472092.qxd 13/9/03 6:01 PM Page 1735 Neurochemical Research, Vol. 28, No. 11, November 2003 (© 2003), pp. 1735–1742 Stress and Plasticity in the Limbic System* Robert M. Sapolsky1 (Accepted March 4, 2003) The adult nervous system is not static, but instead can change, can be reshaped by experience. Such plasticity has been demonstrated from the most reductive to the most integrated levels, and understanding the bases of this plasticity is a major challenge. It is apparent that stress can alter plasticity in the nervous system, particularly in the limbic system. This paper reviews that sub- ject, concentrating on: a) the ability of severe and/or prolonged stress to impair hippocampal- dependent explicit learning and the plasticity that underlies it; b) the ability of mild and transient stress to facilitate such plasticity; c) the ability of a range of stressors to enhance implicit fear conditioning, and to enhance the amygdaloid plasticity that underlies it. KEY WORDS: Stress; hippocampus; glucocorticoids; amygdala; LTP; LTD. INTRODUCTION level, it is the truly revolutionary finding that learning, environmental enrichment, even exercise can stimulate In 1967’s The Graduate, Dustin Hoffman, embark- neurogenesis. As perhaps the most important cornerstone ing on life postcollege, was given some unwanted career of such plasticity, these instances of experience-dependent advice—plastics. And the field of neural plasticity has modification of the nervous system can occur throughout yet to recover fully from this setback. the lifetime. We all responded to that bit of advice with a snicker, Carl Cotman has made seminal contributions to this based on the pejorative view of “plastic” as artificial, un- topic, helping to make it one of the most exciting branches natural, cold, unyielding. And the problem is that neural of neuroscience. But neural plasticity has a dark side. It plasticity traditionally implies anything but that. Instead, is not the banality of “plastics,” but instead, the undesir- it is a good thing. Specifically, it is a field built around able realm where “neural plasticity” means that experi- the fact that experience can alter the nervous system adap- ence is causing involution, impairment, and damage to the tively, enhancing function in self-perpetuating ways. At nervous system. This can include impairment of LTP, the level of the synapse, this is the world of long-term retraction of dendritic processes, inhibition of neurogene- potentiation and related electrophysiological phenomena. sis, and even the death of neurons. At the cytoarchitectural level, it is the demonstration that In principle, this need not be particularly interesting, neurons can respond to the proper stimuli by forming new the fact that there can be “good” and “bad” aspects to synapses, by elaborating dendritic processes. At the cellular neural plasticity. For example, there can be “good” and “bad” aspects to, say, the neurobiological consequences of things that we humans can ingest. Thus, ingest a well- * Special issue dedicated to Dr. Carl Cotman. balanced diet and you promote proper neural develop- 1Departments of Biological Sciences, and of Neurology and Neuro- logical Sciences, Stanford University, Gilbert Laboratory, MC 5020, ment; ingest a diet with vast excesses of alcohol and you Stanford, California 94305-5020. Tel: 650-723-2649; E-mail: sapol- promote neuron death. This is a fairly unsubtle contrast. [email protected] What is fascinating in the realm of the adaptive and 1735 0364-3190/03/1100–1735/0 © 2003 Plenum Publishing Corporation 472092.qxd 13/9/03 6:01 PM Page 1736 1736 Sapolsky maladaptive features of experience-dependent neural plas- the stress range also disrupt spatial memory perform- ticity is how similar the experiences can be in bringing ance (6–8). Such impaired performance could reflect about quite contrasting outcomes. Depending on changes impairment of the initial consolidation of the spatial in the magnitude or duration of the stimulus, the indi- information and/or the retrieval of it. Recent work sug- vidual who is experiencing the stimulus, or the part of gests that it is the retrieval component that is most sen- the brain being considered, the outcome can be neural sitive to the disruptive effects of GCs (9,10). plasticity of the “good” kind or of the “bad.” An emerging literature demonstrates that GCs can In this review I will first consider the basic find- disrupt hippocampal-dependent declarative memory per- ings regarding how one aspect of experience—the expe- formance in humans as well. Some of these studies exam- rience of stress—can have adverse effects on neural ine humans treated chronically with exogenous GCs to plasticity. I will then consider some parameters in which control an autoimmune, or inflammatory disorder, or an stress does not always have such adverse effects and immune cancer (11,12). Moreover, declarative memory can even promote versions of the classically “good” performance in Cushing’s syndrome patients (in which forms of neural plasticity. GCs are hypersecreted secondary to any of a number of types of tumors) is impaired (13). A fascinating literature of aged humans demonstrate that those whose basal GC levels increase most dramatically with age over time, or DISCUSSION increase most dramatically in response to an acute stres- sor, have the poorest declarative memory performance Glucocorticoids and Their Adverse Effects on (14–21). Finally, treatment of healthy volunteers with Hippocampal-Dependent Cognition exogenous GCs in the range used in clinical medicine Glucocorticoids (GCs) are the adrenal steroid hor- impairs declarative memory performance as well (22–29). mones secreted in response to stress. The hormones are As with the rodent studies, the impaired performance in central to successfully coping with a major physical stres- the hippocampal-dependent tasks could represent impair- sor (such as fleeing a predator), as they mobilize stored ment of consolidation and/or retrieval; as with rodents, it energy, increase cardiovascular tone, and suppress costly appears as if the retrieval component is most sensitive anabolism (such as growth, tissue repair, reproduction, (30). As important controls, a number of these studies digestion and immunity) for more auspicious times. How- demonstrating impairment of hippocampal-dependent cog- ever, if the exposure to GCs is prolonged, there are a nition also demonstrated that hippocampal-independent variety of pathological outcomes that become more likely, cognition remained intact (11,29). including insulin-resistant diabetes, hypertension, Thus stress and/or exposure to elevated GC concen- immunosuppression, and reproductive impairments (1). trations disrupt hippocampal-dependent cognition while These deleterious consequences include adverse sparing hippocampal-independent cognition, in both effects in the nervous system. The most dramatic ones rodents and humans. occur in the hippocampus, a primary GC target, with ample quantities of corticosteroid receptors. Mechanisms Underlying These Adverse GC Effects At the most integrated level, sustained stress or exposure to GCs can impair aspects of hippocampal- There is considerable information regarding the dependent cognition. Memory is not a monolithic mechanisms contributing to these disruptive GC actions. phenomenon; instead, there are a number of types of As an initial critical observation, GCs and stress impair memory, with the medial temporal lobe, and particularly the synaptic plasticity essential to hippocampal-dependent the hippocampus within it, playing a critical role in cognition. Thus stress disrupts long-term potentiation explicit memory (concerned with facts and events) (2). (LTP) and primed burst potentiation (PBP) in a variety Thus stressors as different as a number of weeks of daily of hippocampal cell fields in vivo (31–37), with the sug- restraint stress, brief exposure to the smell of a preda- gestion that PBP is more vulnerable to this effect than tor (a cat), or months of rotating group membership is LTP (38). Moreover, administration of exogenous disrupt spatial memory performance in rats (a classic GCs in a regimen producing circulating concentrations hippocampal-dependent explicit memory task in a typical of stress also disrupt LTP and PBP (36,39–42). rodent) (3–5). The stress-induced GC secretion in these In addition, both premortem stress and in vitro GC expo- instances appears to mediate the disruptive effects. As sure disrupt LTP in hippocampal slices in vitro (38,43). evidence, similar time courses of administration of There are two receptors for GCs found in the brain exogenous GCs producing circulating levels typical of (with ample concentrations of both in the hippocampus), 472092.qxd 13/9/03 6:01 PM Page 1737 Stress and Plasticity in the Limbic System 1737 with the high-affinity mineralocorticoid receptor (MR) gesting the reversible atrophy of processes seen in the occupied heavily under basal conditions, and the low animal studies. affinity glucocorticoid receptor (GR) occupied heavily Another somewhat controversial route by which only during major stressors. Heavy occupancy of GR stress and GCs may impair cognition has emerged in mediates these disruptive effects of stress and GCs upon recent years. The acceptance by the neuroscience com- LTP (39,42,44). Such GR occupancy leads to increased munity that the early, heretical reports of adult neuroge- calcium conductance in hippocampal neurons; this in nesis in the hippocampus are true represents a revolution turn leads to prolonged