Long Term Potentiation (LTP)

Introduction:One area that seems to separate humans from other animals is intelligence. Part of that

intelligence may be attributed to the size of the human brain, since an increase in brain size seems to be a key anatomical difference separating humans from apes and also modern homo sapiens from earlier hominids, as judged by fossils. Yet any knowledgeable discussion of intelligence, perception of emotions, and ability to reason will require further advancement of the knowledge of brain function.

Early studies of neuron mechanisms focused on the neurons involved in transmitting sensory information of our five senses back to the brain to be processed. This led to a good understanding of the neurotransmitter, acetylcholine. How it is released by an excited cell, diffuses across the synapse to a neighboring cell and binds to the acetylcholine receptor on the adjacent neuron. This alters the conformation of the receptor, also a sodium channel, to allow the free flow of sodium ions into the cell, which in its normal state has a higher (-) ion concentration inside than out. This depolarization, via the release of Ca2+ ions from vesicles, results in a wave of depolarization that travels like an electric current through the axon to the other end of the neuron. Almost instantaneously causing the release of acetylcholine from this cell etc. Cholinergic receptors explain the basic mechanism for the transmission of information from the senses to the brain and any return messages. These often result in behaviors requiring muscle activity as a response. If we smell something good to eat, we move closer to investigate. If we hear frightful sounds we might move to put distance between us and the source. In this type of neuron function we are not that much different from other animals. In fact chemical insecticides used to protect crops are frequently acetylcholinesterase inhibitors that interfere with the nerve-brain communication pathway and result in paralysis.

Cognition, at least what we understand of it, appears to involve the same basic mechanism of electrical and chemical transmission between neurons. However, the brain contains a vast array of specialized neurons, containing a variety of neurotransmitters, and their corresponding receptors, that help us to sort out vast arrays of sensory input, emotional responses, memories of prior events and associations, to name just a few of the factors involved in brain function. Although humans seem to have an edge in the ability to reason, the biochemical mechanisms for all of the previously mentioned brain functions are not vastly different in other mammals. Since it is unethical to do research on human subjects, a lot of what we know comes from the studies of these mechanisms in animals. In fact the similarities that we do observe in brain function lead many to question whether it is ethical to perform such studies on animals. Nonetheless, it is currently accepted practice to use laboratory animals to study brain function. You’ve all heard the story of Pavlov’s dogs, which learned to associate a bell with dinner, such that the bell alone could cause them to salivate in anticipation of their meal, even if it didn’t come.

Some years ago, an MIT researcher performed an experiment that caught the imagination of the press, in part because of the associations made by the researchers choice of terminology for the mice used in the study. (Tsien - Nature, Sept. 2, 1999 & Science, Nov 10, 2000) They were called Doogie mice after the name of the title character in a TV show about a precocious teenager who attained his M.D. degree at the age where most students would just be finishing middle school. The show’s name was Doogie Howser, M.D. (I can’t recall seeing any more than 10 minutes of one or two shows, but nonetheless, I remembered the name).

The research showed that over-expression (by virtue of additional copies) of a gene called NR2B, which encodes the NMDA (N-methyl-D-Aspartate) Receptor, in mice allowed them to learn faster than average mice. This of course requires explaining what we mean by learning. Although memorization typically gets a bad name in pedagogical circles, the concept of intelligence and learning without memory is like the concept of playing basketball without a ball. All intelligence requires memory. Memory, is basically the association of events, sensory input,

emotions, thoughts or ideas, or combinations of these. Learning is the process by which these associations are established, and can sometimes be maintained over years. (Just for fun, try to think back to the very first thing in your life you can remember and recall as much as you can about the event. It might be interesting to check with your parents or older siblings to see how accurate your memory of the event matches theirs, assuming its something they may also remember). I still recall high school cross country training and races at Schenley Park in Pittsburgh, every time I smell the odor of fallen leaves in the fall. Without that input, I rarely think of those experiences. Thus I always associate the smell of rotting leaves with running (in a very pleasant way).

People often discuss short-term vs. long-term memory in talking about tests and examinations. If you take a single class and do nothing with the subject matter for a long time after, you tend to lose most of your knowledge from the class. However, if you learn something and then immediately go out and make repetitive use of the information, you tend to hang unto that knowledge for a longer period of time. The theory of how we retain associations for long periods of time is called long term potentiation (LTP). The Doogie mouse is only one of many studies focusing on components of this process.

In simplistic terms LTP theory credits brain neurons for housing memories. These neurons make multiple connections with other brain neurons involved with receiving sensory input, like vision from the optic nerve, for example. The more input cells connected to the same memory cell are stimulated the stronger the connections between these cells become, and the longer lasting the associations between those inputs are, to the extent that at a much later time, one of the inputs alone will allow us to recreate the emotions from the other input(s). However, if the senses connected with a memory cell are not stimulated within a short time range, then it is likely that the dendritic connections between these cells will be deconstructed and reformed to other cells in a somewhat random fashion. In essence it is not unlike the immune system where we make thousands T cells carrying one unique, randomly produced antibody anchored to its membrane. If that T-cell never encounters an antigen that it binds to, it will eventually die like any other cell. Other immune cells with a different, random antibody molecule will take its place. Likewise if multiple sensory inputs are wired to the same ‘memory’ neuron and those sensory inputs are not used, the connections will eventually be broken apart and new random connections will be made. However, if an immune cell binds to a foreign antigen it sets in motion a process to make many more immune cells that will make and distribute soluble forms of the antibody it contained. Likewise if two different sensory inputs in the brain happened to be wired to the same ‘memory’ neuron when both of those sensory inputs get stimulated the connections are strengthened rather than diminished. The more frequent the sensory input (still temporally associated) or the more intense the simultaneous sensory input, the longer lasting will the association become. This is why you should study outside of class!

The extent to which the initial connections are ‘hard-wired’ by genetic information vs. how much is ‘learned’ by the random connections fortified by experiences as we grow is hotly debated. The old philosophical assumption that humans are born with ‘blank-slate’ brains is surely incorrect. However, the notion that our knowledge level is 100% genetically predetermined is equally absurd.

What makes this theory experimentally difficult to test, even in animals, is that we are not just talking about connections of one cell with a dozen or so of its nearest neighbors. A single neuron in the brain can receive connections from 10,000 – 30,000 other cells via its dendrites. The axons making these connections may be as long a meter or as short as 1mm. There are ways of checking electrical and metabolic activity in different parts of the brain following controlled sensory input, but as far as I know now, you cannot verify cell to cell connections except by autopsy. Finding volunteers for such research is problematic!

However, there are current methods utilizing non-invasive procedures like MRI that can assess the metabolic activity of neurons by gauging their metabolism level. Thus when certain environmental stimuli are presented to a subject, the areas of the brain that are most active can be

determined. The biochemistry of brain function will be a very active and productive area of research in the upcoming decades.. The Doogie Mouse Experiment

The ‘smart’ mice in Tsien’s experiment had extra copies of the gene encoding the NMDA receptor added. But, how do you test the intelligence of a mouse? Tsien performed the following three types of experiments for this purpose.

1) Place a mouse in a cage with three objects. The time the mouse spends ‘interacting’ with each object is charted. Now change one of the objects. If the mouse has a good memory it will be already familiar with the usual objects and will thus spend more time ‘interacting’ with the newer object. The Doogie mice did indeed spend more time examining the newer object relative to control mice.

2) Place a mouse in a certain environment. This environment may be supplemented with a particular sensory input, like a tone. Now shock the mouse while in this environment. After a certain number of days return the mouse to the same environment. The mouse that remembers the shock will display higher levels of anxiety. The Doogie mice began to display anxiety signs after fewer shock exposures that the control mice. However, if the shocks were discontinued they also lost their anxiety more quickly than control mice that had developed the same level of anxiety due to longer exposure.

3) Throw a mouse into a pool with a submerged platform. The mice swim randomly until they find the platform. The number of tries before the mouse immediately swims to the platform is noted. The Doogie mice averaged 3 tries while the control mice averaged 6.

The added NMDA receptors of the Doogie mice were designed to be deactivated by the drug doxycycline, which could be added to their drinking water. Deactivating the added receptors after the learning took place did not remove the gained ability. This supports the aspect of learning theory that involves the increase in the number of dendritic connections between the associated memory cells.

I will insert a few notes of caution here. The NMDA receptor gene is not the intelligence gene. Mice with enhanced genetic predisposition for nerve growth factors were also better in developing LTP. The list could probably extend quite a bit. It should also be noted that making associations too easily or remembering too well might be detrimental to survival. If fear motivates action it can enhance survival. If fear causes a type of paralysis, because we can’t deal with the intensity of perceived memory, it might be detrimental to survival. Scientists should display extreme caution before judging that increased expression of ‘intelligence’ genes is a good thing.

Long Term Potentiation – The basic mechanismsCertain molecular aspects of neuron functions and memory are beginning to clear up a bit.

For a review of this topic see Malenka & Nicoll – Science (1999) September 17, volume 285 1870-1874

The hippocampus is the brain’s center for forming new memories. The parts of the neurons that are stimulated by the release of neurotransmitters from neurons that receive sensory information in the brain are called dendrites. They are referred to as post-synaptic. The cell releasing the neurotransmitter does so through the ends of its axons. These are referred to a presynaptic. The presynaptic axons make and release neurotransmitters (when stimulated by sensory input). The post-synaptic dendrites contain the receptors that the neurotransmitters bind to. Neurons are connected by short synapses between the axons and dendrites. This space is typically on the order of 75 å, on the order of the thickness of a lipid bilayer. The dendrites of the memory neurons contain two kinds of receptors involved in memory. The AMPA receptors respond to the neurotransmitter, glutamate, or the agonist, -amino-3-hydroxy-5-methyl-4-isoxazoleproprionic acid (AMPA). They contain a channel permeable to Na+ and K+ that mostly create inward flow of Na+ ions upon activation. This depolarization is the reason for STP (short

term potentiation). If you only stimulate these receptors the associations made between the stimulating sensory neurons will be short lived, decaying to baseline levels in about 5-20 minutes.

The second type of receptor is the NMDA (N-methyl-D-aspartate) receptor. These receptors contain a Ca++ channel, which is typically blocked by Mg2+ under normal conditions and contributes very little to the basal postsynaptic response. However, if the AMPA receptors produces a significant enough depolarization, the Mg2+ is removed and Ca2+ influx into the cell joins the Na+ influx, triggering LTP. Supporting this model is the fact that NMDA receptor antagonists have little effect on STP but block LTP. Further support of the Calcium role comes from the fact that LTP can also be blocked by the addition of Ca++ chelators.

Sub-threshold concentrations of Ca++ can either cause STP or LTD (long term depression) in which the effects of LTP are reversed.

The signal transduction cascade set in motion by the calcium influx in LTP is not well understood. What does seem clear however, is that one major component of this cascade is -calcium calmodulin-dependent Protein Kinase II. (CaMKII). One of the key explanations of the longer term effects of LTP, is that once CaMKII is phosphorylated on Thr-286 it no longer requires Ca++ for activation. Thus the signal transduction cascade remains active long after the [Ca++] has returned to baseline. In addition the AMPA receptors appear to be both upregulated and phosphorylated, so that a future stimulatory event elicits a stronger basal response from them. This will enhance the response of these memory cells to normal level stimulation in the future. In other words your response to a particular sensory input might be much greater than someone else who has not had the same previous experiences that you have. We will not discuss other components implicated in this cascade.

But how does this account for memory of events many years into the future? The working model involves the building of additional dendritic spines between memory and sensory cells that have undergone LTP (Muller, Tony, and Buchs – Hippocampus (2000) 10(5), 596-604). In essence these cells become hard wired together like cable attachments between a network of computers. It is known that the spines containing the dendrites that make the post-synaptic connections between cells can be rapidly built, causing an increase in the postsynaptic density (PSD). With disuse, they can be taken apart. The stronger the LTP, the more highly connected the associated cells.

Summary - Chronology of LTP:1. Stimulus triggers activation of AMPA receptors. Na+ & K+ enter the cell reducing the cell’s

polarization (E). This produces the basal synaptic transmission.2. If the stimulation has sufficient duration/intensity the depolarization results in activation of

the voltage gated NMDA receptors. Mg++ is released from its position blocking the NMDA receptor channel. Ca++ enter the cell ([Ca++] ). Evidence: NMDA receptor antagonists block the generation of LTP but have no effect on basal synaptic transmission.

3. Associativity occurs in part because the strong activation of one set of synapses depolarizes adjacent regions of the dendritic tree.

4. CaMKII (-calcium-calmodulin-dependent protein kinase II) phosphorylates and causes clustering of the AMPA receptors. AMPA receptor response in the post-synaptic cell is also enhanced, due to up-regulation of the receptor. Synapses that had expressed only NMDA receptors now contain AMPA receptors as well.

5. CaMKII is also auto-phosphorylated at Thr-286. Because of the auto-phosphorylation, Ca++ elevation is no longer required for its activation which is what makes the cells activation long-term. LTD (long-term depression) involves the dephosphorylation of the AMPA receptors and CaMKII.

6. A retrograde messenger (unknown but possibly NO, CO, arachidonic acid, or platelet activating factor) released from the post-synaptic cell modifies the pre-synaptic function of the signaling cell. The cell is now responding more readily to neurotransmitters, receiving

more neurotransmitter, and maintaining its responsive state for a longer period of time. In addition a single stimulus is linked to a different stimulus that need not be present.

7. A signal transduction cascade is stimulated such that the neuron actively constructs new dendrites and connects them to the cells that carried the original stimuli signals. These stimuli now become ‘linked’ such that a only one of the stimuli present in the environment will elicit a memory of other associated stimuli.