Overview: Cooperativity and Associativity

Long-term potentiation (LTP) is the collective name for synaptic plasticity processes in which brief (less than one second) episodes of intense synaptic activity lead to an enhancement of synaptic efficacy lasting hours to weeks, or longer. LTP in many regions has associative induction properties based on a requirement for coincident presynaptic and postsynaptic activity. This requirement is similar to that proposed on theoretical grounds by Donald Hebb in 1949 for a synaptic modification involved in learning and memory, known as the Hebb synapse. Because of its longevity and associative induction properties, scientists regard LTP as the prime neuronal model for learning and memory (see Figure 1).

Scientists first described LTP in the rabbit hippocampal formation, and later in the neocortex and a variety of other regions in the vertebrate (including human) nervous system. It has mostly been found for excitatory synapses in principal cells. These synapses are typically spine synapses; that is, located on small protuberances (spines) on the dendrite. Excitatory action in these synapses is mediated by glutamate acting on AMPA receptors located postsynaptically, and LTP is seen as an increase of this AMPA receptor-mediated transmission.

Associative LTP: A History

T. Lo/mo reported in a short note in 1966 that a few seconds of repetitive synaptic activation (10 to 15 hertz) led to a prolonged potentiation (LTP) of excitatory action on granule cells in the dentate gyrus of the hippocampus. In 1973, T. V. P. Bliss and A. R. Gardner-Medwin reported that LTP could last for weeks and that its induction appeared to depend on the number of activated presynaptic fibers, or rather the number of activated synapses. This dependence on the co-activation of many presynaptic fibers (cooperativity), and the restriction of LTP to the synapses of the activated fibers (input specificity), were demonstrated by several laboratories in the late 1970s. LTP induction was also found to be associative: A brief activation of a weak synaptic input that did not induce LTP by itself did so when occurring in close temporal contiguity with brief activation of a separate strong synaptic input to the same target neurons. Thus, LTP can form an associative connection between a weak input and a strong one or, alternatively seen, an association between a weak input and the response elicited by the strong one. In the mid-1980s several groups showed that LTP induction requires coincident spike activity of the presynaptic terminal and of the target postsynaptic neuron; that is, they demonstrated the existence of Hebb synapses. This requirement for postsynaptic depolarization of sufficient strength to evoke postsynaptic spike activity explains cooperativity. The requirement for coincident presynaptic activity explains associativity.

In 1983, G. L. Collingridge and colleagues showed that blocking another glutamate receptor, the NMDA receptor, prevented LTP induction. In the same year Dingledine described that NMDA channels permeate calcium ions, and G. Lynch and colleagues found that a rise in postsynaptic calcium was necessary for LTP induction. Moreover, several groups showed that the NMDA receptor was coupled to a voltage-sensitive channel; that is, it needed, in addition to glutamate, membrane depolarization to open. These results, together with the strict dependence of the induction on coincident presynaptic and postsynaptic activity, led H. Wigström and B. Gustafsson in 1985 to propose that LTP is initiated as a consequence of calcium influx through NMDA receptor channels co-localized with the AMPA receptors on the postsynaptic spine membrane. The NMDA receptor would act as a coincidence (and cooperativity) detector because of its need for both transmitter binding and membrane depolarization for activation. The co-localization of the NMDA and AMPA receptor channels on the subsynaptic spine membrane, together with a restricted localization of the rise in postsynaptic calcium due to the spine location of the synapse, would secure input specificity for LTP.

LTP Induction: Modulation of Threshold

Although many studies use unnatural stimulus conditions to induce LTP, such as a one-second synchronous activation of afferents at high frequency, physiological stimulus patterns have also been shown to be effective in inducing it. Brief burst stimulation that simulates the 5 to 7 hertz hippocampal EEG wave activity (theta rhythm) effectively produces LTP. This is because this stimulus pattern depresses inhibitory circuits and thereby enables a brief burst to produce substantial postsynaptic depolarization. High frequency presynaptic activation is not necessary for LTP induction. LTP can be induced in synapses activated by single stimuli at low frequency (such as 0.2 hertz) provided that this activation is associated with sufficient postsynaptic depolarization. Experimentally this depolarization can be provided by current-induced action potentials in the cell body that either passively spread or actively back-propagate into the dendritic tree where the synapses are located. Under physiological conditions when action potential activity is generated by synaptic excitation, the necessary depolarization will in addition be provided by the passive spread from active synapses. The synaptic excitation may actually play a triple role for generating the necessary conditions for LTP induction. First, by generating local depolarization in the synaptic region. Second, by generating action potentials in the cell body region that may back-propagate into the dendrite. Third, by its depolarization of the dendritic membrane facilitate back-propagation into the active dendritic region.

The threshold for inducing LTP is controlled by a number of factors. These include metaplasticity, neuromodulators, and pharmacological agents that all, directly or indirectly, interfere with the opening of NMDA receptor channels. Metaplasticity (plasticity of synaptic plasticity), a concept introduced by W. C. Abraham in 1995, is a change in induction threshold of LTP induced by prior synaptic, or cellular, activity, that is not necessarily expressed as change in efficiacy of normal synaptic transmission. One possible mechanism would be a prolonged activity-dependent change in NMDA receptor-mediated signaling, such as a LTP of that signaling. In fact, such activity-dependent changes have been reported. Neuromodulators, such as norepinephrine, acetylcholine, serotonin, dopamine, glutamate (via its metabotropic receptors), and neuropeptides regulating attention, motivation, emotion, and wakefulness, will all modulate LTP induction. As these neuromodulators exert their neuromodulatory role by their action on synaptic transmission and on cellular excitability, the NMDA receptor dependent requirements for LTP induction will concurrently be affected. Similarly, pharmacological agents that affect excitatory or inhibitory synaptic transmission, or cellular excitability, will alter the threshold for LTP induction. For example, drugs that enhance inhibitory transmission such as ethanol and benzodiazepines will impair the generation of LTP.

LTP Expression

Independently of the induction intensity (above threshold) LTP is established with a similar time course and within less than a minute. Its duration is, however, variable, depending on the intensity and/or the repetition of the synaptic activity. Thus, following weak activation it may decay within minutes whereas an intense activation can lead to a LTP lasting weeks. The more prolonged LTP given by more intense synaptic activation may not only be explained by a greater extent of NMDA receptor activation but also by the concurrent activation of voltage-gated calcium channels and/or other receptor/channel systems. Based on the effect of protein synthesis inhibitors, LTP has been divided in an early phase (E-LTP, less than a few hours), and a late LTP (L-LTP). L-LTP is thus supposed to rely on synthesis of new proteins whereas E-LTP relies on modification of pre-existing ones.

What aspect of synaptic transmission is actually modified in NMDA receptor-dependent LTP? This issue has been intensively debated since the mid-1980s. Despite numerous studies, the question of whether there is a more efficient release of glutamate, or an increase in AMPA receptor channel number or efficiency, is still unsettled. A complicating factor may be that NMDA receptor-dependent LTP is based on different expression mechanisms dependent on experimental conditions, for example cultured versus intact tissue, or on other factors such as brain region and animal age.

Nonassociative LTP

Long-term synaptic plasticity exists not only in the form of NMDA receptor-dependent LTP but also as nonassociative forms not relying on NMDA receptor activation. The foremost example of such NMDA receptor-independent LTP, first described by E. W. Harris and C. W. Cotman in 1986, is that in the synapse that connects granule cells in the dentate gyrus with CA3 pyramidal neurons in the hippocampus. In contrast to NMDA receptor-dependent LTP its expression appears undisputed whereas its induction mechanism is debated. A number of studies by R. A. Nicoll and colleagues in the 1990s have established that NMDA receptor-independent LTP is presynaptically expressed as an increased efficiacy of transmitter release. These studies have also indicated its induction to be noncooperative, only related to presynaptic calcium accumulation. On the other hand, studies by D. Johnston and colleagues have indicated a cooperative induction, related to postsynaptic calcium influx via voltage-gated calcium channels and release from internal calcium stores. A similar NMDA receptor-independent LTP is also described for the synapse that connects granule cells and Purkinje cells in the cerebellum.

LTP on Interneurons

Glutamatergic afferent fibers do not only make excitatory synapses with principal cells, but also with interneurons. Scientists have debated whether these interneuronal synapses exhibit LTP. It would appear that LTP does not exist in most of these synapses. However, LTP with associative properties has been found in some interneuron types. Interestingly, this associative induction was found not to rely on NMDA receptor activation as in the principal cells, indicating the existence of other mechanisms for coincidence detection.

Conclusion

LTP denotes forms of synaptic plasticity with associative as well as nonassociative induction (i.e., they depend on temporal contiguity between activity in different pathways and on activity in a single pathway, respectively). Non-associative LTP will allow for a more efficient transmission in specific pathways that are intensely used, irrespective of activity in others. Associative LTP on the other hand strengthens synapses in a manner that relies on contiguity of activity in those neurons that are connected via the modifiable synapse. This Hebbian modification rule is a powerful device used in neural network models of nonsupervised learning to produce, for example, self-organizing capabilities.

See also:GLUTAMATE RECEPTORS AND THEIR CHARACTERIZATION; HEBB, DONALD

Bibliography

Abraham, W. C., and Bear, M. F. (1996). Metaplasticity: The plasticity of synaptic plasticity. Trends in Neurosciences 19 (4), 126-130.

Bailey, C. H., Giustetto, M., Huang, Y. Y., Hawkins, R. D., and Kandel, E. R. (2000). Is heterosynaptic modulation essential for stabilizing Hebbian plasticity and memory? Nature Reviews Neuroscience 1 (1), 11-20.

Gustafsson, B., and Wigstrom, H. (1988). Physiological mechanisms underlying long-term potentiation. Trends in Neurosciences 11 (4), 156-162.

Hanse, E., and Gustafsson, B. (1994). Onset and stabilization ofNMDA receptor-dependent hippocampal long-term potentiation. Neuroscience Research 20 (1), 15-25.

Johnston, D., Williams, S., Jaffe, D., and Gray, R. (1992). NMDA-receptor-independent long-term potentiation. Annual Review of Physiology 54, 489-505.

Linden, D. J. (1999). The return of the spike: Postsynaptic action potentials and the induction of LTP and LTD. Neuron 22, 661-666.

Malenka, R. C., and Nicoll, R. A. (1999). Long-term potentiation—a decade of progress? Science 285, 1,870-1,874.

McBain, C. J., Freund, T. F., and Mody, I. (1999). Glutamatergic synapses onto hippocampal interneurons: Precision timing without lasting plasticity. Trends in Neurosciences 22 (5), 228-235.

Sanes, J. R., and Lichtman, J. W. (1999). Can molecules explain long-term potentiation? Nature Neuroscience 2 (7), 597-604.

BengtGustafsson

EricHanse