERVE CELL - U. of Colorado)

At rest, in the absence of action potentials or any other activity, the inside of a neuron has a voltage of about -65mV compared to the outside. Neural signals are changes in this resting potential. (Ref.: Neuronal Membrane at Rest: Ionic Basis of the Membrane Resting Potential)

The resting potential results from the establishment of ionic gradients across the cell membrane for potassium, sodium, chloride, and calcium. Most membranes are permeable principally to potassium and sodium. The diffusion pressure for each ion is balanced by an electrical force from the voltage that develops across the membrane. The voltage that exactly balances each ion is the equilibrium potential for that ion. The resting potential is a weighted average of the equilibrium potentials for the ions to which the membrane is permeable. The net driving force on a given ion is the difference between the membrane potential and its equilibrium potential. (ref.) 

The above references get into the key equations that govern ionic behaviour: the Nernst Equation calculates the numerical value of the equilibrium potential, and the Goldman Equation calculates the membrane potential as a weighted average of the equilibrium potentials.  Not going into these here :)  

From the above, you can see that membranes are REALLY important. You need a membrane to separate two different environments with different characteristics in order to do any 'work'. Stuart Kauffman has some very interesting perspectives on this -  I've added a new NanoBiologyNotes blog entry on Membranes that touches on these ideas.  

Most neurons are interconnected indirectly with each other.   An electrical 'action potential' flows down a neuron's axon until it reaches a terminating 'bouton'  - a bulb containing mitochondria and tiny bubbles called vesicles that are filled with one or more kinds of neurotransmitters.  The action potential causes the vesicles to fuse with the axon's membrane and releaseits payload of neurotransmitters into the synaptic gap that seperates it from another neuron's dendrites or soma. (We'll call this second neuron the 'receiving neuron'). These neurotransmitters bond with ion channels in the membrane of the receiving neuron, causing a conformational change in the shape of each of the ion channels which allows ions that are present in the fluid surrounding the neurons to flow into the receiving neuron, or ions that are inside of the neuron to flow out, depending on the type of ion channel that has been opened. A single neuron can have tens of thousands of these types of indirect connections with other neurons. If enough sending neurons trigger at rougly the same time, a large enough number of ion channels in the receiving neuron will be open that the ionic concentration of the receiving neuron will hit a threshold that initiates one or more action potentials to rush through the receiving neuron to it's axon. Roughly speaking.

But, uh, where do all of these ions come from?  What establishes and maintains the correct concentration of ions outside of the neurons??

This simple question led me into some fascinating stuff on morphobiology, ph and H+ gradients, genetic toolkits and so much interesting material I'm starting up another blog to focus on it.  But, back to the original question...what's up with the ions?

Zhi-Qi Xiong and Janet L. Stringfer (Baylor College of Medicine) published a nice introduction to what is going on to establish the right environment in the brain for neurons to work in the Journal of Neurophsiology (ref.) (I've removed the references from the text to improve readability - please see the original for these):

Neuronal activity is associated with a rise in the extracellular potassium concentration ([K⁺]_o) caused by efflux of potassium during action potential repolarization. Neuronal activity, in the absence of clearance mechanisms, would cause the [K⁺]_o to rise in seconds to values that would abolish all electrical activity. However, it is known that during intense evoked neuronal activity or spontaneous epileptiform activity in the cortex  and hippocampus  [K⁺]_o rises to a ceiling level of 10-12 mM from a resting level of 3 mM. The occurrence of a plateau, or ceiling, level during continued neuronal activity suggests that [K⁺]_o is actively cleared from the extracellular space.

Clearing of excess [K⁺]_o is believed to occur by diffusion, active uptake by neurons and glia, or passive uptake by glia. There is evidence that the rate of potassium release during repetitive neuronal activity is faster than the rate at which it would diffuse away. Although neurons can take up potassium, uptake by astrocytes is believed to play a major role in regulation of [K⁺]_o. Glia are thought to be required for the normal fine tuning of [K⁺]_o and for the recovery of pathologically elevated [K⁺]_o. Glia have been shown to increase their internal potassium concentration when [K⁺]_o is increased and release it once the [K⁺]_o decreases. They are also believed to remove potassium by spatial buffering through the glial syncytium. According to the spatial buffering hypothesis, potassium released from active neurons enters glial cells, possibly through inwardly rectifying potassium channels. Potassium is then redistributed through the network of glial cells and leaves at sites of low [K⁺]_o. Spatial buffering can be directly demonstrated in the drone retina as a result of a fortunate spatial arrangement of neuronal and glial structures. However, the exact role of glial spatial buffering in other parts of the brain and during times when the regulation systems are significantly stressed (i.e., during synchronous epileptiform activity) is not clear. It has been argued that spatial buffering has no role in situations of elevated [K⁺]_o.

There's a nicely written article (Astrocytes) by Pierre J. Magistretti and Bruce R. Ransom that provides a good overview of the role played by glial cells:

• The astrocyte is a ubiquitous type of glial cell that is defined in part by what it lacks: axons, action potentials, and synaptic potentials. Astrocytes greatly outnumber neurons, often 10:1 and occupy 25% to 50% of brain volume (1–3). Although these cells are anatomically obvious, their functions have been difficult to determine. Discoveries in the last 25 years, however, have revealed some of their functions and established the essential nature of interactions between neurons and astrocytes for normal brain function.
  
• One of the best-established functions of astrocytes is regulation of brain [K]o. Astrocytes are also likely to participate in the regulation of extracellular pH. 
  
• Neural activity can rapidly increase [K]o, which is tightly regulated to a resting level of about 3 mM (25). A single action potential increases the instantaneous [K]o by0.75mM(26). The increase in [K]o is proportional to the intensity of neural activity but has a so-called ‘‘ceiling’’ level of accumulation of 10 to 12 mM (27,28), which is only exceeded under pathologic conditions (29). If diffusion alone were responsible for dissipating K released from neurons, it is easily calculated that extracellular K accumulation would exceed 10 mM during normal neural activity, whereas measured increases in [K]o are in the range of 1 to 3mMindicating powerful control mechanisms (30). Homeostatic control of [K]o is needed because brain [K]o can influence transmitter release (31), cerebral blood flow (32), ECS volume (33,34), glucose metabolism (35), and neuronal activity (36). Unchecked increases in [K]o act as an unstable positive feedback loop increasing excitability.
  
• Astrocytes expedite the removal of evoked increases in [K]o and limit its accumulation to a maximum level of 10 to 12mM, the ceiling level seen with intense activity such as epileptic discharge (37,38). Neurons, and perhaps blood vessels, also participate in [K]o regulation, but glial mechanisms are probably most important. Two general mechanisms of astrocyte K removal have been proposed
  (39): 1) net K uptake into astrocytes (by transport mechanisms and/or Donnan forces) and 2) K redistribution through astrocytes, which is known as K spatial buffering. The relative importance of these two mechanisms of [K]o regulation remains an open question and may depend on the nature of the [K]o increase as well as brain region (38). If glial cells take up K during neural activity and release it thereafter, a transient increase in glial [K]i should result. Astrocyte [K]i does transiently increase during neural activity and has a similar time course to the K lost from active neurons and the increase in [K]o, indicating that the K released from neurons is passing by way of the ECS into glial cells (40–42). Uptake of K into glial cells depends on the glial Na pump (38,42–44), an anion transporter that cotransports K and Na with Cl (43) and Donnan forces that propel KCl into glial cells in the face of elevated [K]o (42) (Fig 10.1). It has not been determined with certainty which of these mechanisms is quantitatively most important for K uptake. The astrocyte Na pump, however, is exquisitely sensitive to elevations of [K]o. Even a 1 mM increase in [K]o activates the Na pump in these cells indicating, perhaps, that this is the major mechanism of K sequestration (44). Neurons, of course, must eventually reaccumulate K lost during activity using their Na pump, but only glial cells show net accumulation
  of K (Fig. 10.1). It is interesting to note that the neuronal Na pump is not sensitive to small increases in [K]o and is probably activated mainly by increases in intracellular [Na] (45).

Another key mechanism that plays an important role in regulating the ionic concentrations is something called the "Na⁺-K⁺-Cl^- cotransporter" (ref.)

• Little is known regarding how Na⁺-K⁺-Cl^- cotransporter activity is regulated in the CNS. Glutamate, N-methyl-d-aspartate, and the metabotropic glutamate receptor agonist t-ACPD significantly stimulate cotransporter activity in neurons (Sun and Murali, 1998, 1999). Cotransporter activity in cortical neurons and astrocytes is elevated when intracellular Ca⁺⁺ increases in the presence of high [K⁺]_o(Schomberg et al., 2001; Su et al., 2000). Because both high [K⁺]_o and elevated extracellular glutamate play important roles in ischemic cell damage, the authors hypothesize that stimulation of the cotransporter in neurons may contribute to overload of intracellular Na⁺ and Cl^- and cell swelling during ischemia. Several studies suggest that the Na⁺-K⁺-Cl^- cotransporter may be involved in ischemic cerebral cell damage. Twenty-four hours of hypoxia decreases cellular adenosine triphosphate (ATP) content and reduces Na⁺-K⁺-ATPase activity, while significantly increasing the Na⁺-K⁺-Cl^- cotransporter activity in rat brain capillary endothelial cells (Kawai et al., 1996). Significant reduction of brain edema by the Na⁺-K⁺-Cl^- cotransporter and Cl^- channel inhibitor torasemide or its derivative also has been observed in focal cerebral ischemia and traumatic brain injury (Staub et al., 1994; Le Bars et al., 1996). However, no study has yet directly demonstrated a role of the cotransporter in ischemic neuronal damage.

The ion pump Na⁺/K⁺-ATPase is a third key mechanism:

• ref. These findings suggest that potassium redistribution by glia only plays a minor role in the regulation of [K⁺]_o in this model. The major regulator of [K⁺]_o in this model appears to be uptake via a Na⁺/K⁺-ATPase, most likely neuronal.

 

And the Blood-brain barrier is also important, since it limits the movement of potassium through the walls of brain capillaries (more info here.)