TY - JOUR
AU - Geinisman, Yuri
AB - Abstract An important problem in the neurobiology of memory is whether cellular mechanisms of learning and memory include the formation of new synapses or the remodeling of existing ones. To elucidate this problem, numerous studies have examined alterations in the number and structure of synapses following behavioral learning and hippocampal long-term potentiation (LTP), which is viewed as a synaptic model of memory. The data reported in the literature and obtained in this laboratory are analyzed here to evaluate what kind of structural modification is likely to account for synaptic plasticity associated with learning and memory. It has been demonstrated that LTP induction elicits the formation of additional synapses between activated axon terminals and newly emerging dendritic spines. Similarly, some forms of learning have been shown to increase the number of synapses. Although many ultrastructural studies examining the effect of LTP or learning failed to find a change in total synapse number, this population measure might not detect an increase in a small proportion of synapses established by activated terminals. LTP and learning have also been shown to induce a remodeling of synapses. This process is proposed to involve the transformation of certain synaptic subtypes into more efficacious ones, including the conversion of ‘silent’ synapses into functional synapses. It appears, therefore, that cellular mechanisms of learning and memory are likely to include both synaptogenesis and synapse remodeling. Introduction It has long been believed that changes in the number or structure of synapses may represent a substrate of memory formation following learning. Such changes were postulated to involve the establishment of new synaptic connections or a remodeling of existing synapses that makes them more efficacious (Ramón y Cajal, 1893; Tanzi, 1893). Since a number of previous studies have reported increases in the synaptic numerical density after acquisition of new behaviors [reviewed by Greenough and Bailey and by Bailey and Kandel (Greenough and Bailey, 1988; Bailey and Kandel, 1993)], the existing dogma is that learning involves net synaptogenesis. The results of our studies, however, are inconsistent with this dogma. They show that neither the N-methyl-d-aspartate (NMDA) receptor-mediated form of hippocampal long-term potentiation (LTP), which is widely regarded as a synaptic model of memory (Bliss and Collingridge, 1993), nor hippocampus-dependent associative learning is accompanied by a net gain in total synapse number (Geinisman et al., 1991, 1996a, 2000). The aim of the present report is to summarize the findings of our work and to compare them with those of others in order to evaluate whether cellular mechanisms of learning and memory are likely to include synaptogenesis or synapse remodeling, or both. Structural Synaptic Alterations Associated with the NMDA Receptor-mediated Form of Hippocampal LTP The essence of LTP is a remarkably persistent enhancement of synaptic responses resulting from brief, repetitive activation of an excitatory afferent monosynaptic pathway by high-frequency trains of electrical pulses (Bliss and Gardner-Medwin, 1973; Bliss and Collingridge, 1993). LTP has been studied most intensively at excitatory hippocampal synapses formed by Schaffer collaterals on CA1 pyramidal cells and by perforant path fibers on dentate gyrus granule cells. The NMDA receptor-mediated form of LTP observed at these synapses is characterized by such basic properties as synapse specificity, cooperativity and associativity [for a review see (Bliss and Collingridge, 1993)]. The latter property appears to be especially pertinent to the postulated significance of hippocampal LTP as a synaptic model of memory. Associativity means that a weak synaptic input, which fails to induce LTP when stimulated alone, can evoke LTP if it is stimulated concurrently with a separate convergent input. This provides a possible physiological mechanism for encoding association between two stimuli that nearly concurrently activate the same population of target neurons. It is generally assumed, therefore, that the NMDA receptor-mediated form of hippocampal LTP may represent a cellular analogue of associative learning and long-term memory. Since the LTP phenomenon is extremely durable, it is likely to be supported by changes in the number and/or structure of synapses that may model structural synaptic alterations resulting from behavioral learning. The validity of this suggestion will be evaluated in the present review by analyzing the available data regarding structural synaptic modifications that accompany hippocampal LTP and by comparing them with those that have been shown to occur as a consequence of memory formation after behavioral learning. Stability of the Total Number of Synapses following LTP The results of our LTP studies were reported in previous publications (Geinisman et al., 1991, 1993, 1996a). Since these publications provide a detailed description of experimental procedures used and of protocols for tissue preparation and synapse quantitation, only a brief account is given below. Young adult rats were chronically implanted with stimulating electrodes into the right medial perforant path and recording electrodes into the hilus of the ipsilateral dentate gyrus. Potentiated animals received medial perforant path stimulation that was carried out on each of four consecutive days and consisted of fifteen 20 ms bursts of 400 Hz pulses repeated at 0.2 Hz. The extent of potentiation was assessed with single pulses delivered at 0.2 Hz. A similar stimulation procedure was used for control rats except that all stimuli were delivered at 0.2 Hz. Each control rat was matched with a respective potentiated rat according to the total amount of current delivered. For electron microscopy, the animals were perfused transcardially with paraformaldehyde-glutaraldehyde fixatives either 1 h (seven control and seven potentiated rats) or 13 days (nine control and nine potentiated rats) after the last stimulation to study the effects of the induction or the maintenance phase of LTP, respectively. Tissue was treated with OsO4, embedded in Araldite and used to obtain serial ultrathin sections that were stained with uranyl acetate and lead citrate. Synapses were analyzed in the middle molecular layer (MML) of the dentate gyrus since it was a synaptic field of the stimulated axons. Estimates of the number of MML synapses per postsynaptic granule cell were obtained with the unbiased method of double disector (Gundersen, 1986). A major finding of our studies is that the induction and maintenance phases of LTP are not associated with significant changes in the total number of synapses per postsynaptic neuron (Figure 1). This finding is consistent with the results of most electron microscopic studies showing that the total number of synapses per unit area and volume of tissue or per postsynaptic neuron remains stable after LTP induction in the dentate gyrus (Wenzel and Matthies, 1985; Desmond and Levy, 1986, 1990; Schuster et al., 1990; Weeks et al., 1998) and CA1 subfield (Lee et al., 1980, 1981; Chang and Greenough, 1984; Chang et al., 1991; Sorra and Harris, 1998). Similarly, conventional and two-photon confocal microscopy of these hippocampal subregions has not revealed LTP-induced alterations in the total number of spines per unit length of parent dendrites (Hosokawa et al., 1995; Andersen and Soleng, 1998; Engert and Bonhoeffer, 1999). In contrast, electron microscopic studies of the dentate gyrus performed in the Andersen laboratory have reported LTP-associated increases in the density of spines attached to dendritic segments and in the proportion of bifurcated spines (Trommald et al., 1990, 1996; Andersen and Soleng, 1998). These studies, however, have serious drawbacks. Potentiated groups consisted of only two (Trommald et al., 1996), three (Trommald et al., 1990) or four (Andersen and Soleng, 1998) rats. A total of 19 or 24 dendritic segments were examined in four potentiated and six control animals, respectively (Andersen and Soleng, 1998). A total of 20 or six bifurcated spines were sampled from the entire potentiated and control material, respectively (Trommald et al., 1996). It is not clear whether such limited sampling resulted in representative estimates of spine number. In any event, the strikingly consistent observation of the electron microscopic studies available in the literature is that the total number of synapses is not changed by hippocampal LTP. Synapse Restructuring Associated with the Induction Phase of LTP Taking into account the data described above, it was necessary to explore the possibility that the relative abundance of some morphological varieties of synapses is selectively changed as a consequence of LTP induction. Examination of synapses in the MML of the rat dentate gyrus showed that they are morphologically heterogeneous (Geinisman et al., 1987b, 1991; Geinisman, 1993). Two major synaptic categories are represented by axodendritic synapses involving dendritic shafts and axospinous ones involving dendritic spines. The axodendritic category consists of symmetrical and asymmetrical synaptic subtypes that have presynaptic density and postsynaptic density (PSD) of equal thickness or exhibit a relatively thicker PSD, respectively. The axospinous category includes perforated synapses showing a discontinuous PSD profile in some serial sections and non-perforated ones exhibiting continuous PSD profiles in all consecutive sections. Perforated axospinous synapses may be further subdivided into several distinct morphological subtypes that are described and illustrated below. Nonperforated axospinous synapses have typically much smaller overall dimensions than perforated ones. There are, however, some atypical nonperforated synapses that are unusually large. Their morphological characteristics, with the exception of a continuous PSD, are similar to those of perforated synapses. The parameter of number per postsynaptic neuron was estimated separately for all these morphological varieties of synapses. Increase in the Ratio of Perforated to Nonperforated Axospinous Synapses following LTP Induction Trends toward an increase (+17%) in the number of perforated axospinous synapses and a decrease (–14%) in the number of nonperforated axospinous synapses were observed in the MML of potentiated rats examined 1 h following the last high-frequency stimulation (Geinisman et al., 1991). Although both trends were not statistically significant, they suggested that LTP induction could change the ratio of perforated to nonperforated synapses. This ratio was, therefore, estimated and found to be significantly and markedly (by 41%) increased as a consequence of LTP induction (Figure 2). A similar observation was made in studies that used organotypic slice cultures of the hippocampus to examine the effect of LTP induction on calcium-labeled and, hence presumably activated, CA1 synapses (Buchs and Muller, 1996; Toni et al., 1999). The proportion of labeled perforated synapses was found to be significantly increased 5–30 min after potentiating stimulation. These findings appear to be of special interest in view of the recent discovery of ‘silent’ hippocampal synapses. Electro-physiological experiments have revealed that a proportion of hippocampal synapses exhibit functional NMDA receptors but not functional AMPA receptors (Issaks et al., 1995; Liao et al., 1995). This makes such synaptic contacts postsynaptically silent in that they do not generate a synaptic response to a release of a neurotransmitter at resting membrane potentials because of the blockade of NMDA receptor channels by extracellular magnesium. Correspondingly, immunocytochemical studies have provided evidence for the existence of hippocampal synapses that exhibit only NMDA, but not AMPA, receptor immunoreactivity (Desmond and Weinberg, 1998; He et al., 1998; Petralia et al., 1999; Takumi et al., 1999; Racca et al., 2000). It is a lack of AMPA receptors and not their inactive state that accounts for this phenomenon (Nusser et al., 1998; Takumi et al., 1999). Silent synapses acquire AMPA-type responses after LTP induction (Isaac et al., 1995; Liao et al., 1995), indicating that they may be transformed into active synapses. The proportion of perforated axospinous synapses expressing AMPA receptors is at least twice as large as that of nonperforated ones (Desmond and Weinberg, 1998). Therefore, the observed increase in the ratio of perforated to nonperforated synapses may be a morphological manifestation of the conversion of silent synaptic contacts into functional ones during the induction phase of LTP. Selective Increase in the Number of Perforated Axospinous Synapses with Multiple, Completely Partitioned Transmission Zones after LTP Induction The induction phase of LTP was also found by us to be associated with another structural synaptic modification (Geinisman et al., 1993). Those perforated axospinous synapses that have multiple, completely partitioned transmission zones were markedly increased in number (by 53%) in potentiated rats relative to control animals at 1 h after the last stimulation (Figure 3). This change was highly selective since no other synaptic subtype, including axodendritic asymmetrical synapses (Figure 3), exhibited a significant change in number at the time point examined. Perforated axospinous synapses with multiple transmission zones are distinguished by specific morphological features. Figure 4 presents electron micrographs of such a synapse. The presynaptic axon terminal is seen in all sections as a single profile. The postsynaptic spine head exhibits profiles of finger-like extensions, or spinules, which emerge from the spine head at discontinuities in the perforated PSD and invaginate the axon terminal. A three-dimensional reconstruction of this synapse reveals that its axon terminal has three protrusions (Figure 5a). The surface of the spine head facing the axon terminal is concave, and the cavity of the postsynaptic element is divided into three compartments (Figure 5b), each one being filled with a corresponding axon terminal protrusion. The compartments are bordered by two spine partitions identifiable in single sections as profiles of spinules. These partitions completely separate three transmission zones from each other. Each transmission zone is formed presynaptically by a separate axon terminal protrusion and delineated postsynaptically by a separate PSD segment. Three distinct PSD segments (Figure 5c) mark the location of the transmission zones. The unique structural characteristics of such synapses suggest that they may be especially efficacious. Their multiple transmission zones may function as independent units, provided that there is a diffusion barrier in the synaptic cleft and that each separate PSD segment is associated with an activated or newly inserted receptor cluster (Edwards, 1995). Under these conditions, an amplification of impulse transmission would be expected to occur. Following LTP induction via perforant path stimulation, each granule cell acquires some 70 additional axospinous synapses with multiple, completely partitioned transmission zones (Figure 3). Activation of ~400 axospinous synapses by perforant path stimulation is sufficient to evoke a response of a granule cell in the dentate gyrus (McNaughton et al., 1981). A substantial and selective increase in the number of presumably more efficacious axospinous synapses with multiple transmission zones may represent a structural substrate of the marked augmentation of synaptic responses which characterizes the induction phase of LTP. Other electron microscopic studies have also shown that the effects of LTP induction on synaptic morphology include a selective increase in the numerical density of only certain synaptic subtypes. Such a change was found to involve synapses on dendritic spines that exhibit a concave configuration (Desmond and Levy, 1986), spinules (Schuster et al., 1990) or a perforated PSD (Buchs and Muller, 1996; Toni et al., 1999). These observations appear to be compatible with our data because axospinous synapses with multiple, completely partitioned transmission zones are located on concave spines, have spine partitions and are characterized by a perforated PSD of the segmented variety. In the CA1 subfield, however, an increase in the numerical density after LTP induction was also reported to selectively involve synapses on dendritic shafts and sessile spines (Lee et al., 1980, 1981; Chang and Greenough, 1984; Chang et al., 1991). Therefore, the data available in the literature suggest that there may be some differences between the dentate gyrus and CA1 subfield with respect to the pattern of synapse remodeling associated with LTP induction. Model of Synapse Restructuring during the Induction Phase of LTP It has been proposed that LTP may elicit the conversion of some morphological subtypes of synapses into others (Chang and Greenough, 1984; Desmond and Levy, 1986, 1990). The existence of distinct morphological subtypes of axospinous synapses (Geinisman, 1993) is consistent with this idea and suggests a hypothetical model of synapse restructuring that may account for synaptic plasticity associated with the induction phase of LTP (Figure 6). According to this model, LTP induction initiates a sequence of structural synaptic modifications which commences with the enlargement of typical small nonperforated synapses (Figure 6a) and their conversion into atypically large nonperforated ones (Figure 6b). This is followed by the consecutive formation of perforated synapses that have initially a focal spine partition with a fenestrated PSD (Figure 6c), then a sectional partition with a horseshoe-shaped PSD (Figure 6d) and finally a complete partition(s) with a segmented PSD (Figure 6e). Synapses of the latter subtype have multiple (2–4) transmission zones instead of only a single one, as is usual, and an increase in their number after LTP induction may result in an augmentation of synaptic transmission. The proposed model implies that an increase in the number of axospinous synapses with multiple, completely partitioned transmission zones should be associated with a corresponding loss of other synaptic subtypes. Although a precise correspondence of this kind was not observed, axospinous nonperforated synapses exhibited a trend toward a decrease in numbers (Geinisman et al., 1991). This suggests that additional perforated synapses having multiple, completely partitioned transmission zones may evolve from nonperforated axospinous synapses. There are, however, recent reports indicating that LTP induction may be followed not only by a restructuring of synapses, but also by synaptogenesis. That this, in fact, may be the case is suggested by an elegant study (Engert and Bonhoeffer, 1999) that analyzed apical dendrites of CA1 pyramidal neurons in organotypic slice cultures using two-photon microscopy in combination with a local superfusion technique. The latter approach allowed the investigators to block transmitter release (by a solution containing a high concentration of cadmium and a low concentration of calcium) everywhere except in a small area (~30 μm in diameter) that was superfused by normal extracellular solution. Due to this, the site of synaptic activation and LTP induction was restricted to a small region of the postsynaptic dendrite. Continuous two-photon imaging of activated regions of the postsynaptic dendrites demonstrated the appearance of newly formed spines no earlier than 30 min after LTP induction. While new spines usually appeared within the superfusion area, a disappearance of spines occurred predominantly outside the superfusion zone. Although the total number of spines per unit length of activated dendritic segments was not altered by LTP, the extent of new spine appearance significantly correlated with the degree of synaptic enhancement. Moreover, this structural change did not occur when LTP induction was pharmacologically blocked, which proves its specific association with LTP. The emergence of new spines caused by LTP may be followed by the formation of synapses involving them. This suggestion was substantiated by an electron microscopic study of hippocampal organotypic slice cultures (Toni et al., 1999). In this study, calcium labeling was used to identify CA1 stratum radiatum synapses that were probably activated by potentiating stimulation. It was shown that the proportion of labeled multiple synapse boutons, which form mature synapses on two or more spines, was markedly increased at 45–120 min after LTP induction. In the vast majority of cases, postsynaptic spines contacting a labeled multiple synapse bouton originated from the same dendrite. Because these changes were not observed when non-labeled synapses were examined or when LTP induction was pharmacologically blocked, they were interpreted as a manifestation of LTP-induced synaptogenesis. Another important aspect of the study by Toni et al. (Toni et al., 1999) is that its results indicate that the processes of synapse restructuring and synaptogenesis may be complementary to one other in supporting synaptic enhancement after LTP induction. The authors observed a gradual increase in the proportion of labeled perforated synapses at 5, 15 and 30 min after cessation of potentiating stimulation. Then the proportion of these synapses decreased at 45 min and reached control values at 60 min. On the other hand, the proportion of labeled multiple synapse boutons remained stable at 5–30 min after stimulation. An initial increase in this measure was detected at 45 min and was followed by a threefold increase at 60 min. It is possible that the formation of additional perforated synapses observed early after LTP induction may be due to their rapid remodeling from existing nonperforated synapses. A synapse restructuring which rapidly leads to the formation of presumably more efficacious synapses with perforated PSDs may provide a transitory structural substrate of synaptic enhancement over a period of time necessary for synaptogenesis to occur and for newly formed synapses to mature. It is also conceivable, however, that the formation of new axospinous synapses may be followed by their remodeling from small nonperforated synapses into large perforated ones, including those with multiple, completely partitioned transmission zones. Synapse Restructuring Associated with the Maintenance Phase of LTP Both the ratio of perforated to nonperforated axospinous synapses and the number of axospinous synapses with multiple, completely partitioned transmission zones returned to control levels in potentiated rats examined by us 13 days after the last high frequency stimulation (Figures 2 and 7). These observations are consistent with those of Toni et al. (Toni et al., 1999), who also found that an increase in the proportion of perforated axospinous synapses, which occurred early after LTP induction, was only transitory. Additionally, our data showed that the maintenance phase of LTP was accompanied by an increase in the number of synaptic contacts that involved a subtype of axodendritic, rather than axospinous, synapses (Geinisman et al., 1996a). Selective Increase in the Number of Asymmetrical Axodendritic Synapses during LTP Maintenance Only asymmetrical axodendritic synapses were significantly, though moderately (by 28%), increased in number in potentiated rats at the 13 day post-stimulation interval (Figure 7). The number of other synaptic subtypes was not significantly changed during LTP maintenance. The pattern of synaptic plasticity which characterizes the maintenance phase of LTP includes two basic phenomena: a decay of the maximum degree of synaptic enhancement that is observed during the LTP induction phase and, simultaneously, the retention of a lesser degree of synaptic augmentation for a relatively long period of time. The results of our experiment suggest that the two phenomena may be accounted for by different structural synaptic modifications. At the 13 day test interval following the last high-frequency stimulation, the slope of the excitatory postsynaptic potential (EPSP) was still significantly potentiated relative to baseline, but only by 51 ± 9%, whereas the extent of the EPSP slope potentiation at the 1 h interval reached 111 ± 25% in the same animals (Geinisman et al., 1996a). The decay of potentiation of synaptic responses during the maintenance phase of LTP may be a consequence of a loss of additional axospinous synapses with multiple transmission zones that were acquired soon after LTP induction (compare Figures 3 and 7). The retention of a low level of synaptic enhancement during the maintenance phase of LTP may be supported by a moderate increase in the number of asymmetrical axodendritic synapses (Figure 7). The latter are supposed to be excitatory in function and to have a relatively high strength due to their strategic location directly on dendritic shafts, rather than on spines. Moreover, the observed addition of some 30 asymmetrical axodendritic synapses per granule cell (Figure 7) resulted in a 1% increase in the total number of synaptic contacts involving each granule cell in the MML. Activation of only 1–5% of the total number of synapses in the MML is required to evoke granule cell discharge (McNaughton et al., 1981). Thus, the magnitude of the increase in the number of asymmetrical axodendritic synapses during LTP maintenance appears to be sufficient to exert a measurable facilitating effect on the amplitude of synaptic responses elicited from the population of dentate granule cells. Possible Relationship between Structural Synaptic Modifications Associated with Induction and Maintenance Phases of LTP Although the increases in the number of synapses observed following LTP induction and during LTP maintenance involve different synaptic subtypes, these changes may be related to each other. Such a possibility is suggested by the existence of a special subtype of perforated synapse that appears to be transitional between axospinous and axodendritic ones. An example of synapse belonging to this subtype is demonstrated in Figure 8. In some serial sections (Figure 8a,b,c) the postsynaptic element of this synapse can be unequivocally identified as a spine since it contains a characteristic floccular material, lacks mitochondria and microtubules, and is associated with a spine apparatus. The postsynaptic spine has no neck, and it is seen in the following sections (Figure 8d,e) to be retracted into the parent dendrite so that its upper surface becomes continuous with that of the dendrite. At this level, the postsynaptic element is represented by a dendritic shaft, rather than by a spine. Thus, the presynaptic terminal forms a perforated asymmetrical synapse that has a transmission zone extending from the spine (Figure 8b) to its parent dendrite (Figure 8e). It is conceivable that, under certain conditions, the postsynaptic spine may be completely retracted into the parent dendrite, and the perforated asymmetrical synapse formerly associated with the spine becomes an axodendritic one. In fact, some asymmetrical axodendritic synapses of the MML do have a perforated PSD (Geinisman et al., 1987a), and they may evolve from perforated axospinous synapses. Model of Synapse Restructuring during LTP Maintenance Our data have suggested a hypothetical model of structural synaptic modifications that may be responsible for synaptic plasticity characteristic of the maintenance phase of LTP (Figure 9). According to this model, some additional axospinous synapses with multiple, completely partitioned transmission zones (Figure 9a), which were formed during LTP induction, are converted into asymmetrical axodendritic synapses during LTP maintenance. In the process of this restructuring, the postsynaptic spine loses its neck (Figure 9b) and is gradually retracted into the parent dendrite (Figure 9c) until it levels with the dendritic surface. Consequently, the perforated axospinous synapse that involved the spine becomes an asymmetrical axodendritic synapse (Figure 9d). Its perforated PSD may then be remodeled into a nonperforated one (Figure 9e). Other additional axospinous synapses with multiple transmission zones are proposed to be transformed into typical nonperforated ones (Figure 9j). Such restructuring may involve the consecutive formation of axospinous perforated synapses that lack spine partitions and exhibit a segmented (Figure 9f), horseshoe-shaped (Figure 9g) and fenestrated (Figure 9h) PSD. Final steps include the formation of atypical (Figure 9i) and then typical (Figure 9j) nonperforated axospinous synapses. These structural synaptic modifications may underlie both the decay of potentiation (the sequence from a through b, c and d to e) and the sustained retention of a low level of synaptic enhancement (the sequence from a through f, g, h and i to j) which are characteristic of the maintenance phase of LTP. It cannot be ruled out that the observed increase in the number of asymmetrical axodendritic synapses might result, at least in part, from synaptogenesis. However, it is not known if the process of synaptogenesis remains active during LTP maintenance. Future experiments are necessary to clarify this issue. Synapse Restructuring Associated with Behavioral Learning Search for Learning-induced Changes in Synapse Number A number of electron microscopic studies available in the literature report that the numerical density of synapses is increased in relevant areas of the vertebrate brain as a consequence of learning [reviewed by Greenough and Bailey and by Bailey and Kandel (Greenough and Bailey, 1988; Bailey and Kandel, 1993)]. We attempted to extend these observations by utilizing a stereological method for obtaining unbiased estimates of total synapse number in a brain region (Geinisman et al., 1996b). The behavioral paradigm of trace eyeblink conditioning was used because a brain region crucial for mediating this form of associative learning has been identified. Lesions of the hippocampus have been shown to prevent acquisition of the trace eyeblink conditioned response in rabbits (Solomon et al., 1986; Moyer et al., 1990). Moreover, electrophysiological studies have also demonstrated that trace eyeblink conditioning is followed by increases in the synaptic responsiveness and postsynaptic excitability of pyramidal neurons in the CA1 subfield of the rabbit hippocampus (de Jonge et al., 1990; Moyer et al., 1996; Power et al., 1997; Geinisman et al., 2000). These learning-induced alterations in neuronal function are intrinsic to the hippocampus since they have been observed in hippocampal slices isolated from the rest of the brain. In our experiment, which was described in detail elsewhere (Geinisman et al., 2000), young adult rabbits were trained in pairs and received either trace eyeblink conditioning or pseudo-conditioning. Conditioned rabbits were given daily 80 trial sessions to a criterion of 80% conditioned responses in a session. During each trial, the conditioned stimulus (tone) and the unconditioned stimulus (corneal airpuff) were presented with a stimulus-free, or trace, interval of 500 ms. Control rabbits were pseudoconditioned by equal numbers of random presentations of the same stimuli. Nine pairs of conditioned and pseudo-conditioned animals were examined. Brain tissue was taken for morphological analyses 24 h after the last training session. Synapses were quantified in the CA1 stratum radiatum. The results showed that the mean total number (± SE) of synapses in this layer was virtually the same in the groups of conditioned (23250 ± 915 synapses × 106) and pseudo-conditioned (23267 ± 869 synapses × 106) animals. Similarly, separate quantitative analyses of various synaptic subtypes revealed that their total numbers did not change significantly after trace eyeblink conditioning. Our data are in agreement with the observations of those earlier ultrastructural studies that found no evidence for a learning-related alteration in total synapse number based on the estimation of synaptic numerical density. The latter parameter remained stable in the pertinent regions of the vertebrate brain examined under a variety of experimental conditions: in the visual cortex of rabbits following visual discrimination training (Vrensen and Nunes Cardozo, 1981); in the hyperstriatum ventrale nucleus of chicks after imprinting (Bradley et al., 1981; Horn et al., 1985); in the dentate gyrus of rats following one-way active avoidance conditioning (Van Reempts et al., 1992); and in the dentate gyrus or CA1 subfield of rats after spatial learning in the Morris water maze (Rusakov et al., 1997). These results, taken together, are consistent with the notion that the formation of long-term memories following learning of new behaviors may not necessarily involve a net synaptogenesis. Work from the laboratories of Rakic and Goldman-Rakic supports this idea [for a review see (Bourgeois et al., 1999)]. Their estimates of synaptic numerical density in five major neocortical areas of the macaque monkey brain reveal no ultrastructural sign of net synaptogenesis over the entire period of adulthood in spite of continuous acquisition and accumulation of long-term memories. In marked contrast, other electron microscopic studies reported estimates of synaptic numerical density that are indicative of a learning-induced increase in the total number of synapses. A change of this kind was found to occur in pertinent brain regions of different vertebrate species as a consequence of various forms of learning: in the subfield CA1 of rats after brightness discrimination conditioning in a Y-maze (Wenzel et al., 1980); in the robustus archistralis nucleus of canaries following acquisition of male-like singing behavior by females treated with testosterone (DeVoogd et al., 1985); in the chick paleostriatal complex after passive avoidance conditioning (Stewart et al., 1987; Hunter and Stewart, 1989; Doubell and Stewart, 1993; Stewart and Rusakov, 1995); and in the motor and cerebellar cortices of rats after motor skill learning (Black et al., 1990; Kleim et al., 1996, 1997). Additionally, a quantitative electron microscopic analysis of dendritic spines revealed that their total number was increased in the rat dentate gyrus after the learning of a passive avoidance task (O'Malley et al., 1998). Similar results were obtained with the aid of confocal microscopy. The number of spines was shown to increase per unit length of basal, but not apical, dendrites of CA1 pyramidal cells following spatial training of rats in a complex environment (Moser et al., 1994, 1997). These observations have led to the conclusion that the process of learning is accompanied by a net synaptogenesis (Greenough and Bailey, 1988; Bailey and Kandel, 1993; Andersen and Soleng, 1998). The discrepancy between the morphological data that demonstrated or that failed to demonstrate a net increase in total synapse number following behavioral learning may be explained in a number of ways. One possibility is that some of the previous results were affected by the use of inadequate methodologies available at the time. In many studies cited above, synaptic profiles were identified in single ultrathin sections, although an unequivocal identification of synapses can only be achieved if synaptic contacts are visualized in consecutive serial sections (Geinisman et al., 1996b). Moreover, counts performed in single sections provide estimates of numbers that are biased by several factors, including the size, shape and orientation of synapses, the section thickness, truncation and overprojection (Gundersen, 1986). The magnitude and direction of the biases were not taken into account, although they might be different in conditioned and control animals. An additional methodological problem of the previous studies is that their data were most often expressed in terms of the numerical density of synapses per unit area or volume of tissue. Such density measures, as opposed to the parameter of total number, are influenced by changes in tissue volume that result from processing for microscopy. It was not determined whether or not these changes were the same in conditioned and control animals. Another possible reason for the diversity of the results is suggested by the dynamic nature of the learning phenomenon. In our study, for example, synapses were quantified only at one time point relative to behavioral acquisition. If stereological analyses were performed at different time points along the learning curve, an increase in synapse number resulting from trace eyeblink conditioning might have been detected as well. It is obvious that future studies aimed at the elucidation of learning-induced changes in synapse number should probe the time course of acquisition of new behaviors. Learning-induced Enlargement of PSD Dimensions When the total number of synapses was not altered by learning, other structural synaptic modifications were observed. They include increases in the number of all axospinous perforated synapses (Vrensen and Nunes Cardozo, 1981) or in their variety that involves concave spines (Van Reempts et al., 1992), as well as in the spatial clustering of synapses (Rusakov et al., 1997). The results of previous studies based on the analysis of single sections have also indicated that the length of PSD profiles increases following acquisition of a new behavior (DeVoogd et al., 1985; Horn et al., 1985; Van Reempts et al., 1992; Stewart and Rusakov, 1995). We measured the length of PSD profiles on electron micrographs of consecutive sections through each sampled synapse to obtain estimates of the PSD area in conditioned and pseudoconditioned rabbits (Geinisman et al., 2000). The results showed that nonperforated axospinous synapses had a larger PSD area in conditioned animals (30.3 ± 0.8 nm2 × 103) than in pseudoconditioned ones (27.5 ± 0.9 nm2 × 103) and that the difference between the group mean values was statistically significant (P = 0.0129, two-tailed t-test for matched pairs). The area of nonperforated PSDs was additionally estimated in a group of unstimulated control rabbits. The group mean value (26.1 ± 1.0 nm2 × 103) was not significantly different from that of pseudoconditioned controls. Since the area of nonperforated PSDs was not altered by pseudoconditioning, the relative change in this parameter observed in conditioned animals, as compared with pseudoconditioned animals, reflected an actual conditioning-induced enlargement of the PSD area. The PSD contains signal transduction proteins, such as postsynaptic receptors and ion channels (Ziff, 1997; Kennedy, 1998). The learning-induced increase in the PSD area may reflect an addition of these components, especially AMPA receptors. The ratio between AMPA and NMDA receptors is directly proportional to the PSD size in nonperforated axospinous synapses from the rat CA1 stratum radiatum (Nusser et al., 1988; Takumi et al., 1999; Racca et al., 2000), and the AMPA receptor number regresses to zero when a PSD diameter is <180 nm (Takumi et al., 1999). It seems reasonable, therefore, to suppose that small nonperforated PSDs may increase in size following learning due to the insertion of AMPA receptors, which is necessary for silent synapses lacking such receptors to become active. Support for this supposition comes from the analysis of our data in terms of distributions of nonperforated axospinous synapses with regard to the area of their PSDs (Geinisman et al., 2000). Comparison of conditioned and pseudoconditioned rabbits showed that the proportion of nonperforated synapses with PSDs that fell into the smallest size category (PSD area < 20 nm2 × 103) was decreased in the conditioned group (Figure 10). Conversely, the proportions of those nonperforated synapses that had PSDs belonging to larger size categories were increased in conditioned animals (Figure 10). Thus, only the smallest nonperforated PSDs — ones that probably lack AMPA receptors — were increased in their area in conditioned animals. If this change is accompanied by the insertion of AMPA receptors, it may indicate that the enlargement of nonperforated PSDs following trace eyeblink conditioning represents a structural correlate of the conversion of silent synapses into functional ones. Such a transformation would lead to an augmentation of synaptic efficacy, which is believed to be necessary for learning (Tanzi, 1893; Konorski, 1948; Hebb, 1975). An intriguing observation made in our study is that the learning-induced increase in the PSD area selectively involved only nonperforated axospinous synapses. None of the other synaptic subtypes exhibited an alteration of this kind (Geinisman et al., 2000). A characterization of the nonperforated synaptic subtype provides a plausible explanation for this observation. Nonperforated axospinous synapses are twice as likely as perforated ones to lack AMPA receptors (Desmond and Weinberg, 1998). The ratio of nonperforated to perforated synapses in the rabbit CA1 stratum radiatum is 7.4:1 (Geinisman et al., 2000). Thus, the pool of silent axospinous synapses lacking AMPA receptors appears to consist primarily of those synapses that have a relatively small nonperforated PSD. Provided AMPA receptors were inserted into PSDs of silent synapses following trace eyeblink conditioning, the resulting increase in the PSD area would be detected predominantly, if not exclusively, in nonperforated axospinous synapses. Comparison of Structural Synaptic Alterations Associated with Hippocampal LTP and Behavioral Learning The data reviewed above show that LTP and behavioral learning are accompanied by structural modifications of synaptic connectivity and that there are certain similarities between the two phenomena in terms of the patterns of these modifications. The formation of new spines and synapses associated with them has been demonstrated to occur following LTP induction (Engert and Bonhoeffer, 1999; Toni et al., 1999). There is no direct evidence that such a process also takes place during acquisition of new behaviors; however, an increase in the total number of synapses in pertinent brain areas, which indicates net synaptogenesis, appears to be a typical feature of many forms of behavioral learning (Greenough and Bailey, 1988; Bailey and Kandel, 1993; Andersen and Soleng, 1998). An essential characteristic of LTP-induced synaptogenesis is that it involves only those newly emerging spines that are contacted by axon terminals activated by potentiating stimulation (Toni et al., 1999). By analogy, synaptogenesis-promoting effects of behavioral learning may also be restricted to activated terminals. This would explain why no change in total synapse number was detected not only in the majority of ultrastructural LTP studies, but also in a number of electron microscopic studies exploring structural synaptic modifications associated with behavioral learning. An examination of entire synaptic populations might not reveal an increase in the number of those synapses that are established by activated terminals because the latter constitute only a small proportion of all presynaptic boutons. A restructuring of synapses, possibly coupled with their turnover, emerges as the other basic form of structural synaptic plasticity that is common to LTP and learning. Of special importance in this regard are the observations suggesting that the conversion of silent axospinous synapses into functional ones may be a determinant of synapse restructuring associated with both hippocampal LTP (Issaks et al., 1995; Liao et al., 1995) and behavioral learning (Geinisman et al., 2000). During postnatal development, such a conversion occurs on a large scale in the hippocampus (Durand et al., 1996; Wu et al., 1996; Liao et al., 1999; Petralia et al., 1999). If this process is a prerequisite for successful learning, developmental learning disabilities may be due in part to an impairment of the mechanisms that govern the transformation of silent synapses into active ones. Whether or not this assumption is proven to be correct, there is a growing body of evidence indicating that the cellular mechanisms of memory formation following behavioral learning are likely to include both the establishment of new synaptic connections and the remodeling of existing and newly formed synapses. Notes The expertise of Drs L. deToledo-Morrell, J.F. Disterhoft, H.J.G. Gundersen, F. Morrell and M.J. West in designing and conducting the experiments described here, as well as in discussing their results is gratefully acknowledged. The author wishes to thank Drs M.A. Beatty, R.E. Heller, M.D. McEchron, R.F. Parshall, I.S. Persina, J.M. Power, M. Rossi and E.A. Van der Zee for help with data collection and analyses, as well as Dr R.W. Berry for critical comments on the manuscript. This work was supported in part by grants 5 RO1 NS34582 from NINDS and 1 RO1 AG17139 from NIA. Address correspondence to Yuri Geinisman, Department of Cell and Molecular Biology, Northwestern University Medical School, 303 East Chicago Avenue, Chicago, IL 60611, USA. Email: yurig@nwu.edu. Figure 1. View largeDownload slide  Total number of synapses per neuron in the dentate middle molecular layer of control and potentiated rats examined 1 h or 13 days after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. Figure 1. View largeDownload slide  Total number of synapses per neuron in the dentate middle molecular layer of control and potentiated rats examined 1 h or 13 days after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. Figure 2. View largeDownload slide  Ratio of perforated to nonperforated axospinous synapses in the dentate middle molecular layer of control and potentiated rats examined 1 h or 13 days after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. **P < 0.01, two-tailed Mann-Whitney U-test. Figure 2. View largeDownload slide  Ratio of perforated to nonperforated axospinous synapses in the dentate middle molecular layer of control and potentiated rats examined 1 h or 13 days after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. **P < 0.01, two-tailed Mann-Whitney U-test. Figure 3. View largeDownload slide  Number of axospinous perforated synapses with multiple transmission zones (AsMTZ) and of axodendritic asymmetrical synapses (AdA) in the dentate middle molecular layer of control and potentiated rats examined 1 h after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. *P < 0.05, two-tailed Mann-Whitney U-test. Figure 3. View largeDownload slide  Number of axospinous perforated synapses with multiple transmission zones (AsMTZ) and of axodendritic asymmetrical synapses (AdA) in the dentate middle molecular layer of control and potentiated rats examined 1 h after the last stimulation [data from (Geinisman et al., 1991, 1996a)]. Bars show group means ± SE. *P < 0.05, two-tailed Mann-Whitney U-test. Figure 4. View largeDownload slide  Electron micrographs of consecutive ultrathin sections (a-f) demonstrating a perforated axospinous synapse with three completely partitioned transmission zones. All synaptic profiles containing a PSD are presented. The presynaptic axon terminal (labeled AT in f) is apposed by the postsynaptic spine head (labeled SP in a). Profiles of spinules (arrows), spine neck (arrowheads) and spine apparatus (asterisks) are observed. Scale bar: 0.25 μm. Figure 4. View largeDownload slide  Electron micrographs of consecutive ultrathin sections (a-f) demonstrating a perforated axospinous synapse with three completely partitioned transmission zones. All synaptic profiles containing a PSD are presented. The presynaptic axon terminal (labeled AT in f) is apposed by the postsynaptic spine head (labeled SP in a). Profiles of spinules (arrows), spine neck (arrowheads) and spine apparatus (asterisks) are observed. Scale bar: 0.25 μm. Figure 5. View largeDownload slide  Reconstructions of the synapse shown in Figure 4. (a) A three-dimensional reconstruction of the presynaptic axon terminal having three separate protrusions (arrows). The image has been rotated by 190° and 35° around the x and the z axis, respectively. (b) A three-dimensional reconstruction of the postsynaptic spine head exhibiting two partitions (arrows). The image has been rotated at 15° and 25° angles around the x and the z axis, respectively. (c) A two-dimensional reconstruction of the PSD consisting of three separate segments. Figure 5. View largeDownload slide  Reconstructions of the synapse shown in Figure 4. (a) A three-dimensional reconstruction of the presynaptic axon terminal having three separate protrusions (arrows). The image has been rotated by 190° and 35° around the x and the z axis, respectively. (b) A three-dimensional reconstruction of the postsynaptic spine head exhibiting two partitions (arrows). The image has been rotated at 15° and 25° angles around the x and the z axis, respectively. (c) A two-dimensional reconstruction of the PSD consisting of three separate segments. Figure 6. View largeDownload slide  Diagram illustrating a model of synapse restructuring associated with the induction phase of LTP as described in the text. The schematic shows the following subtypes of axospinous synapses: (a) typical (small) and (b) atypical (large) non-perforated synapses; perforated synapses that exhibit (c) a focal spine partition and fenestrated PSD, (d) a sectional partition and horseshoe-shaped PSD, or (e) a complete spine partition(s) and segmented PSD. Figure 6. View largeDownload slide  Diagram illustrating a model of synapse restructuring associated with the induction phase of LTP as described in the text. The schematic shows the following subtypes of axospinous synapses: (a) typical (small) and (b) atypical (large) non-perforated synapses; perforated synapses that exhibit (c) a focal spine partition and fenestrated PSD, (d) a sectional partition and horseshoe-shaped PSD, or (e) a complete spine partition(s) and segmented PSD. Figure 7. View largeDownload slide  Number of axospinous perforated synapses with multiple transmission zones (AsMTZ) and of axodendritic asymmetrical synapses (AdA) in the dentate middle molecular layer of control and potentiated rats examined 13 days after the last stimulation [data from (Geinisman et al., 1996a)]. Bars show group means ± SE. *P < 0.05, two-tailed Mann-Whitney U-test. Figure 7. View largeDownload slide  Number of axospinous perforated synapses with multiple transmission zones (AsMTZ) and of axodendritic asymmetrical synapses (AdA) in the dentate middle molecular layer of control and potentiated rats examined 13 days after the last stimulation [data from (Geinisman et al., 1996a)]. Bars show group means ± SE. *P < 0.05, two-tailed Mann-Whitney U-test. Figure 8. View largeDownload slide  Electron micrographs of consecutive ultrathin sections (a-e) demonstrating a special subtype of perforated synapse (arrows). The synaptic contact involves both a neckless spine (a-c) associated with a spine apparatus (asterisks) and its parent dendrite (d,e) due to a partial retraction of the spine into the dendrite. Scale bar: 0.25 μm. Figure 8. View largeDownload slide  Electron micrographs of consecutive ultrathin sections (a-e) demonstrating a special subtype of perforated synapse (arrows). The synaptic contact involves both a neckless spine (a-c) associated with a spine apparatus (asterisks) and its parent dendrite (d,e) due to a partial retraction of the spine into the dendrite. Scale bar: 0.25 μm. Figure 9. View largeDownload slide  Diagram illustrating a model of synapse restructuring associated with the maintenance phase of LTP as described in the text. The schematic shows the following synaptic subtypes: axospinous perforated synapses that have multiple transmission zones and involve (a) a spine with a neck, (b) a neckless spine or (c) a spine partially retracted into a parent dendrite; (d) axodendritic perforated synapse; (e) axodendritic nonperforated synapse; axospinous perforated synapses that lack spine partitions and exhibit (f) a segmented, (g) horseshoe-shaped or (h) fenestrated PSD; (i) atypical and (j) typical axospinous nonperforated synapses. Figure 9. View largeDownload slide  Diagram illustrating a model of synapse restructuring associated with the maintenance phase of LTP as described in the text. The schematic shows the following synaptic subtypes: axospinous perforated synapses that have multiple transmission zones and involve (a) a spine with a neck, (b) a neckless spine or (c) a spine partially retracted into a parent dendrite; (d) axodendritic perforated synapse; (e) axodendritic nonperforated synapse; axospinous perforated synapses that lack spine partitions and exhibit (f) a segmented, (g) horseshoe-shaped or (h) fenestrated PSD; (i) atypical and (j) typical axospinous nonperforated synapses. Figure 10. View largeDownload slide  Comparison of conditioned and pseudoconditioned rabbits with regard to the number of nonperforated axospinous synapses falling into various size categories of the PSD area [data from (Geinisman et al., 2000)]. Figure 10. View largeDownload slide  Comparison of conditioned and pseudoconditioned rabbits with regard to the number of nonperforated axospinous synapses falling into various size categories of the PSD area [data from (Geinisman et al., 2000)]. References Andersen P, Soleng AF ( 1998) Long-term potentiation and spatial training are both associated with the generation of new excitatory synapses. Brain Res Rev  26: 353–359. Google Scholar Bailey CH, Kandel ER ( 1993) Structural changes accompanying memory storage. Annu Rev Physiol  55: 397–426. Google Scholar Black JE, Isaaks KR, Anderson BJ, Alcantara AA, Greenough WT ( 1990) Learning causes synaptogenesis, whereas motor activity causes angiogenesis, in cerebellar cortex of adult rats. Proc Natl Acad Sci USA  87: 5568–5572. Google Scholar Bliss TVP, Collingridge GL ( 1993) A synaptic model of memory: long-term potentiation in the hippocampus. Nature  361: 31–39. Google Scholar Bliss TVP, Gardner-Medwin AR ( 1973) Long-lasting potentiation of synaptic transmission in the dentate gyrus of the unanaesthetized rabbit following stimulation of the perforant path. J Physiol (Lond)  232: 357–374. Google Scholar Bourgeois J-P, Goldman-Rakic P, Rakic P (1999) Formation, elimination, and stabilization of synapses in the primate cerebral cortex. In: Cognitive neuroscience. A handbook for the field, 2nd edn (Gazzaniga MS, ed.), pp. 23–32. Cambridge, MA: MIT Press. Google Scholar Bradley P, Horn G, Bateson P ( 1981) An electron microscopic study of chick hyperstiatum ventrale. Exp Brain Res  41: 115–120. Google Scholar Buchs P-A, Muller D ( 1996) Induction of long-term potentiation is associated with major ultrastructural changes of activated synapses. Proc Natl Acad Sci USA  93: 8040–8045. Google Scholar Chang F-LF, Greenough WT ( 1984) Transient and enduring morphological correlates of synaptic activity and efficacy change in the rat hippocampal slice. Brain Res  309: 35–46. Google Scholar Chang F-LF, Isaaks KR, Greenough WT ( 1991) Synapse formation occurs in association with the induction of long-term potentiation in two-year-old rat hippocampus in vitro. Neurobiol Aging  12: 517–522. Google Scholar de Jonge MC, Black J, Deyo RA, Disterhoft, JF ( 1990) Learning-induced afterhyperpolarization reductions in hippocampus are specific for cell type and potassium conductance. Exp Brain Res  80: 456–462. Google Scholar Desmond NL, Levy WB ( 1986) Changes in the numerical density of synaptic contacts with long-term potentiation in the hippocampal dentate gyrus. J Comp Neurol  253: 466–475. Google Scholar Desmond NL, Levy WB ( 1990) Morphological correlates of long-term potentiation imply the modification of existing synapses, not synaptogenesis, in the hippocampal dentate gyrus. Synapse  5: 139–143. Google Scholar Desmond NL, Weinberg RJ ( 1998) Enhanced expression of AMPA receptor protein at perforated axospinous synapses. NeuroReport  9: 857–860. Google Scholar DeVoogd TJ, Nixdorf B, Nottenbohm F ( 1985) Synaptogenesis and changes in synaptic morphology related to acquisition of new behavior. Brain Res  329: 304–308. Google Scholar Doubell TP, Stewart MG ( 1993) Short term changes in the numerical density of synapses in the intermediate and medial hyperstriatum ventrale following one-trial passive avoidance learning. J Neurosci  13: 2230–2236. Google Scholar Durand GM, Kovalchuk Y, Konnereth A ( 1966) Long-term potentiation and functional synapse induction in developing hippocampus. Nature  381: 71–75. Google Scholar Edwards FA ( 1995) LTP — a structural model to explain the inconsistencies. Trends Neurosci  18: 250–255. Google Scholar Engert F, Bonhoeffer, T ( 1999) Dendritic spine changes associated with long-term synaptic plasticity. Nature  399: 66–70. Google Scholar Geinisman, Y. ( 1993) Perforated axospinous synapses with multiple, completely partitioned transmission zones: probable structural intermediates in synaptic plasticity. Hippocampus  3: 417–434. Google Scholar Geinisman, Y, Morrell F, deToledo-Morrell L ( 1987) Synapses on dendritic shafts exhibit a perforated postsynaptic density. Brain Res  422: 352–356. Google Scholar Geinisman, Y, Morrell F, deToledo-Morrell L ( 1987) Axospinous synapses with segmented postsynaptic densities: a morphologically distinct synaptic subtype contributing to the number of profiles of ‘perforated’ synapses visualized in random sections. Brain Res  423: 179–188. Google Scholar Geinisman Y, deToledo-Morrell L, Morrell F ( 1991) Induction of long-term potentiation is associated with an increase in the number of axospinous synapses with segmented postsynaptic densities. Brain Res  566: 77–88. Google Scholar Geinisman Y, deToledo-Morrell L, Morrell F, Heller RE, Rossi M, Parshall RF ( 1993) Structural synaptic correlate of long-term potentiation: formation of axospinous synapses with multiple, completely partitioned transmission zones. Hippocampus  3: 435–446. Google Scholar Geinisman Y, deToledo-Morrell L, Morrell F, Persina IS, Beatty MA ( 1996) Synapse restructuring associated with the maintenance phase of hippocampal long-term potentiation. J Comp Neurol  368: 413–423. Google Scholar Geinisman Y, Gundersen HJG, Van der Zee E, West MJ ( 1996) Unbiased stereological estimation of the total number of synapses in a brain region. J Neurocytol  25: 805–819. Google Scholar Geinisman Y, Disterhoft JF, Gundersen HJG, McEchron MD, Persina IS, Power JM, Van der Zee EA, West MJ ( 2000) Remodeling of hippocampal synapses following hippocampus-dependent associative learning. J Comp Neurol  417: 49–59. Google Scholar Greenough WT, Bailey CH ( 1988) The anatomy of a memory: convergence of results across a diversity of tests. Trends Neurosci  11: 142–147. Google Scholar Gundersen HJG ( 1986) Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R. Thompson. J Microsc (Lond)  143: 219–223. Google Scholar He Y, Janssen WGM, Morrison JH ( 1998) Synaptic coexistence of AMPA and NMDA receptors in the rat hippocampus: a postembedding immunogold study. J Neurosci Res  54: 444–449. Google Scholar Hebb DO (1975) The organization of behavior. New York: Wiley. Google Scholar Horn G, Bradley P, McCabe BJ ( 1985) Changes in the structure of synapses associated with learning. J Neurosci  5: 3161–3168. Google Scholar Hosokawa T, Rusakov DA, Bliss TVP, Fine A ( 1995) Repeated confocal imaging of individual dendritic spines in the living hippocampal slice: evidence for changes in length and orientation associated with chemically induced LTP. J Neurosci  15: 5560–5573. Google Scholar Hunter A, Stewart MG ( 1989) A quantitative analysis of the synaptic development of the lobus parolfactorius of the chick (Gallus domesticus). Exp Brain Res  78: 425–434. Google Scholar Isaac JTR, Nicoll RA, Malenka RC ( 1995) Evidence for silent synapses: implications for the expression of LTP. Neuron  15: 427–434. Google Scholar Kennedy MB ( 1998) Signal transduction molecules at the glutamatergic postsynaptic membrane. Brain Res Rev  26: 243–257. Google Scholar Kleim JA, Lussing E, Schwarz ER, Comery TA, Greenough WT ( 1996) Synaptogenesis and fos expression in the motor cortex of the adult rat after motor skill learning. J Neurosci  16: 4529–4535. Google Scholar Kleim JA, Vij K, Ballard DH, Greenough WT ( 1997) Learning-dependent synaptic modifications in the cerebellar cortex of the adult rat persist for at least four weeks. J Neurosci  17: 717–721. Google Scholar Konorski J (1948) Conditioned reflexes and neuron organization. Cambridge: Cambridge University Press. Google Scholar Lee KS, Schottler F, Oliver M, Lynch G ( 1980) Brief bursts of high-frequency stimulation produce two types of structural change in rat hippocampus. J Neurophysiol  44: 247–258. Google Scholar Lee KS, Oliver M, Schottler F, Lynch G (1981) Electron microscopic studies of brain slices: the effects of high frequency stimulation on dendritic ultrastructure. In: Electrophysiology of isolated mammalian CNS preparations (Kerkut GA, Wheal HW, eds), pp. 181–211. New York: Academic Press. Google Scholar Liao D, Hessler NA, Malinow R ( 1995) Activation of postsynaptically silent synapses during pairing-induced LTP in CA1 region of hippocampal slice. Nature  375: 400–404. Google Scholar Liao D, Zhang X, O'Brian R, Ehlers MD, Hungair RL ( 1999) Regulation of morphological postsynaptic silent synapses in developing hippocampal neurons. Nature Neurosci  2: 37–43. Google Scholar McNaughton BL, Barnes CA, Anderson P ( 1981) Synaptic efficacy and EPSP summation in granule cells of rat fascia dentata studied in vitro. J Neurophysiol  46: 952–966. Google Scholar Moser M-B, Trommald M, Andersen P ( 1994) An increase in dendritic spine density on hippocampal CA1 pyramidal cells following spatial learning in adult rats suggests the formation of new synapses. Proc Natl Acad Sci USA  91: 12673–12675. Google Scholar Moser M-B, Trommald M, Egeland T, Andersen P ( 1997) Spatial training in a complex environment and isolation alter the spine distribution differentially in rat CA1 pyramidal cells. J Comp Neurol  380: 373–381. Google Scholar Moyer JR Jr, Deyo RA, Disterhoft JF ( 1990) Hippocampectomy disrupts trace eye-blink conditioning in rabbits. Behav Neurosci  104: 243–252. Google Scholar Moyer JR Jr, Thompson LT, Disterhoft JF ( 1996) Trace eyeblink conditioning increases CA1 excitability in a transient and learning-specific manner. J Neurosci  16: 5536–5546. Google Scholar Nusser Z, Lujan R, Laube G, Roberts JDB, Molnar E, Somogyi P ( 1998) Cell type and pathway dependence of synaptic AMPA receptor number and variability in the hippocampus. Neuron  21: 545–559. Google Scholar O'Malley A, O'Connel C, Regan CM ( 1998) Ultrastructural analysis reveals avoidance conditioning to induce a transient increase in hippocampal dentate spine density in the 6 h post-training period of consolidation. Neuroscience  87: 607–613. Google Scholar Petralia RS, Esteban JA, Wang Y-X, Partridge JG, Zhao H-M, Wenthold RJ, Malinow R ( 1999) Selective acquisition of AMPA receptors over postnatal development suggests a molecular basis for silent synapses. Nature Neurosci  2: 31–36. Google Scholar Power JM, Thompson LT, Moyer JR Jr, Disterhoft JF ( 1997) Enhanced synaptic transmission in CA1 hippocampus after eyeblink conditioning. J Neurophysiol  78: 1184–1187. Google Scholar Racca C, Stephenson FA, Streit P, Roberts JDB, Somogyi P ( 2000) NMDA receptor content of synapses in stratum radiatum of the hippocampal CA1 area. J Neurosci  20: 2512–2522. Google Scholar Ramón y Cajal S ( 1893) Neue Darstellung vom histologischen Bau des Centralnerevensystem. Arch Anat Physiol  319–428. Google Scholar Rusakov DA, Davies HA, Harrison E, Diana G, Richter-Levin G, Bliss TVP, Stewart MG ( 1997) Ultrastructural synaptic correlates of spatial learning in rat hippocampus. Neuroscience  80: 69–77. Google Scholar Schuster T, Krug M, Wenzel J ( 1990) Spinules in axospinous synapses of the rat dentate gyrus: changes in density following long-term potentiation. Brain Res  523: 171–174. Google Scholar Solomon PR, Van der Schaaf EV, Thompson RF, Weisz DJ ( 1986) Hippocampus and trace conditioning of the rabbit's classically conditioned nictitating membrane response. Behav Neurosci  100: 729–744. Google Scholar Sorra KE, Harris KM ( 1998) Stability in synapse number and size at 2 h after long-term potentiation in hippocampal area CA1. J Neurosci  18: 658–671. Google Scholar Stewart MG, Rusakov DA ( 1995) Morphological changes associated with stages of memory formation in the chick following passive avoidance training. Behav Brain Res  66: 21–28. Google Scholar Stewart MG, Csillag A, Rose SPR ( 1987) Alterations in synaptic structure in the paleostriatal complex of the domestic chick, Gallus domesticus, following passive avoidance training. Brain Res  426: 69–81. Google Scholar Takumi Y, Ramírez-León V, Laake P, Rinvik E, Ottersen P ( 1999) Different modes of expression of AMPA and NMDA receptors in hippocampal synapses. Nature Neurosci  2: 618–624. Google Scholar Tanzi E ( 1893) I fatti i le induzione nell'odierna istologia del sistema nervoso. Riv Sperim Freniatr Med  19: 419–472. Google Scholar Toni E, Buchs P-A, Nikonenko I, Bron CR, Muller D ( 1999) LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature  402: 421–425. Google Scholar Trommald M, Vaaland JL, Blackstad TW, Andersen P (1990) Dendritic spine changes in rat dentate granule cells associated with long-term potentiation. In: Neurotoxicity of excitatory amino acids (Guidotti A, ed.), pp. 163–174. New York: Raven Press. Google Scholar Trommald M, Hulleberg G, Andersen P ( 1996) Long-term potentiation is associated with new excitatory spine synapses on rat dentate granule cells. Learn Mem  3: 218–228. Google Scholar Van Reempts J, Dikova M, Webrouck L, Clincke G, Borgers M ( 1992) Synaptic plasticity in rat hippocampus associated with learning. Behav Brain Res  51: 179–183. Google Scholar Vrensen G, Nunes Cardozo J ( 1981) Changes in size and shape of synaptic connections after visual training: an ultrastructural approach to synaptic plasticity. Brain Res  218: 79–97. Google Scholar Weeks ACW, Ivanco TL, LeBoutillier JC, Racine RJ, Petit TL ( 1998) The degree of potentiation is associated with synaptic number during the maintenance of long-term potentiation in the rat dentate gyrus. Brain Res  798: 211–216. Google Scholar Wenzel J, Matthies H (1985) Morphological changes in the hippocampal formation accompanying memory formation and long-term potentiation. In: Memory systems of the brain (McGaugh IL, Lynch G, eds), pp. 151–170. New York: Guilford. Google Scholar Wenzel J, Kammerer E, Frotscher M, Kirsche W, Matthies H, Wenzel M ( 1980) Electronmicroscopical and morphometrical investigations on the plasticity of synapses in the hippocampus of the rat following conditioning. J Hirnforsch  21: 647–654. Google Scholar Wu G-Y, Malinow R, Cline HT ( 1996) Maturation of a central glutamatergic synapse. Science  274: 972–976. Google Scholar Ziff EB ( 1997) Enlightening the postsynaptic density. Neuron  19: 1163–1174. Google Scholar © Oxford University Press
TI - Structural Synaptic Modifications Associated with Hippocampal LTP and Behavioral Learning
JF - Cerebral Cortex
DO - 10.1093/cercor/10.10.952
DA - 2000-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/structural-synaptic-modifications-associated-with-hippocampal-ltp-and-7gKLC3Sgto
SP - 952
EP - 962
VL - 10
IS - 10
DP - DeepDyve
ER -