The genetic code is decoded in all living cells in two major "phases. First it is transcribed from the DNA into a specific messenger RNA (mRNA) molecule. This process is termed 'transcription'. The code is then read from the mRNA molecule to produce the copy of the corresponding protein. This process, which takes place in a complex cellular factory called ribosome, is termed 'translation'. In practice, a group of ribosomes ('polyri-bosome') performs the task on each mRNA. There are many complex steps in protein synthesis (Alberts et al. 1994), but most of the details need not concern us here. Rather, we will concentrate on those aspects of protein synthesis that have become a major focus of research in the biology of learning and memory.
It has all started with the notion that the brain develops with experience, and that, similarly to all "developmental processes, this involves growth. This is an old idea, which has become a tenet of modern neurobiology (Hebb 1949; "zeitgeist). Growth means, among others, synthesis of proteins, which are the major constituents of all living cells. It became evident that experience is indeed accompanied by alteration in protein synthesis in the nervous system (Hamberger and Hyden 1945). Eventually, the proposal was explicitly made that the formation of new memory traces depend on new protein synthesis (Monne 1949; Katz and Halstead 1950). The experimental proof followed (RNA synthesis—Dingman and Sporn 1961; protein synthesis— Flexner et al. 1963; Agranoff and Klinger 1964).
Over the years, two types of "methods have been used to demonstrate the role of protein synthesis in memory. The first method involved intervention with the metabolism of the brain, by the use of antibiotics that inhibit RNA or protein synthesis (inference of function from the dysfunction, "method). After the initial successes (Flexner et al. 1963; Agranoff and Klinger 1964), scores of laboratories became enthusiastically immersed in the new paradigm, injecting protein synthesis inhibitors into the brain of behaving animals ranging from goldfish to rodents. In spite of occasional worries about the exact target of the antibiotics (e.g. Kyriakis et al. 1994), the overwhelming conclusion was that protein synthesis during or up to a few hours after training, but not afterwards, is required for the "consolidation of long-term memory, but not for its "acquisition, short-term retention, or "retrieval once the long-term memory had been formed (Davis and Squire 1984; Dudai 1989; Rose 1995 b).
In the early days of the protein-synthesis-inhibition-of-memory experiments, the inhibitors were injected into wide areas of the brain, the whole brain, or even the whole body. This created a problem, because the experimenters could not prove that the effect is on neurons, and that the targets have anything to do with the specific memorized task. This difficulty was resolved only in the mid-80s. During that period, new preparations were developed, in which identified memory-subserving neurons are studied in isolation ("Aplysia), and cellular analogues of learning are analysed in brain tissue ("long-term potentiation). Another wave of protein-synthesis-inhibition-of-memory studies soon followed. These experiments finally proved that the initial conclusions about the role of RNA and protein synthesis in consolidation were basically correct (e.g. Montarolo et al. 1986, Linden 1996). The new preparations also permitted the effective use of a complementary approach, based on correlation rather than perturbation: identification of neuronal gene products that are induced by experience ("immediate early genes, "late response genes).
The combination of the two aforesaid approaches has led to the following textbook cellular "model of long-term memory: "stimuli that exceed a certain threshold or "coincide in appropriate combinations, operate on the relevant "receptors in the target synapses, and activate "intracellular signal transduction cascades. The latter activate constitutive transcription factors (TFs), and induce the transcription of additional TFs as well as other types of immediate early genes.1 The TFs trigger phases of gene expression, culminating in the induction of expression of late response genes. The products of the induced genes ultimately induce and embody persistent alterations in the synapses and neurons that encode the memory. Further, whereas neuronal alterations that are based on post-translational modifications are constrained by the limited life span of the modified protein molecule (which is commonly anywhere between a few minutes to a few weeks), the modulation of gene expression could render the alterations immune to the molecular turnover of individual protein molecules. The mechanisms of transcriptional regulation by extracellular signals in neurons are basically similar to those that operate in non-neuronal cells (Hill and Treisman 1995). This suggests, by the way, that the specificity of the "engram should be searched for at higher "levels of organization of the brain ("reduction).
The textbook model is, as usual, too simplistic. The complications are of two types. One type relates to the role of the newly synthesized proteins in the context of memory: Is it permissive, or causal? Do these proteins play a direct role in modifying the use-dependent "internal representation? This issue surfaces in several discussions in this book (e.g. "CREB, "homeostasis, "late response genes); it will not be further elaborated here. The other type of issue refers to the specificity of the process and to cellular economy. Isn't the modulation of cell-wide gene expression by only one or a few synapses remarkably nonparsimonious? And how would synaptic specificity be preserved, if at all?
The solution may lie in the intricacies of a multipha-sic mechanism, which is both synapse-specific and cell-wide (Figure 58) (Dudai and Morris 2000). It appears that the activated synapse is somehow 'tagged', possibly by post-translational modification of synaptic protein(s), or by reorganization of such proteins (Katz and Halstead 1950; Dudai 1989; Frey and Morris 1997; Martin et al. 1997a). This results in a new local synaptic configuration. It could also attract proteins from other parts of the cell. In addition, the stimulus activates constitutive transcription factors, such as CREB, and induces immediate early gene expression, some of which encode inducible transcription factors, others different types of proteins, including enzymes, cytoskeletal elements, and growth factors. The synapse itself contains the full translation apparatus and is capable of synthesizing proteins on location, from mRNA which is delivered from the nucleus (Steward and Levy 1982; Rao and Steward 1991; Weller and Greenough 1993; Martin et al. 1997a; Steward et al. 1998; Huber et al. 2000). These locally synthesized proteins strengthen the tagging of the synapse, and/or serve as retrograde messages, which travel to the cell body and inform the nucleus about the change (Casadio et al. 1999). This results in modulation of gene expression in the nucleus, and in the production of new mRNAs and proteins, that are funnelled to the tagged synapse. All in all, the process is hence assumed to involve intimate co-ordination between the synapse and the nucleus, which probably optimizes the exploitation of the metabolic resources of the neuron and the specificity of the long-term synaptic change (Dudai and Morris 2000).
We still have a long way to go before we fully understand the mechanisms and roles of synaptic tagging and step-wise synaptic consolidation. We are bound for surprises on the way. For example, it has been reported that the wave of protein synthesis that is triggered by a salient event in one synapse, is capable of affecting the registration of activity in adjacent synapses as well. This
Fig. 58 A highly simplified 'model of the role of protein synthesis in the production of long-term synaptic modifications that are assumed to subserve the 'acquisition and 'consolidation and contribute to the 'persistence of long-term memory. (a) Protein synthesis involves transcription in the nucleus and translation on ribosomes in the cytosol (the inner cellular space outside the nucleus). (b) A teaching stimulus ('teacher'), which is sufficiently salient to induce long-term memory, activates membrane receptors and their downstream 'intracellular signal transduction cascades. This results in the activation of constitutive transcription factors (CTFs), such as *CREB, and in the induction of the expression of 'immediate early genes, which encode inducible transcription factors as well as a variety of other proteins. Immediately after the stimulation, the activated 'synapse is tagged in a way that differentiates it from the nonstimulated synapses. The teaching stimulus also induces local protein synthesis in the synapse (the ribosomes in the synapse are not shown in (a) for simplicity). The proteins that are synthesized on location might contribute to the tagging of the activated synapse. They might also comprise or generate a retrograde signal, which travels to the nucleus. The expression of certain genes in the informed nucleus is now modulated by sets of TFs. (c) The newly synthesized proteins travel from the cell body to the activated, tagged synapse, and contribute to its lasting modification. Protein synthesis hence fulfils multiple roles at multiple sites and times after the training, and is controlled by both the activated synapse and the cell body and nucleus.
means that synapses are tuned to the history of each other (Frey and Morris 1997). This could subserve the cellular encoding of "context, "generalization, even "priming. Whether the synaptic phenomenology indeed contributes to the behavioural phenomena of context encoding, generalization or priming, is currently an intriguing yet unresolved issue.
Selected associations: Consolidation, CREB, Development, Immediate early genes, Late response genes
1On what transcription factors are, and on the distinction between constitutive and inducible TFs, see in 'immediate early genes.
Was this article helpful?