If synapses simply continued to increase in strength as a result of LTP, eventually they would reach some level of maximum efficacy, making it difficult to encode new information. Thus, to make synaptic strengthening useful, other processes must selectively weaken specific sets of synapses. Long-term depression (LTD) is such a process. In the late 1970s, LTD was found to occur at the synapses between the Schaffer collaterals and the CA1 pyramidal cells in the hippocampus. Whereas LTP at these synapses requires brief, high-frequency stimulation, LTD occurs when the Schaffer collaterals are stimulated at a low rate—about 1 Hz—for long periods (10–15 minutes). This pattern of activity depresses the EPSP for several hours and, like LTP, is specific to the activated synapses (Figure 25.12). Moreover, LTD can erase the increase in EPSP size due to LTP, and, conversely, LTP can erase the decrease in EPSP size due to LTD. This complementarity suggests that LTD and LTP reversibly affect synaptic efficiency by acting at a common site.

Figure 25.12

Long-term synaptic depression in the hippocampus. Low-frequency stimulation (1 per second) of the Schaffer collateral axons causes a long-lasting depression of synaptic transmission. (After Mulkey and Malenka, 1992.)

LTP and LTD at the Schaffer collaterals-CA1 synapses actually share several key elements. Both require activation of NMDA-type glutamate receptors and the resulting entry of Ca²⁺ into the postsynaptic cell. The major determinant of whether LTP or LTD arises appears to be the amount of Ca²⁺ in the postsynaptic cell: Small rises in Ca²⁺ lead to depression, whereas large increases trigger potentiation. As noted above, LTP is at least partially due to activation of CaMKII, which phosphorylates target proteins. LTD, on the other hand, appears to result from activation of Ca²⁺-dependent phosphatases that cleave phosphate groups from these target molecules (see Chapter 8). Evidence in support of this idea is that phosphatase inhibitors prevent LTD, but have no effect on LTP. The different effects of Ca²⁺ during LTD and LTP may arise from the selective activation of protein phosphatases and kinases by low and high levels of Ca²⁺. It may be that LTP and LTD phosphorylate and dephosphorylate the same set of regulatory proteins to control the efficacy of transmission at the Schaeffer collateral-CA1 synapse.

A somewhat different form of LTD has also been implicated in motor learning in the cerebellum, which mediates the coordination, acquisition, and storage of complex movements (see Chapter 19). LTD of synaptic inputs onto cerebellar Purkinje cells was first described in the early 1980s. The role of cerebellar LTD in motor learning remains controversial; it has nonetheless been a useful model system for understanding the cellular mechanisms of long-term synaptic plasticity.

Although LTD decreases the efficacy of synapses in both the cerebellum and hippocampus, the properties of LTD at these two sites are different. Purkinje neurons in the cerebellum receive two distinct types of excitatory input: climbing fibers and parallel fibers (Figure 25.13A). In the cerebellum, LTD is associative in that it arises when climbing fibers and parallel fibers are activated at the same time. In this circumstance, LTD selectively reduces the strength of transmission at the parallel fiber synapse (Figure 25.13B).

Figure 25.13

Long-term synaptic depression in the cerebellum. (A) Experimental arrangement. Synaptic responses were recorded from Purkinje cells following stimulation of parallel fibers and climbing fibers. (B) Pairing stimulation of climbing fibers (CF) and parallel (more...)

Activity of climbing fiber and parallel fiber synapses results in activation of two distinct intracellular signal transduction pathways in the postsynaptic Purkinje cell. In the first pathway, glutamate released from the parallel fiber terminals activates at two types of receptors, the AMPA-type and metabotropic glutamate receptors (see Chapter 7). Glutamate binding to the AMPA receptor results in membrane depolarization, whereas binding to the metabotropic receptor initiates a second messenger cascade that produces inositol trisphosphate (IP₃) and activates an intracellular enzyme, protein kinase C (PKC) (see Chapter 8). This second pathway is initiated by climbing fiber activation, which causes a large influx of Ca²⁺ through voltage-gated channels and subsequent increase in intracellular Ca²⁺ concentration. The conjoint activation of these two intracellular pathways leads to LTD: Through an as-yet-unknown mechanism, Ca²⁺ interacts with PKC to decrease the postsynaptic response of AMPA receptors to glutamate at the parallel fiber synapses (Figure 25.13C). Thus, in contrast to LTD in the hippocampus, cerebellar LTD requires the activity of a protein kinase, rather than a phosphatase, and does not involve Ca²⁺ entry through the NMDA type of glutamate receptor (which is not present in mature Purkinje cells).

In summary, studies of synaptic plasticity have shown some of the strategies used to change synaptic function in various brain regions. Behavioral plasticity requires activity-dependent synaptic changes in an assembly of interconnected neurons. These circuit changes are the basis not only of learning, memory, and other forms of plasticity, but also some pathologies. Thus, abnormal patterns of neuronal activity, such as those that occur in epilepsy, can stimulate abnormal changes in synaptic connections that may further increase the frequency and severity of seizures (Box C). Despite the substantial advances in understanding the cellular and molecular bases of some forms of plasticity, how selectively changing synaptic strength encodes memories or other complex behavioral modifications in the mammalian brain is simply not known.

Box C

Epilepsy: The Effect of Pathological Activity on Neural Circuitry.