views updated


Most neural connections exhibit synaptic plasticity, increases or decreases in synaptic efficacy. Several distinct forms of synaptic plasticity exist, differing in both their induction requirements and time course of expression. Synaptic plasticity allows for dynamic modification of neural circuitry that can act on time scales ranging from milliseconds to potentially lifetimes, and has been implicated in a wide range of neural and behavioral phenomena including learning and memory. A body of literature has developed demonstrating that the ability to induce synaptic plasticity is itself modifiable; that is, plasticity is plastic. This is referred to as metaplasticity, a higher-order form of synaptic plasticity. Metaplasticity's critical feature is that, once instantiated, it modifies the ability of subsequent activity to alter synaptic efficacy, fundamentally altering the rules that govern when and how changes are expressed.

Metaplasticity has been observed in a variety of brain systems and across different species, suggesting that it may be a ubiquitous feature of synaptic operation. Of particular interest, metaplasticity is observed in brain structures that have been implicated in learning and memory (e.g., hippocampus, amygdala, cortex), suggesting that it may play a key role in the regulation of information processing. A state of metaplasticity can be created by the same factors that can induce plasticity itself: the intrinsic activity of a neuron (homosynaptic factors) as well as through the actions of hormones and neuromodulators (heterosynaptic factors). To distinguish between these two causal factors, a distinction can be made between activity-dependent metaplasticity, which refers to states created homosynaptically; and modulatory metaplasticity, which refers to states created heterosynaptically.

Activity-Dependent Metaplasticity

The intrinsic activity of a neuron has long been known to result in synaptic plasticity. The duration of the induced change, as well as the type of plasticity involved, depends upon the frequency, duration, and pattern of activity. For example, high frequency activation can produce long-term potentiation (LTP), an enhancement of synaptic strength that can last for hours. LTP has emerged as a leading cellular mechanism underlying learning, and is perhaps best characterized in the hippocampus. It is now understood that the ability to induce LTP is not a static feature of hippocampal neurons, but instead is strongly influenced by the recent activation history of the neuron. For example, a low frequency priming burst, which is insufficient for producing LTP, was found to increase the subsequent activity requirements for inducing LTP. This effect was transient, synapse-specific, and depended (as does LTP) on postsynaptic N-methyl-D-aspartate receptor (NMDAR) activation.

It is now understood that the induction of plasticity at hippocampal synapses is more complex than first appreciated. Depending upon the activity of the neuron, the expression of plasticity can be bidirectional, either exhibiting a long-term enhancement (LTP), or a long-term decrement (long-term depression; LTD). For example, 900 pulses at 1-3 Hz produces LTD, 900 pulses at 10 Hz results in no lasting change, and 900 pulses at 50 Hz produced LTP. These effects are blocked by the application of an NMDA receptor antagonist, suggesting a relatively straightforward relationship between the activity of a presynaptic neuron, postsynaptic Ca++ entry, and the direction of change. These data suggest the existence of a crossover point, where activity below this point produces LTD, activity at this point (e.g., 10 Hz) produces no change, and activity above this point produces LTP. This crossover point is represented as the modification threshold theta (θm) proposed as a theoretical neural plasticity algorithm in artificial neural networks in what is now commonly known as the BCM rule.

An important theoretical feature of θm is that it represents a sliding modification threshold, the level of which depends on the recent average activity of the synapse. This has been confirmed experimentally in the hippocampus, where prior activation of a synapse was found to modify activity thresholds for LTP and LTD (i.e., modify θm). For example, high frequency priming stimuli increases the threshold activity level required for LTP (modifies θm due to an increase in average activity), making it less likely that LTP will be induced and in parallel increasing the ability to induce LTD. Conversely, low frequency priming stimuli modifies θm to decrease the threshold activity requirement for LTP; thus reducing the probability of LTD. A behavioral means of altering θm was demonstrated in the developing visual cortex by Bear and Rittenhouse. In these experiments, early visual deprivation (which would decrease the average activity of cortical neurons) facilitated the induction of LTP compared to light-reared animals of similar age. These effects were reversed by as little as two days of visual experience, demonstrating the importance of neural activity in adjusting the modification threshold.

The underlying cellular mechanisms regulating plasticity are only partially understood. One possibility is that Ca++ entry during NMDAR activation may activate a molecular cascade leading to an alteration of the NMDAR itself. Because the NMDAR is a critical point of calcium entry into the postsynaptic neuron, these changes would significantly alter the properties of activity-dependent synaptic plasticity. In addition to signaling through the NMDAR, activation of meta-botropic glutamate receptors (mGluRs) has also been shown to induce metaplasticity. For example, activation of group one mGluRs has been shown to facilitate LTP. The effects of group two mGluRs seem more complex, with reports of group two mGluR activation facilitating the induction of LTD in the dentate gyrus in vivo, but inhibiting LTD and facilitating LTP in area CA1 of the hippocampus in vitro. Thus, it appears that there are multiple regulatory mechanisms underlying activity-dependent metaplasticity, suggesting that there may be multiple modes of control over long-term synaptic plasticity.

Two possible functions for activity-dependent metaplasticity have been proposed. First, it may provide a mechanism to promote stability within a neural network by maintaining synaptic parameters within some optimal operating range. For example, increased levels of activity (which lead to enhanced synaptic efficacy) would concurrently raise the threshold for subsequent enhancement, thus applying a brake on facilitation. This would prevent saturation of the neural network at some ceiling level. Further, the sliding threshold would lead to the promotion of synaptic decrement, perhaps providing a means for resetting synaptic weights within a network. Conversely, low activity at a synapse would slide θm in the opposite direction, promoting synaptic enhancement and preventing saturation at some floor level. Second, meta-plasticity may provide a functional means for integrating neural activity over extended periods of time. This function may have an especially important role in regulating synaptic connectivity in development, allowing experience occurring over extended periods of time to exert top-down control over the refinement of synaptic connectivity.

Modulatory Metaplasticity

Modulatory metaplasticity is the regulation of synaptic plasticity through the actions of extrinsic factors such as neuromodulators and hormones. It is important to distinguish modulatory metaplasticity from the myriad number of other effects that neuromodulators can have on the process of synaptic communication. Modulatory metaplasticity refers to the regulation of the induction of synaptic plasticity. It differs from a more generalized modulatory scaling of synaptic efficacy, which would affect all aspects of synaptic transmission and plasticity in equal fashion. A second characteristic of this form of metaplasticity is that it does not require the activity of the synapse whose plasticity is being regulated. This feature distinguishes modulatory metaplasticity from the activity-dependent forms described above.

A primary example of modulatory metaplasticity has been described in the synaptic regulation of the defensive siphon withdrawal reflex (SWR) in the marine invertebrate Aplysia. The SWR is subject to multiple forms of regulation including a dynamic form based upon the recent tactile experience of the animal. For example, a local water disturbance (turbulence) can produce a transient inhibition of the SWR that is a consequence of an elevation of reflex threshold, essentially increasing the signal requirement for reflex initiation in the face of greater environmental noise. This form of behavioral regulation can be mediated by an activity-dependent form of synaptic enhancement called short-term plasticity (STP) that is expressed by identified inhibitory interneurons in response to tactile stimulation. This transient elevation of inhibitory synaptic strength reduces the ability to activate excitatory interneurons in the SWR circuit; thus imposing a requirement for greater sensory input in order to activate these neurons (and subsequently the reflex). That specific forms of STP expressed by these inhibitory neurons could be significantly suppressed by tail shock and the subsequent release of serotonin was an important finding. This regulation of STP had direct behavioral consequences: Following tail shock, the same tactile experience that normally produced reflex inhibition instead resulted in either no change or a trend towards reflex enhancement. Thus modifying the capacity for STP fundamentally altered a basic regulatory response normally exhibited by the animal.

Stress and Metaplasticity

A rapidly growing body of literature demonstrates that behavioral stress can regulate LTP and serve as an example of modulatory metaplasticity in the mammalian brain. Studies have demonstrated that acute and severe behavioral stress (e.g., by restraining and shocking the animal) produces a marked impairment of LTP. Interestingly, the same stress that blocked the induction of LTP facilitates the induction of LTD, suggesting that stress may in fact be capable of modifying θm. Consistent with this, more mildly stressful situations can in fact produce a facilitation of LTP, suggesting that the level of stress is analogous to the level of average activity in determining the direction of θm modification. Prime candidates for mediating these effects of stress on synaptic plasticity are the corticosteroids released by the adrenal cortex in times of stress. The hippocampus contains an abundance of corticosteroid receptors which are of two types: (1) a high-affinity mineralocorticoid receptor (MR); and (2) a lower-affinity glucocorticoid receptor (GR). Differential binding to these receptor types may underlie the differential effects of mild and severe stress. MR agonists facilitate LTP, whereas GR agonists impair LTP and facilitate the induction of LTD. Therefore low levels of corticosteroids would primarily act through the high-affinity MR receptor and facilitate LTP, with higher levels leading to greater activation of the GR receptor leading to suppression of LTP.

The cellular mechanism underlying this form of modulatory metaplasticity may be similar to that of activity-dependent metaplasticity, in that both appear to involve Ca++ influx into the postsynaptic cell. Corticosteroids have been shown to induce a rise in Ca++ influx, the levels of which may depend upon levels of corticosteroids. In a similar fashion as proposed for activity-dependent metaplasticity, low levels of Ca++ entry produced by mild stress may lead to a functional shift in θm leading to a promotion of LTP, whereas more severe stress results in greater levels of Ca++ entry that shifts θm in the opposite direction, impairing LTP and promoting LTD. Thus an extrinsic modulator may regulate plasticity in fundamentally the same way as was observed with intrinsic activity. It is important to note that other factors besides the corticosteroids have been implicated in stress-induced regulation of LTP, including the cytokine interleukin-1beta and endogenous opioids.


It is now understood that the ability to induce many different forms of synaptic plasticity is not a static feature of a neuron, but is regulated both by the recent activity history of the synapse, as well as by the presence of a number of neuromodulators. While the detailed means by which regulation is achieved can be diverse, most seem to share a common feature in that they act through the same signaling pathways that induce plasticity, most notably through Ca++ signaling in the neuron. The apparent complexity of regulation poses a fundamental challenge in trying to relate synaptic events to behavioral output, because the direction of plasticity and even the capacity for induction may vary based upon a number of intervening factors. Clearly, metaplasticity endows the nervous system with additional degrees of freedom in the dynamic regulation of neural circuits and systems, adding yet another dimension to the already remarkable complexity of the brain.



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

Bear, M. F., and Rittenhouse, C. D. (1999). Molecular basis for induction of ocular dominance plasticity. Journal of Neurobiology 41, 83-91.

Bienenstock, E. L., Cooper, L. N., and Munro, P. W. (1982). Theory for the development of neuron selectivity: Orientation specificity and binocular interaction in visual cortex. Journal of Neuroscience 2, 32-48.

de Kloet, E. R., Oitzl, M. S., and Joels, M. (1999). Stress and cognition: Are corticosteroids good or bad guys? Trends in Neuroscience 22, 422-426.

Dudek, S. M., and Bear, M. F. (1992). Homosynaptic long-term depression in area CA1 of hippocampus and effects of N-methyl-D-aspartate receptor blockade. Proceedings of the National Academy of Sciences of the United States of America 89, 4,363-4,367.

Fischer, T. M., Blazis, D. E., Priver, N. A., and Carew, T. J. (1997). Metaplasticity at identified inhibitory synapses in Aplysia. Nature 389, 860-865.

Huang, Y. Y., Colino, A., Selig, D. K., and Malenka, R. C. (1992). The influence of prior synaptic activity on the induction of long-term potentiation. Science 255, 730-733.

Kim, J. J., and Yoon, K. S. (1998). Stress: Metaplastic effects in the hippocampus. Trends in Neuroscience 21, 505-59.

Kirkwood, A., Rioult, M. C., and Bear, M. F. (1996). Experience-dependent modification of synaptic plasticity in visual cortex. Nature 381, 526-528.

Quinlan, E. M., Olstein, D. H., and Bear, M. F. (1999). Bidirectional, experience-dependent regulation of N-methyl-D-aspartate receptor subunit composition in the rat visual cortex during postnatal development. Proceedings of the National Academy of Sciences of the United States of America 96, 12,876-12,880.

Thomas J.Carew

Thomas M.Fischer