Learning Mechanisms: The Case for CaM-KII

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2001-2002
DOI: 10.1126/science.276.5321.2001

What are the long-lasting changes that occur in your brain as you learn new information? It is generally thought that memory is due to persistent modification of the strength of synapses, the structures that communicate information from one neuron to the next. One such modification is the long-term potentiation (LTP) that occurs at the synapses of the CA1 region of the hippocampus. This type of modification is of particular interest because it has associative properties that match that of learning itself (we come to associate the smell and taste of food), and because the hippocampus is important for long-term memory. The report by Barria et al. on page 2042 in this issue (1) is a significant step forward in our understanding of the persistent biochemical modifications that underlie this form of LTP.

The major new finding of Barria et al. is that the postsynaptic receptors (a subtype of glutamate receptor known as AMPA receptors) that mediate excitatory synaptic transmission at this synapse become phosphorylated after LTP induction and stay phosphorylated for at least 1 hour thereafter. The authors also show that phosphorylation of AMPA receptors in an expression system enhances the responses of these receptors to glutamate. Together with previous findings (2), these results provide strong evidence for a simple, postsynaptic mechanism for enhancing synaptic transmission during LTP.

The initial triggering event of LTP, believed to be a brief rise in postsynaptic calcium, results in the phosphorylation of AMPA receptors. What causes this modification and how is it maintained for at least an hour after the initial triggering event? A large body of evidence suggests that calmodulin-dependent protein kinase II (CaM-KII) is a critical player in LTP, and it has special properties that make it an attractive candidate for exhibiting persistent changes and serving as a memory molecule (3). CaM-KII is localized in the postsynaptic density, directly adjacent to the channels that mediate synaptic transmission (see the figure). The work of Barria et al. suggests that CaM-KII controls AMPA receptors directly, because the phosphorylation of AMPA receptors after LTP induction occurs at a site that can be phosphorylated by CaM-KII and is blocked by an inhibitor of CaM-KII. Theoretical (4) and experimental (5) studies suggest that the maintenance of the AMPA receptor phosphorylation may be due to the ability of CaM-KII to maintain its activity for long periods after its initial activation by calcium. The kinase may accomplish this by calcium-dependent autophosphorylation of the threonine residue at position 286, which renders its activity independent of calcium. This autocatalytic process could maintain the “on” state of the kinase for long periods, perhaps indefinitely. Indeed, Barria et al. provide support for this mechanism by showing that CaM-KII stays persistently phosphorylated at the 286 site for at least 1 hour after LTP induction [but see (6)]. Other kinases may also contribute to this persistent phosphorylation (7), as inhibitors of CaM-KII have failed to depress established LTP (8).

Where LTP happens.

The postsynaptic side of the synapse (arrows) is the site of persistent biochemical modifications that maintain LTP. [Photo courtesy of Kristen Harris]

A simple and direct role for CaM-KII in triggering and perhaps maintaining LTP is supported by studies in which CaM-KII activity was acutely increased either with viral transfection (9), injection of the active enzyme (10), or injection of calcium and calmodulin (11). In these cases synaptic transmission is enhanced and LTP is occluded. However, recent work (12) with transgenic mice, in which constitutively active CaM-KII was chronically expressed, appears to contradict this view of the role of CaM-KII in LTP. Such mice exhibit no increase in synaptic transmission, and normal LTP can still be generated (although there is a shift in the frequency dependence of LTP). These results led to the conclusion that CaM-KII is not part of the direct signal transduction cascade responsible for generating LTP but only modulates this machinery.

Can these conflicting data be reconciled? One possibility is that CaM-KII does directly mediate the generation of LTP but can also modulate the sensitivity of this transduction pathway. In particular, chronic CaM-KII activity (12) could decrease the sensitivity of this pathway (perhaps by phosphatase activation). This view is a special case of the “sliding threshold” model (13) in which a requisite component of the signal transduction pathway used to generate LTP (that is, CaM-KII) can itself also act to modulate the pathway. This modulation may occur primarily as a result of long-term genetic modifications and not of more acute perturbations. A homeostatic compensatory regulation of this kind may be an example of a general principle in biology that serves to establish a functionally relevant dynamic range for signal transduction. Simply put, if a pathway is continually activated it will tend to decrease its sensitivity; if it is not activated it will increase its sensitivity.

Clear examples of how difficult it is to interpret the results of genetic modification of transduction cascades comes from work on rod phototransduction (14). The advantage of this system is that the molecular basis of the cascade is known, so one can assess whether genetic modifications produce easily interpretable effects. Transgenic mice expressing a constitutively active transducin that mediates phototransduction would be expected to show saturation of the subsequent transduction cascade. However, a compensatory reduction in downstream enzymes prevents saturation. Another example comes from mice in which the inhibitory subunit of cyclic guanosine 3′,5′-monophosphate (cGMP)-phosphodiesterase is deleted. This would be expected to lead to a drop in cGMP concentration. Surprisingly, cGMP concentration is elevated as a result of an unanticipated disappearance of the catalytic subunits of the enzyme. In many other cases, modification of rod proteins leads to cell-specific degeneration. These examples indicate that extreme caution must be used in interpreting results obtained by genetic modification of signal transduction pathways, especially when genetic and anatomical controls are not feasible.

In theory, compensatory changes might be minimized by inducible genetic modification. For instance, heat shock promoters have been used in Drosophila to produce changes in gene expression within 30 min. The elegant new methods (15) for inducible gene expression in mice are comparatively slow (2 weeks) and may not eliminate the problems of compensatory change.

With the new results from Barria et al., a strong case is emerging for the importance of CaM-KII in the direct control of synaptic strength and, by implication, in the storage of information (16). CaM-KII, as part of an important synaptic signal transduction pathway, may also control homeostatic pathways, which only become apparent during chronic manipulations, such as those that occur with current methods for genetic manipulations.


View Abstract

Stay Connected to Science

Navigate This Article