Journal of Neuroscience Research 58:42–61 (1999)

The Plasticity–Pathology Continuum: Defining a Role for the LTP Phenomenon

Jill C. McEachern1* and Christopher A. Shaw1,2 1Department of Physiology, University of British Columbia, Vancouver, Canada 2Departments of Ophthalmology and Neuroscience, University of British Columbia, Vancouver, Canada

Long-term potentiation (LTP) is the most widely Key words: homeostasis; receptor regulation; kin- studied form of neuroplasticity and is believed by dling; age-dependence; neuroplasticity many in the field to be the substrate for learning and memory. For this reason, an understanding of the INTRODUCTION mechanisms underlying LTP is thought to be of fundamental importance to the neurosciences, but a Since its discovery by Bliss and Lomo (1973), the definitive linkage of LTP to learning or memory has phenomenon of long-term potentiation (LTP) has domi- not been achieved. Much of the correlational data nated the empirical and theoretical search for the synaptic/ used to support this claim is ambiguous and controver- cellular basis of learning and memory. Many thousands of sial, precluding any solid conclusion about the func- articles and chapters have been written about the diverse tional relevance of this often artificially induced form subtypes of the phenomenon and the myriad characteris- tics that describe it (for references, see McEachern and of neuroplasticity. In spite of this fact, the belief that Shaw, 1996a,b). Synaptic potentiations that appear to be LTP is a mechanism subserving learning and/or LTP-like have been identified in every imaginable neural memory has become so dominant in the field that the circuit and subpopulation of both vertebrates and inverte- investigation of other potential roles or actions of brates. And, whereas many of these experiments have LTP-like phenomena in the nervous system has been been phenomenological, often mundane, a large number seriously hindered. The multiple subtypes of the of experiments on LTP have been technological tours de phenomena and the myriad molecules apparently force that are models of intellectual precision and rigor. involved in these subtypes raise the possibility that Throughout it all, the underlying assumption, stated or observed forms of LTP may represent very different not, is that LTP is the synaptic basis of memory and that types of modification events, with vastly different when we understand the mechanisms behind it we will consequences for neural function and survival. A hold in our grasp the Holy Grail of the neurosciences. relationship between LTP and neuropathology is sug- Judging by a recent commentary (Stevens, 1998), many gested in part by the fact that many of the molecular neuroscientists not only accept that LTP or something processes involved in LTP induction or maintenance much like it is the basis of learning and memory, but, are the same as those activated during excitotoxic further, some think that the final elucidation of the events in neurons. In addition, some LTP subtypes are underlying mechanisms is within reach. To question clearly induced by pathological stimuli, e.g., anoxic either of these assumptions is to question the faith in the LTP. Such data raise the possibility that LTP is part of ‘‘LTP equals memory’’ orthodoxy that so clearly domi- a continuum of types of neural modification, some nates the field of neuroplasticity research. Until recently, leading to beneficial alterations such as may occur in relatively few investigators have chosen to query the learning and others that may be primarily pathologi- basic assumptions in the field, a rather astonishing fact cal in nature, as in kindling and excitotoxicity. In this given the amount of time and treasure invested in sorting article, we introduce a plasticity–pathology con- out the mechanisms of LTP. Indeed, it is our contention tinuum model that is designed to place the various that the field appears to be in stasis: new molecules are forms of neural modification into proper context. In vitro and kindling receptor regulation studies are used Contract grant sponsor: British Columbia Health Research Foundation; to provide a basis for evaluating the specific synaptic/ Contract grant sponsor: MITACS Centre of Excellence. cellular response modification along the continuum of *Correspondence to: Jill C. McEachern, 2177 Wesbrook Mall, Univer- events, from beneficial to detrimental, that will be sity of British Columbia, Vancouver, BC, Canada, V6T 1Z3. induced by a particular stimulus. J. Neurosci. Res. E-mail: [email protected] 58:42–61, 1999. ௠ 1999 Wiley-Liss, Inc. Received 10 May 1999; Accepted 17 May 1999

௠ 1999 Wiley-Liss, Inc. LTP and the Plasticity–Pathology Continuum 43 constantly being tossed into the breach, but fundamental In the following sections we will briefly attempt to assumptions remain largely unquestioned. For example, deal with these issues. For a more comprehensive over- What if LTP has little or nothing to do with memory? Or, view of the vast LTP literature, the reader is referred to What if it serves other processes as well? If LTP is not the the reviews cited in the article. basis for memory, what is? And, not least, If not memory, then what is LTP? We have previously (McEachern and Shaw, 1996a,b) LTP CHARACTERISTICS PERTINENT published two reviews on the LTP phenomena and laid TO A MODEL OF LEARNING OR MEMORY open to scrutiny the fundamental assumptions in the field. Permanence of LTP Since then, others have repeated these queries and Although with some induction methods LTP may be proposed alternatives (Shors and Matzel, 1997; for a long-lasting, it is not permanent (Jeffery et al., 1990; review, see also Cain, 1997). Moreover, whereas many Abraham et al., 1994, 1995). This fact argues against LTP investigators have responded in a positive way to the acting in itself as a lasting physical change underlying challenges to orthodoxy, others have taken refuge in the long-term memory. Impermanence is not necessarily a repetition of familiar litanies or in overt statements of fatal strike against LTP as a memory mechanism, how- faith (Abraham, 1997). ever. The problem can be circumvented in a variety of A fundamental goal of our reviews was to address ways, for example: (a) a relatively transient form of LTP selected sources of controversy in the LTP literature in an could, before decaying, trigger lasting alterations (e.g., effort to (1) assess the progress that had been made morphological) that then act as the substrate for memory; toward understanding the underlying mechanisms of LTP (b) the function of the hippocampus in memory may be and (2) sift compelling correlations to learning or memory most significant initially; subsequently its role may from vagaries of experimental technique. What we discov- diminish and that of other structures, possibly the cortex, ered as a result of our literature review was that LTP may become more important. Hippocampal LTP may be research, already a complex undertaking, was further sufficiently stable to reflect the time course of this type of complicated by the use of vastly different experimental relationship; however, studies of cortical LTP indicate techniques and preparations, often in disregard of vari- that it is not permanent either (Doyere et al., 1993); (c) ables that are absolutely crucial to experimental outcome. studies of human memory indicate that memories are Some of these practices included the failure to control for neither permanent nor invariant across the lifespan the effects of test stimulus intensity by measuring input/ (Bartlett, 1932), and mechanisms for reinvigorating mol- output curves, failure to account for age-dependent ecules involved in synaptic modifications have been variability, and lack of attention to nonspecific effects of proposed and may provide the basis for the longevity of drugs, gene deletion, temperature, and motor activity. certain memories (Kavanau, 1994). If LTP is periodically These factors, combined with the high level of contradic- refreshed or, alternatively, acts as a transient triggering tion in the field, left little room to draw firm conclusions mechanism for other changes, then it would be more about either the function or the underlying mechanisms of compatible with the durability of memory. LTP. In our view, the situation has not changed since that time, and to date the extent of homology between observed types of LTP, the site and nature of the Depotentiation/Reversal of LTP alteration, the temporal progression and duration of the The properties of LTP depotentiation/reversal are potentiation, and the relationship of LTP to learning and also pertinent to the questions of permanence and the memory all remain issues of contention. suitability of LTP as a memory mechanism. LTP induced Nevertheless, it is hard to circumvent the fact that in vitro and in vivo in the hippocampus (Dudek and Bear, changes in synaptic strength and/or number seem to be a 1993; Larson et al., 1993; Martin, 1998) and in vivo in the prerequisite for a memory model based on a non- afferent system to the prefrontal cortex (Burette et al., increasing population of neurons, despite the difficulty of 1997) can be reversed by various patterns of low- proving it. Certain attributes of LTP make it an attractive frequency stimulation and within time windows varying candidate as a memory mechanism: It can be relatively from only minutes to the longest experimental time point durable, has associative properties (but see Gallistel, measured (hours). In one of these experiments (Larson et 1995), can be induced by stimuli approximating endog- al., 1993), LTP reversal was possible with theta- enous activity patterns, and similar increases in synaptic frequency stimulation. The ability of the theta rhythm, a response may occur as a result of learning. However, a pattern of neural activity that accompanies certain forms model of learning or memory with LTP as its substrate of natural activity in the behaving animal, to both induce will need to account for a variety of other properties of the and reverse LTP suggests that if LTP is to serve in phenomenon. memory then either safeguard must protect against LTP 44 McEachern and Shaw reversal except in special conditions. Or, as we suggest, rendered inconclusive due to nonspecific effects of the LTP is meant to persist only transiently before being manipulations in question. Pharmacological interventions reversed by ongoing neural activity and regulatory pro- with NMDA receptor antagonists that block both LTP and cesses. Alternatively, it has been suggested that the return learning or memory in certain tasks seemed to provide a of a subset of potentiated synapses to baseline activity promising correlation between molecular and behavioral levels could function to enhance detail in newly encoded neuroplasticity (e.g., Morris et al., 1986; Robinson et al., representations (Larson et al., 1993) or to prevent ‘‘satura- 1989; Staubli et al., 1989). However, upon close inspec- tion’’ of synaptic plasticity (Doyere et al., 1993). What- tion, the drugs used in these studies were found to cause ever the postulated function of LTP reversal, a theory of insurmountable nonspecific effects on neural activity LTP-based long-term memory must be capable of account- (Keith and Rudy, 1990; Hargreaves and Cain, 1992). As a ing for the instability of a process previously considered result, no rightful claim can be made that only learning or essentially irreversible during the maintenance phase. memory and the underlying molecular substrates were specifically affected. LTP Distribution Similarly, various genetic deletion models were created in which animals mature without a gene encoding The prototype of LTP is the N-methyl-D-aspartate a protein considered important in neuroplastic processes, (NMDA) receptor-dependent version first discovered in for instance, a receptor subtype or a protein kinase (Grant the hippocampus. Joining this ‘‘classic’’ form are various et al., 1992; Silva et al., 1992a,b; Aiba et al., 1994). NMDA- and non-NMDA-dependent forms in the hippo- Again, an LTP–learning/memory link was proclaimed campus and many other structures including, in mam- based on the corresponding failure of measures of both in mals, the spinal cord (Randic et al., 1993), superior the mutants. And once again, the nonspecific effects cervical ganglion (Burgos et al., 1994), and medial prevent such a conclusion: thorough testing of the geniculate nucleus (Gerren and Weinberger, 1983). The mutants in various experiments demonstrated severe discovery that LTP can be induced in structures not changes in morphological, electrophysiological, and be- conventionally associated with higher vertebrate memory havioral attributes as a result of gene deletion during function means either that the definition of ‘‘memory’’ development (Grant et al., 1992; Butler et al., 1993; Aiba must be broad enough to encompass nontypical forms or et al., 1994; McNamara, 1994; Stevens et al., 1994). This that, if LTP occurs physiologically, it serves functions should not be all that surprising given that genes and their other than or in addition to memory. We believe the latter products directly and indirectly mediate multiple func- to be the case. tions and regulatory processes during development and in the mature brain and that gene deletion disrupts all of Specificity of LTP these. An additional confound of developmental gene Innumerable protocols have been found to induce deletion is that it will not be possible to untangle effects and/or block LTP. Included are a myriad of electrical and of deletion on normal development from the effects on pharmacological manipulations and even pathological adult nervous system function and plasticity. However, events such as anoxia (Gozlan et al., 1995; for references, these experiments continue to be cited as proof that LTP see also McEachern and Shaw, 1996a,b). This raises the is the substrate of learning and/or memory despite the question of where the specificity lies in a mechanism that, ambiguity of their interpretation. if relevant only to learning or memory, should presum- Now, as when our original review was written, we ably be quite selective in how it is induced. take exception to the dogmatic adherence by many in the field to the view that a relationship between LTP and learning/memory has been established unequivocally. In Relevance of LTP to Learning/Memory our opinion, this stance has largely prevented other The property of LTP that has perhaps received possible roles for LTP-like plasticity from being exam- relatively less overall experimental attention than it ined. For example, correlations between LTP and certain warrants, especially given its importance, is its potential pathological forms of neuroplasticity may be as striking relevance to behavior, generally taken to refer to learning as any between LTP and learning or memory; this is an and/or memory. We previously raised the following area of inquiry that may have had more attention if not for points on this subject (McEachern and Shaw, 1996a,b), preconceptions in the field. which to our knowledge remain unanswered by those In the following section, we attempt to find the who continue to make definitive statements about the place LTP might hold in the context of plastic modifica- connection between LTP and learning or memory: much tions underlying beneficial behavioral changes, and then of the experimental evidence cited as the strongest we look more closely at the reasons for considering a role support for a link between LTP and learning or memory is for LTP in neuropathology. LTP and the Plasticity–Pathology Continuum 45

RELATIONSHIP OF LTP AND LTD ern and Shaw, 1996a), they may be regulated by similar TO PLASTICITY AND PATHOLOGY: basic governing principles. A CONTINUUM MODEL During development, LTD-like and LTP-like synap- A close look at the literature led us to believe that tic modifications may be involved in the generation and the terms LTP and long-term depression (LTD) are used pruning of synapses. We will expand this idea later in the as a catch-all to encompass a spectrum of activity- article. Similar plastic processes may also be activated in potentiating and -depressing phenomena of different response to injury, in this context serving a compensatory functions, durations, and underlying biochemical sub- role, perhaps spurring sprouting and reconnection in strates. Although artificially induced forms of LTP and damaged neural tissue, possibly through the action of LTD may serve many functions, or none, in the behaving neurotrophic factors. In addition to developmental and animal, researchers may have tapped into a range of compensating functions, several roles are possible for naturally occurring processes that serve to adjust synaptic LTP-like and LTD-like phenomena in the processes strength in a multitude of ways and that contribute to a underlying the variety of forms of learning and memory. variety of processes, both beneficial and pathological. Synaptic alterations of short duration could function to LTP- and LTD-like alterations in synaptic efficacy may be transiently alter synaptic gain, thereby affecting the employed throughout the central nervous system (CNS) balance of neuronal interactions in a circuit. More lasting in whatever context is required, including neuronal effects might act in the capacity of temporary or persistent migration, synaptic connection, and setting response ‘‘templates’’ or representations, possibly depending on levels during early development; laying down sensory factors including the specialization of the structure or and motor schemata; repairing damage; and recording circuit and the strength and significance of the stimulus experience. Although we believe there also is a link of (note that a ‘‘motivated’’ behavioral state has been found some forms of LTP and LTD to neuropathology (see to increase the durability of LTP; Seidenbecher et al., below), we briefly discuss in the following section some 1995). These ideas will be treated in more detail in a characteristics and possible beneficial roles for some of forthcoming article (Shaw and McEachern, in press). the experimentally observed LTP- and LTD-like effects. Modification of synaptic elements and processes including receptors and neurotransmitter release would be ideally suited for short- and intermediate-term effects. ‘‘Beneficial’’ Plasticity The most durable modifications might involve recruit- It may be that included under the rubrics of ‘‘LTP’’ ment of cascade elements that cause lasting alterations and ‘‘LTD’’ there are an entire range of potentiation and through changes in gene expression and protein synthe- depression effects with durations ranging from very short sis. Modification of neuronal function might include to quasi-permanent, each maintained by unique modifica- functional or morphological connection of previously tions at sequential steps along a biochemical cascade (for noninteracting neurons or disconnection of formerly full details, see McEachern and Shaw, 1996a,b). Perhaps associated neurons. Suppression of inhibition through at each stage there is a threshold that must be met to tetanic stimulation or treatment with ␥-aminobutyric acid proceed to the next, and failing this, the potentiation/ subtype A (GABAA) blockers does in fact show latent depression decays or is actively re-regulated to baseline excitatory connections in hippocampal CA3 (Miles and level. The individual players in the cascades underlying Wong, 1987), thus providing a precedent for at least one potentiation and depression might differ in neurons of of these ideas. different phenotype, or even within neurons, as a function of the eliciting stimulus, contribution of modulators, etc. A spectrum of potentiating and depressing effects of Pathological Plasticity different durations and subserved by different biochemi- The discussion above demonstrates that LTP/LTD cal cascades would give the system a high degree of phenomena may have a role in beneficial neuroplastic flexibility to perform the multitude of plastic modifica- modifications during development, in response to injury, tions required in the brain while minimizing cross-talk and in learning and memory. The latter aspects make LTP between them. For example, different biochemical path- and LTD the current prima donnas of synaptic plasticity ways in a single neuron could underlie damage-induced research; however, models of activity-dependent synaptic plasticity and learning-induced plasticity, and each would alterations such as kindling also provide insight into be activated and affected only by the modulators feeding lasting changes in brain function. In this section, we build specifically into the appropriate molecular cascade. De- a rationale for the proposal that certain forms of neuroplas- spite the overburdening number of different molecules ticity, specifically, LTD, LTP, and kindling, fall along a involved in the different cascades (see Table 1 in McEach- continuum of related neuroplastic and neuropathological 46 McEachern and Shaw events that at each extreme of the continuum results in al., 1994), neurotrophic factors (Sazgar et al., 1995), gene neuronal death. transcription (Qian et al., 1993), and synthesis of new protein (Abraham and Otani, 1991; Bliss and Collin- The Kindling Paradigm gridge, 1993) have been implicated in all three forms of neuroplasticity. In addition, the different forms of neuro- Kindling is a model of progressively developing plasticity and cellular response to trauma all appear to electrographic and behavioral seizure expression induced stimulate activation of some of the same immediate early by electrical or chemical stimulation of the brain. Be- genes and gene products (Hughes et al., 1999). cause the progressive and persistent changes in kindling- Additional lines of support indicate an overlap of induced synaptic potentiation and behavior were sugges- biochemical mechanisms subserving LTD and LTP. First, tive of characteristics considered desirable for memory, it there is a parallel time course of onset and maintenance, first drew interest as a potential mechanism of long-term and, more convincingly, saturation of LTP can be fully memory (Goddard and Douglas, 1975). However, other reversed by subsequent induction of LTD, and vice versa, aspects of the phenomenon were incompatible with such without affecting the maximal level of either. Were the a role. The increased neural excitability culminates in initial induction mechanisms distinct, at most a partial seizures, neuronal damage, and behavioral disturbances reversal would be achieved, and the attainable ceiling for similar to those of human temporal lobe epileptics, and each would be altered by serial inductions (Dudek and kindling is now considered a model of the ‘‘pathological Bear, 1993). plasticity’’ that may be involved in human epilepsy. The relationship between LTP and kindling is a subject of ongoing investigation. The discovery that Plasticity–Pathology Relationship electrical (Sutula and Steward, 1986) and chemical Interestingly, many of the characteristics and mo- kindling are accompanied by a lasting potentiation of lecular mechanisms involved in LTP and LTD, which are measures of synaptic efficacy in a manner thought by considered substrates of beneficial behavioral outcomes some to resemble ‘‘classic’’ LTP (Ben-Ari and Gho, such as learning and memory, are strikingly similar to the 1988) led to the proposal that LTP might serve as the underlying pathological changes that may lead to epilep- cellular mechanism of kindling (Slater et al., 1985; togenesis. The concept of a link between physiological Baudry, 1986; Collingridge and Bliss, 1987). In addition, and pathological plasticity is not new. Parallels have the morphological synaptic modifications accompanying previously been recognized between LTP and neuronal some forms of LTP and kindling are similar (Geinisman degeneration (Lynch and Seubert, 1989); among LTP, cell et al., 1991). However, a variety of sources have reported migration, trophic interactions, seizures, and ischemia discrepancies between the two phenomena; the most (Ben-Ari, 1995); and between mechanisms that mediate important in our view is the fact that repeated induction of neuronal sprouting during normal development and those LTP without an afterdischarge does not induce kindling described in Alzheimer’s disease (Cotman et al., 1990; (Sutula and Steward, 1986). Neill, 1995). These and other observations indicate that LTP Parallels among LTD, LTP, and kindling exist at alone is insufficient to explain the mechanisms of kin- many levels. For example, similar patterns of high- dling. Kindling is, however, a multifaceted phenomenon frequency stimulation trains lead in each case to persis- measured in part by behavioral criteria including motor tent alterations in synaptic activity levels. Further, the seizures. Nevertheless, it is evident that LTP can contrib- mechanisms involved in induction and maintenance of all ute to kindling. When LTP is first induced in the perforant three phenomena include elements of the glutamatergic path, the number of stimulations subsequently required to excitatory transmission system and associated intracellu- induce kindling is diminished (Sutula and Steward, lar signaling cascades. Recruitment of the NMDA recep- 1987). It has also been discovered that kindling can be tor (but note that NMDA receptor-independent forms of both prevented from developing and reversed by low- each exist) and an elevation of internal calcium ion frequency stimulation similar to that used to induce LTD concentration are often important in the induction phase, or depotentiation of LTP (Weiss et al., 1995). Like LTP, and ␣-amino-3-hydroxy-5-methyl-4-isoxazole-propio- there are multiple forms of kindling with distinct charac- nate (AMPA) receptor activity contributes to the expres- teristics (Lothman and Williamson, 1994), possibly reflect- sion of each (Sloviter, 1989; Dingledine et al., 1990; ing different protocols and sites of delivery of kindling Velisek and Mares, 1991; Vreugdenhil and Wadman, stimulation and subserved by cascades with unique 1992; Bliss and Collingridge, 1993). The activity of molecular components or interactions. metabotropic glutamate receptors (Bashir et al., 1993; In view of the mutual facilitation between LTP and Linden, 1994), protein kinases (Ono et al., 1994; Stevens kindling, we speculate that the induction of certain forms et al., 1994), phosphatases (Moia et al., 1994; Mulkey et of LTP can contribute directly to kindling of abnormal LTP and the Plasticity–Pathology Continuum 47 activity or can trigger processes that increase susceptibil- ity to kindling under the influence of additional factors. Interpolating back to the overlap between mechanisms of LTP and LTD, we are presented with the conundrum that activation of an initially similar subset of the cellular ‘‘machinery,’’ i.e., the glutamatergic system, can produce widely divergent effects on synaptic activity. These range from depressed activity (LTD) and an intermediate degree of enhancement (LTP), to the pathological case of excessive, synchronous firing (kindling). The physiologi- cal basis for the divergence in outcome is unknown and cannot be simply related to the magnitude of the rise in internal calcium (McEachern and Shaw, 1996a,b). Based on observed similarities among LTD, LTP, and kindling, a Fig. 1. Schematic depiction of the proposed relationships continuum of mechanistically related synaptic effects can among long-term depression (LTD), long-term potentiation be proposed and are simplistically represented in Figure 1. (LTP), and kindling. In this scheme, LTD and LTP are conventionally induced by the ‘‘appropriate’’ stimulation para- digms. Each is reversible to a prestimulus baseline (No PLASTICITY-INDUCED NEURONAL Change), and both are interconvertible via this neutral point. In DEGENERATION this scheme, kindling is viewed as an intesification of processes that give rise to LTP. In turn, as kindling leads to increasing Kindling neural activity, it may lead to the death of affected neurons by The kindling process can result in neuronal death. excitotoxic processes. The damage can be mediated by the kindling stimulation itself when the protocol employs intense stimulations and short interstimulus intervals (rapid kindling) or can be excitatory receptor activation and higher internal Ca2ϩ caused or exacerbated by seizure activity (Wasterlain and and Naϩ levels would leave the neuron vulnerable to Shirasaka, 1994). degenerative processes. Such is the case when response magnitude at a synapse is persistently increased and maintained via increased glutamatergic system function; Pathological LTP as a result there is the potential to unleash the cascade of A persistent enhancement of synaptic activity level excitotoxic/apoptotic processes just described. In addi- as a result of induction of some forms of LTP may also tion, synaptic activity appears to increase the generation trigger changes that, especially in combination with other of free radicals in neuronal mitochondria (Bindokas et al., stressors (see below), contribute to neuronal degenera- 1998). Maintenance of this form of potentiation, in tion. At this point it becomes necessary to be more precise particular long term, could result in neural dysfunction or in our treatment of the ‘‘umbrella term’’ LTP because the death. Experimental support exists for the idea that LTP specific underlying mechanism of response enhancement based on an increased magnitude of glutamatergic activa- in LTP is very relevant from the standpoint of possible tion may be pathological. Yamada (1998) studied the pathological outcome. Neuronal damage is well known to effects of benzoylpiperidine 1 (1-BCP) on hippocampal occur as a result of excitotoxic overactivation of receptors neurons in culture. 1-BCP increases field excitatory mediating excitatory neurotransmission (Cavazos et al., postsynaptic potential peak and duration in hippocampal 1994; Lipton and Rosenberg, 1994; Wasterlain and Shi- slices and also facilitates LTP and improves learning and rasaka, 1994) and the concomitant increase in intracellu- memory performance in rats and humans. Yamada found lar ion levels. Excessive influx of sodium ions (Naϩ) can that AMPA receptor function potentiated by 1-BCP pull in water molecules and result in lysis of neuronal resulted in AMPA receptor-mediated excitotoxic death of membranes (Obrenovitch and Urenjak, 1997), and high a significant proportion of the hippocampal neurons. Free intracellular calcium (Ca2ϩ) levels activate potentially radical-mediated oxidative stress may also be a causal fatal degradative processes (Olney, 1978; Choi, 1988; mechanism of neural degeneration in many of the age- Lipton and Rosenberg, 1994). Neuronal degeneration related neurological disorders (for references, see Bains may also proceed through extended activation of ‘‘death and Shaw, 1997). genes’’ reputed to figure in apoptosis (Bissonette et al., Regulatory processes designed to cause the decay 1992; Represa et al., 1995), including local dendritic of this type of LTP may be important safeguards against a apoptosis (Sugimori et al., 1995). Therefore, we would pathological outcome. In contrast to the possibly detrimen- predict that a neural change that results in greater tal result of an increase in synaptic response magnitude, 48 McEachern and Shaw we predict that response potentiation achieved by activat- ments of the glutamatergic system and the related internal ing ‘‘new infrastructure’’ (e.g., creation of new synapses ionic buildup are involved in both plastic alterations such or activation of previously quiescent ones; Isaac et al., as LTD and LTP and pathological alterations such as 1995) is the only way to improve neuronal response kindling and excitotoxic/apoptotic neural degeneration. properties without pushing cells closer to the line be- Therefore, the neural ‘‘plasticity machinery’’ is not dis- tween physiological and pathological levels of excitation. tinct from the ‘‘pathology machinery,’’and neural dysfunc- Other means of synaptic potentiation, including an in- tion and death need not necessarily result from changes creased magnitude or duration of postsynaptic response, a that are alien to the normal brain (e.g., plaques and greater ‘‘probability of hits’’ (decreased failure rate at a tangles). More importantly, the fact that the same molecu- synapse), and a decreased threshold (successfully respond- lar machinery is involved in both neuroplastic and ing to a weaker stimulus), would all seem to run the neuropathological alterations means that the ability of danger of building up excessive ionic fluxes and free neurons to undergo molecular and functional modifica- radical production. This is particularly true because LTP has a positive feed-forward nature, in that strong synaptic tion (plasticity) has a price: it leaves them vulnerable to activation resets the response to a higher level. As such, pathology. Susceptibility to pathological changes would there is an inherent danger that the process could cycle be exacerbated by induction of a form of LTP that beyond limits of neuronal tolerance. This would have the produces a persistent increase in postsynaptic response greatest likelihood of inducing pathology when (a) main- magnitude via increased ionic fluxes and would be tained long term, (b) induced in neurons with selective mitigated or avoided by the decay of this type of synaptic vulnerability, due, for example, to a low complement of potentiation. Improved response would instead be main- calcium-binding proteins, etc., and (c) combined with any tained by a morphological alteration. However, we must other stressor that compromised energy metabolism or emphasize that morphological changes can have the same antioxidant defense (trauma, hypoxia, etc.). negative impact as other forms of neuroplasticity when Does this mean that the brain avoids the use of these inappropriately induced. To illustrate, we refer to two ‘‘potentially dangerous’’ modes of strengthening syn- articles that have detailed a specific change in synaptic apses? Presumably, this would depend on the margin of morphology, the so-called perforated synapse. One re- safety separating the physiological from the pathological view (Geinisman et al., 1991) has noted that perforated range of excitation. For example, What is the difference synapse formation follows LTP induction. The other in magnitude of glutamate activation leading to Ca2ϩ and ϩ (Wolff et al., 1995) has discussed the dynamics of Na fluxes that causes excitotoxic damage compared synaptic stabilization, during which the development of with that measured in baseline physiological activity or the perforated synapse is a transient event that sometimes following LTP induction? A large difference in magnitude results in the loss of the affected synapse. If perforated would suggest that neuronal protective mechanisms such synapse creation underlies both LTP and synaptic prun- as neurotransmitter uptake, ion pumps, and Ca2ϩ buffer- ing, the possibility that LTP reflects a pathological ing would be sufficient to prevent damage except in the face of other stressors. Conversely, a small margin of process must be considered. safety would make it unlikely that such high-risk methods A greater chance of causing detrimental effects no of synaptic enhancement would be feasible as permanent doubt exists when plasticity is induced by artificial substrates for learning or memory. However, a transient stimulation paradigms because they can most likely potentiation that initiates ‘‘downstream’’ changes and overcome many of the normal safeguards that would be then decays would be less likely to have pathological present in vivo. Protective responses may be bypassed in consequences, although these might occur if induction any number of ways, for example, by using the GABAA coincided with severe neural trauma (e.g., anoxia). It is antagonist bicuculline to depress inhibition, by employ- this type of transient potentiation that we believe may be ing non-physiological stimulation and/or pathways in the useful for beneficial changes, for example, in the early absence of normal connections and context, etc. In stages of a synaptic mechanism that acts as a building addition, there is the possibility that what is being studied block for learning or memory. in such experiments is the brain’s response to injury, caused by electrode implantation or tissue sectioning. In contrast, natural learning- or memory-induced plasticity IMPLICATIONS OF SHARED MOLECULAR in the intact animal will reflect a very tightly controlled MECHANISMS IN NEUROPLASTICITY balance between the dynamics of appropriately changing AND NEUROPATHOLOGY neural response and the regulatory processes that prevent There are some very significant implications of a those changes from going beyond the range of cellular plasticity–pathology relationship. As we have seen, ele- tolerance. LTP and the Plasticity–Pathology Continuum 49

PATHOLOGICAL EXTREMES OF THE PLASTICITY–PATHOLOGY CONTINUUM The preceding section presented the case for recog- nizing the negative side of neural modifications such as kindling and some forms of LTP. Therefore, we extend the continuum of plasticity–pathology relationships to include neuronal degeneration as an extreme outcome that may occur at both ends of the spectrum when plastic alterations go awry. Whereas some forms of LTP might increase the likelihood that a neuron will succumb to disorders of overexcitation, persistent depression of neu- Fig. 2. The plasticity–pathology continuum. The speculative ral activity could have equally detrimental effects. At the scheme shown in Figure 1 has been extended to encompass extreme low end of activity, we hypothesize that there is a both excitotoxic cell death due to overactivation and a form of ‘‘disuse degeneration’’ phenomenon resulting from se- disuse degeneration. This figure elaborates on a continuum of verely diminished or inappropriate activity in a neuron events that include prolonged synaptic responses as in long- that has been deprived of normal afferent or target term depression (LTD), long-term potentiation (LTP), and kindling. In this scheme, events such as LTP induced by anoxia activation as a consequence of functional and/or morpho- (bent arrow; Gozlan et al., 1995) are part of a process that can logical disconnection (Fig. 2). This type of perturbation lead to neuronal death. As in Figure 1, LTP and kindling fall could result in the loss of the benefits of neurotrophic along a continuum of excitatory events whose outcome may be factors, molecules that play a key role in neuronal excitotoxic cell death. At the other end of the continuum, survival and repair (Oppenheim, 1989), and whose tro- prolonged LTD leads to a form of disuse degeneration. phic action may be derived from appropriate neural activity. Although speculative, this process has a precedent safety between physiological and detrimental decreases of sorts in the developmental elimination of neurons that in neural activity. make insufficient or inappropriate synaptic contacts We note also that hyperexcitation and disuse pro- (Clarke, 1985; Oppenheim, 1989). In ocular dominance cesses may occur simultaneously for interactive neural plasticity, for example, geniculostriate afferents from an populations in disease states, with both processes leading occluded eye become disconnected from their target to neural degeneration, albeit with different underlying neurons in cortical layer IV during a ‘‘critical period’’ in molecular mechanisms and time courses. This is because early development (Movshon and Van Sluyters, 1981; degeneration of neurons either by excitotoxic/apoptotic Collingridge and Singer, 1990). More importantly from processes or through disuse could lead to further disuse the perspective of our model, apoptosis also appears to degeneration in synaptically associated cells that become occur in adult neurons as a result of abnormal afferent disconnected. This process is perhaps reflected in the input caused by a noninvasive manipulation. Adult mice progressive nature of many neurodegenerative diseases. housed on a slowly revolving (0.8 revolutions/min) Sprouting of new connections to deprived neurons may turntable had significant numbers of dying neurons in the retard the loss of cells due to underactivity. Alternatively, vestibular nuclei after only 48 hr compared with non- they may cause further pathology (Wasterlain and Shi- rotated controls (Mitchell et al., 1995). Observations rasaka, 1994), possibly by contributing to abnormal from another source may also bear upon the idea of activation patterns. Certain circuits (e.g., cortex) appear disuse-related degeneration. The apoptosis of specific to possess strong safeguards designed to prevent inappro- neurons in the dentate gyrus occurring as a result of priate plastic modifications from occurring. This could adrenalectomy can be prevented by low-intensity electro- serve the dual function of limiting superfluous informa- convulsive stimulation, which increases the expression of tion storage and susceptibility to pathological disruptions. mRNAs for certain trophic factors (Masco et al., 1995). The rescue of deprived neurons by compensatory overac- tivity is a feature that would be predicted by our model. WHAT DETERMINES WHETHER THE We speculate that disuse degeneration may occur as OUTCOME OF NEURAL MODIFICATION LTD devolves to a more severe curtailment of appropriate IS BENEFICIAL OR PATHOLOGICAL? or relevant activity (Fig. 2). As was true for a postulated We postulate that the effect along the continuum LTP–pathology link, the move from LTD to pathology that results from a given modification event depends on a would depend on its underlying mechanism, persistence, variety of factors, including the nature of the inducing the coincidence of additional stressors, and the margin of stimulus, the selective vulnerability of the neurons, and 50 McEachern and Shaw the ability of the circuit to re-regulate to maintain safe both in hippocampal neurons particularly sensitive to limits of activity. seizure activity-induced degeneration (Sloviter, 1989) and in motor neurons that degenerate in the early stages of amyotrophic lateral sclerosis (Alexianu et al., 1994). Stimulation Effects Circuit characteristics that determine selective vulnerabil- The nature of the stimulus may include characteris- ity may include developmental changes in the number tics of intensity, frequency, duration, number of repeti- and type of connections and sprouting of new connec- tions, level of postsynaptic depolarization, magnitude of tions as a reaction to injury. rise in internal Ca2ϩ and Naϩ, and spatiotemporal pattern of stimulation, not to mention as yet unknown factors. Trauma Stimulus-dependent induction of synaptic response changes ranging from LTD to LTP has been observed by Any perturbation that affects normal neural func- various groups. Dudek and Bear (1992) found that tion, especially by compromising energy state and main- stimulation at a frequency of 1–3 Hz produced a lasting tenance of membrane potentials (e.g., ischemia, anoxia, synaptic depression in hippocampal CA1 neurons, at 10 etc.), will have great influence on the outcome of a plastic Hz there was no net change in response, but stimulation at alteration. Under these conditions, even normal levels of 50 Hz produced a persistent potentiation. Similarly, the neural activity can cause neural degeneration by excitotox- same tetanic stimulus that induces LTP can induce LTD icity (Meldrum, 1993) or activation of a cell death under conditions of (a) low postsynaptic depolarization program (Gozlan et al., 1995); the effects of prolonged (Artola et al., 1990), (b) decreased extracellular Ca2ϩ synaptic potentiation would be expected to be worse. (Mulkey and Malenka, 1992), or (c) synaptic activation delayed relative to postsynaptic depolarization. As an Age added complication, the nature of the plastic outcome Many of the traits of neurons and circuits, their also depends on the previous history of the activity in the ability to undergo plastic modification, and their vulner- synapse (e.g., Freeman, 1981, 1983), a previous observa- ability and response to injury are age dependent. Below, tion recently named ‘‘metaplasticity’’(Abraham and Tate, we provide a partial list; a more detailed presentation is 1997). The next stages in the continuum are achieved provided in the Appendix. Among differences between with stronger stimulation. Kindling is induced by stimula- young and adult CNSs are fluctuations in synaptic tion trains in a frequency range comparable to that used to number and distribution (Aghajanian and Bloom, 1967; induce LTD and LTP but at a greater intensity (typically Blue and Parnavelas, 1983), growth and/or retraction of defined by the capability to elicit an afterdischarge) and dendritic branches (Movshon and Van Sluyters, 1981), with a greater number of repetitions. Neuronal degenera- modifications of the postsynaptic density (Rao et al., tion results from rapid kindling and frequent perforant 1998), and a vast array of changes in pre- and postsynap- path stimulation by using long trains and a short inter- tic receptor characteristics (Shaw, 1996). Such wide- stimulus interval, also in the same frequency range spread differences in ‘‘hardware’’ and function result in (Sloviter, 1987, 1989). Therefore, the plastic outcome in a significant age-related differences in plasticity. Age- circuit is uniquely dependent on variations in induction dependent changes in magnitude and/or ease of induction stimulus that are for the most part poorly understood. have been documented for LTD (Mulkey and Malenka, 1992; Dudek and Bear, 1993), LTP (Harris and Teyler, 1984; Bronzino et al., 1994), and kindling (Moshe et al. SELECTIVE VULNERABILITY OF NEURONS: 1993; Trommer et al., 1994; Geinisman et al., 1995), with NEURON AND CIRCUIT CHARACTERISTICS, greater plasticity reported in young than in adult or aged TRAUMA, AGE, AND ABILITY TO REGULATE animals. It is also a well-established observation that the Neuron and Circuit Characteristics type and severity of neuronal injury that results from The predisposition of neurons or neuronal circuits various forms of CNS trauma are age dependent (Moshe, to exhibit plastic or pathological alterations may also be 1998; Villablanca et al., 1998). To illustrate, a number of dictated by neuronal phenotype and circuit characteris- features make the immature hippocampus hyperexcitable tics. Several phenotypical attributes of neurons may make and, hence, highly susceptible to kindling and epileptogen- them selectively vulnerable to pathological outcomes of esis (Wasterlain and Shirasaka, 1994), including an modification. This includes factors such as the specific NMDA receptor isoform with decreased voltage depen- complement of neurotransmitter receptors, ion channels, dence (Ben-Ari et al., 1988), an excess of recurrent Ca2ϩ-binding proteins, and free-radical scavenging mecha- excitatory connections (Swann et al., 1991), and the early nisms. For example, a paucity of two calcium-binding excitatory action of GABA (Cherubini et al., 1990). proteins, calbindin and parvalbumin, has been observed Unique age-dependent attributes of the young brain can LTP and the Plasticity–Pathology Continuum 51 also confer resistance to insult. For example, the neuronal functional decrease of each receptor type, and exposure to damage resulting from seizures is less severe relative to the depolarizing agent veratridine decreases binding to that produced in the adult as a result of a greatly receptors that promote excitation and increases binding to diminished metabolic rate and the immaturity of certain inhibitory receptors (Fig. 3; for a summary of these data aspects of the excitatory neurotransmission system (Wast- and model of ionotropic receptor regulation, see Shaw et erlain and Shirasaka, 1994; Moshe, 1998). The fact that al., 1994). These alterations would tend to return synaptic such age dependence exists in recovery from injury may function to the level that existed prior to the imposed also reflect the ability of the young nervous system to perturbations. rebuild and compensate for damage in a manner unavail- From an engineering perspective, this and several able in the adult. These observations make it all the more other key aspects of receptor regulation suggest that adult remarkable that age is often not treated as a critical neurons are trying very hard to prevent long-term changes variable in LTP experiments, a lack that severely compro- in synaptic response magnitude: (a) as stated, the regula- mises the task of understanding plastic and pathological tion of any receptor is in a direction that opposes its processes in the CNS and their relevance to behavior. function, such that stimuli that activate the receptor serve In summary, these examples begin to show the as triggers leading to decreases in functional receptor significance of the summation and interaction of poten- number; (b) excitatory and inhibitory receptor popula- tially detrimental and protective characteristics for deter- tions, e.g., GABAA and AMPA receptors, oppose each mining the degree of susceptibility to pathology in a other not only in the type of current generated but also by neuron or circuit, particularly when compounded by any heterologous receptor regulation, e.g., AMPA and cellular additional stressor (e.g., physical trauma, hypoxia, etc.) depolarizations increase GABAA receptor number and that compromises energy metabolism and maintenance of vice versa; and (c) the opposing actions of protein kinases membrane potentials. and phosphatases on receptor regulation ensure that receptor modifications induced by one are reversed by the ABILITY OF NEURONS TO RE-REGULATE other. PLASTIC CHANGES TO MAINTAIN SAFE LEVELS OF ACTIVITY Receptor Regulation in the Young An additional factor that may determine whether a Regulatory modifications in response to these stimuli particular modification leads to pathology is the ability of are not uniform across development. Rather, the magni- the affected neurons to rebalance activity to a safe level. A tude and even direction of regulation can change in an common response of dynamic systems to perturbation is age-dependent manner, such that the regulation of several homeostatic regulation, and the wide range of homeo- receptors in response to the depolarizing agent veratridine static processes observed throughout biology may act as is in the opposite direction in cortical slices from rats protective mechanisms in this regard. Determining how younger than 15 days old relative to adults. For example, the brain maintains homeostasis while undergoing change in young rat cortex, depolarizing stimulation resets AMPA and understanding how it fails are fundamental chal- receptors to higher levels and reduces GABAA receptor lenges of neuroplasticity. In the next section, we address binding, resulting in a positive feedforward, antihomeo- the specifics of the dilemma of controlled versus patho- static (or ‘‘homeodynamic’’) state (Borroni et al., 1996; logical change in the context of ionotropic receptor see also Shaw et al., 1994; Shaw, 1996). The results from regulation. immature animals complement other findings that sug- gest that the balance between excitation and inhibition MODEL OF HOMEOSTATIC PROCESSES: favors excitation in the young brain (Ben-Ari et al., 1988; IONOTROPIC RECEPTOR REGULATION Cherubini et al., 1990; Swann et al., 1991). They are also in good accord with known characteristics and mecha- Evidence for and against modification of various nisms of LTP, wherein strong synaptic activation resets measures of receptor function has been reported for LTP the response to a higher level, possibly through changes and kindling (for references, see McEachern and Shaw, in receptor function mediated by kinase activity (Shaw et 1996a,b), but no definitive conclusion has been reached al., 1994). The relatively greater magnitude and ease of in either area. In general, however, receptor regulatory induction of LTP (Harris and Teyler, 1984) and kindling mechanisms in several systems appear to act to maintain (Moshe et al., 1983) in young versus adult or aged homeostasis in the face of changing input. animals would also arise as a natural consequence of a balance tipped in favor of excitation. Receptor Regulation in the Adult The above discussion illustrates that the pattern of In living cortical slices from adult rats, agonist response to excitatory stimulation is opposite in young stimulation of AMPA and GABAA receptors results in a and adult animals: whereas the homeodynamic resetting 52 McEachern and Shaw

differences in functional relevance that could account for the distinct response patterns.

RELATIONSHIP OF RECEPTOR REGULATION TO LTP AND SYNAPTOGENESIS: ADULT VERSUS YOUNG Adult Regulation and LTP Based on strong similarities between receptor regu- lation and characteristics of LTD and LTP, we believe that receptor modifications may form the basis of these plastic phenomena. The parallels include the rapid interconvert- ibility of LTP and LTD, the opposing actions of kinases and phosphatases on each, and the opposing action of GABA on LTP induction, all of which conform to common principles of receptor regulation. However, LTP and LTD can be induced in adult neurons. This point is seemingly at odds with the regulation data showing only homeostatic regulation of adult receptor numbers. Clearly, in spite of their superficial resemblance, tetanic stimuli producing LTP and LTD and those stimuli that induce receptor regulation are not equivalent at the level of Fig. 3. Basic mechanisms of ionotropic receptor regulation in a cellular signaling. Perhaps the temporal and/or spatial ‘‘living’’cortical slice preparation. The top panel shows data from an pattern of the stimulation paradigms used to induce in vitro cortical slice preparation of adult rat in which some slices homeodynamic alterations fuel the escape from homeo- have been exposed for 30 min to 100 µM AMPAat 37°C. Following static control by uncoupling protective heterologous this incubation, AMPA-treated and control slices were assayed for regulatory mechanisms (i.e., the bidirectional regulatory alterations in AMPA receptor number and affinity by saturation balance between excitatory and inhibitory processes) that binding at 4°C using the antagonist [3H]-CNQX. The resulting normally prevent such changes from occurring nonspecifi- saturation binding curves show that prolonged agonist exposure cally or inappropriately. This is substantiated by the leads to a significant and rapid decrease in AMPA receptor number. observation that LTP is facilitated by and, in some cases, The bottom panel shows the percentage of change from control (0) of AMPAreceptor number following various treatments. Treatments dependent on manipulations that depress inhibition, includ- shown here are both AMPA stimulation and the result of prolonged ing administration of GABA receptor blockers (Artola cellular depolarization by veratridine/glutamate (100 and 10 µM, and Singer, 1987) or an initial ‘‘priming’’shock preceding respectively; both for 120 min). Both treatments led to significant the LTP-inducing train (‘‘primed burst’’ stimulation; decreases in AMPA receptor number, and both were blocked by Larson and Lynch, 1986). Likewise, an important con- specific inhibitors of the regulatory protein kinases (Rp-cAMPs for straint in kindling is selection of an interstimulation protein kinase A and KN62 for calmodulin kinase II). These data interval sufficiently long to minimize rebound inhibitory demonstrate that the normal receptor response in adult cortical processes that counter the effects of the excitatory neurons is one of homeostasis. Increasing stimulation of the receptor stimulation (Mucha and Pinel, 1977). by agonist or of the cell by depolarization downregulates receptor The lability of receptor regulation in the adult, number following the activation of particular protein kinases. Such seemingly designed to promote homeostasis, may suggest regulation might be expected to follow from the increased stimula- tion and response in neural pathways subjected to long-term additional conclusions. First, insofar as LTP and LTD potentiation–inducing stimuli. Statistical significance was deter- arise from such regulation, they are intended to be mined by one-way analysis of variance with Dunnett’s post hoc test relatively transient synaptic events only in the adult and using GraphPad Prism version 2.1. In the top panel, the curves were do not in themselves serve as memory processes. Rather, fit by least squares regression analysis using GraphPad Prism they may initiate cascades that produce long-term alter- (Pasqualotto et al., unpublished, 1999). ations in circuit activity through other means (e.g., morphological). The changes in receptor function then decay to baseline levels, with two advantages: vulnerabil- of neural response to higher levels in young is reminis- ity to neurodegenerative processes is diminished, and cent of LTP, the adult pattern of homeostatic down- flexibility of neural response is restored. Data from regulation of receptors is not consistent with the character- various sources suggest that in the hippocampus and istics of LTP. In the next section we speculate on the cortex in vivo, both artificially induced LTP and learning- LTP and the Plasticity–Pathology Continuum 53 associated synaptic potentiation and depression, are tran- sient in nature (de Jonge and Racine, 1985; Doyere et al., 1993; Abraham et al., 1995; Seidenbecher et al., 1995). In two of these studies (de Jonge and Racine, 1985; Abra- ham et al., 1995), repeated induction of LTP over the course of 5 days did not prolong the potentiation relative to paradigms using fewer stimulation sessions. These facts are consistent with the idea that induced increases in the magnitude of synaptic activity are engineered to return to a baseline homeostatic state, and strengthened response is maintained another way, such as through morphological changes distributed in a functionally rel- evant circuit. The second conclusion to be derived from the Fig. 4. Response of AMPA receptors in rat hippocampus to homeostatic nature of adult receptor regulation is that increasing number of kindling stimulations. Adult male rats certain forms of nondecremental LTP may reflect a form were unilaterally implanted with electrodes in the basolateral of bypassed homeostatic receptor regulation and may amygdala. Experimental animals received different numbers of make a neuron susceptible to pathological processes that kindling stimuli ranging from 5 to 99. Kindling stimuli were of regulation was designed to prevent. LTP has a positive 1-sec duration, 60 Hz, 400 µA, given three times per day (5 feedback property, in that strong activation of the synapse days/week). Control animals had electrodes implanted but resets the response to a higher level. As such, there is an received no stimuli. All animals were killed 36 hr after the last inherent danger that the process could cycle beyond stimulus, and the brains were removed and frozen. Coronal 3 limits of neuronal tolerance, particularly if protective sections from all animals were incubated with [ H]-CNQX to mechanisms are otherwise compromised. We previously label AMPA receptors and processed for in vitro autoradiogra- predicted that the nature of the mechanism underlying phy. The film images were quantified with computerized densitometry. The graph shows the average (N ϭ 3) percentage LTP determines whether extended maintenance is danger- of change from control AMPA receptor levels (100%) in pooled ous (McEachern and Shaw, 1996a,b). For example, samples of left and right dentate gyrus as a function of the potentiation produced by development of new synapses number of kindling stimuli. The greatest change in AMPA or activation of synapses previously silent at normal receptor number was observed during the earliest stages of resting potential (Isaac et al., 1995; Liao et al., 1995) may kindling. The Ϸ35% decrease in binding seen after five be less likely to escape control than LTP mediated by a kindling stimuli is comparable to that measured in the acute postsynaptic response of greater amplitude; the latter regulation experiments described in Figure 3. We speculate that could potentially trigger processes known to lead to the decrease in AMPA receptors reflects an attempt by the neuronal death (Olney, 1978; Choi, 1988; Lipton and system to control the level of neural response by a process of Rosenberg, 1994). Figure 4 shows an example of what we homeostatic receptor regulation. It is notable that, overall, as the believe is a case of failed receptor regulation during the number of stimulations increases, the amount of regulation declines in concert to the gradual appearance of the kindling development of kindling. response. One interpretation of these data is that kindling reflects the failure of homeostatic receptor regulation mecha- Young Regulation and LTP nisms in response to prolonged stimulation. What about receptor regulation and LTP in young neurons? In this case, receptor regulation appears to promote synaptic modifications by shifting neural circuits tic response to afferent stimulation. Studies of cortical to homeodynamic regulation. This may occur in different development in the rat have supported the view that peak ways, including (a) receptor regulation favoring increases synaptogenesis overlaps temporally with maximal levels in excitatory versus inhibitory receptors, perhaps by the of LTP and homeodynamic receptor regulation (Blue and excitatory action of both AMPA and GABA on some Parnavelas, 1983; Perkins and Teyler, 1988). A mutually young neurons; (b) an age-dependent delay in the cou- reinforcing relationship between feed-forward receptor pling of opposing heterologous receptor regulation mecha- modifications and synaptogenesis therefore may occur, nisms; and (c) age-dependent changes in the role and wherein receptor increases stimulate the formation of action of protein kinases and phosphatases (Shaw et al., additional synapses, which in turn fuels the synthesis of 1994). We speculate that positive feed-forward modifica- new receptor protein at the synapse, leading to further tions (of which LTP would be one kind) in the immature juggling of receptor setpoints and balances between brain serve specialized developmental functions such as receptor populations, in a dynamic cycle. We further promoting synaptogenesis and/or setting levels of synap- speculate that a developmental consequence of LTD 54 McEachern and Shaw and/or receptor decreases is to initiate pruning of inappro- lation is the ordinary response to small changes in input in priate or underused synapses. Assuming this cycle oper- adults, what are the special circumstances that allow the ates under constraints of activity dependence, these set point to be altered, as appears to occur during LTD, processes could function to refine and expand existing LTP, and kindling? Does the new set point become a genetically coded circuitry to adapt it to better respond to stable state about which response again fluctuates, or does the idiosyncrasies of the afferent activity that impinges it decay to the original state? If changes in receptor upon it (e.g., from the environment or other brain characteristics are substrates for the maintenance of LTD structures and connections). The cycle would wind down and LTP, does the synapse lose the flexibility to respond with a slowing of synaptogenesis and a switch to the adult to further fluctuations in input level? Or do receptor pattern of homeostatic receptor regulation, both events alterations occur only transiently as part of a cascade, possibly dictated by age-dependent changes in expression which then activates changes farther downstream (e.g., of genes encoding ‘‘young’’transcription factors, kinases/ morphological), which are responsible for the lasting phosphatases, and/or receptor subtypes, etc. Under chang- maintenance of the potentiated/depressed neural activity? ing input conditions, adult homeostatic regulation would The age-dependent induction of LTD and its relevance to act to maintain responses near a level preset during a behavior must also be examined in more detail. critical period of juvenile development, and further refinement of synaptic elements and function would SUMMARY OF FACTORS INVOLVED occur only under precisely defined and controlled condi- IN DETERMINING PLASTIC VERSUS tions. PATHOLOGICAL OUTCOME OF NEURAL MODIFICATIONS Summary of Age-Dependent Regulation and LTP In conclusion, neurons are in no way homogeneous To summarize, synaptic potentiation and depression in their plastic response to stimuli or in their degree of in both young and adult systems may promote morphologi- susceptibility to damage. Some of the factors that deter- cal change in the form of synaptogenesis and synaptic mine whether the response to a given perturbation will be pruning, respectively. However, the nature and the under- a beneficial plastic alteration (such as might underlie lying mechanisms are different. The young nervous learning or memory) or a pathological alteration (such as system is undergoing large-scale changes in connectivity might underlie various neural diseases) include the nature and setting of activity levels in response to afferent of the stimulation and the selective neuronal vulnerability stimulation. Such critical period activity may be facili- due to phenotype, traumatic influences, age, and ability to tated by the young brain’s homeodynamic plasticity- regulate. enhancing regulatory patterns. In contrast, such large- Various possibilities seem to exist regarding the scale changes in the adult would be pathological (except behavioral relevance of LTP. One is that long-term possibly in response to injury). Instead, a homeostatic potentiation of the magnitude of glutamatergic function at pattern of regulation prevails and ensures that neural a synapse is exceedingly risky, and the brain either does response potentiation and depression, possibly main- not make use of this mechanism for beneficial plastic tained by fine-tuning the infrastructure through limited alterations or quickly reverses it. A second is that the synaptogenesis and functional/morphological disconnec- margin of safety between tolerable and pathological tion of synapses, occurs only under appropriate and levels of synaptic activity is sufficiently large that this tightly controlled conditions. Failure of those controls form of synapse strengthening is only pathological in could mean that plastic modifications lead to neural combination with other stressors. A third is that certain dysfunction and/or death, including kindling and neuro- neuronal subsets are uniquely suited for the demands of nal degeneration due to overactivation (for LTP) or potentiated activity through highly efficient ‘‘clean-up’’ underactivation (for LTD). The greater flexibility of mechanisms and/or the ability to replace degenerated plasticity processes in the young, although making it neurons by new cell birth, an effect recently noted to easier for things to go wrong, would at the same time occur in adult hippocampus (Gould et al., 1998). Regions allow for better repair of damage compared with adults. of the hippocampus could, for example, be designed to Although in our view this idea is plausible, it will be act as ‘‘hair-trigger’’ zones for plastic change. Greater necessary to address the following questions: Does mortality of neurons in such structures would be balanced homeostatic regulation in the adult occur as a result of by a relatively high level of cell birth. Replacing neurons physiological deviations in activity, as it does in vitro to in a circuit that transiently triggered the lasting changes counteract large-scale, potentially pathological perturba- that were maintained elsewhere would seemingly be less tions? What are the temporal characteristics of induction complex than replacing cells in a circuit related to and maintenance of receptor alterations in response to ‘‘long-term information storage’’ (e.g., cortical?). A new physiological variations in activity? If homeostatic regu- ‘‘trigger circuit’’ neuron would only be required to take LTP and the Plasticity–Pathology Continuum 55 on relatively general properties of circuit function, whereas speculated to be intimately related to processes that lead complex connections and interrelationships would most to pathological activity and neural degeneration. ‘‘Benefi- likely need to be regained by a new ‘‘information storage cial’’ and ‘‘pathological’’ plastic effects are proposed to circuit’’ neuron. This distinction is speculative and is not lie along a continuum of synaptic alterations that result in meant to be interpreted to the literal extreme that there is a response levels ranging from severe underactivation to strict separation between some kinds of neurons that are excessive overactivation. The interrelationships between modifiable ‘‘disposable’’ neurons and others that are effects are defined by mechanistic similarity, nature of the ‘‘storage bins.’’ However, a certain amount of specializa- inducing stimulus, and selective neuronal vulnerability. tion of function would be a beneficial way to manage the This model is summarized in Figure 5. potentially neuropathological consequences of exhibiting But what is the value of a theoretical model like the a high degree of plasticity. An additional implication is plasticity–pathology continuum? To answer, we would that the search for the most durable changes should not be point to factors including technological advances that in the initial ‘‘trigger’’ site but at downstream locations. have driven much of the research in the fields of The fundamental interactions that determine whether neuroplasticity and neurodegenerative disease toward the neurons undergo plastic or pathological changes may be pursuit of individual ‘‘plasticity-inducing’’ and ‘‘pathol- equivalent to distinct ‘‘states’’ of neural function. As ogy-inducing’’ molecules (Shaw, 1996). In addition to such, they may also be formally similar to ‘‘attractor elaborating important players, these methods have pro- states,’’ a concept derived from and predicted by the vided a glimpse at the high degree of complexity that can mathematical methods of dynamics theory. A fundamen- be derived from burgeoning numbers of known receptor tal principle of dynamics theory is that the final activity of subtypes and enzyme isoforms. The potential molecular any complex interactive system will be highly dependent complexity is from current perspectives essentially infi- on that system’s initial conditions. For each state change nite, even before modulatory and regulatory influences there is a ‘‘bifurcation’’ of possible outcomes, where and systems–level interactions are factored in. Like the reversal of a change would require that prior conditions multiplicity of forms and routes that lead to altered be reinstated. Because the final outcome is dependent on synaptic strength, a particular neurodegenerative disease a precise sequence of bifurcations where certain doors is probably the manifestation of a collection of symptom- close at each step, a return to initial conditions is atically similar conditions arising from diverse, multifac- precluded. This ‘‘sensitive dependence on initial condi- toral origins. Thus, the merit of our model will lie in tions’’ may be closely related to age dependence in our focusing attention on dynamic interrelationships between model. One crucial implication of the latter is that it will cascade elements to extract unifying principles, regard- not be possible under most circumstances to restore the less of the identity of individual molecules. It is the nature properties of young nervous systems to adults, a con- of these fundamental interactions that will determine straint that will be of utmost importance for putative neuronal state change, whether altered synaptic efficacy strategies designed to induce neuroregeneration follow- or neuronal dysfunction or death. ing trauma or developmental disorders. For example, although techniques exist to transfect various gene prod- Predictions ucts into neurons, e.g., leading to the expression of particular ‘‘young’’ kinases or receptors, the induction of The proposed model leads to a number of experimen- these products out of context to other regulatory mecha- tally verifiable hypotheses that are presented in McEach- nisms unique to the immature nervous system may be ern and Shaw (1996a,b). ineffective or, worse, pathological in the adult. This problem is also reflected in the degeneration of fetal Principles, Interactions, and Conclusions mesencephalic neurons transplanted into damaged tissue The implications of a plasticity–pathology con- as replacement therapy for Parkinson’s disease (Boonman tinuum include the following. and Isacson, 1999). The elaboration of our model of a plasticity–pathology continuum by using methods and 1. Various forms of plasticity thought to be benefi- formulas designed to describe dynamical systems seems cial for learning and memory may be closely likely to be a fruitful area for future research. related at a mechanistic level to forms of ‘‘patho- logical plasticity,’’ leading to neural dysfunction and death. THE PLASTICITY–PATHOLOGY CONTINUUM 2. Mechanisms that regulate neuroplastic changes, MODEL: A SUMMARY including regulators of receptor function, are We have outlined a model wherein neuroplastic also likely to play a dominant role in the etiology mechanisms thought to underlie learning and memory are of epilepsy and neurodegenerative diseases. 56 McEachern and Shaw

4. The balance in neuronal activity apparently favors excitation in the immature nervous sys- tem. This corresponds well with the homeody- namic nature of both LTP and the regulation of several receptor types in response to depolariz- ing stimulation during this stage in development. We speculate that LTP- and LTD-like processes promote synaptogenesis and synaptic pruning, respectively, as part of a ‘‘wiring-up’’ process, i.e., ‘‘critical period,’’ in the immature brain. This critical period in the young brain is charac- terized by the ability to undergo large-scale modifications of neural connectivity through addition and/or loss of neurons and synapses in response to altered input. 5. Receptor regulation in the adult acts to maintain homeostatic synaptic function, which may pro- tect against large changes in preset connectivity, thereby allowing only fine response adjustments under strictly specified conditions; a significant change in the number of synapses or neurons in the adult brain is pathological and is produced Fig. 5. The consequences of neural stimulation in the adult only by severe or extended perturbation of brain. In this scheme, we have specified that baseline neural activity. Therefore, there may be potentially activity in the adult brain is modified on by additional stimuli beneficial and detrimental consequences of acting to increase or decrease neural response levels [inducing ‘‘failed’’ or bypassed receptor regulation for long-term potentiation (LTP) or long-term depression (LTD), respectively] in the target neural population. Each form of memory and neuropathology, respectively. neural modification is interconvertible, as illustrated. Neural 6. The principles of dynamics theory will allow modifications have two forms: transient and persistent. In the estimations of probability for remaining in or former, modifications in long-term behavior (e.g., learning) are escaping from any state along the plasticity– manifest as changes in the structure or function of the stimu- pathology continuum. lated neurons or in those of more downstream neural circuits; 7. LTP-like mechanisms are equally likely to con- neurons directly potentiated or depressed show homeostatic tribute to pathological changes as to beneficial receptor regulation that acts to return neural activity to prestimu- plastic changes, an idea conceptualized by a lation levels. Stronger stimuli may lead to persistent modifica- plasticity–pathology continuum model of mecha- tions of neural activity and the failure of receptor regulation. In nistically related effects. this view, persistent LTP leads to excitotoxic cell death, and persistent LTD leads to cell death due to disuse. ACKNOWLEDGMENTS This work was supported by an operating grant Therefore, manipulation of either beneficial plas- from the British Columbia Health Research Foundation ticity or neuropathology, therapeutic or other- (C.A.S.) and a MITACS Centre of Excellence scholarship wise, will have ramifications for the other. (J.C.M.). We thank Drs. J.S. Bains and B.A. Pasqualotto 3. Escape of plastic processes from limits of cellu- for their assistance and Dr. T.J. Teyler for insightful lar tolerance may fuel progression to a pathologi- comments on an earlier version of this manuscript. cal state and eventual neuronal degeneration. The effect along the continuum produced by activity in a synapse is hypothesized to depend REFERENCES on the nature of the stimulus and selective Abraham WC. 1997. Keeping faith with the properties of LTP. Behav neuronal vulnerability decreed by synergistic, Brain Sci 20:614. often age-dependent, interactions of phenotypic, Abraham WC, Otani, S. 1991. Macromolecules and the maintenance of regulatory and external influences. Summation long-term potentiation. In: Morrell F, editor. Kindling and synaptic plasticity: the legacy of Graham Goddard. Boston: of contributory factors could figure prominently Birkhauser. p 92–109. in the susceptibility of neurons to succumb to Abraham WC, Tate WP. 1997. Metaplasticity: a new vista across the deterioration from a plastic to a pathologic state. field of synaptic plasticity. Prog Neurobiol 52:303–323. LTP and the Plasticity–Pathology Continuum 57

Abraham WC, Christie BR, Logan B, Lawlor P, Dragunow M. 1994. Maturation of long-term potentiation in the hippocampal den- Immediate early gene expression associated with the persistence tate gyrus of the freely moving rat. Hippocampus 4:439–446. of heterosynaptic long-term depression in the hippocampus. Burette F, Jay TM, Laroche S. 1997. Reversal of LTP in the Proc Natl Acad Sci USA 91:10049–10053. hippocampal afferent fiber system to the prefrontal cortex in Abraham WC, Masonparker SE, Williams J, Dragunow M. 1995. vivo with low-frequency patterns of stimulation that do not Analysis of the decremental nature of LTP in the dentate gyrus. produce LTD. J. Neurophysiol 78:1155–1160. Mol Brain Res 30:367–372. Burgos GRG, Biali FI, Siri LCN, Cardinali DP. 1994. Effect of Aghajanian GK, Bloom FE. 1967. The formation of synaptic junctions gamma-aminobutyric acid on synaptic transmission and long- in developing rat brain: a quantitative electron microscopic term potentiation in rat superior cervical ganglion. Brain Res study. Brain Res 6:716–727. 658:1–7. Aiba A, Chen C, Herrup K, Rosenmund C, Stevens CF, Tonegawa S. Butler L, Silva A, Tonegawa S, McNamara JO. 1993. Enhanced seizure 1994. Reduced hippocampal long-term potentiation and context- susceptibility yet impairment of kindling development in ␣-cal- specific deficit in associative learning in mGluR1 mutant mice. cium calmodulin kinase II mutant mice. Soc Neurosci Abstr Cell 79:365–375. 19:1030. Alexianu ME, Ho BK, Mohamed AH, Labella V, Smith RG, Appel SH. Cain DP. 1997. LTP, NMDA, genes and learning. Curr Opin Neurobiol 1994. The role of calcium-binding proteins in selective motoneu- 7:235–242. ron vulnerability in amyotrophic lateral sclerosis. Ann Neurol Cavazos JE, Das I, Sutula TP. 1994. Neuronal loss induced in limbic 36:846–858. pathways by kindling: evidence for induction of hippocampal Artola A, Singer W. 1997. Long-term potentiation and NMDA sclerosis by repeated brief seizures. J Neurosci 14:3106–3121. receptors in rat visual cortex. Nature 330:649–652. Cherubini E, Rovira C, Gaiarsa JL, Corradetti R, Ben-Ari Y. 1990. Artola A, Brocher S, Singer W. 1990. Different voltage-dependent GABA mediated excitation in immature rat CA3 hippocampal thresholds for inducing long-term depression and long-term neurons. Int J Dev Neurosci 8:481–490. potentiation in slices of rat visual cortex. Nature 347:69–72. Choi DW. 1988. Glutamate neurotoxicity and diseases of the nervous Bains JS, Shaw CA. 1997. Neurodegenerative disorders in humans: the system. Neuron 1:623–634. role of glutathione in oxidative stress-mediated neuronal death. Clarke PGH. 1985. Neuronal death in the development of the Brain Res Rev 25:335–358. vertebrate nervous system. TINS 8:345–349. Bartlett FC. 1932. Remembering: a study in experimental and social Collingridge GL, Bliss TVP. 1987. NMDA receptors—their role in psychology, London: Cambridge University Press. long-term potentiation. TINS 10:288–293. Bashir ZI, Bortolotto ZA, Davies CH, Berretta N, Irving AJ, Seal AJ, Collingridge GL, Singer W. 1990. Excitatory amino acid receptors and Henley JM, Jane DE, Watkins JC, Collingridge G. 1993. synaptic plasticity. Trends Pharmacol Sci 11:290–296. Induction of LTP in the hippocampus needs synaptic activation Cotman CW, Geddes JW, Ulas J, Klein M. 1990. Plasticity of of glutamate metabotropic receptors. Nature 363:347–350. excitatory amino acid receptors: implications for aging and Baudry M. 1986. Long-term potentiation and kindling: similar biochemi- Alzheimer’s disease. Prog Brain Res 86:55–61. cal mechanisms? Adv Neurol 449:401–410. de Jonge M, Racine RJ. 1985. The effects of repeated induction of Ben-Ari Y. 1995. Activity-dependent forms of plasticity. J Neurobiol long-term potentiation in the dentate gyrus. Brain Res 328:181– 26:295–298. 185. Ben-Ari Y, Gho M. 1988. Long-lasting modifiacation of the synaptic Dingledine R, McBain CJ, McNamara JO. 1990. Excitatory amino acid properties of rat CA3 hippocampal neurones induced by kainic receptors in epilepsy. TIPS 111:334–338. acid. J Physiol [Lond] 404:365–384. Doyere V, Burette F, Redini-Del Negro C, Laroche S. 1993. Long-term Ben-Ari Y, Cherubini E, Krnjevic K. 1988. Changes in voltage potentiation of hippocampal afferents and efferents to prefrontal dependence of NMDA currents during development. Neurosci cortex: implications for associative learning. Neuropsychologia Lett 94:88–92. 3:1031–1053. Bindokas VP, Lee CC, Colmers WF, Miller RJ. 1998. Changes in Dudek SM, Bear MF. 1992. Homosynaptic long-term depression in mitochondrial function resulting from synaptic activity in rat area CA1 of hippocampus and effects of N-methyl-D-aspartate hippocampal slice. J Neurosci 18:4570–4587. receptor blockade. Proc Natl Acad Sci USA 89:4363–4367. Bissonette RP, Echeverri F, Mahboubi A, Green KR. 1992. Apoptotic Dudek SM, Bear MF. 1993. Bidirectional long-term modification of cell death induced by c-myc is inhibited by bcl-2. Nature synaptic effectiveness in the adult and immature hippocampus. J 359:552–554. Neurosci 13:2910–2918. Bliss TVP, Collingridge GL. 1993. A synaptic model of memory: Freeman WJ. 1981. A physiological hypothesis of perception. Perspect long-term potentiation in the hippocampus. Nature 361:31–39. Biol Med 24:561–592. Bliss TVP, Lomo T. 1973. Long-lasting potentiation of synaptic Freeman WJ. 1983. The physiological basis of mental images. Biol transmission in the dentate area of the anaesthetized rabbit Psychiatry 18:1107–1125. following stimulation of the perforant path. J Physiol [Lond] Gallistel CR. 1995. Is LTP a plausible basis for memory? In: McGaugh 232:331–356. JL, Weinberger NM, and Lynch G, editors. Brain and memory: Blue ME, Parnavelas JG. 1983. The formation and maturation of modulation and mediation of neuroplasticity. New York: Oxford synapses in the visual cortex of the rat. II. Quantitative analysis. University Press. p 328–337. J Neurocytol 12:697–712. Geinisman Y, De Toledo-Morrell L, Morrell F. 1991. Structural Boonman Z and Isacson O. 1999. Apoptosis in neural development and synaptic substrates of kindling and long-term potentiation. In: transplantation: role of capsases and trophic factors. Exp Neurol Morrell F, editor. Kindling and synaptic plasticity. Boston: 156:1–15. Birkhauser. p 124–159. Borroni A, Coussens C, and Teyler TJ. 1996. Receptor regulation in Geinisman Y, De Toledo-Morrell L, Morrell F, Heller RE. 1995. synaptic plasticity. In: Shaw CA, editor. Receptor dynamics in Hippocampal markers of age-related memory dysfunction: neural development. Boca Raton: CRC Press. p 239–242. behavioral, electrophysiological and morphological perspec- Bronzino JD, Abuhasaballa K, Austinlafrance RJ, Morgane PJ. 1994. tives. Prog Neurobiol 45:223–252. 58 McEachern and Shaw

Gerren RA, Weinberger NM. 1983. Long-term potentiation in the a rapidly developing, transient kindling and a slowly develop- magnocellular medial geniculate nucleurs of the anaesthetized ing, enduring kindling. Brain Res 649:71–84. cat. Brain Res 265:138–142. Lynch G, Seubert P. 1989. Links between long-term potentiation and Goddard GV, Douglas RM. 1975. Does the engram of kindling model neuropathology. An hypothesis involving calcium-activated the engram of normal long-term memory? Can J Neurol Sci proteases. Ann NY Acad Sci 568:171–180. 2:385–394. Martin SJ. 1998. Time-dependent reversal of dentate LTP by 5Hz Gould E, Tanapat P, McEwen BS, Flugge G, Fuchs E. 1998. Prolifera- stimulation. NeuroReport 9:3775–3781. tion of granule cell precursors in the dentate gyrus of adult Masco D, Sahibzada N, Gale K. 1995. Electroconvulsive shock monkeys is diminished by stress. Proc Natl Acad Sci USA prevents apoptotic and excitotoxic cell death. Soc Neurosci 95:3168–3171. Abstr 21:811. Gozlan H, Khazipov R, Ben-Ari, Y. 1995. Multiple forms of long-term McEachern JC, Shaw CA. 1996a. An alternative to the LTP orthodoxy: potentiation and multiple regulatory sites of N-methyl-D- a plasticity–pathology continuum model. Brain Res Rev 22: aspartate receptors: role of the redox site. J Neurobiol 26:360– 51–92. 369. McEachern JC, Shaw CA. 1996b. Does LTP correlate with memory or Grant SGN, O’Dell TJ, Karl KA, Stein PL, Soriano P, Kandel ER. neuropathology? Or what happens when homeostatic receptor 1992. Impaired long-term potentiation, spatial learning, and regulation fails? In: Shaw CA, editor. Receptor dynamics in hippocampal development in fyn mutant mice. Science 258: neural development. Boca Raton: CRC Press. p 253–301. 1903–1910. McNamara JO. 1994. Cellular and molecular basis of epilepsy. J Hargreaves EL, Cain DP. 1992. Hyperactivity, hyper-reactivity, and Neurosci 14:3413–3425. sensorimotor deficits induced by low doses of the N-methyl-D- aspartate non-competitive channel blocker MK801. Behav Meldrum B. 1993. Amino acids as dietary excitotoxins: a contribution Brain Res 47:23–33. to understanding neurodegenerative disorders. Brain Res Rev Harris KM, Teyler TJ. 1984. Developmental onset of long-term 18:293–314. potentiation in area CA1 of the rat hippocampus. J Physiol Miles R, Wong RKS. 1987. Latent synaptic pathways revealed after [Lond] 346:27–48. tetanic stimulation in the hippocampus. Nature 329:724–726. Hughes PE, Alexi T, Walton M, Williams CE, Dragunow M, Clark RG, Mitchell IJ, Cooper AJ, Brown, GDA, Waters CM. 1995. Apoptosis of Gluckman PD. 1999. Activity and injury-dependent expression neurons in the vestibular nuclei of adult mice results from of inducible transcription factors, growth factors and apoptosis- prolonged change in the external environment. Neurosci Lett related genes within the central nervous system. Prog Neurobiol 198:153–156. 57:421–450. Moshe SL. 1998. Brain injury with prolonged seizures in children and Isaac JTR, Nicoll RA, Malenka RC. 1995. Evidence for silent adults. J Child Neurol 13(Suppl 1):S3–S6. synapses—implications for the expression of LTP. Neuron Moshe SL, Albala BJ, Ackermann RF, Engel J Jr. 1983. Increased 15:427–434. seizure susceptibility of the immature brain. Brain Res 283: Ito K-I, Skinkle KL, Hicks TP. 1999. Age-dependent, steroid-specific 81–85. effects of oestrogen on long-term potentiation in rat hippocam- Moia LJMP, Matsui H, Debarros GA, Tomizawa K, Miyamoto K, pal slices. J Physiol 515:209–220. Kuwata Y,Tokuda M, Itano T, Hatase O. 1994. Immunosuppres- Jeffery KJ, Abraham WC, Dragunow M, Mason SE. 1990. Induction of sants and calcineurin inhibitors, cyclosporin A and FK506, Fos-like immunoreactivity and the maintenance of long-term reversibly inhibit epileptogenesis in amygdaloid kindled rat. potentiation in the dentate gyrus of unanesthetized rats. Mol Brain Res 648:337–341. Brain Res 8:267–274. Morris RGM, Anderson E, Lynch GS, Baudry M. 1986. Selective Kamal A, Biessels GJ, Gispen WH, Urban IJA. 1998. Increasing age impairment of learning and blockade of long-term potentiation reduces expression of long-term depression and dynamic range by an N-methyl-D-aspartate receptor antagonist, AP5. Nature of transmission plasticity in CA1 field of the rat hippocamus. 319:774–776. Neuroscience 83:707–715. Movshon JA, Van Sluyters RC. 1981. Visual neural development. Annu Kavanau JL. 1994. Sleep and dynamic stabilization of neural circuitry: Rev Psychol 32:477–522. a review and synthesis. Behav Brain Res 63:111–126. Mucha RF, Pinel JPJ. 1977. Postseizure inhibition of kindled seizures, Keith JR, Rudy JW. 1990. Why NMDA-receptor-dependent long-term Exp Neurol 54:266–282. potentiation may not be a mechanism of learning and memory: Mulkey RM, Malenka RC. 1992. Mechanisms underlying induction of reappraisal of the NMDA-receptor blockade strategy. Psychobi- homosynaptic long-term depression in area CA1 of the hippo- ology 18:251–257. Larson J, Lynch G. 1986. Induction of synaptic potentiation in campus. Neuron 9:967–975. hippocampus by patterned stimulation involves two events. Mulkey RM, Endo S, Shenolikar S, Malenka RC. 1994. Involvement of Science 232:985–988. a calcineurin/inhibitor-1 phosphatase cascade in hippocampal Larson J, Xiao P, Lynch G. 1993. Reversal of LTP by theta frequency long-term depression. Nature 369:486–488. stimulation. Brain Res 600:97–102. Neill D. 1995. Alzheimer’s disease: maladaptive synaptoplasticity Liao DZ, Hessler, NA, Malinow R. 1995. Activation of postsynapti- hypothesis. Neurodegeneration 4:217–232. cally silent synapses during pairing-induced LTP in CA1 region Norris CM, Korol DL, Foster TC. 1996. Increased susceptibility to of hippocampal slice. Nature 37:400–404. induction of long-term depression and long-term potentiation Linden DJ. 1994. Long-term synaptic depression in the mammalian reversal during aging. J Neurosci 16:5382–5392. brain. Neuron 12:457–472. Obrenovitch TP, Urenjak J. 1997. Altered glutamatergic transmission Lipton SA, Rosenberg PA. 1994. Excitatory amino acids as a final in neurological disorders: from high extracellular glutamate to common pathway for neurologic disorders. N Engl J Med excessive synaptic efficacy. Prog Neurobiol 51:39–87. 330:613–622. Olney JW. 1978. Neurotoxicity of excitatory amino acids. In: McGeer Lothman EW, Williamson JM. 1994. Closely spaced recurrent hippo- EG, Olney JW, McGeer PL, editors. Kainic acid as a tool in campal seizures elicit two types of heightened epileptogenesis: neurobiology. New York: Raven Press. p 95–121. LTP and the Plasticity–Pathology Continuum 59

Ono M, Akiyama K, Tsutsui K, Kuroda S. 1994. Differential changes in Stevens CF. 1998. A million dollar question: does LTP ϭ memory? the activities of multiple protein kinase C subspecies in the Neuron 20:1–2. hippocampal-kindled rat. Brain Res 660:27–33. Stevens CF, Tonegawa S, Wang Y. 1994. The role of calcium- Oppenheim RW. 1989. The neurotrophic theory and naturally occur- calmodulin kinase II in three forms of synaptic plasticity. Curr ring motoneuron death. TINS 12:252–255. Biol 4:687–693. Perkins AT IV, Teyler TJ. 1988. A critical period for long-term Sugimori M, Cherksey BD, Llinas R. 1995. Dendritic apoptosis: a new potentiation in the developing rat visual cortex. Brain Res mechanism for restricted neuroal death. Soc Neurosci Abstr 439:222–229. 21:2019. Qian Z, Gilbert ME, Colicos MA, Kandel ER, Kuhi D. 1993. Sutula T, Steward O. 1986. Quantitative analysis of synaptic potentia- Tissue-plasminogen activator is induced as an immediate early tion during kindling of the perforant path. J Neurophysiol gene during seizure, kindling and long-term potentiation. Na- 56:732–746. ture 361:453–457. Swann JW, Smith KL, Brady RJ. 1991. Age-dependent alterations in Randic M, Jiang MC, Cerne R. 1993. Long-term potentiation and the operations of hippocampal neural networks. Ann NY Acad long-term depression of primary afferent neurotransmission in Sci 627:264–276. the rat spinal cord. J Neurosci 13:5228–5241. Trommer BL, Pasternak JF, Nelson PJ, Colley PA, Kennelly JJ. 1994. Rao A, Kim E, Sheng M, Craig AM. 1998. Heterogeneity in the Perforant path kindling alters dentate gyrus field potentials and molecular composition of excitatory postsynaptic sites during paired pulse depression in an age-dependent manner. Dev Brain development of hippocampal neurons in culture. J Neurosci Res 79:115–121. 18:1217–1229. Velisek L, Mares P. 1991. Increased epileptogenesis in the immature Represa A, Niquet J, Pollard H, Ben-Ari Y. 1995. Cell death, gliosis, hippocampus. In: Klee MR, Lux HD, Speckmann E-J, editors. and synaptic remodeling in the hippocampus of epileptic rats. J Physiology, pharmacology and development of epileptogenic Neurobiol 26:413–425. phenomena. Series 20, Experimental brain research. Berlin: Robinson GS Jr, Crooks GB Jr, Shinkman PG, Gallagher M. 1989. Springer-Verlag. p 183–185. Behavioral effects of MK-801 mimic deficits associated with Villablanca JR, Carlson-Kuhta P, Schmanke TD, Houda DA. 1998. A hippocampal damage. Psychobiology 17:156–164. critical maturational period of reduced brain vulnerability to Sazgar M, Chick BA, Rashid K, Van der Zee CEEM, Diamond J, developmental injury. I. Behavioral studies in cats. Dev Brain Fahnestock M, Racine RJ. 1995. Role of nerve growth factor in Res 105:309–324. kindling and kindling-induced mossy fiber sprouting. Soc Vreugdenhil M, Wadman WJ. 1992. Enhancement of calcium currents Neurosci Abstr 21:1973. in rat hippocampal CA1 neurons induced by kindling epilepto- Seidenbecher T, Balshun D, Reymann KG. 1995. Drinking after water genesis. Neuroscience 49:373–381. deprivation prolongs ‘‘unsaturated’’ LTP in the dentate gyrus of Wagner JJ, Alger BE. 1996. Homosynaptic LTP and depotentiation: do rats. Physiol Behav 57:1001–1004. they differ in name only? Hippocampus 6:24–29. Shaw CA. 1996. Age-dependent expression of receptor properties and Wasterlain CG, Shirasaka Y. 1994. Seizures, brain damage and brain function in CNS development. In: Shaw CA, editor. Receptor development. Brain Dev 16:279–295. dynamics in neural development. Boca Raton: CRC Press. p Weiss SR, Rosen JB, Li H, Heyner T, and Post RM. 1995. Quenching: 3–17. inhibition of development and expression of amygdala kindled Shaw CA, Lanius RA, van den Doel K. 1994. The origin of synaptic seizures with low frequency stimulation. Neuroreport 6:2171– neuroplasticity: crucial molecules or a dynamical cascade? 2176. Brain Res Rev 19:241–263. Wolff JR, Laskawi R, Spatz WB, Missler M. 1995. Structural dynamics Shaw CA and McEachern JC. 2000. Traversing levels of organization: of synapses and synaptic components. Behav Brain Res 66: neural stability and plasticity. In: Shaw CA and McEachern JC, 13–20. editors. Toward a theory of neuroplasticity. Philadelphia: Taylor Yamada, K. 1998. AMPA receptor activation potentiated by the AMPA and Francis Publishers. In press. modulator 1-BCP is toxic to cultured rat hippocampal neurons. Shors TJ, Matzel LB. 1997. Long-term potentiation: what’s learning Neurosci Lett 249:119–122. got to do with it? Behav Brain Sci 20:597–655. Silva AJ, Paylor R, Wehner JM, Tonegawa S. 1992a. Impaired spatial learning in ␣-calcium calmodulin kinase II mutant mice. Science 257:206–211. APPENDIX: AGE-DEPENDENCE AND LTP Silva AJ, Stevens CF, Tonegawa S, Wang Y. 1992b. Deficient Much of the neuroplasticity literature suffers from a hippocampal long-term potentiation in ␣-calcium-calmodulin kinase II mutant mice. Science 257:201–206. fundamental and, we believe, fatal flaw, namely the Slater NT, Stelzer A, Galvan M. 1985. Kindling-like stimulus patterns failure to treat age as a critical variable in studies of LTP induce epileptiform discharges in the guinea pig in vitro phenomenology and possible behavioral relevance. This hippocampus. Neurosci Lett 60:25–31. flaw produces two sets of problems. The first arises as a Sloviter RS. 1987. Decreased hippocampal inhibition and a selective result of the fact that experimental age has an enormous loss of interneurons in experimental epilepsy. Science 235: 73–76. impact on several characteristics of LTP, including ease Sloviter RS. 1989. Calcium-binding protein (calbindin-D28K) and of induction, amount, duration, and induction of LTD parvalbumin immunocytochemistry: localization in the rat hip- versus LTP by a given stimulation. Furthermore, the pocampus with specific reference to the selective vulnerability molecular mechanisms underlying LTP expression differ of hippocampal neurons to seizure activity. J Comp Neurol at different ages. Because experimental age is not stan- 280:183–196. Staubli U, Thibault O, DiLorenzo M, Lynch G. 1989. Antagonism of dard among researchers, any attempt to synthesize the NMDA receptors impairs acquisition but not retention of vast literature into an overall view of the nature and olfactory memory. Behav Neurosci 103:54–60. mechanisms of LTP is precluded. 60 McEachern and Shaw

The second set of problems is due to the likelihood modulate receptor function. This last has as an immediate that neuroplastic modifications serve a developmental consequence a major effect on amount and direction of function in young animals in addition to or even separate receptor regulation. As we have described elsewhere from their actions in the adult CNS. The problem is (Shaw et al., 1994), similar stimuli produce opposite dissociating developmental plasticity subserving geneti- effects on a number of receptor populations as a function cally programmed and activity-dependent ‘‘wiring-up’’ of age. To illustrate, sustained cellular depolarization processes from the concomitant neuroplastic changes reduces excitatory AMPA receptors and increases inhibi- subserving learning and memory in young animals. That tory GABAA receptors in adult rat cortex, yet in young is, it will be difficult to correlate any LTP effect observed cortex the same stimulus increases AMPA and reduces in a young animal specifically to learning and/or memory, GABAA receptor numbers. In other words, the effects of as is often claimed in the literature. The following such stimulation will be to reduce excitability in the adult discussion develops these points in greater detail. cortex and increase it in the young. It is rather difficult to In much of the LTP literature, the assumption that imagine that such age-dependent dynamics could fail to LTP is the substrate of various types of behavioral have a significant impact on the short- and long-term neuroplasticity (e.g., learning and memory) is taken for neuronal responses to LTP-inducing stimulation. The granted (for citations, see McEachern and Shaw, 1996a,b). rather routine need to suppress GABAA receptor function Although we do not accept this view, we will for the before inducing LTP in adult neural circuits would seem moment allow it to stand. Just how important is the to support this point (e.g., Artola and Singer, 1987). variable of age to neuroplasticity? We note, first, that Although this example has provided details only for age-dependent forms of neuroplasticity have been widely age-dependent differences in the regulation of receptors, documented in a great number of mammalian neural equally profound modifications of features of neuronal circuits and systems (for a review, see Shaw et al., 1994). structure, biochemistry, and function can also be found in Early postnatal life often contains various ‘‘critical peri- abundance. ods,’’ and it is in just such periods that the dynamics of Without question, such differences must be ad- neuronal modification appear to be most pronounced. It is dressed by those attempting to find the underlying quite widely acknowledged that age-dependent modifica- mechanism and physiological role of LTP and must call tions of neuronal activity and function occur in large part into question any simple assumption that LTP in its because of the very dramatic alterations in neuronal various forms and underlying mechanisms can be studied structure and biochemistry that occur during develop- independently of age. Given this, it is remarkable that ment and aging. Included in a partial list of differences many in the LTP field choose to proceed as if this were between young and adult CNSs are fluctuations in not the case. As cited by us (McEachern and Shaw, synaptic number and distribution (Aghajanian and Bloom, 1996a,b), some workers in the field routinely ignore the 1967; Blue and Parnavelas, 1983), growth and/or retrac- issue altogether by mixing experimental ages or even tion of dendritic branches (Movshon and Van Sluyters, failing to cite them. 1981), modifications of the postsynaptic density (Rao et Young animals are preferred for neuroplasticity al., 1998), and a vast array of changes in pre- and research because of the greater ease of manipulation and postsynaptic receptor characteristics (Shaw, 1996). To preparation of tissue from these animals, and because of expand upon just the latter, age-dependent receptor the apparently greater viability of the tissue in acute modifications can involve a range of rather fundamental experiments. In addition, young brain tissues usually features, each of which would be expected to have a produce more robust, clearer LTP. The reasons for this are major impact on neuronal functions involved in activity- not clear but may arise due to less anoxic damage during dependent modifications. Of these, receptor modifica- preparation, a legitimately greater level of LTP, or both. tions leading to changes in cell sensitivity, e.g., number What makes this situation even more remarkable is that and affinity, are obvious. In addition, however, many workers in the field should not be ignorant of the receptor populations undergo rather profound alterations widespread descriptions of age-dependent LTP (Harris in overall regional and cellular distribution. In concert and Teyler, 1984; Kamal et al., 1998; Ito et al., 1999), with alterations in number, changing distributions affect which appears to display properties in the young that the ratios of the various populations, leading inevitably to differ in important ways from those in adults. The net alterations in cellular response properties. Further, chan- result of all of this is that there is no way to determine nel kinetics may be developmentally affected, often by from the current literature if LTP in adult neurons is changes in subunit composition. Not least, second- closely related to that in younger animals in induction messenger type and coupling to receptors appear to be in parameters, amount, mechanism, function, or durability. flux during development, as do the activities of the Rather, the few studies that do exist in which young and various protein kinases and phosphatases that act to adult LTP are compared suggest significant differences LTP and the Plasticity–Pathology Continuum 61

(McEachern and Shaw, 1996a,b). Regarding induction, background of evolving genetic instruction in dynamic some workers would clearly hope that age differences interplay with experience. Developmental alterations of could be resolved simply, e.g., by the notion that optimum the nervous system involve large-scale remodeling of parameters of stimulation differ as a function of age while synapses and neural cells, during which both synapses leaving the basic mechanisms and function the same. This and neurons are added and removed as an animal may, in fact, be true, but experiments have not to our develops. In contrast, large-scale neural death and remod- knowledge been done to demonstrate it. The absence of eling in adulthood is considered to be a consequence of such experiments leaves open the real prospect that LTP pathology and, although it now seems that neurons are mechanisms and function studied in developing animals born continuously in some regions of the nervous system may bear little to no relation to that in adults. As such, the throughout life (Gould et al., 1998), it will be obvious that nature and mechanisms of LTP in young animals should replacing some neurons in a circuit is not the same as the not be presented as a general template of neuroplasticity, large-scale process of circuit formation in the young as is the case when age is not reported. A concerted effort brain. It is a well-established observation that the type should also be made to more systematically compare LTP induced at different developmental stages. Recent studies and severity of neuronal injury that results from various have made an effort to address this issue (Norris et al., forms of CNS trauma is age dependent (Moshe, 1998; 1996; Kamal et al., 1998; Ito et al., 1999). Villablanca et al., 1998). The fact that such age depen- The attempt to formulate strong correlations be- dence exists in response to injury may reflect the ability tween LTP and learning or memory based on studies in of the young nervous system to rebuild and compensate young animals is problematic because aspects of develop- for damage in a manner unavailable in the adult. ment add many nonspecific variables to the equation, This discussion should have made clear that the only one of which is activity-dependent neuroplasticity. It variable of age is not only not trivial but is in fact an is clear that developmental alterations in neuronal func- absolute ‘‘show-stopper’’ for the task of understanding tion are not solely activity dependent. Rather, neuroplas- plastic and pathological processes in the CNS and their tic modifications in development play out against a relevance to behavior.