Cyclic AMP Signaling Has a Role in Long-Term Sensitization
In all forms of learning practice makes perfect. Repeated experience converts short-term memory into a long-term form. In Aplysia the form of long-term memory that has been most intensively studied is long-term sensitization. Like the short-term form, long-term sensitization of the gill-withdrawal reflex involves changes in the strength of connections at several synapses, including those between sensory and motor neurons. However, it also involves the growth of new synaptic connections.
Five spaced training sessions (or repeated applications of serotonin) over approximately 1 hour produce long-term sensitization and long-term synaptic facilitation lasting 1 or more days; continued spaced training over several days produces sensitization that persists for 1 or more weeks. Long-term sensitization, like the short-term form, requires protein phosphorylation that is dependent on increased levels of cAMP (Figure 66–6).
Long-term sensitization involves synaptic facilitation and the growth of new synaptic connections.
A. Long-term sensitization of the gill-withdrawal reflex of Aplysia following repeated tail shocks involves long-lasting facilitation of transmitter release at the synapses between sensory and motor neurons.
B. Long-term sensitization of the gill-withdrawal reflex leads to persistent activity of PKA, resulting in the growth of new synaptic connections. Repeated tail shock leads to more pronounced elevation of cAMP, producing long-term facilitation (lasting 1 or more days) that outlasts the increase in cAMP and recruits the synthesis of new proteins. This inductive mechanism is initiated by translocation of PKA to the nucleus (pathway 1), where PKA phosphorylates the transcriptional activator CREB-1 (cAMP response element binding protein 1) (pathway 2). CREB-1 binds cAMP regulatory elements (CRE) located in the upstream region of several cAMP-inducible genes, activating gene transcription (pathway 3). PKA also activates the mitogen-activated protein kinase (MAPK), which phosphorylates the transcriptional repressor CREB-2 (cAMP response element binding protein 2), thus removing its repressive action. One gene activated by CREB-1 encodes a ubiquitin hydrolase, a component of a specific ubiquitin proteasome that leads to the proteolytic cleavage of the regulatory subunit of PKA, resulting in persistent activity of PKA, even after cAMP has returned to its resting level (pathway 4). CREB-1 also activates the expression of the transcription factor C/EBP, which leads to expression of a set of unidentified proteins important for the growth of new synaptic connections (pathway 5).
The conversion of short-term memory into long-term memory, called consolidation, requires synthesis of messenger RNAs and proteins in the neurons in the circuit. Thus specific gene expression is required for long-term memory. The transition from short-term to long-term memory depends on the prolonged rise in cAMP that follows repeated applications of serotonin. This leads to prolonged activation of PKA, allowing the catalytic subunit of the kinase to translocate into the nucleus of the sensory neurons. It also leads indirectly to activation of a second protein kinase, the mitogen-activated protein kinase (MAPK), a kinase commonly associated with cellular growth (see Chapter 11). Within the nucleus the catalytic subunit of PKA phosphorylates and thereby activates the transcription factor CREB-1 (c AMP r esponse e lement b inding protein 1), which binds a promoter element called CRE (c AMP r ecognition e lement) (Figures 66–6 and 66–7).
Regulation of histone acetylation by serotonin, CREB-1, and CBP.
A. Under basal conditions the activator CREB-1 (here in complex with CREB-2) occupies the binding site for CRE (cAMP recognition element) within the promoter region of its target genes. In the example shown here, CREB-1 binds to the CRE within the C/EBP promoter. In the basal state CREB-1 binding is not able to activate transcription because the TATA box, the core promoter region responsible for recruiting RNA polymerase II (Pol II) during transcription initiation, is inaccessible because the DNA is tightly bound to histone proteins in the nucleosome.
B. Serotonin (5-HT) activates PKA, which phosphorylates CREB-1 and indirectly enhances CREB-2 phosphorylation by MAPK, causing the repressor to dissociate from the promoter. This allows CREB-1 to form a complex at the promoter with CREB binding protein (CBP). Activated CBP acetylates specific lysine residues of the histones, causing them to bind less tightly to DNA. Along with other changes in chromatin structure, acetylation facilitates the repositioning of the nucleosome that previously blocked access of the Pol II complex to the TATA box. This repositioning allows Pol II to be recruited to initiate transcription of the C/EBP gene. (TBP, TATA binding protein).
To turn on gene transcription, phosphorylated CREB-1 recruits a transcriptional coactivator, CREB-binding protein (CBP), to the promoter region. CBP has two important properties that facilitate transcriptional activation: it recruits RNA polymerase II to the promoter, and it functions as an acetyltransferase, adding acetyl groups to certain lysine residues on its substrate proteins. One of the most important substrates of CBP are the histone proteins, which are components of nucleosomes, the fundamental building blocks of chromatin. The histones contain a series of positively charged basic residues that strongly interact with the negatively charged phosphates of DNA. This interaction causes DNA to become tightly wrapped around the nucleosomes, much like string is wrapped around a spool, thereby preventing necessary transcription factors from accessing their gene targets.
The binding of CBP to CREB-1 leads to histone acetylation, which causes a number of important structural and functional changes at the nucleosome level. For example, acetylation neutralizes the positive charge of lysine residues in the histone tail domains, decreasing the affinity of histones for DNA. Also, specific classes of transcriptional activators can bind to acetylated histones and facilitate the repositioning of nucleosomes at the promoter region. Together these and other types of chromatin modifications serve to regulate the accessibility of chromatin to the transcriptional machinery, and thus enhance the ability of a gene to be transcribed. As we will see in Chapter 67, a mutation in the gene encoding CBP underlies Rubinstein-Taybi syndrome, a disorder associated with mental retardation.
The turning on of transcription by PKA also depends on its ability to indirectly activate the MAPK pathway (see Chapter 11). MAPK phosphorylation of the transcription factor CREB-2 relieves an inhibitory action of CREB-2 on transcription (Figure 66–6B). The combined effects of CREB-1 activation and relief of CREB-2 repression induces a cascade of new gene expression important for learning and memory (Figure 66–7).
The presence of both a repressor (CREB-2) and an activator (CREB-1) of transcription at the first step in long-term facilitation suggests that the threshold for long-term memory storage can be regulated. Indeed, we see in everyday life that the ease with which short-term memory is transferred into long-term memory varies greatly with attention, mood, and social context.
Two of the genes expressed in the wake of CREB-1 activation and the consequential alteration in chromatin structure are important in the early development of long-term facilitation. One is a gene for ubiquitin carboxyterminal hydrolase. The other is a gene for a transcription factor, CAAT box enhancer binding protein (C/EBP), a component of a gene cascade necessary for synthesizing proteins needed for the growth of new synaptic connections (Figure 66–6, 66–7).
The hydrolase, which facilitates ubiquitin-mediated protein degradation (see Chapter 3), helps enhance activation of PKA. Protein kinase A is made up of four subunits; two regulatory subunits inhibit two catalytic subunits (see Chapter 11). With long-term training and the induction of the hydrolase, approximately 25% of the regulatory subunits are degraded in the sensory neurons. As a result, free catalytic subunits can continue to phosphorylate proteins important for the enhancement of transmitter release and the strengthening of synaptic connections, including CREB-1, long after cAMP has returned to its basal level (Figure 66–6B). Formation of a constitutively active enzyme is therefore the simplest molecular mechanism for long-term memory. With repeated training a second-messenger kinase critical for short-term facilitation can remain persistently active for up to 24 hours without requiring a continuous activating signal.
The second and more enduring consequence of CREB-1 activation is the activation of the transcription factor C/EBP. This transcription factor forms both a homodimer with itself and a heterodimer with another transcription factor called activating factor. Together these factors act on downstream genes that trigger the growth of new synaptic connections that support long-term memory.
With long-term sensitization the number of presynaptic terminals in the sensory neurons in the gill-withdrawal circuit doubles (Figure 66–8). The dendrites of the motor neurons also grow to accommodate the additional synaptic input. Thus long-term structural changes in both post- and presynaptic cells increase the number of synapses.
Long-term habituation and sensitization involve structural changes in the presynaptic terminals of sensory neurons.
A. Long-term habituation leads to a loss of synapses, and long-term sensitization leads to an increase in number of synapses. When measured either 1 day (shown here) or 1 week after training, the number of presynaptic terminals (or boutons) relative to control levels is increased in sensitized animals and reduced in habituated animals. The drawings below the graph illustrate changes in the number of synaptic contacts. Swellings or varicosities on sensory neuron processes are presynaptic terminals. (Adapted, with permission, from Bailey and Chen 1983.)
B. Fluorescence images of the axon of a sensory neuron contacting a motor neuron in culture before (left) and 1 day after (right) five brief exposures to serotonin. The resulting increase in varicosities simulates the synaptic changes associated with long-term sensitization. Prior to serotonin application no presynaptic varicosities are visible in the outlined area (control). After serotonin the growth of several new varicosities is apparent (arrows), indicative of formation of new synapses. Boutons can be seen at the arrows in the right, some of which contain a fully developed zone, identified by the asterisk, or have small, immature active zones. Scale bar = 50 μm. (Reproduced, with permission, from Glanzman, Kandel, and Schacher 1990.)
Long-term habituation, in contrast, leads to pruning of synaptic connections. The long-term inactivation of the functional connections between sensory and motor neurons reduces the number of terminals of each sensory neuron by one-third (Figure 66–8A).
Long-Term Synaptic Facilitation Is Synapse Specific
A typical neuron in the mammalian brain makes 10,000 synapses with a wide range of target cells. It is therefore generally thought that long-term memory storage should be synapse specific—that is, only those synapses that actively participate in learning should be enhanced. However, the finding that long-term facilitation involves gene expression—which occurs in the nucleus, an organelle that is far removed from a neuron's synapses—raises some fundamental questions regarding information storage.
Is long-term memory storage indeed synapse specific, or do the gene products recruited during long-term memory storage alter the strength of every presynaptic terminal in a neuron? And if long-term memory is synapse specific, what are the cellular mechanisms that enable the products of gene transcription to selectively strengthen just some synapses and not others?
Kelsey Martin and her colleagues addressed these questions regarding long-term facilitation by using a cell culture system consisting of an isolated Aplysia sensory neuron with a bifurcated axon that makes separate synaptic contacts with two motor neurons. The sensory neuron terminals on one of the two motor neurons were activated by focal pulses of serotonin, thus mimicking the neural effects of a shock to the tail. When only one pulse of serotonin was applied, those synapses showed short-term facilitation. The synapses on the second motor neuron, which did not receive serotonin, showed no change in synaptic transmission (Figure 66–9).
Long-term facilitation of synaptic transmission is synapse-specific.
(Adapted, with permission, from Martin et al. 1997.)
A. The experiment uses a single presynaptic sensory neuron that contacts two postsynaptic motor neurons (A and B). The pipette on the left is used to apply five pulses of serotonin (5-HT) to a sensory neuron synapse with motor neuron A, initiating long-term facilitation at these synapses. The pipette on the right is used to apply one pulse of 5-HT to a sensory neuron synapse with motor neuron B, allowing this synapse to make use of (capture) new proteins produced in the cell body in response to the five pulses of 5-HT at the synapses with motor neuron A. The image at the right shows the actual appearance of the cells in culture.
B. 1. One pulse of 5-HT applied to the synapses with motor neuron A produces only short-term (10 min) facilitation of the excitatory postsynaptic potential (EPSP) in the neuron. By 24 hours the EPSP has returned to its normal size. There is no significant change in EPSP size in cell B. 2. Application of five pulses of 5-HT to the synapses with cell A produces long-term (24 hour) facilitation of the EPSP in that cell but no change in the size of the EPSP in cell B. 3. However, when five pulses of 5-HT onto the synapses with cell A are paired with a single pulse of 5-HT onto the synapses with cell B, cell B displays long-term facilitation and an increase in EPSP size after 24 hours.
When five pulses of serotonin were applied to the same synapses, those synapses displayed both short-term and long-term facilitation, and new synaptic connections were formed with the motor neuron. Again the synapses that did not receive serotonin showed no enhancement of synaptic transmission (Figure 66–9B). Thus both short-term and long-term synaptic facilitation are synapse specific and manifested only by those synapses that receive the modulatory serotonin signal.
But how are the nuclear products able to enhance transmission at certain synapses only? Are the newly synthesized proteins somehow targeted to only those synapses that receive serotonin? Or are they shipped out to all synapses but used productively for the growth of new synaptic connections only at those synapses that have been activated—or marked—perhaps by only a single pulse of serotonin?
To test this question Martin and her colleagues again selectively applied five pulses of serotonin to the synapses made by the sensory neuron onto one of the motor neurons. This time, however, the synapses with the second motor neuron were simultaneously activated by a single pulse of serotonin (which by itself produces only short-term synaptic facilitation lasting minutes). Under these conditions the single pulse of serotonin was sufficient to induce long-term facilitation and growth of new synaptic connections at the contacts between the sensory neuron and the second motor neuron. Thus application of the single pulse of serotonin onto the synapses at the second branch enabled those synapses to use the nuclear products produced in response to the five pulses of serotonin onto the synapses of the first branch, a process called capture.
These results suggest that newly synthesized gene products, both mRNAs and proteins, are delivered by a fast axonal transport mechanism to all the synapses of a neuron but are functionally incorporated only at synapses that have been tagged or marked by previous synaptic activity.
For a synapse to use the new proteins and mRNAs for long-term facilitation, it must first be marked by serotonin. Although one pulse of serotonin at a synapse is insufficient to turn on new gene expression in the cell body, it is sufficient to allow that synapse to make productive use of new proteins generated in the soma in response to the five pulses of serotonin at another synapse. This idea, developed by Martin and her colleagues for Aplysia and independently by Frey and Morris for the hippocampus in rodents, is called synaptic capture or synaptic tagging.
These findings raise the question, what is the nature of the synaptic mark that allows the capture of the gene products for long-term facilitation? When an inhibitor of PKA was applied locally to the synapses receiving the single pulse of serotonin, those synapses could no longer capture the gene products produced in response to the five pulses of serotonin (Figure 66–10). This indicates that phosphorylation mediated by PKA is required for capturing the long-term process.
Long-term facilitation requires both cAMP-dependent phosphorylation and local protein synthesis. (Adapted, with permission, from Casadio et al., 1999.)
A. Five pulses of serotonin (5-HT) are applied to the synapses on motor neuron A and a single pulse is applied to those of cell B. Inhibitors of protein kinase A (Rp-cAMPS) or local protein synthesis (emetine) are applied to synapses on cell B.
B. Rp-cAMPS blocks the capture of long-term facilitation completely at the synapses on neuron B. Emetine has no effect on the capture of facilitation or the growth of new synaptic connections measured 24 hours after 5-HT application, but by 72 hours it fully blocks synaptic enhancement. The outgrowth of new synaptic connections is retracted and long-term facilitation decays after 1 day if capture is not maintained by local protein synthesis. (Rp-cAMPS, Rp-diaster-eomer of adenosine cyclic 3′,5′-phosphorothioate.)
In the early 1980s Oswald Steward discovered that ribosomes, the machinery for protein synthesis, are situated locally at the synapse in addition to being present in the cell body. Martin examined the importance of local protein synthesis in long-term synaptic facilitation by applying a single pulse of serotonin together with an inhibitor of local protein synthesis onto one set of synapses while simultaneously applying five pulses of serotonin to the other set of synapses. Normally long-term facilitation and synaptic growth would persist for up to 72 hours in response to synaptic capture. In the presence of the protein synthesis inhibitor, synaptic capture could still generate long-term synaptic facilitation at the synapses exposed to only one pulse of serotonin for at least 24 hours (Figure 66–10B), but synaptic growth and facilitation at these synapses collapsed after 24 hours, indicating that the maintenance of learning-induced synaptic growth requires new local protein synthesis at the synapse.
Martin and her colleagues thus found that regulation of protein synthesis at the synapse plays a major role in controlling synaptic strength at the sensory-to-motor neuron connection in Aplysia. As we shall see in Chapter 67, local protein synthesis is also important for the later phases of long-term potentiation of synaptic strength in the hippocampus.
These findings indicate there are two distinct components of synaptic marking in Aplysia. The first component, lasting about 24 hours, initiates long-term synaptic plasticity and synaptic growth, requires transcription and translation in the nucleus, and recruits local PKA activity, but does not require local protein synthesis. The second component, which stabilizes the long-term synaptic change after 72 hours, requires local protein synthesis at the synapse. How might this local protein synthesis be regulated?
Long-Term Facilitation Requires a Prion-Like Protein Regulator of Local Protein Synthesis for Maintenance
The fact that mRNAs are translated at the synapse in response to marking of that synapse by one pulse of serotonin suggests that these mRNAs may initially be dormant and under the control of a regulator of translation recruited by serotonin. Joel Richter found that in Xenopus (frog) oocytes the maternal mRNAs have a short tail of adenine nucleotides, poly(A), at their 3′ end and are silent until activated by the cytoplasmic polyadenylation element binding protein (CPEB), which binds to a site on mRNAs and recruits poly(A) polymerase, leading to the elongation of the poly(A) tail. Kausik Si and his colleagues found that serotonin increases the local synthesis of a novel, neuron-specific isoform of CPEB in Aplysia sensory neuron processes (Figure 66–11). The induction of CPEB is independent of transcription but requires new protein synthesis. Blocking CPEB locally at an activated synapse blocks the long-term maintenance of synaptic facilitation at the synapse but not its initiation and maintenance for 24 hours.
CPEB may be a self-perpetuating switch of protein synthesis at axon terminals and synapse-specific growth.
According to this model (based on Bailey, Kandel, and Si, 2004) five pulses of serotonin (5-HT) set up a signal that goes back to the nucleus to activate synthesis of mRNA. Newly transcribed mRNAs and newly synthesized proteins made in the cell body are then sent to all terminals by fast axonal transport. However, only those terminals that have been marked by exposure to at least one pulse of serotonin can use the proteins productively to grow new synapses and produce long-term facilitation. The marking of a terminal involves two components: (1) protein kinase A (PKA), which is necessary for the immediate synaptic growth initiated by the proteins transported to the terminals, and (2) phosphoinositide 3 kinase (PI3 kinase), which initiates the local translation of mRNAs required to maintain synaptic growth and long-term facilitation past 24 hours. Some of the mRNAs at the terminals encode CPEB, a regulator of local protein synthesis. In the basal state CPEB is thought to exist in a largely inactive conformation as a soluble monomer that cannot bind to mRNAs. Through some as yet unspecified mechanism activated by serotonin and PI3 kinase, some copies of CPEB convert to an active conformation that forms aggregates. The aggregates function like prions in that they are able to recruit monomers to join the aggregate, thereby activating the monomers. The CPEB aggregates bind the cytoplasmic polyadenylation element (CPE) site of mRNAs. This binding recruits the poly(A) polymerase machinery and allows poly(A) tails of adenine nucleotides (A) to be added to dormant mRNAs. The polyadenylated mRNAs can now be recognized by ribosomes, allowing the translation of these mRNAs to several proteins. For example, in addition to CPEB, this leads to the local synthesis of N-actin and tubulin, which stabilize newly grown synaptic structures.
How might CPEB stabilize the late phase of long-term facilitation? Most biological molecules have a relatively short half-life (hours to days) compared to the duration of memory (days, weeks, even years). How then can the learning-induced alterations in the molecular composition of a synapse be maintained for such a long time? Most hypotheses rely on some type of self-sustained mechanism that can somehow modulate synaptic strength and synaptic structure.
Si and his colleagues found that the neuronal isoform of Aplysia CPEB indeed appears to have self-sustaining properties that resemble those of prion proteins. Prions were discovered by Stanley Prusiner, who demonstrated that these proteins were the causative agents of Jacob-Creutzfeldt disease, a terrible neurodegenerative human disease, and of mad cow disease. Prion proteins can exist in a soluble form and an aggregated form that is capable of self-perpetuation. Aplysia CPEB also has two conformational states, a soluble form that is inactive and an aggregated form that is active. This switch involves an N-terminal domain of CPEB that is rich in glutamine, similar to prion domains in other proteins.
In a naïve synapse CPEB exists in the soluble, inactive state, and its basal level of expression is low. However, in response to serotonin the local synthesis of CPEB increases until a threshold concentration is reached that switches CPEB to the aggregated, active state, which is then capable of activating the translation of dormant mRNAs. Once the active state is established, it becomes self-perpetuating by recruiting soluble CPEB to aggregates. Dormant mRNAs, made in the cell body and distributed cell-wide, are translated only at synapses with active CPEB.
Because the activated CPEB is self-perpetuating, it could promote a self-sustaining, synapse-specific long-term molecular change and provide a mechanism for the stabilization of learning-related synaptic growth and the persistence of memory storage (Figure 66–11). This proposed mechanism, albeit self-perpetuating, is different from conventional prion mechanisms, which are pathogenic (the aggregated state of most prion proteins causes cell death). By contrast, CPEB is a new form of a prion-like protein. It is a functional prion; the active self-perpetuating form of the protein does not kill cells but rather has an important physiological function.