Homer Bdnf Nt3 Mkp1 Ssat

FIGURE 5 Transcriptional regulation pathways controlling the expression of synaptic plasticity-associated genes. See text for discussion.

occurs with L-LTP, these interesting findings support the idea that late waves of altered gene expression contribute to memory consolidation involving hip-pocampal neurons.

Thus, we see that at least two (zif268 and krox 20) and probably several more transcription factor-encoding genes are up-regulated in L-LTP. Overall, these findings that transcription factors are targets of the gene expression system of L-LTP and long-term memory suggest that plasticity-associated gene regulation will be quite complex. The findings indicate the likelihood that the initial triggering of altered gene expression with L-LTP-inducing stimulation sets off secondary waves of altered gene expression. The potential for exponential expansion of altered gene expression in L-LTP, along with combinatorics for secondary signals impinging on these mechanisms, appears somewhat daunting. However, parsing out these complex pathways will be necessary to answer a fundamental question in biology—how neuronal cell surface activity impinges upon the genome.

Signaling Molecules

A second category of L-LTP-associated genes encode signaling molecules. As with the transcription factors regulated in L-LTP, increased production of signaling molecules and regulators of signal transduction cascades suggest the triggering of a variety of post-genome secondary effects by L-LTP-inducing stimulation.

One of the most interesting molecules in this category is BDNF, which we have already discussed several times. BDNF gene expression increases with L-LTP-inducing stimulation, along with that of a related growth factor, Neurotrophin-3 (46). This likely occurs via CRE-regulated expression as the BDNF gene has the necessary sequence in its upstream region. BDNF is particularly intriguing because, as we have already discussed, it is a modulator of LTP induction and is furthermore capable of triggering lasting increases in synaptic strength in its own right. Thus increased BDNF expression could play a role in modulating subsequent plasticity in a temporal integration fashion, or itself trigger lasting plastic change. In the limit, BDNF, which couples to the ERK/CREB cascade, could trigger a self-perpetuating feedback mechanism for maintaining increased synaptic strength perpetually. Even though this idea is quite speculative, it provides an appealing example of a potential self-reinforcing mechanism that could survive protein turnover and last the lifetime of the animal.

Another signaling molecule gene regulated by L-LTP-inducing stimulation codes for tissue Plasminogen Activator (t-PA). t-PA is a secreted protease that has the capacity to modulate the extracellular matrix structure. t-PA knockout mice have defects in L-LTP (47). t-PA could potentially serve in a long-term regulatory role through increasing active products in the extracellular space via its known role of converting pro-hormones into hormones. Another molecule that has been proposed to play a similar role is Matrix Metalloprotease-9 (MMP-9), whose gene is regulated in an activity-dependent fashion and which cleaves extracellular matrix molecules (48). Both t-PA and MMP-9 might also play a role in regulating the structure of the synaptic region as well.

In the last chapter, we discussed the fact that polyamines like spermine and spermi-dine can directly modulate NMDA receptor function. In this context, it is interesting to note that a synthetic enzyme that is involved in their production, SSAT (spermidine/ spermine N1-acetyltransferase), is also up-regulated in L-LTP (49). This potentially provides yet another positive feedback signal that might be used by the neuron for long-term temporal integration. However, lest we get too carried away with the positive feedback loop concept, I note that not all gene targets in L-LTP will promote subsequent activity-dependent plasticity.

The MAP Kinase Phosphatase MKP-2 also is up-regulated in L-LTP. This dual-specificity phosphatase dephospho-rylates phospho-ERK, providing a potential negative feedback mechanism limiting activation of the ERK/CREB pathway (35, 50).

Thus, no clear model for the role of L-LTP-associated genes affecting signal transduction emerges at this time. However, some interesting possibilities exist, and, as I mentioned earlier, the available data are consistent with diverse secondary effects downstream of altered gene regulation. As with the transcription factor category of targets, the regulation of signaling molecules with L-LTP similarly suggests that complicated cascades of biochemical sequelae will be triggered by L-LTP-inducing stimulation.

Structural Proteins

In contrast to the signaling molecules described earlier, the few structural proteins that have been identified as targets in L-LTP bring us refreshingly back around to tried-and-true functional players in synaptic transmission. Most straightforward in this context is the AMPA receptor itself (51). Mike Browning's group has shown that L-LTP is associated with increased AMPA receptor synthesis at the protein level (although the genetic basis for this has not been worked out). This translates in a conceptually straightforward way into a mechanism for increasing synaptic strength.

The metabotropic receptor scaffolding protein HOMER has also been identified as a gene regulated in L-LTP (11, 52, 53). This is interesting because this protein is a part of the synaptic spine structural complex interacting with metabotropic glutamate receptors, as we discussed in Chapter 6. In addition, mGluRs have been implicated as controlling dendritic spine protein synthesis as we discussed in the last chapter, and increased HOMER expression might play a role in facilitating or regulating local protein synthesis as one component of a mechanism for maintaining increased synaptic strength.

This last idea is not as far-fetched as it might sound considering that another target of L-LTP-associated gene expression is involved in the same types of processes— Arc. We have already discussed Arc in the context of local dendritic protein synthesis (see Chapter 7 and reference 54 for a review). Arc mRNA is rapidly induced by LTP-inducing stimulation. Moreover, as we will discuss in the next section, this mRNA is selectively localized to recently active synaptic regions and is subject to selective expression by local protein synthesis mechanisms. Arc is a cytoskeleton-associated protein that may be involved in stabilizing structural changes at potentiated synapses, as inhibition of Arc synthesis disrupts maintenance of LTP (13).

Thus, it is palatable to think that increased expression of AMPA receptors, HOMER, and Arc might contribute to a stable increase in synaptic strength as a mechanism for L-LTP expression. As more pieces of the puzzle become available, the applicability of this model will become clear.

While we are at a very early stage with these types of studies, some tantalizing themes have begun to emerge from the identified gene targets in L-LTP. First, transcription factors are up-regulated, suggesting that secondary waves of gene expression will be involved in L-LTP induction and maintenance. Second, structural proteins at the synapse appear to be targets that are up-regulated, suggesting that the hypothesis that altered gene expression is a component of triggering and maintaining long-term structural changes at the synapse in L-LTP (55, 56). We will return to this idea in more general terms in the last section of the chapter. Finally, signaling molecules and modulators of plasticity-related signal transduction mechanisms are targets of altered gene expression in L-LTP. Specifically, both positive and negative feedback mechanisms are triggered with L-LTP-inducing stimulation.

Particularly intriguing molecules in this context are molecules like BDNF, which themselves are capable of triggering long-term change and promoting activity-dependent long-term change. This suggests the interesting idea that positive feedback mechanisms that provide a self-reinforcing component necessary for maintaining synaptic potentiation perpetually might be triggered. We will return to this idea in a theoretical context in the last chapter of the book.

E. mRNA Targeting and Transport

L-LTP, a protein-synthesis-dependent phenomenon, can be selectively induced at particular synapses, or at least particular dendritic regions (see Box 4 in the last chapter, for example). Considering these data leads us to the hypothetical necessity for localization of newly synthesized mRNAs (or proteins) at recently activated synapses, so that the altered mRNA expression necessary for L-LTP be manifest appropriately at these potentiated synapses. This brings us to the final question that we will think about in the context of the cell biology of L-LTP. How is it that the new gene products get expressed at the right place among the multitude of synapses in a pyramidal neuron dendritic tree? It is known that L-LTP is synapse-specific— how is it that the gene products get targeted to the right synapses in order to increase synaptic strength appropriately?

This process is still mysterious but is actively under investigation. However, several useful insights are already available, in large part from studies of regulation of induction and expression of the mRNA for Arc/Arg3.1. The research groups of Deitmar Kuhl, Paul Worley, and Ozzie Steward have pioneered this area of investigation, and the following discussion derives largely from their work.

It is useful to think of the Arc messenger RNA as a prototype for studying the regulation of the distribution of new gene products in the neuron (reviewed in references 57 and 58). As we have already discussed, Arc/Arg3.1 was discovered as a gene product induced by activity in general and LTP in particular (13, 54, 59). Arc, like many of the other L-LTP-associated gene products we have been discussing, is an immediate early gene. Like the other IEGs, its increased expression is transient, which facilitates experimental study because there is a low pre-stimulus level of its mRNA. Finally and most importantly, once Arc mRNA is expressed, the mRNA is selectively localized to active synapses, or synapses that have been potentiated (see Figure 6). Thus, Arc regulation likely represents a microcosm of activity-dependent, selectively targeted gene products in the neuron.

Returning to our original question, how is it that mRNAs end up at the right synapses? Two broad possibilities present themselves. One possibility is that mRNAs leave the nucleus with "addresses" that send them to the right spot. The second possibility is that mRNAs are sent throughout the neuron and are selectively captured at the appropriate synapses. Studies of Arc indicate that the second scenario is the correct one. Newly synthesized Arc mRNA is distributed throughout the dendrites, presumably by diffusion but also potentially by mRNA carrier proteins associated with the cytoskeleton. The Arc mRNA is then "captured," or concentrated, at active synapses.

Thus, the targeting process appears to be not just locally initiated and controlled, but also dependent upon biochemical processes restricted to a particular synaptic region. The mRNA is not targeted specifically to a predefined dendritic region as it leaves the nucleus. There is no address on the mRNA when it leaves the nucleus—the mRNA is sent out globally and sequestered locally. If this is the general mechanism for neuronal mRNA targeting (and there is no reason to think it isn't), this type of mechanism eliminates an enormously complex trafficking problem. mRNAs do not have to

FIGURE 6 Activity-dependent Arc expression and dendritic localization. Newly synthesized Arc mRNA is selectively targeted to dendritic domains that have been synaptically activated. The photomicrographs illustrate the distribution of Arc mRNA as revealed by in situ hybridization in (A) nonactivated dentate gyrus; (B), 2 hours after a single electroconvulsive seizure and (C), and after delivering high-frequency trains to the medial perforant path over a 2-hour period. Note the uniform distribution of Arc mRNA across the dendritic laminae after an ECS and the prominent band of labeling in the middle molecular layer after high-frequency stimulation of the perforant path. (D) Schematic illustration of the dendrites of a typical dentate granule cell and the pattern of termination of medial perforant path projections. HF, hippocampal fissure; gcl, granule cell layer. Figure from Steward and Worley, (57).

FIGURE 6 Activity-dependent Arc expression and dendritic localization. Newly synthesized Arc mRNA is selectively targeted to dendritic domains that have been synaptically activated. The photomicrographs illustrate the distribution of Arc mRNA as revealed by in situ hybridization in (A) nonactivated dentate gyrus; (B), 2 hours after a single electroconvulsive seizure and (C), and after delivering high-frequency trains to the medial perforant path over a 2-hour period. Note the uniform distribution of Arc mRNA across the dendritic laminae after an ECS and the prominent band of labeling in the middle molecular layer after high-frequency stimulation of the perforant path. (D) Schematic illustration of the dendrites of a typical dentate granule cell and the pattern of termination of medial perforant path projections. HF, hippocampal fissure; gcl, granule cell layer. Figure from Steward and Worley, (57).

have to be pre-addressed—local demands can dictate their disposition.

An additional implication of this finding is that induction signals from the dendrites to the nucleus, such as those processes we discussed in Section II.A, can be "unaddressed" as well. The synapse-to-nucleus signal does not have to have a return address on it, a signal arriving at the genome can originate from any synapse or dendritic region and its point of origin is immaterial. The genome simply has to respond appropriately to the signal with an increased readout, and local activity-dependent mechanisms in the den-drites will assure that the right synapses capture the product.

The mechanism for the appropriate capture of mRNAs at potentiated synapses is unknown, but the basic phenomenon is, of course, highly reminiscent of the synap-tic tagging observation described in the last chapter (Box 4). As we discussed earlier, a parsimonious model for synaptic capture is to simply invoke persisting signals already localized at potentiated synapses, such as autonomously active CaMKII or PKC, as the root of the capture signal. A similar mechanism has already been proposed to operate in site-specific facilitation of Aplysia sensory neuron synapses by Wayne Sossin and his colleagues (60). This idea is speculative but is appealing because it allows for a straightforward transition of E-LTP into L-LTP at predefined synapses.

One aspect of the targeting and capture mechanism that seems clear is that an mRNA binding protein of some sort must be involved. After all, a priori it is clear that an mRNA must bind to something in order to have its diffusion restricted to a particular domain. Several interesting candidate molecules are known mRNA binding proteins and thus might participate in mRNA capture at dendrites (reviewed in references 58 and 61). One appealing possibility is the fragile X mRNA binding protein (FMRP) that we discussed in the last chapter. Activity-dependent synthesis of FMRP might provide a mechanism for mRNA capture. Another candidate that at a minimum is involved in dendritic mRNA trafficking (if not synapse-specific localization) is the staufen protein (62). Staufen is an mRNA binding protein localized to dendrites that is involved in mRNA

targeting in a variety of systems. A final possibility is the cytoplasmic polyadeny-lation element binding proteins (CPEB, not to be confused with the transcription factor C/EBP; see (33). The CPEBs are mRNA binding proteins that also regulate translation of mRNAs. Again, I emphasize that the mechanisms for targeting are unknown and that these are just possibilities that have been identified as potential players.

Though the mechanisms are unknown, it is clear that the specifically localized Arc mRNA is converted into a localized increase in Arc protein. There is no need to invoke a specific mechanism for this—it can simply be accomplished by local protein synthesis as we discussed at the end of the last chapter in the context of dendritic protein expression. Arc might be handled by the same mechanisms that are used for the synthesis of proteins from mRNAs that are expressed constitutively such as CaMKII and MAP2.

What is it that the Arc protein does once it is made at the right synapses? Although the function of Arc is unknown, there are tantalizing clues. It is known from various studies that Arc is part of the NMDAR supramolecular complex that we discussed in Chapter 5. Also, Arc is known to be a cytoskeleton-associated protein, which is part of how it got its name (Activity-regulated cytoskeleton-associated protein). Thus, Arc may be part of a local receptor-complex stabilizing structure, or even a component of the specific NMDAR/ AMPAR stabilizing mechanism that was described in the last chapter in Figure 12. This is, of course, speculative, but it at least allows the possibility of directly translating an increase in Arc protein into an increase in synaptic strength. Many additional components of the synapse are almost certainly involved as well. Specifically relevant in this context is the observation from Michael Browning's group of increased synthesis of AMPA receptor protein in L-LTP (51).

However, in summary, Arc is an interesting proof-of-principle molecule.

Studies of Arc have demonstrated that activity-dependent mechanisms of mRNA localization exist in neurons. They have given important initial insights into some of the strategies the neuron uses to solve the problem of targeting altered gene products to the right synapses, and highlight an important role for localized protein synthesis at activated synapses. What remains is to figure out the detailed biochemical mechanisms for these processes and to increase our understanding of how local protein synthesis gets converted into potentiated synaptic transmission locally. Also, it is clear that identifying the molecular identity of the "synaptic tag" is a high priority, in order to understand the bridging mechanisms for the E-LTP to L-LTP transition.

F. Effects of the Gene Products on Synaptic Structure

Consideration of Arc as a molecule contributing to L-LTP also introduces a number of very important considerations about constraints on maintenance mechanisms for very long-lasting L-LTP. The half-life of the Arc protein is a few hours— think about the implications of this fact. The Arc message increases transiently (about 1/2 hour) and returns to a low basal level. The mRNA is concentrated at the right synapses and translated into protein, which is broken down with a half-life of a few hours. In the absence of Arc mRNA, which has already decayed to basal levels, this protein cannot be replenished. Thus, if Arc (or any other protein with similar regulation and kinetics) participates in synaptic potentiation, this mechanism can only allow potentiation to be maintained for less than a day. Induction of Arc or any similar protein is clearly not a maintenance mechanism for very long-lasting events.

This type of thought experiment is prima facie evidence that there are multiple phases of L-LTP, because LTP is clearly able to be maintained for many days in vivo.

Any Arc-dependent process will have decayed within about 5 protein half-lives, or something on the order of 24 hours at most. Some other molecular process must maintain later stages of L-LTP in vivo.

Even though we have discussed this concept specifically in the context of Arc because it is well-studied and known to be necessary for L-LTP (13), the generalization applies to any protein involved in very long-lasting effects in neurons. This consideration highlights the fact that triggering altered gene expression and protein synthesis is likely an induction mechanism for long-lasting changes, but not necessarily a maintenance mechanism. All proteins have a finite half-life, thus maintaining long-lasting change requires an ongoing process. As we will discuss in the last chapter of the book, in biochemical systems the route for achieving persisting effects in the face of breakdown and resynthesis of the component molecules involved is via a self-reinforcing chemical reaction. This is a controlled positive feedback loop wherein the activated molecule promotes its own resynthesis.

Given the constraint of protein turnover, how is it that you get really long-lasting changes in synaptic strength in L-LTP? I will speculate wildly here about how this might happen, looking to the studies of L-LTP that we have been discussing for clues. The model I will present has two components: a change in synaptic structure that results from altered expression of structural proteins and a positive feedback mechanism to maintain these changes in the face of protein turnover.

The Structural Change

I am going to leave the structural change ill-defined. It could be an increased number of AMPA receptors at a pre-existing spine, with its multitudinous receptor binding partners and cytoskeletal/PSD associations. It could be a new spine that splits off by spine fission, as has been proposed by a number of investigators (reviewed in 56). It could be the growth of entirely new spines from the dendritic shaft (55, 56, 63-65). These are variations on the same issue from a biochemical perspective—doubling the surface area of a spine is basically placing the same demand on the protein synthetic machinery as splitting off a second spine or growing a new spine from the dendrite. It's not that the question of which of these actually happens is uninteresting, or that they are even functionally identical, it's just that for the purposes of the present discussion they are equivalent in their impact. However, for ease of presenting the model, I will just take the case of increasing the postsynaptic size (equivalent to an increased number of receptors) in a pre-existing spine.

The mechanisms we talked about in Sections II.C and II.E have gotten us to the point of having a spine with increased AMPA receptor protein, hypothetically through Arc-dependent signaling mechanisms. What happens when the Arc signal decays and all of the newly synthesized AMPA receptors and the like are undergoing continual breakdown and resynthesis? The new, larger PSD complex must have some way to perpetuate itself in its new, larger, state. This requires some ongoing feedback signal that says "stay big." This could happen by any of a number of specific mechanisms, but this basic point is the important concept for this section of the chapter. Maintaining an increased synaptic connection over many protein half-lives, in the face of complete breakdown and resynthesis of all the protein components that make up that synapse, requires a self-reinforcing maintenance signal.

The specific mechanisms for this self-perpetuating signal are completely unknown at this point. The possibilities fall into two broad categories. One possibility is that there is an ongoing activity integration mechanism that keeps the size of the synapse scaled to its pre-existing level. Variations of this idea are such things as a mechanism integrating the average spine

FIGURE 7 Hypothetical mechanisms contributing to activity-dependent changes in synaptic structure. Both local effects and nuclear signaling are likely involved in triggering and maintaining very long-term structural changes in synaptic spines. See text for additional discussion.

calcium level (which would vary depending on the net synaptic activity; it would be larger in a potentiated synapse) or a mechanism reading out the amount of receptor amino termini present and maintaining it at a constant level. These are just specific examples for illustrative purposes, there are, of course, many other possibilities.4

The second general possibility is that a unique molecule is introduced into a potentiated synapse that makes a unique signal that is self-perpetuating. This

4One of the advantages of the spine fission/new spine models is that the necessity for this type of scaling mechanism is obviated. A synapse, once formed, must simply keep making itself unless it gets a new signal to go away. However, I don't prefer this idea at present because it requires that a doubling of synaptic strength involves a doubling of the number of synaptic spines. Quantitatively large morphological changes such as this have not generally been observed to date, although there are clearly examples of increased synaptic contacts with L-LTP inducing stimulation (reviewed in reference 66).

molecule must, of course, have the capacity to in parallel increase synaptic size through impinging on protein synthesis, the receptor insertion machinery, and the like.

I have a modest preference for this second model because there are some appealing specific candidate molecules that can perform this function, that have been implicated experimentally as playing a role in L-LTP. These are the cell adhesion molecules, specifically the integrins and cadherins (see Figures 7 and 8). While I will not belabor the case for these molecules because my entire model for long-lasting LTP is so speculative, there are a number of appealing findings in this context. First, blocking the function of either cadherins or integrins blocks L-LTP (67-71). Second, this category of molecules functions in transducing signals from one cell directly to another, allowing for direct presynaptic-to-postsynaptic communication. Third,

FIGURE 8 Signaling mechanisms utilized by integrins and cadherins. Integrins and cadherins, cell surface adhesions molecules, are prominent regulators of the cytoskeleton. Both families of molecules serve as cytoskeletal anchors and also regulate various signal transduction mechanisms. One role of activating these signal transduction processes is dynamically to regulate cytoskeletal structure and function. Via other pathways such as ERK, they might regulate local protein synthesis.

FIGURE 8 Signaling mechanisms utilized by integrins and cadherins. Integrins and cadherins, cell surface adhesions molecules, are prominent regulators of the cytoskeleton. Both families of molecules serve as cytoskeletal anchors and also regulate various signal transduction mechanisms. One role of activating these signal transduction processes is dynamically to regulate cytoskeletal structure and function. Via other pathways such as ERK, they might regulate local protein synthesis.

integrins couple to MAPK cascades in many cells to trigger local changes at the submembranous compartment, similar to what would be required for allowing them to communicate locally with the protein synthesis machinery at synapses. This local control of protein synthesis could allow for both pleiotropic effects on synaptic function and for self-reinforcing stimulation of their own synthesis. ERK coupling could be also used to communicate with the genome if that were necessary, for example for allowing the nucleus to integrate total metabolic needs for ongoing gene expression across a large number of synapses. Fourth, these molecules classically function in regulating cell morphology by controlling the cytoskeleton (reviewed in reference 72; see Figure 8); in fact, in most cells, the molecules are involved in maintaining long-lasting morphological differentiation.

Thus, while the case for integrins and cadherins in maintaining long-lasting synaptic change is somewhat circumstantial at this point, these molecules appear to have all the attributes necessary to serve as self-perpetuating signals at dendritic spines.

Thus, the basic model for maintaining perpetual synaptic strengthening is as follows. Specific integrin and cadherin molecules have their mRNA expression increased and targeted to the right synapse a la Arc. The synthesis and insertion of these molecules into a synapse (where they did not previously exist) leads to the following consequences: morphological changes and increased contact with the presynaptic terminal; enhanced ongoing AMPA receptor insertion; and localized self-stimulation of their ongoing synthesis and membrane insertion via stimulation of

ERK or other signal transduction pathways. This last point fulfils the requirement for generating a self-perpetuating signal that can outlast the triggering events and maintain the potentiated state indefinitely.

While I have presented this brief model using specific molecules as examples, please keep in mind that the model is more in the vein of a thought experiment than a specific hypothesis. It serves to illustrate some of the important functions that must be subserved in maintaining a life-long change in synaptic strength, especially the necessity of a self-perpetuating signal that can impinge upon synaptic transmission. I happen to think that the specific candidate molecules I outlined are appealing possibilities, but they are after all just candidates.

Was this article helpful?

0 0
Advanced Memory Techniques

Advanced Memory Techniques

A course in techniques and skills for mentalists, magicians and students. For students, improve your grades with less effort! But this book is also.... The ideal for any stage mentalist or magician by establishing credibility of amazing skills with an easy to follow instructional book on using the amazing power of your memory.

Get My Free Ebook

Post a comment