In neuroscience, long-term potentiation (LTP) is the strengthening (or potentiation) of the connection between two nerve cells which lasts for an extended period of time (minutes to hours in vitro and hours to days and months in vivo). LTP can be induced experimentally by applying a sequence of short, high-frequency stimulations to nerve cell synapses. The phenomenon was discovered in the mammalian hippocampus by Terje LÝmo in 1966 and is commonly regarded as the cellular basis of memory.
Early theories of learning
Santiago Ramůn y Cajal proposed that memories might be stored in the connections between nerve cells.
By the turn of the 19th century, neurobiologists had good reason to believe that memories were generally not the product of new nerve cell growth. Scientists generally believed that the number of neurons in the adult brain (roughly 1011) did not increase significantly with age. With this realization came the need to explain how memories were created in the absence of new cell growth.
Among the first neuroscientists to suggest that learning was not the product of new cell growth was the Spanish anatomist Santiago Ramůn y Cajal. In 1894 he proposed that memories might be formed by strengthening the connections between existing neurons to improve the effectiveness of their communication. Hebbian theory, introduced by Donald Hebb in 1949, echoed Ramůn y Cajal's ideas, and further proposed that cells may grow new connections between each other to enhance their ability to communicate:
- Let us assume that the persistence or repetition of a reverberatory activity (or "trace") tends to induce lasting cellular changes that add to its stability.... When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased. (Hebb, The organization of behavior)
Similarly, memories may be forgotten through the weakening or loss of connections. For example, a man might be startled by the sound of a car alarm outside. Sensory cells in the ear record the sound and send it to the brain where it activates neurons that control the man's muscles. But as the blaring alarm continues, those connections are weakened so that the alarm no longer causes the man to be startled.
These theories about memory formation were unfortunately foresighted. Neuroscientists were simply not yet equipped with the neurophysiological techniques necessary for elucidating the biological underpinnings of learning in animals. These skills would not come until the latter half of the 20th century, at about the same time as the discovery of long-term potentiation.
Discovery of long-term potentiation
LTP was first discovered in the rabbit hippocampus. In humans the hippocampus is located in the medial temporal lobe
LTP was first observed by Terje LÝmo in 1966 in the Oslo, Norway, laboratory of Per Andersen (12740104). There, LÝmo conducted a series of neurophysiological experiments exploring the role of the hippocampus in the rabbit short-term memory. Targeting the synapses between granule cells of the perforant pathway and those of the dentate gyrus, LÝmo elicited excitatory postsynaptic potentials (EPSPs) from dentate gyrus cells by stimulating the perforant pathway. He observed that a high-frequency train of stimulation produced larger, prolonged EPSPs compared to the responses evoked by a single stimulation. This phenomenon was soon dubbed "long-term potentiation".
Timothy Bliss, who joined the Andersen laboratory in 1968, collaborated with LÝmo in 1973 to publish the first characterization of LTP in rabbit hippocampus.
Types of LTP
Since its original discovery in the rabbit hippocampus, LTP has been observed in a variety of other neural structures, including the cerebral cortex, cerebellum, amygdala, and many others. The underlying mechanisms of LTP are generally conserved across these different regions, but there are subtle differences in LTP's precise molecular machinery between sites. Very broadly, there are two types of LTP, associative and nonassociative.
Associative LTP is the molecular analog of associative learning (e.g. classical conditioning). It is the strengthening of the connection between two neurons that have been simultaneously active. To detect the simultaneous activity of the pre- and postsynaptic cells, associative LTP requires a so-called coincidence detector. In many parts of the brain where associative LTP is observed, the NMDA receptor (NMDAR) fills the role of coincidence detector. At rest, the NMDAR's calcium channel is blocked by magnesium; the blockade is relieved only after strong postsynaptic depolarization (6325946). The calcium channel is also ligand-gated, so that it only opens when presynaptically-released glutamate binds the receptor. When the NMDAR opens, calcium floods the postsynaptic cell triggering associative LTP.
NMDAR-dependent LTP has been demonstrated in the hippocampus, particularly in the Schaffer collaterals and perforant pathway, and several other brain regions including parts of the amygdala (9403688) and cerebral cortex (2446147).
There are several types of associative LTP that do not depend on NMDA receptors. NMDAR-independent LTP has been observed, for example, in the amygdala, where it depends instead on voltage-gated calcium channels (10575047).
Nonassociative LTP is brought about by the repeated application of one stimulus (whereas in associative LTP there are at least two stimuli). At nonassociative synapses, such as those involved in habituation and sensitization, persistent stimulation of the synapse triggers an influx of calcium into the postsynaptic cell. As in associative LTP, calcium signals the beginning of long-term potentiation, but the precise mechanisms of nonassociative LTP are still unknown.
Properties of LTP
NMDA receptor-dependent LTP classically exhibits four main properties: rapid induction, cooperativity, associativity, and input specificity:
- LTP can be rapidly induced by applying one or more brief tetanic stimuli to a presynaptic cell. (A tetanic stimulus is a high-frequency sequence of individual stimulati.)
- LTP can be induced either by strong tetanic stimulation of a single pathway, or cooperatively via the weaker stimulation of many. It is explained by the presence of a stimulus threshold that must be reached in order to induce LTP.
- Associativity refers to the observation that when weak stimulation of a single pathway is insufficient for the induction of LTP, simultaneous strong stimulation of another pathway will induce LTP at both pathways. There is some evidence that associativity and cooperativity share the same underlying cellular mechanism (see Synaptic tagging).
- Once induced, LTP at one synapse is not propagated to adjacent synapses; rather LTP is input specific.
Phases of LTP
LTP is often divided into two phases, an early, protein synthesis-independent phase (E-LTP) that lasts between one and five hours, and a late, protein synthesis-dependent phase (L-LTP) that lasts from days to months (10575022). Broadly, E-LTP produces short-lived synaptic facilitation by making existing postsynaptic glutamate receptors (e.g. AMPA receptors) more sensitive to glutamate.
Conversely, L-LTP results in a pronounced facilitation of the postsynaptic response largely through the synthesis of new proteins. These proteins include glutamate receptors (e.g. AMPAR), transcription factors, and structural proteins that enhance existing synapses and form new connections. There is also considerable evidence that late LTP prompts the postsynaptic synthesis of a retrograde messenger that diffuses to the presynaptic cell increasing the probability of neurotransmitter vesicle release on subsequent stimuli.
Simultaneous pre- and postsynaptic activity initiates NMDA receptor-dependent LTP.
E-LTP can be induced experimentally by applying a few trains of tetanic stimulation to the connection between two neurons (104675870). Through normal synaptic transmission, this stimulation causes the release of neurotransmitters, particularly glutamate, from the presynaptic terminal onto the postsynaptic cell membrane, where they bind to neurotransmitter receptors embedded in the postsynaptic membrane. Though a single presentation of the stimulus is usually not sufficient to induce LTP, repeated presentations cause the postsynaptic cell to be progressively depolarized. In NMDAR-dependent synapses, this progressive depolarization relieves the magnesium blockade of the NMDA receptor. When the next stimulus is applied, glutamate binds the NMDA receptor and calcium floods the postsynaptic cell, rapidly increasing the intracellular concentration of calcium. It is this rapid rise in calcium concentration that induces E-LTP.
Beyond calcium's critical role in the induction of E-LTP, few downstream molecular events leading to the expression and maintenance of E-LTP are known with certainty. Yet there is considerable evidence that E-LTP induction depends upon the activity of several protein kinases, including calcium/calmodulin-dependent protein kinase II (CaMKII) (9114065)(2549638), protein kinase C (PKC) (9114065)(2549638), protein kinase A (PKA) (10844013), mitogen-activated protein kinase (MAPK) (9235897)(11145972), and tyrosine kinases (10545144).
Postsynaptically, the early phase of LTP is expressed primarily through the enhancement of receptor/channel sensitivity. In NMDA-dependent LTP in the CA1 hippocampus, the endogenous calcium chelator calmodulin rapidly binds calcium as a result of NMDAR opening (2549423). The calcium-calmodulin complex directly activates CaMKII which 1) phosphorylates voltage-gated potassium channels increasing their excitability (10541462); 2) enhances the activity of existing AMPA receptors; and 3) phosphorylates intracellular AMPARs and activates Syn GAP (a Ras GTPase activating protein) and the MAPK cascade, facilitating the insertion of AMPARs into the postsynaptic membrane (14993459).
PKA serves a role similar to that of CaMKII, but PKA's effects are more broad. PKA's activity is enhanced during LTP induction by elevated levels of cAMP as a result of calcium's activation of adenylyl cyclase-1 (10844013). Like CaMKII, PKA phosphorylates voltage-dependent potassium channels and also calcium channels enhancing their excitability to future stimuli. Additionally, PKA phosphorylates intracellular AMPAR stores, facilitating their insertion postsynaptically (14993459). PKA may also enhance AMPAR delivery via activation of the MAPK cascade (10541462).
While LTP is induced postsynaptically, it is partially expressed presynaptically. One hypothesis of presynaptic facilitation is that enhanced CaMKII activity during early LTP gives rise to CaMKII autophosphorylation and constitutive activation. Persistent CaMKII activity then stimulates NO synthase, leading to the enhanced production of the putative retrograde messenger, NO. Since NO is a diffusable gas, it freely diffuses across the synaptic cleft to the presynaptic cell leading to a chain of molecular events that facilitate the presynaptic response to subsequent stimuli. (See Retrograde signaling for discussion about the identity of the retrograde messenger.)
Late LTP can be experimentally induced by a series of three or more trains of tetanic stimulation spaced roughly 10 minutes apart. Unlike early LTP, late LTP requires gene transcription (8066450)(7923378) and protein synthesis (3401749), making it an attractive candidate for the molecular analog of long-term memory (7923378).
The synthesis of gene products is driven by kinases which in turn activate transcription factors that mediate gene expression. cAMP response element binding protein-1 (CREB-1) is thought to be the primary transcription factor in the cascade of gene expression that leads to prolonged structural changes to the synapse enhancing its strength (11378299). CREB-1 is both necessary (2141668) and sufficient (9790528) for late LTP. It is active in its phosphorylated form and induces the transcription of so-called immediate-early genes, including c-fos and c-jun (11306573). Ultimately, the products of CREB-1-mediated transcription and protein synthesis give rise to new building materials for the synaptic connection between pre- and postsynaptic cell.
During L-LTP, constitutively active CaMKII activates a related kinase, CaMKIV. Additionally, enhanced Ca2+ levels during late LTP increase cAMP synthesis via adenylyl cyclase-1, further activating PKA and resulting in the phosphorylation and activation of MAPK (10964936). Facilitated by cAMP (11378299), both CaMKII and CaMKIV translocate to the cell nucleus along with PKA and MAPK (mediated by PKA)(9808472), where they phosphorylate CREB-1 (11378299)(10964936)(9920677).
There is also some evidence that L-LTP is mediated in part by nitric oxide (NO) (10575022). In particular, NO may activate guanylyl cyclase, leading to the production of cyclic GMP and activation protein kinase G (PKG), which phosphorylates CREB-1 (10575022). PKG may also cause the release of Ca2+ from ryanodine receptor-gated intracellular stores, increasing the Ca2+ concentration which activates other previously mentioned kinase cascades to further activate CREB-1 (12205148).
Since NMDAR-dependent postsynaptic induction gives way to presynaptic expression of LTP, the postsynaptic cell must produce a so-called retrograde messenger that travels from the postsynaptic cell to the presynaptic cell to enhance the presynaptic cell's response to future stimuli through enhanced neurotransmitter release (8421494)(2164158). While the molecular identity of the retrograde messenger is not yet known definitively, it is possible that the messenger is nitric oxide (8978607)(9932440).
Several studies point to retrograde messengers besides nitric oxide. These include, for example, carbon monoxide (11331380) and platelet-activating factor (8906580)(8114914).
The gene expression and protein synthesis that mediate the long-term changes of LTP generally take place in the cell body, but LTP is synapse-specific; LTP induced at one synapse does not propagate to adjacent inactive synapses. Therefore, the cell is posed with the difficult problem of synthesizing plasticity-related products in the nucleus and cell body, but ensuring they only reach synapses that have received LTP-inducing stimuli.
The synthesis of a "synaptic tag" at a given synapse after LTP-inducing stimuli may serve to capture plasticity-related proteins shipped cell-wide from the nucleus (9020359). Studies of LTP in the marine snail Aplysia californica have implicated synaptic tagging as a mechanism for the input-specificity of LTP (9428516)(10535740). Given two widely separated synapses, an LTP-inducing stimulus at one synapse drives several signaling cascades (described previously) that initiates gene expression in the cell nucleus. At the same synapse (but not the unstimulated synapse), local protein synthesis creates a short-lived (less than three hours) synaptic tag (9020359). The products of gene expression are shipped globally throughout the cell, but are only captured by synapses that express the synaptic tag. Thus only the input receiving LTP-inducing stimuli is potentiated, demonstrating LTP's input-specificity.
Synaptic tagging may also give rise to LTP's associativity. If one synapse is excited with LTP-inducing stimulation while a separate synapse is only weakly stimulated, both synapses will undergo LTP, while weak stimulation alone is insufficient to induce LTP at either synapse. While weak stimuli are unable to induce gene expression in the cell nucleus, they do prompt the synthesis of a synaptic tag. Simultaneous strong stimulation of a separate pathway, capable of inducing nuclear gene expression, then prompts the production of plasticity-related proteins, which are shipped cell-wide. With both synapses expressing the synaptic tag, both capture the protein products resulting in the induction of LTP in both the strongly stimulated and weakly stimulated pathways.
The synaptic tag hypothesis may also explain LTP's cooperativity. While weak stimulation of a single pathway is insufficient to induce LTP, the simultaneous weak stimulation of two pathways is sufficient. As noted previously, weak stimulation initiates the synthesis of a synaptic tag, but is insufficient to trigger late LTP and thus CREB-1-mediated gene expression. But simultaneous weak input converges on kinases that sufficiently activate CREB-1 thereby inducing the synthesis of plasticity-related proteins, which are shipped out cell-wide as described previously. Since a synaptic tag has been synthesized at both synapses, both capture the products of gene expression and both are subsequently potentiated.
In addition to the signalling pathways described above, hippocampal LTP can be modulated by a variety of molecules. For example, the steroid hormone estradiol is one of several molecules that enhances LTP by driving CREB-1 phosphorylation and subsequent dendritic spine growth (9920677). Additionally, β-adrenergic receptor agonists such as norepinephrine contribute to the protein synthesis-dependent late phase of LTP (12770561). Nitric oxide synthase also plays an important role, leading to the up-regulation of nitric oxide and subsequent activation of guanylyl cyclase and PKG, as described previously (10575022). Similarly, activation of dopamine receptors give rise to dopamine and enhances the PKA cascade (1833673)(8922403).
LTP and behavioral memory
The mere fact that cultured synapses can undergo long-term potentiation when stimulated by electrodes says little about LTP's relation to memory. Several studies have provided some insight as to whether LTP is a requirement for memory.
Richard Morris provided some of the first evidence that LTP was indeed required for the formation of memories  (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=2869411). He tested the spatial memory of two groups of rats, one whose hippocampi were bathed in the NMDA receptor blocker APV, and the other acting as a control group. (Incidentally, the hippocampus, where LTP was originally observed, is required for spatial learning.) Both groups were then subjected to the Morris water maze, in which rats were placed into a pool of murky water and tested on how quickly they could locate a platform hidden beneath the water's surface.
Rats in the control group were able to locate the platform and escape from the pool, whereas the ability of APV-treated rats to complete the task was significantly impaired. Moreover, when slices of the hippocampus were taken from both groups of rats, LTP was easily induced in controls, but could not be induced in the brains of APV-treated rats. This provided some evidence that the NMDA receptor — and thus LTP — was somehow involved with at least some types of learning and memory.
Similarly, Susumu Tonegawa has demonstrated that a specific region of the hippocampus, namely CA1, is crucial to the formation of spatial memories  (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=8980239). So-called place cells located in this region are responsible for creating "place fields" of the rat's environment, which may be roughly equated with maps of the rat's surroundings. The accuracy of these maps determines how well a rat learns about its environment, and thus how well it can navigate about it.
Tonegawa found that by impairing the NMDA receptor, specifically by genetically removing the NMDAR1 subunit in the CA1 region, the place fields generated were substantially less specific than those of controls. That is, rats produced faulty spatial maps when their NMDA receptors were impaired. As expected, these rats performed very poorly on spatial tasks compared to controls, providing more support to the notion that LTP is the underlying mechanism of spatial learning.
Enhanced NMDA receptor activity in the hippocampus has also been shown to produce enhanced LTP and an overall improvement in spatial learning. Joe Tsein produced a line of mice with enhanced NMDA receptor function by overexpressing the NR2B subunit in the hippocampus  (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11640933). These mice, nicknamed "Doogie mice" after the precocious doctor Doogie Howser, had larger long-term potentiation and excelled at spatial learning tasks, once again suggesting LTP's involvement in the formation of hippocampal-dependent memories.
- Deadwyler SA, Dunwiddie T, Lynch G. "A critical level of protein synthesis is required for long-term potentiation." Synapse. 1987;1(1):90-5. PMID 3505366
- Frey U, Morris RG. "Synaptic tagging and long-term potentiation." Nature. 1997 Feb 6;385(6616):533-6. PMID 9020359
- Bennett MR. "The concept of long term potentiation of transmission at synapses." Prog Neurobiol. 2000 Feb;60(2):109-37. PMID 10639051 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=10639051&dopt=Abstract)
- Collingridge GL, Kehl SJ, McLennan H. "Excitatory amino acids in synaptic transmission in the Schaffer collateral-commissural pathway of the rat hippocampus." J Physiol. 1983 Jan;334:33-46. PMID 6306230 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=6306230&dopt=Abstract)
- Martin SJ, Grimwood PD, Morris RG. "Synaptic plasticity and memory: an evaluation of the hypothesis." Annu Rev Neurosci. 2000;23:649-711. PMID 10845078 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=10845078)
- McNaughton BL, Douglas RM, Goddard GV. "Synaptic enhancement in fascia dentata: cooperativity among coactive afferents." Brain Res. 1978 Nov 24;157(2):277-93. PMID 719524 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=719524&dopt=Abstract)
- Murphy GG, Glanzman DL. "Enhancement of sensorimotor connections by conditioning-related stimulation in Aplysia depends upon postsynaptic Ca2+." Proc Natl Acad Sci U S A. 1996 Sep 3;93(18):9931-6. PMID 8790434 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8790434&dopt=Abstract) (full text PDF (http://www.pnas.org/cgi/reprint/93/18/9931.pdf))
- Morris RG, Anderson E, Lynch GS, Baudry M. "Selective impairment of learning and blockade of long-term potentiation by an N-methyl-D-aspartate receptor antagonist, AP5." Nature. 1986 Feb 27-Mar 5;319(6056):774-6. PMID 2869411 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=2869411)
- McHugh TJ, Blum KI, Tsien JZ, Tonegawa S, Wilson MA. "Impaired hippocampal representation of space in CA1-specific NMDAR1 knockout mice." Cell. 1996 Dec 27;87(7):1147-8. PMID 8980239 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=8980239)
- Tang YP, Wang H, Feng R, Kyin M, Tsien JZ. "Differential effects of enrichment on learning and memory function in NR2B transgenic mice." Neuropharmacology. 2001 Nov;41(6):779-90. PMID 11640933 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11640933)
- Hebb, D.O. (1949) The organization of behavior. Wiley, New York.