The selfish brain: competition for energy resources

Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 www.elsevier.com/locate/neubiorev

Review

The selfish brain: competition for energy resources A. Petersa,*, U. Schweigerb, L. Pellerine, C. Hubolda, K.M. Oltmannsb, M. Conradc, B. Schultesa, J. Bornd, H.L. Fehma a

Department of Internal Medicine, University of Luebeck, Ratzeburger Allee 160, D-23538 Germany b Psychiatry and Psychotherapy, University of Luebeck, Ratzeburger Allee 160, D-23538 Germany c Institute of Mathematics, University of Luebeck, Ratzeburger Allee 160, D-23538 Germany d Institute of Neuroendocrinology, University of Luebeck, Ratzeburger Allee 160, D-23538 Germany e Institut de Physiologie, Universite de Lausanne, 7 Rue du Bugnon, 1005 Lausanne, Switzerland Received 1 December 2003; revised 12 March 2004; accepted 17 March 2004

Abstract The brain occupies a special hierarchical position in the organism. It is separated from the general circulation by the blood-brain barrier, has high energy consumption and a low energy storage capacity, uses only specific substrates, and it can record information from the peripheral organs and control them. Here we present a new paradigm for the regulation of energy supply within the organism. The brain gives priority to regulating its own adenosine triphosphate (ATP) concentration. In that postulate, the peripheral energy supply is only of secondary importance. The brain has two possibilities to ensure its energy supply: allocation or intake of nutrients. The term ‘allocation’ refers to the allocation of energy resources between the brain and the periphery. Neocortex and the limbic-hypothalamus-pituitary – adrenal (LHPA) system control the allocation and intake. In order to keep the energy concentrations constant, the following mechanisms are available to the brain: (1) high and low-affinity ATP-sensitive potassium channels measure the ATP concentration in neurons of the neocortex and generate a ‘glutamate command’ signal. This signal affects the brain ATP concentration by locally (via astrocytes) stimulating glucose uptake across the blood-brain barrier and by systemically (via the LHPA system) inhibiting glucose uptake into the muscular and adipose tissue. (2) Highaffinity mineralocorticoid and low-affinity glucocorticoid receptors determine the state of balance, i.e. the setpoint, of the LHPA system. This setpoint can permanently and pathologically be displaced by extreme stress situations (chronic metabolic and psychological stress, traumatization, etc.), by starvation, exercise, infectious diseases, hormones, drugs, substances of abuse, or chemicals disrupting the endocrine system. Disorders in the ‘energy on demand’ process or the LHPA-system can influence the allocation of energy and in so doing alter the body mass of the organism. In summary, the presented model includes a newly discovered ‘principle of balance’ of how pairs of high and low-affinity receptors can originate setpoints in biological systems. In this ‘Selfish Brain Theory’, the neocortex and limbic system play a central role in the pathogenesis of diseases such as anorexia nervosa and obesity. q 2004 Elsevier Ltd. All rights reserved. Keywords: ATP, adenosine triphosphate; KATP, ATP-sensitive potassium channels; Naþ/Kþ-ATPase, sodium potassium dependent adenosine triphosphatase; BBB, blood–brain barrier; LHPA, limbic-hypothalamus-pituitary –adrenal; SNS, sympathetic nervous system; MR, mineralocorticoid receptors; GR, glucocorticoid receptors; VMH, ventromedial hypothalamus; PVN, paraventricular nucleus; LH, lateral hypothalamus; ARC, arcuate nucleus; CRH, corticotropin-releasing hormone; ACTH, adrenocorticotropin; POMC, pro-opiomelanocortin; a-MSH, a-melanocyte-stimulating hormone; MC, melanocortin; NPY, neuropeptide Y; GABA, g-amino-butyric acid; BDNF, brain-derived neurotrophic factor; NMDA, N-methyl-D -aspartate; AMPA, amino-3-hydroxy-5-methyl-4-isoxazol propionate; LTP, long-term potentiation; LTD, long-term depression; CREB, cAMP responsive element binding

Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Physiological glucose regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Setpoints in the brain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. Setpoint of brain ATP regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2. Setpoint of limbic-hypothalamic-pituitary – adrenal system regulation . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3. Homeostasis: brain ATP and the LHPA system in balance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

* Corresponding author. Tel.: þ 49-451-500-3546; fax: þ49-451-500-4807. E-mail address: [email protected] (A. Peters). 0149-7634/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.neubiorev.2004.03.002

144 147 147 147 152 153

144

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

2.2. Load of the brain-supplying regulatory system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. Malnutrition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Psychological stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Sleep and the consolidation of setpoints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1. Stressors and the limbic system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2. Stress reactions and the limbic system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.3. Memory formation during sleep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Pathological glucose regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Hypoglycemia unawareness (type 1 diabetes mellitus). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Anorexia nervosa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Obesity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Type 2 diabetes mellitus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction How does the human organism control its energy supply? The answer to this question is the key to treating many diseases: obesity and the so-called metabolic syndrome with diabetes mellitus, hyperlipoproteinemia, hypertension and cardiovascular diseases belonging to these disorders. Gynecological diseases including polycystic ovaries or psychiatric disorders such as depression or eating disorders are also associated with disrupted regulation of energy supplies. Two different processes can be distinguished that regulate energy metabolism: energy supply (appetite, intake of foods) and allocation (assignment). The various organs of the body must compete for the allocation of a limited number of energy resources. The brain occupies a special position amongst all the organs concerning energy metabolism. It is the central organ for regulating energy supply, and it is able to receive information about the peripheral organs via peripheral (e.g. hepatic) sensors and their afferent neuronal pathways. Conversely, it can also control the functions of many peripheral organs, e.g. the skeletal musculature, the heart, the gastrointestinal tract or the sexual organs, via its efferent nerve pathways. It is probable that this control is not just restricted to physical movements and the function of many inner organs, but that it also includes the regulation of energy metabolism. The neuronal discharge and release of neurotransmitters and neuropeptides requires exceptionally large amounts of energy [1]. The energy consumption of the brain, related to its small proportion of the entire body mass, is much larger than the energy consumption of all other organs (e.g. muscle). The proportion of energy consumed by the human brain exceeds the proportion found in all other known species. This fact may be relevant for the origin of characteristics and disorders of metabolism found primarily in humans, e.g. obesity. The brain is separated from the general circulation by the blood –brain barrier. Specific substrates (such as glucose and lactate) or hormonal signals (such as insulin or leptin) are transported exclusively by specific

155 156 165 166 167 168 168 168 169 170 171 172 173 173

transportation mechanisms across the blood – brain barrier [2,3]. Thus, the transfer of substrates and hormones into the brain is very strictly controlled. The capacity of the brain to store energy is extremely limited, but maintenance of the energy supply to the brain is of prime importance to the survival of the whole organism. It is not therefore surprising that the energy content immediately available to the brain, i.e. in the form of adenosine triphosphate (ATP), is strictly regulated within extremely narrow boundaries. The brain is almost exclusively dependent on the metabolization of glucose. As such, selection of substrates by the brain is highly specific, while peripheral organs (muscle) can metabolize glucose, fat or proteins. Fatty acids can not traverse the blood –brain barrier. Only in special situations, such as with hypo or hypernutrition, does the organism produce significant amounts of alternative substrates such as ketones or lactate that can traverse the blood – brain barrier and assume a role in supplying energy to the brain. Finally, the brain is able to memorize information about its control actions and their subsequent effects, and to learn from the outcomes. It can use its plasticity to optimize its control behavior. Overall, therefore, the unique position of the brain is characterized by 1. 2. 3. 4. 5. 6.

its physical barrier properties, its high energy consumption, its low energy storage capacity, its substrate specificity, its plasticity, and its ability to record information from and to control peripheral organs.

In order to account for the idiosyncrasies of the brain’s energy supply and to establish the meaning of these for the entire organism, we propose here a new paradigm for the regulation of energy supply in the organism: † The brain prioritizes adjustment of its own ATP concentration. For this reason it activates its stress

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

system and in so doing competes for energy resources with the rest of the organism (allocation). † The brain then alters the appetite (food intake) so that it can alleviate the stress system and return it to a state of rest. With these two postulates, the brain simultaneously represents the highest regulatory authority and the consumer with the highest priority. The brain looks after itself first. Such selfishness is reminiscent of an earlier concept in which the brain’s selfishness was addressed with respect to addiction [4]. We chose our title by analogy but applied it in a different context, i.e. the competition for energy resources. During stress and times of shortage it safeguards its own supply even at the expense of all the other organs. The brain’s obligation to alleviate its stress system in a second step and allow it to return to a state of rest is not trivial. From a regulatory-theoretic standpoint we presume that the stress system is adjusted around a so-called setpoint at which it is at a state of rest. In the second step the brain therefore pursues the objective of satisfying its own energetic needs and those of the entire organism on a long-term basis in the most economic way possible. The regulation of the mass of the various body compartments such as the adipose tissue is then considered to be a secondary objective with this paradigm. According to traditional paradigms the brain regulates body mass by changing the intake of foods. Maintenance of blood glucose within narrow limits is also of key importance for maintaining health. The ‘lipostatic’ theory was originally formulated by Kennedy 1953 [5]. Jeffrey Friedman and coworkers of the New York Rockefeller University supported this view in 1994 with their ground-breaking finding of the hormone leptin [6]. With leptin, a hormone was discovered in fat and muscle tissue that sends a feedback signal to the brain so that the brain is informed about the status quo of peripherally stored energy. Most researchers considered this to be a closed regulatory system in which the absorption of nutrients is the regulator, body mass is the controlled parameter, and leptin is the feedback signal. Notably, before leptin was discovered, the research team of Stephen Woods and Daniel Porte at the University of Washington, Seattle, presented compelling evidence for insulin being an adiposity signal [7,8]. With the ‘glucostatic’ theory, blood glucose is considered to be the regulated parameter in the center of the regulatory system and it is assumed that endocrine changes (for example insulin, glucagon, growth hormone, and cortisol) and behavioral changes are mainly responsible for maintaining the concentration of blood glucose within narrow limits. The implicit assumption that an adequate energy supply to the brain automatically results from the constant behavior of the fat reserves and the blood glucose is common to both the glucostatic and the lipostatic theory. Another common feature is the assumption that with obesity a defect can be traced to the closed feedback loop. It can indeed be shown that with most overweight people leptin is not able to restrict the intake of foods. This phenomenon has been termed

145

‘leptin resistance’. Such a leptin resistance is found both as an inherited phenomenon with monogenetic defects [9,10] and as an acquired phenomenon after overfeeding [11]. A large number of neurotransmitters, neuropeptides and their receptors that mediate the leptin effect in the brain, e.g. anorexigens such as ‘Melanocyte Stimulating Hormone’ (a-MSH), have been studied in detail over the last few years [12]. The phenomenon of leptin resistance has as such been well described, but its origin has so far escaped explanation. The glucostatic and the lipostatic theories have explicitly or implicitly provided the basis for a large number of research strategies and therapeutic interventions for diabetes mellitus, obesity and other diseases. Against this, however, a range of observations have accumulated that can not be satisfactorily explained by these views and research approaches: † If healthy people are advised in a study to overeat considerably over a period of months, they do increase substantially in weight during this time, but within a few months they can return again to their initial body weight [13]. Clinical experience on the other hand shows that although many people show good body mass regulation at the start of their life, in later life (e.g. in the third decade), their body mass increases. If these people then attempt to reduce their body weight by dieting, the ‘yo-yo’ effect then sets in, and one gets the impression that body weight is regulated at a new, raised virtual setpoint [14]. Phenomena such as the yo-yo effect show that the system of body mass regulation is more complex than previously assumed. If only a simple defect within the regulatory system for weight regulation exists, such persons should be able to return to and maintain their initial body weight with their normal nutrition after a diet. However, the body mass often exceeds the previous maximum. The fact that only few people succeed in reaching and maintaining their initial body weight means that the traditional view that changes can be found within the assumed closed loop of the body mass regulatory system (e.g. single or multiple gene mutations) is too simple. † The study of metabolic, endocrine and behavioral phenomena in repeated hypoglycemia has shown that the brain has mechanisms for protecting its functionality actively within certain limits despite the existence of very variable blood glucose concentrations. The energy supply of the brain therefore represents more than just a by-product of the energy supply of the whole organism. † If the energy supply of the brain is threatened, lipostatic signals do not play any significant role in behavioral regulation: ravenous hunger with hypoglycemia occurs independently of the adipose tissue mass of the organism. † Traditional treatment concepts of type 2 diabetes mellitus are derived from the glucostatic theory and aim at normalization of blood glucose concentrations. The United Kingdom Prospective Diabetes Study showed that ‘tight’ blood glucose control results in a reduction in

146

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

the risk of microvascular but not of macrovascular diseases [15]. No effects on the overall mortality were observed. As side effects of such concepts using hypoglycemic agents or insulin, undersupply of the brain (recurrent neuroglucopenic comas) or oversupply of fat stores (body mass gain) occurred [15]. Peter G. Kopelman from the Bartholomew’s and the Royal London School of Medicine commented in the editorial that the ‘inevitable rise in glycosylated HBA1c witnessed throughout the study period, despite strict glycemic control, emphasizes the need for a better understanding of the pathogenesis of type 2 diabetes in susceptible individuals’ [16]. † Traumatization and psychiatric conditions such as depressive or eating disorders lead to modifications in the stress hormone system and various central transmitter systems. They can also lead to considerable increases and also reductions in body fat, even where defects in

the fundamental mechanisms of lipostasis or glucostasis have not yet been observed until now. These observations cast doubt on the priority of lipostatic signals in particular. † Despite intense research and the outstanding methodology that is now available, genetic defects have been able to explain only a small proportion of obesity and diabetes cases up until now. The observed obesity epidemic throughout the entire industrialized world illustrates this [17,18]. The fact that people of a similar genetic background under defined environmental conditions remain of normal weight or develop excessive overweight early on, however (e.g. Nauruans or Pima Indians) [19], supports a significant role of genetic factors. The traditional view fails to consider that a disorder might also lie outside the feedback system for weight regulation, e.g. in a higher-ranking regulatory system providing it with commands.

Fig. 1. The ‘Fishbone Model’ of glucose metabolism. The cerebral cortex sends a ‘glutamate cinnabd’ signal to the subordinate regulatory subsystems: 1. the allocation sybsystem assigns glucose via the glucose transporter 1 (GLUT1) to the brain, and via GLUT4 to the muscle and adipose tissue (yellow arrow). 2. The appetite regulatory subsystem controls the total amount of glucose available for allocation (red arrow). The energy content of the brain and peripheral tissues is measured with multiple sensors. The limbic-hypothalamic-pituitary-adrenal (LHPA) system, which includes the sympathetic nervous system, plays a decisive role in allocating glucose. The activity of the LHPA-system is indicated by the serum cortisol concentration. Feedback signals on the energy status in the brain (glucose), the peripheral organs (leptin), and on the activity of the LHPA system (cortisol) act on the various hierarchical levels of the system, i.e. the cerebral cortex, the limbic system and the hypothalamic sites for allocation (ventromedial hypothalamus) and intake (lateral hypothalamus) of foods. † Cortical balance. If the brain-ATP is too low, the glutamate command signal is stimulated in the cerebral cortex via high-affinity ATP-sensitive potassium channels; if the brain-ATP is too high, it is suppressed via low-affinity ATP-sensitive potassium (KATP) channels. In this way the system strives for a balance whereby the opposing effects of high-affinity and low-affinity KATP channels are of the same magnitude. † Limbic balance. If the serum cortisol is too low, the LHPA system is stimulated via high-affinity brain mineralocorticoid receptors (MR); if the serum cortisol is too high, it is suppressed via low-affinity brain glucocorticoid receptors (GR). Here, the LHPA system strives to achieve a balance whereby the stimulating and suppressing feedback signals are of the same magnitude. † Allocation. If the energy content is too great in the muscle and adipose tissue, leptin activates the ventromedial hypothalamus (VMH) that allocates glucose to the brain; if the energy content of the brain is too large, the brain ATP suppresses the VMH, so that glucose is allocated more to the musculature and adipose tissue. Thus, the allocation-subsystem strives for a balance whereby the feedback signals from the brain and the periphery are of the same magnitude. † Appetite. If the energy content is too low in peripheral tissues, the appetite stimulating lateral hypothalamus (LH) is activated via NPY; if the energy content is too large in the periphery, the LH is inhibited via a-MSH. The NPY- and a-MSH-signals are filtered in the arcuate nucleus (ARC) and conveyed only under certain circumstances to the LH. The key feedback-signal for regulating the intake of foods is brain glucose. If the stimulatory and inhibitory feedback-signals in the cerebral cortex, in the LHPA system, and in the hypothalamus are balanced, the organism achieves a state of energetic homeostasis. Coordinates indicate positions in the model that are referred to in the text.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

While most research continues to focus on crucial hypothalamic circuits, a small group of scientists have already broken new ground, since recent work has clearly shown that ingestive behavior is influenced by a widely distributed neural network, which includes the caudal brainstem, limbic and cortical structures [20 –22]. The paradigm proposed by us places the regulation of ATP-concentration in the brain at the focal point. The brain initially adjusts its own ATP-concentration by burdening its own stress system and competing for energy resources within the body. The brain changes eating behavior so that it can then alleviate the stress system and return it to a state of balance. The regulatory principles of this paradigm have been formulated mathematically as a dynamic system and graphically illustrated in the form of a so-called ‘fishbone model’ [23] (an overview is given in Fig. 1, more details are explained in chapter 2). Readers and authors are faced with a dilemma regarding the needs of simplicity and complexity, i.e. between merely a suggestive and an explicit representation of specific mechanisms. The fishbone model has a simple but not a trivial structure: it represents a hierarchically organized system with a forward pathway (similar to the spine of a fish) and multiple paired stimulatory and inhibitory feedback pathways (the fishbones). Flow charts of complicated control systems can be simplified by mathematical transformations [24]. The most simple model for allocating energy resources to 2 organs, e.g. to the brain and muscle, has a ‘fishbone’ like structure. Such a special model structure is suitable for dealing with different levels of complexity. Is the model oversimplified or too abstract? One point of view might be that important hormones (e.g. resistin, ghrelin) escape mention here so that the true complexity of energy metabolism is not delved into. We reviewed the literature and indeed often found two or more biological mechanisms for each individual component in the mathematical model. As such there appears to be much redundancy in glucose regulation. Redundant signal pathways can be added to the fishbone model (new fishbones) without changing the basic model structure. The activation of the sympathetic nervous system is mediated by leptin and insulin as well. In the model, the hormone leptin conveys a signal to the brain that energy has been stored in peripheral organs, particularly in the adipose tissue, and is not therefore available at that time as a substrate for the brain. Correspondingly, leptin conveys a signal to certain hypothalamic neurons [25] and in this way invokes an increase in sympathetic nervous system activity and thereby an increased allocation of glucose to the brain. Insulin sends a similar signal analogous to this. Insulin in the same way informs the brain that glucose is stored and unavailable for supplying the brain. Correspondingly, insulin can influence the same hypothalamic neurons in the same manner [26], so that the sympathetic nervous system is stimulated and the appropriation of glucose by the brain is ensured. This example shows that leptin and insulin transmit related or similar signals to the brain. There may be

147

distinguished differences in the timing of their feedback signals, however, in principle they transmit redundant messages. The stimulatory insulin feedback pathway can be integrated into the fishbone model without changing its fundamental structure. Only the degree of redundancy, and not the relevancy of the model, is changed through such additions. Is the model too complex or explicit? We have in fact refrained from including a large number of biological mechanisms that might also fulfill functions in the model. A list of various possible redundant signals was presented in an earlier manuscript [23]. However, we decided to assign only a single functional mechanism and a single anatomical structure to a single signal pathway in the model. Leptin acts for example as a ‘substitute’ for a class of signals that contains insulin amongst other elements, and which can fulfill all the functions described in the model. We are aware that there might be a better selection for such a substitute, and that in the future hormones might be discovered that fulfill this function better and so have a greater biological relevance than the ones mentioned here. This may likewise account for the selection of brain structures referred to in this paper. The limbic system and the hippocampus for example are extremely complex structures per se, supporting many other specific functions not relevant here, and of course, those relevant here may in part be fulfilled by other redundant structures. We are also aware that the assignments proposed here might be the subject of some debate, but we feel that the specificities of the model presented are less important than the general basic principle proposed here for energy metabolism. We followed the advice that ‘everything should be made as simple as possible, but not simpler’ [27]. The newly presented theory regarding the regulation of energy supply is only valid within a certain scope. For example, many experiments that are cited here in support of the model have only been carried out under special experimental conditions in in-vitro or in animal studies, but have not yet been confirmed in humans. Also, many studies in humans cited here have only been performed in men but not in women. Several hypotheses can be derived from the presented model. In the future, testing of such hypotheses shall allow a redefinition of the scope within which the theory is valid, whether it be broadened or narrowed. In this review article we would like to apply the ‘selfish brain theory’ to offer new explanations for phenomena which until now have escaped clarification.

2. Physiological glucose regulation 2.1. Setpoints in the brain 2.1.1. Setpoint of brain ATP regulation 2.1.1.1. Measurement with two receptors. How does the brain maintain ATP constant at a specific concentration? To answer this we propose a principle whereby the brain controls this concentration using high-affinity and low-affinity

148

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

ATP-sensitive potassium channels. ATP-sensitive potassium channels (KATP) belong to a special class of ion channels that couple bioenergetic metabolism to membrane-excitability [28]. KATP is present not only at neurons and neuroendocrine cells, but also on many other cell types, such as those of skeletal and smooth muscle [29,30]. These KATP channels are closed by intracellular ATP. While the energy-rich ATP closes these potassium channels, the low-energy adenosine diphosphate (ADP) can open the ATP-sensitive potassium channels. For this reason the intracellular ratio of ATP to ADP is a key regulator for the functional state of the ATP-sensitive potassium channel. ATP and ADP bind to specific parts of the KATP channel: at the ‘nucleotide binding domain’ of the socalled sulfonylurea receptor (SUR), that together with the actual channel pores forms a single morphological unit [31]. The SUR protein belongs to the ‘ATP binding cassette (ABC) family’ [32]. The KATP channel therefore represents a membranous, molecular structure that fulfills the regulatory-theoretic criteria of an ‘energy sensor’ (or more simply an ‘ATP sensor’). If one provides an excitatory neuron with sufficient energy reserves, i.e. a high intracellular ATP to ADP ratio, these membranous KATP channels are closed. With closed KATP channels a potassium efflux from the cell is prevented via this ion channel, which enables depolarization. Calcium flows into the cell interior. The neuron releases neurotransmitter (such as the excitatory amino acid glutamate) or neuropeptides (such as the neurotrophin ‘brain-derived neurotrophic factor’; BDNF) from its nerve endings. If the energy content of the neuron is high enough, the KATP channels allow a neuronal excitation. If on the other hand a fall in intracellular ATP content occurs, the KATP channels are opened, the neuron is hyperpolarized (and thereby electrically stabilized), and its function is deactivated. The KATP channels therefore also have a cytoprotective function: with energy deficiency the function of the cell is turned off and the residual energy is saved for structural maintenance of the cell [33–35]. Interestingly enough, there are two different types of KATP channels: those with high-affinity and low-affinity ATPbinding sites. These ATP-binding properties allow them to be assigned to two subtypes, i.e. SUR1 and SUR2 [36–39]. With low intracellular ATP content the high-affinity ATP-sensitive potassium channels are mainly occupied, and are closed as a result. These high-affinity ATP-sensitive potassium channels are found in the cortex and in many other brain areas on excitatory neurons [40,41]. Such neurons are able to be electrically active with a low ATP content. However, if the ATP-concentration declines to a very low and thereby critical concentration for survival of the neuron, these high-affinity ATP-sensitive potassium channels no longer bind adequately. The corresponding KATP channels are then opened and the cell function is deactivated. The high-affinity ATP-sensitive potassium channels in the neocortex play an essential role in protecting against seizures and neuronal damage [42]. In contrast, with high intracellular ATP content the lowaffinity ATP-sensitive potassium channels are also

occupied. There are KATP channels in the entire cortex [33,43 – 47], where they are localized both presynaptically and postsynaptically [48]. In some brain areas presynaptic KATP channels reduce the liberation of g-amino-butyric acid (GABA): e.g. in the hippocampus [49] and in the substantia nigra [50 – 53]. It is worthy of note that both low-affinity and high-affinity ATP-sensitive potassium channels have been found in human neocortex [54]. Although it has not been confirmed in any single experiment, we do presume from current data that in human neocortex there are also presynaptic, low-affinity ATP-sensitive potassium channels that reduce the GABAergic tone. This assumption is also supported by the clinical observation that with progressive energy deficiency in the brain there is initially an excitatory stage with a raised seizure tendency, followed by a calming of the cortex. These findings are consistent with a presynaptically mediated GABAergic tone, which with a slight energy deficit can be reversed via low-affinity ATP-sensitive potassium channels [53]. If one assumes that the high-affinity ATP-sensitive potassium channels are located on excitatory neurons, while the low-affinity ATP-sensitive potassium channels are localized on inhibitory neurons, this distribution pattern leads to the following dynamic behavior: with critically reduced ATP both the excitatory and inhibitory neuron populations are functionally inactive. This phenomenon has been described as a ‘global silencing’ of the cerebral cortex [55]; its clinical correlate is the ‘hypoglycemic’, or better said the ‘neuroglucopenic’ coma. With low, but non-critical ATPcontent in both neuronal populations, ATP binds almost exclusively to the high-affinity ATP-sensitive potassium channels, i.e. to those on the excitatory neurons that release glutamate. Contrastingly, with high cerebral ATP concentrations the inhibitory neurons also become active, i.e. those that exert an inhibitory effect on the excitatory population. All in all, a biphasic activity pattern results for the excitatory neuronal population that depends on intracellular ATP content (Fig. 3a). Of decisive importance is the fact that the balance between excitatory and inhibitory neuronal populations changes depending on brain ATP concentration. At low brain ATP concentrations the glutamatergic population is dominantly active, while at high ATP concentrations the activity of the GABA ergic population predominates. An effective regulatory system for brain ATP can be described with the following overall principle:

1. ATP binds to high- and low-affinity ATP-sensitive potassium channels. 2. Bound high affinity ATP-sensitive potassium channels permit glutamatergic neuronal activity, while bound low-affinity ATP-sensitive potassium channels permit GABA-ergic activity. 3. Glutamatergic neurons raise brain ATP, while GABA-ergic neurons lower it.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

Up to now we have demonstrated the first and the second rule. In the next chapter we shall explain the third rule and how the glutamate command signal promotes an increase in the brain’s energy content. These three simple rules regarding the interplay between ATP, the two different affinity ATP-sensitive potassium channels and the glutamatergic and GABAergic neuronal populations describe a secure regulatory system that balances the brain ATP around a certain concentration. This concentration can be described as a ‘balance setpoint’ for brain ATP. 2.1.1.2. Astrocytic ‘energy on demand’. The brain can supply itself by requesting energy firstly from the body periphery and secondly from the environment. For this purpose the brain must invest considerable expense, e.g. activate its stress systems or acquire new food resources in order to actually procure the requested amount of energy. If there is not an adequate food supply (such as during times of starvation), the brain has no other possibility but to compete for energy resources within the organism. How does the brain compete with the body for energy resources? The brain controls the allocation of glucose between the brain on the one hand and the musculature and adipose tissue on the other. In order to allocate glucose to itself, the brain must open the blood – brain barrier for glucose and cut off the supply to peripheral tissues. As mentioned above, it is the glutamatergic neuronal population that activates the allocation of glucose to the brain. Although all neurons independently of the type of neurotransmitter released (glutamate or GABA) use energy, it has only been verified for the glutamatergic population that they also serve for energy replenishment [56]. GABAergic neurons on the other hand do not mediate any such allocation of glucose to the brain [57], but instead inhibit the glutamatergic neurons with the help of their transmitters and only consume energy. Which molecular mechanisms can glutamate utilize to enhance energy substrate availability for parenchymal cells? The astrocyte, a specific type of glial cell, plays a key role in allocating glucose across the blood – brain barrier. The principle ‘energy on demand’ has been used to describe the (local) response of astrocytes to glutamatergic activity in order to provide lactate to active neurons as an energy substrate [56]. Glutamate that is freed at the synapse upon excitation is rapidly removed again to allow subsequent transmission events. Astrocytes enclose most glutamatergic synapses and collect released glutamate with a highly efficient and specific transporter system. Transporters are driven by the electrochemical sodium gradient, a fact that leads to a tight coupling between glutamate and sodium uptake [58]. The astrocyte is now confronted with two tasks: the recovery of glutamate and the restoration of the sodium gradient. The gradient is restored by the activation of the sodium- and potassium-dependent adenosine triphosphatase

149

(Naþ/Kþ-ATPase) [59]. Glutamate is converted into glutamine which is released by astrocytes and taken up by the neuronal terminal. There it is enzymatically converted again into glutamate so that the neuronal glutamate pool is refilled again. There is no ATP exchange between astrocytes and neurons, so that each cell type must secure its own energy supply. The end-feet of the astrocytes are equipped with specific transporter molecules, i.e. glucose transporter 1 (GLUT1), and enclose practically all the capillary walls within the brain. A close morphological and cytological relationship exists between astrocytes and cerebral capillaries. In this way the preconditions for a functional coupling between synaptic activity and glucose uptake are fulfilled: glutamate activates its glutamate transporter and stimulates glucose uptake into astrocytes [60,61]. Glucose is broken down in this process to lactate, which is then released and made available as an energy source for neighboring neurons [62,63]. The energy that arises during the glycolytic breakdown of glucose to lactate is used in the astrocytes to support the activity of the Naþ/Kþ-ATPase and to convert glutamate into glutamine [59], while in the neuron lactate utilization will be employed for closing the postsynaptic KATP channels and for excitation [64]. This cascade of molecular events represents a direct mechanism for the coupling between synaptic glutamate release and glucose allocation to the neuron via the blood –brain barrier and the astrocytes. 2.1.1.3. Systemic ‘energy resource request’. How can the brain prevent glucose uptake into muscle and adipose tissue? Peripheral glucose uptake can be restricted through activation of the limbic-hypothalamic-pituitary – adrenal (LHPA) system. The LHPA system is a neuroendocrine system closely associated with stress in mammals [65]. This system allows a rapid reaction to stressful stimuli and ultimately guarantees a return to homeostasis via complex feedback mechanisms. Hierarchically, the limbic system therefore represents the highest authority in the control of stress reactions. In the limbic system there are two core regions that carry out this control: the hippocampus and the amygdala. These limbic neurons project with axons directly or via the VMH into the paraventricular nucleus (PVN). Here, the sympathetic nervous system is activated and neuropeptides are formed and released such as corticotropin-releasing-hormone (CRH) and vasopressin. These releasing hormones stimulate adrenocorticotropin (ACTH) release into the general blood circulation within the pituitary. ACTH ultimately stimulates the release of cortisol from the adrenal cortex. The sympathetic nervous system projects with its efferent nerve pathways to the adrenal medulla where it stimulates the liberation of adrenaline. The sympathetic system also innervates the pancreatic b cells [66] where it suppresses insulin release [67 – 69], as well as the musculature and adipose tissue where it suppresses

150

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

the uptake of glucose [70 –72]. In this way, the LHPA system can increase the glucose concentration in the blood. In the limbic system, energy needs, in addition to activating the ‘energy on demand’ signal (local), also trigger a systemically effective ‘energy resource request’ signal (for the whole brain), with glutamate being the mediator in both cases. In addition to that direct limbic mechanism requesting energy (internal sensing or detector area), other parts of the cortex will also signal their needs to the limbic system. Thus, the limbic system might act both as a detector and transducer of global brain energy needs. What effects do cortical glutamatergic neurons have on limbic neurons? Patricia Molina and coworkers of the Louisiana State University in New Orleans were able to show recently that primarily glutamate receptors of the NMDA subtype mediate the activation of the LHPA system with brain glucose deficiency [73]. The NMDA receptor plays a key role for the pyramidal cells of the limbic system and has a function not only for setting the tone of the LHPA system, but also in memory formation (see the chapter ‘memory formation during sleep’). The stimulation of other subtypes of glutamate receptors also brings about a strong activation of the LHPA system [74,75]. The above-mentioned team also succeeded in establishing a link between cortical glutamatergic activity and the activation of stress systems. It is well known that stress systems can restrict the allocation of glucose to muscle and adipose tissue. In summary, cortical glutamatergic neuronal populations are apparently able to adjust the allocation of glucose to the brain by favoring glucose utilization in brain while impeding it in muscle and adipose tissue. The cerebral cortex sends the ‘glutamate command’ signal to both of its regulatory subsystems that control glucose allocation and appetite. Energy supply for the brain results from the activity of the two regulatory subsystems. Brain ATP binds to low- and high-affinity ATP-sensitive potassium channels as a feedback signal. High-affinity ATP-sensitive potassium channels increase the cortical glutamatergic tone and in so doing the glutamate command signal. Low-affinity ATP-sensitive potassium channels increase the cortical GABAergic tone and in so doing suppress the glutamate command signal. In this way the primary regulatory system strives for a cortical balance between glutamatergic and GABAergic neuronal activity at which the ATP concentrations are optimal. How does the brain request energy resources from the environment? The LH is a key region of the brain that controls appetite and eating behavior [76]. Feeding and fasting is not simply controlled by a hypothalamic center, but rather by quite a large network of neurons located at many different sites (thalamus, subcortical nuclei, hypothalamus, brainstem, and medulla). Signals originating in the LH appear to reach other brain sites by first descending to the parabrachial nucleus [77]. Here, we assign the LH as one representative

anatomical site to the functional component ‘appetite’ in the model. Glutamate is a potent stimulus that stimulates neurons in the LH to increase appetite and initiate food intake [78,79]. The LH, however, is under the direct influence of the limbic system. Upon cortical excitation, multiple locally effective ‘glutamate command’ signals from cortical neurons are integrated within the limbic system. The limbic system functions as a transducer between this integrated glutamate command signal and setpoints of subordinate hypothalamic systems: one setpoint signal is conveyed via neuronal pathways to the VMH (allocation), and another is conveyed to the LH (appetite). The limbic system transduces the signals to the VMH and the LH differently. While the signal to the VMH is adapted under certain conditions, i.e. amplified or suppressed [80,81], the signal to the LH is rather robust and less altered. As an example, recurrent hypoglycemia leads to attenuation of VMH-mediated counter-regulation (e.g. adrenaline, glucagon)[82], but not to an attenuation of hunger (LH) [83]. The limbic system also coordinates the order in which the VMH or LH are activated. The VMH mobilizes glucose for the brain within seconds, but an activation of the LH only leads after a delay (and only with sufficient food intake) to an increase in glucose supply to the brain. The limbic system therefore conveys the ‘energy resource request’ signal first to the VMH, whereas the LH is inhibited by this signal [84]. If the output to the VMH is weak, the appetite controlling LH is disinhibited. The allocation controlling VMH is therefore ranked higher than the appetite controlling LH, whereby these two components have reciprocal functions [85 –87]. The activated neurons of the LH also provide ‘orexigenic’ (i.e. appetite increasing) neuropeptides via projections to different parts of the brain [88]. These orexigen-secreting neurons increase the drive for feeding and ultimately also have an influence on complex behavioral patterns (e.g. purchasing behavior for foodstuffs) related to feeding. Fig. 2a summarizes the command principle once again: cortical glutamatergic neuronal populations release glutamate upon excitation. On the one hand the glutamate command signal triggers an astrocytic ‘energy on demand’ process. On the other hand the glutamate command signals input into the limbic system, where they are transduced into ‘energy resource request’ signals. These setpoint signals are conveyed to the subordinated hypothalamus. Here the VMH (allocation) and the LH (appetite) are stimulated. The VMH increases the proportion of circulating glucose to be assigned to the brain, while the LH strives to increase the total amount of circulating glucose. Allocation and food supply therefore determine the proportion of glucose that is assigned to the brain. The amount of glucose available to the brain influences the amount of ATP available to it. With low ATP in the brain the high-affinity ATP-sensitive potassium channels are closed for the most part (i.e. those channels

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

151

152

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

enabling the stimulation of glutamatergic neuronal populations), and these demand further energy. If the brain ATP on the other hand is high, the low-affinity ATP-sensitive potassium channels are also closed, and the GABAergic neuronal population decreases the brain’s energy demand. In this way a regulatory system results that resembles the principle of supply and demand in a free market economy, and which is able to regulate brain ATP around a specific balance setpoint. 2.1.2. Setpoint of limbic-hypothalamic-pituitary –adrenal system regulation How does the brain regulate the activity of its LHPAsystem? We propose that it regulates this with the aid of highaffinity and low-affinity brain corticosteroid receptors. Two types of corticosteroid receptors are known in the brain. Starting in the year 1968 with the milestone paper of Bruce McEwen at the Rockefeller University in New York [89], a large number of researchers have since managed to characterize the two brain receptor subtypes both biochemically and functionally. The type I or MR in the brain resembles the MR in the kidney and has a high specificity for selectively binding cortisol, the primarily active glucocorticoid in humans [65]. In the brain, the MR is localized most densely in the limbic system, i.e. in the hippocampus and in the amygdala, where it binds cortisol with high affinity. Contrastingly, the type II or GR binds cortisol with a low affinity. The presence of GR receptors has been confirmed in many brain regions, including the limbic system, the hypothalamus, and the pituitary. MR binds cortisol with a 10-fold higher affinity than does the GR. These receptor properties allow MR and GR to regulate LHPA system activity. MR is bound with low cortisol concentrations and develops its effects mainly during the evening nadir of the cortisol circadian profile. At high cortisol concentrations, e.g. after morning awakening or during a stressful incident, MR also bind cortisol, but the bound GR dominates in its effect and is decisively involved

in ensuring that the LHPA system returns to homeostasis. In fact, three years later the group showed in vivo that peripheral injection of a larger dose of a glucocorticoid reduced hippocampal firing activity [90]. Pyramidal cells in the hippocampus and the amygdala express both MR as well as GR receptors [91]. One known function of these limbic MR and GR receptors is to modify memory storage and retrieval [92,93]. Both receptors are formed in the cell nucleus and are then released into the cytosol of the neuron. Cortisol traverses the external cell membrane of the neuron without a specific transporter and binds in the cytosol with high affinity to MR and with low affinity to GR. The cortisol concentration as well as the number of MR and GR present in the cytosol determine how many cortisol molecules bind to GR and MR. Only cortisol bound MR and GR complexes can traverse the nuclear membrane and reach the cell nucleus where the cortisolbound receptors form dimers with one another [94 – 96]: MR – MR homodimers, MR – GR heterodimers and GR – GR homodimers. The homodimer MR –MR binds to a ‘glucocorticoid responsive element’ (GRE) in the genome. The other dimer-types compete with the MR –MR homodimer for these GRE binding sites in the genome and inhibit its effect. What influences do MR and GR have on the activity of the LHPA system and with that the secretion of cortisol? We assume that cortisol-bound MR stimulates the LHPA system while cortisol-bound GR prevents this stimulation. This assumption is supported by a unique study in which cortisol effects over a very broad range of cortisol concentrations were illustrated [97]. In this study patients with Cushing’s disease, who had had both adrenal glands removed completely, were infused with cortisol. During infusion, serum cortisol climbed from very low concentrations continuously to very high concentrations. With low cortisol concentrations there was a marked increase in ACTH secretion, while with high cortisol concentrations there was a marked decline. This finding is consistent with an MR-mediated stimulation and a GR-mediated suppression of the LHPA-system. In this investigation a bell-shaped

R Fig. 2. (A) The primary regulatory system for brain ATP regulation. The cerebral cortex sends the ‘glutamate command’ signal to both of its regulatory subsystems that control glucose allocation and appetite. Energy supply for the brain results from the activity of the two regulatory subsystems. Brain ATP binds to low- and high-affinity ATP-sensitive potassium channels as feedback signal. High-affinity ATP-sensitive potassium channels increase the cortical glutamatergic tone and in so doing the glutamate command signal. Low-affinity ATP-sensitive potassium channels increase the cortical GABAergic tone and in so doing suppress the glutamate command signal. In this way the primary regulatory system strives for a cortical balance between glutamatergic and GABAergic neuronal activity at which the ATP concentrations are optimal. (B) The LHPA system as a regulatory subsystem of brain ATP regulation. The LHPA-system restricts the GLUT4-mediated glucose uptake into muscle and adipose tissue and with this increases the GLUT1-mediated glucose uptake into the brain. Cortisol is the feedback-signal for the LHPA-system. Cortisol binds in the limbic system to high-affinity mineralocorticoid (MR) and low-affinity glucocorticoid (GR) receptors. With low cortisol concentrations MR stimulate the LHPA-system, and with high cortisol concentrations GR suppress its activity. In the hierarchically subordinate hypothalamus (PVN) only GR-receptors act inhibitorily at high cortisol concentrations. The activity of the LHPAsystem determines the allocation of glucose to the brain and the periphery. In this way the LHPA-system defines the setpoint for regulation of body mass. (C) Leptin and its amplifier. Leptin conveys the feedback-signal regarding energy status in the adipose and muscle tissues to the hypothalamic VMH where leptin stimulates the allocation of glucose to the brain. An amplification mechanism for leptin activity is localized in the arcuate nucleus (ARC). Here, at low leptin concentrations, the appetite stimulating NPY is primarily produced, while at high leptin concentrations the appetite suppressing a-MSH is mainly produced. a-MSH stimulates the allocation centre (VMH) and thereby amplifies the direct effect of leptin, while NPY on the other hand suppresses the effect of leptin on the VMH.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

dose-response relationship was shown for cortisol in humans, analogous with a similar relationship found by other investigators in numerous experiments investigating the effects of cortisol on the excitability of neurons in the hippocampus [98,99]. There is a debate as to whether limbic MR act in a stimulatory or inhibitory way on the LHPA-system. Pharmacological interventions with MR-inhibitors result in elevated basal glucocorticoid concentrations, possibly suggesting an inhibitory effect of MR [100]. However, it must be considered that with such interventions the underlying process is not a simple ligand – receptor interaction, and that ‘heterodimerization’ (see above) or the ‘autoregulation’ (see below) of MR and GR can cause paradoxical effects (as have clearly been demonstrated in literature [101,102]). Thus, conclusions based on pharmacological inhibition may be erroneous. Processes like heterodimerization and autoregulation are so-called ‘nonlinear’ [24], and it is this very nonlinear property that makes an experimental analysis difficult, but at the same time makes the LHPA-system particularly stable. Neurons of the limbic system are the starting point for the stimulation of the LHPA system. Here, MR and GR regulate the expression and transcription of a large number of genes. One group of genes controls the behavior of ion channels (e.g. calcium channels), a second gene group regulates ligand-bound ion channels (e.g. glutamate receptor coupled channels) and a third group influences G protein-coupled receptors. Ronald de Kloet and Marian Joe¨ls at the University of Amsterdam/Leiden, Netherlands, discovered many such corticosteroid effects and described them in a number of comprehensive reviews [98,99]. Thus, MR has the ability to influence the excitability of limbic neurons via the expression and transcription of a variety of gene products. MR and GR modulate amongst other things glutamate-mediated signal input [103]. Here we focus on the effects of MR and GR on limbic neurons that stimulate the hypothalamic neurons. The neurons stimulate via direct or indirect neuronal pathways the hypothalamic release of CRH and vasopressin. The latter release-hormones activate the formation of proopiomelanocortin (POMC) peptide in the pituitary, from which ACTH is cleaved. Pituitary ACTH is secreted and stimulates the adrenal release of cortisol. Therefore, in this model MR promotes and GR inhibits the release of cortisol via a range of intermediate steps. Circulating cortisol is metabolized in the liver and eliminated with a half-life of about 120 min. The clearance function for cortisol corresponds to an elimination of the first order, i.e. the clearance rate of cortisol is proportional to its concentration. The higher the cortisol is in the serum, the higher is its hepatic elimination. ACTH has a half-life of about 20 min and CRH a half-life of about 9 min. Individuals who no longer have adrenal glands, e.g. patients with Addison’s disease, can no longer produce any cortisol themselves; in such individuals cortisol is removed from

153

the circulation according to its half-life: after 2, 4, and 6 h it is reduced to 1/2, 1/4 and 1/8. This means that without adrenal production and secretion of cortisol more than 85% of the cortisol is already eliminated from the circulation after 6 h. In the cortisol circadian profile it falls continuously to a minimum in the evening from a morning maximum after awakening. However, this drop-off rate is much slower than that of the hepatic cortisol clearance. The slow reduction in cortisol over the day therefore requires a continuous release of cortisol from the adrenal gland which slows the fall in cortisol. One can see that the limbic system has to stimulate the hypothalamic center continuously in order to prevent a rapid reduction in serum cortisol. The stimulatory effect of the limbic system must be even greater if one considers that at hierarchically lower levels CRH, ACTH and cortisol are still subject to a GR-mediated feedback-inhibition (Fig. 3b). We propose the following general principle to illustrate how the activity of the LHPA system is regulated: 1. Cortisol binds with high affinity to MR and lower affinity to GR. 2. Cortisol bound MR and GR assemble into three forms of dimers: MR –MR, MR – GR or GR – GR. 3. MR –MR homodimers stimulate the LHPA system and thereby cortisol secretion, while GR interferes with this effect. These three simple rules regarding the interplay between cortisol, the two differing affinity receptors MR and GR and the various MR and GR homo- and heterodimers describe a control system that regulates cortisol secretion around a setpoint. This concentration can be designated as a balance-setpoint for the activity of the LHPA system, which in humans is usually achieved during the evening. The reader will surely notice at this point that the regulation principle underlying brain ATP regulation and LHPA system regulation is the same. It would not be surprising if during evolution a reliable and simple regulatory principle that has proven its worth with one aspect of metabolism should also be encountered in other areas. 2.1.3. Homeostasis: brain ATP and the LHPA system in balance The hierarchical positions of brain ATP regulation and LHPA regulation are different. The brain ATP regulation has the highest biological priority. It therefore represents a primary regulatory system. This primary regulatory system for brain ATP regulation operates with the glutamate command signal. This signal is conveyed to its two regulatory subsystems, i.e.: (1) to the LHPA system, and (2) to the appetite-regulating LH. The LHPA system determines the allocation of glucose to the brain and the body periphery while the LH is essential for eating behavior. Thus, the brain has two ways of fulfilling its demand

154

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

Fig. 3. (A) Setpoint for brain ATP regulation. The rate of change in brain ATP over time [dATP/dt] (ordinate) depends on the brain ATP itself (abscissa). High affinity ATP-sensitive potassium channels on glutamatergic neurons are closed at low brain ATP concentrations so that the neurons can become functionally active (green function). Low affinity ATP-sensitive potassium channels on GABAergic neurons permit functional activity only at higher brain ATP concentrations (red function). Since the GABAergic neurons are inhibitory towards glutamatergic neurons, a reduction of glutamatergic neuron activity occurs at higher brain ATP-concentrations (green function). Inset on the upper right: Dependency of the energy balance of glutamatergic neurons on brain ATP is shown here: glutamatergic neurons stimulate glucose transport across the blood–brain barrier using the ‘energy on demand’ signal. These neurons require

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

for energy. On the one hand it can alter glucose allocation, i.e. the percentage of glucose transported across the blood – brain barrier, and on the other it can alter food intake, i.e. the total amount of glucose available for distribution. If sufficient energy resources are available to the organism, the brain can request energy via both regulatory subsystems (i.e. allocation and appetite). This means that the greater the allocation to the brain the less food intake is necessary or vice versa, the greater the amount of food consumed by the organism the less allocation to the brain is necessary. The reciprocal relationship between allocation and food intake required to satisfy the energy needs of the brain is represented graphically in Fig. 5a. This reciprocal relationship between allocation and required food intake can be mathematically derived as follows. Glucose uptake into the brain is b [g min21 kg21] while m represents glucose uptake into muscle and fat [g min21 kg21]. The ratio between the two glucose uptake rates is defined as allocation: Allocation ¼

b m

ð1Þ

If B is designated as the mass of the brain [kg] and M that of the muscle/fat [kg], the required intake of foods is: Necessary nutrient uptake ¼ bB þ mM

ð2Þ

If one inserts m from Eq. (1) into Eq. (2), the following relationship between food intake and allocation results:   M Necessary nutrient uptake ¼ b B þ ð3Þ Allocation If one assumes that the brain keeps its ATP content constant, the variable b in Eq. (3) is regulated within very narrow limits and kept almost constant. The mass of the brain B is a constant parameter. Eq. (3) is represented in Fig. 5a. All values of this function are characterized by the fact that the brain ATP concentration is held at a constant concentration, whereby the high- and low-affinity ATP-sensitive potassium

155

channels are balanced. Depending on the magnitude of allocation, a food quantity arises from this relationship that the brain requires to fulfill its demand for energy. There is a substantial difference between the two regulatory subsystems for allocation and food intake. The LHPA system that determines allocation can be burdened in unusual crisis situations, e.g. in times of starvation, but it always strives to return to its resting balance. This resting balance is designated as the so-called MR – GR brain corticosteroid balance. In Fig. 5a all the points are represented in a second function, in which MR and GR are balanced for the LHPA subsystem. A special situation therefore occurs in which both highand low-affinity ATP-sensitive potassium channels are in a state of balance, i.e. whereby both the brain ATP is constant and the MR and GR are in a state of balance, meaning that the LHPA system is at a resting state. At exactly this intersection point the energy metabolism is in a state of homeostasis, graphically depicted in Fig. 5a. If brain ATP regulation and the LHPA system are in a state of balance, a certain required food intake results from that. If this food intake can be realized, the organism can remain stable in this metabolic equilibrium state. The body mass is, however, already adequately set by this balance. The idea of an independent system that regulates body mass therefore becomes superfluous. Basically, this balance-setpoint represents an ideal equilibrium state which in fact is rarely achieved. The organism is instead continuously exposed to stresses and the nutrient supply is variable so that it must continually strive to achieve this ideal balance state. 2.2. Load of the brain-supplying regulatory system Loads can put stress on the brain-supplying regulatory system. Does the newly proposed paradigm comply with our knowledge on how the organism reacts to these situations?

R energy themselves for their excitation. The green function (D energy) shows how glutamatergic neurons provide energy for themselves and for GABAergic neurons depending on the brain ATP. GABAergic neurons on the other hand are not able to promote glucose transport across the blood–brain barrier in this way; instead they only consume energy. At low brain ATP concentrations it is the glutamatergic neurons that mobilize glucose and increase brain ATP content that are mostly active; at high brain ATP-concentrations GABAergic neurons that only consume energy and thereby lead to a lowering of brain ATP-content are mostly active. The setpoint for brain ATP regulation is found at the intersection point of the green and red functions (upper panel); here, the rate of change in brain ATP is equal to zero and the regulating system is at a state of balance (lower panel). (B) Setpoint of the LHPA-system. The rate of change in cortisol over time ½dCortisol=dt (ordinate) depends on the cortisol concentration itself (abscissa). High-affinity mineralocorticoid receptors (MR) are active at low cortisol concentrations and stimulate the LHPA system and with that adrenal cortisol production and release. Low-affinity glucocorticoid receptors (GR) are active only at high cortisol concentrations and inhibit the LHPA system so that adrenal cortisol production and secretion are decreased (green function). The hepatic clearance rate of cortisol depends on the cortisol concentration itself (red function). The setpoint of the LHPA-system is found at the intersection point between the green and red functions (upper panel); here, the rate of change of cortisol is equal to zero and the LHPA system is at a state of balance (lower panel). (C) The leptin amplifier in the arcuate nucleus. The neuronal activity of NPY and POMC neurons in the ARC (ordinate) depends on the leptin concentration (abscissa). Leptin inhibits the activity of NPY neurons so that at low leptin concentrations the NPY neurons are spontaneously active. Leptin stimulates the POMC neurons so that at moderate leptin concentrations they are activated. NPY and POMC neurons are glucose responsive and feature ATPsensitive potassium channels that are opened at high leptin concentrations; for such reasons these neurons become deactivated at high leptin concentrations. The ARC neurons project into the VMH. POMC neurons act inhibitorily while NPY neurons act in a stimulatory manner in the VMH. The combined output of both neuronal populations to the VMH is illustrated in the lower panel. With low leptin concentrations the inhibitory NPY neurons predominate, at moderate leptin concentrations the stimulatory a-MSH predominate, and at high leptin concentrations both neuronal populations are inactivated. It is worth noting that leptin at high concentrations can no longer activate the ARC neurons, and these neurons therefore appear to be ‘leptin resistant’.

156

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

From the aspect of competition for energy resources, two types of stress are possible: a pending energy deficiency in the brain and an excessive glucose-utilizing body mass. According to the ‘selfish brain paradigm’, the brain must be informed continuously about the magnitude of these two stressors in the form of feedback-signals. Its integrating centers receive feedback signals for this purpose from all brain areas as well as from the glucose-utilizing muscle and adipose tissues. The feedback signal from the brain itself is ATP, while the feedback signal from muscle and adipose tissue is leptin [12]. Leptin is formed and secreted in fat tissue and musculature in a manner closely coupled with the glucose uptake of these tissues [104,105]. Leptin conveys a signal describing the quantity of peripherally stored energy; it is closely correlated with body mass. From the standpoint of the new paradigm, leptin can be understood as a ‘load signal’ that informs the brain of the size of the metabolic stressors, i.e. the muscle and fat mass competing for glucose. If this load signal is interrupted, as is the case with leptin receptor defects, a key stimulus is missing for the allocation of glucose to the brain. The development of db=db mice expressing a leptin receptor defect has confirmed that the brain mass in the first postnatal weeks develops only slowly and inadequately, while the body mass increases disproportionately [106]. Leptin can therefore be assigned as a class of cytokine due to its functional and biochemical properties and can be understood as a load signal targeted towards the brain. Why is the regulatory system burdened by increasing body mass? The more food an organism consumes, the larger becomes the peripheral mass that must compete with the brain for glucose. With increasing body mass, leptin increases as an indicator of this ‘metabolic load’. Leptin stimulates the sympathetic nervous system in the hypothalamus and in so doing the allocation of glucose to the brain [107 – 109]. This functional feedback between intake of foods and glucose allocation is mediated via the feedback effect of leptin (see Fig. 5b). 2.2.1. Malnutrition 2.2.1.1. Metabolic stressors. In this chapter we basically repeat the principles of the model while focusing on its dynamic behavior. We also provide more insight into biological details by assigning one representative specific metabolic or neuroendocrine mechanism as well as one specific anatomical site to each component of the fishbone model. There are 14 components (flow-chart arrows) in Fig. 1 which are referred to, e.g. as model a1 –a2 in the following text. Mechanisms and neuroanatomical structures involved are explained with the help of a case study on malnutrition oriented towards the studies of Per Opstad [110]: Case 1: The 25-year old Olaf goes on a 10-day wilderness expedition to the mountains of Norway as

part of a ranger training exercise. On the 5th day he loses all his provisions through an accident. He manages to survive the remaining 5 day journey without practically any intake of foods, although during this time he loses 4 kg in body mass, and arrives exhausted in the training camp before indulging in a heavy meal. In the subsequent 2 weeks his food intake is also increased until his original body mass returns. In healthy individuals the brain ATP concentration is strictly regulated so that a marked reduction in ATP is not to be expected during a 5 day fasting period [111]. Nevertheless, the brain is able to measure even only a tiny reduction in brain ATP. As already mentioned in the chapter ‘balance-setpoints’, cortical high- and low-affinity ATP-sensitive potassium channels play a decisive role. With a tiny reduction in brain ATP, only the low-affinity ATP-sensitive potassium channels react while the highaffinity ATP-sensitive potassium channels remain closed. A minor activation of the low-affinity ATP-sensitive potassium channels reduces the GABAergic tone in the entire cerebral cortex. The balance between active glutamatergic and GABAergic neurons is displaced with a slight ATP deficit to the benefit of glutamatergic excitation (see Fig. 4a). Glutamate is taken up by the astrocytes where it stimulates glucose uptake, and this in turn is closely coupled with the transport of glucose via the blood – brain barrier (Fig. 4e) (model a2 – e2). GLUT1 transports glucose both through the luminal and abluminal cell membranes of the cerebral endothelial cells. Activation of the glutamate receptors has not only this rapid effect on GLUT1, but it also exerts a prolonged stimulatory effect on the expression of GLUT1 mRNA [112]. Glutamate therefore facilitates the passage of glucose across the blood – brain barrier in a number of cortical regions and can in this way correct a reduction in brain ATP partly or even completely. In parallel a glutamatergic tone in the cerebral cortex ensures via a series of intermediary steps that the musculature utilizes fatty acids instead of glucose. Glutamate stimulates the glutamate receptor on limbic neurons via projections that innervate the limbic system from various cortical regions (see Fig. 4b) [73] (model a2 –b2). According to Larry Swanson’s topographical model of cerebral hemisphere organization, both hippocampus and amygdala pyramidal cells contribute to triple descending projections—with excitatory, inhibitory, and disinhibitory components—extending to specific parts of the hypothalamus (VMH and PVN) [113]. Excitatory components include the amygdala basolateral complex and the hippocampal CA1 – 3 fields; inhibitory components include the amygdala central nuclei and the lateral septum; and disinhibitory components include the bed nuclei stria terminalis and the medial septal/diagonal band complex [113]. Signals from hippocampus and amygdala, which are transmitted via these multiple descending pathways, have been shown to affect

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

157

Fig. 4. (A) Cerebral cortex. Excitatory neurons produce the neurotransmitter glutamate and the neuropeptide brain-derived neurotrophic factor (BDNF). These neurons project into the limbic system where they release glutamate and BNDF. They are permanently under feedback control: (1) Postsynaptic high-affinity ATP-sensitive potassium channels are localized on these neurons. These channels are themselves closed at low glucose concentrations and ensure the functional activity of the excitatory neurons. (2) Low affinity ATP-sensitive potassium channels are localized presynaptically on GABAergic neurons. These channels are closed at high glucose concentrations so that the nerve endings release the inhibitory neurotransmitter GABA that inhibits the excitatory neurons. Low glucose concentrations act permissively, and high glucose concentrations inhibitorily on the activity of cortical excitatory neurons. (B) The limbic system. Excitatory neurons are localized in the core regions of the limbic system. These neurons project with their neuronal pathways into the ventromedial hypothalamus or into the paraventricular nucleus. They are stimulated by cortical glutamate that binds to the membranous glutamate receptors (Glu-R). These excitatory limbic neurons are subject to the influence of presynaptic GABAergic glucoresponsive neurons. The excitatory limbic pyramidal cells produce two important types of proteins in their cell nucleus: Under the influence of cortisol and its two receptors MR and GR they form proteins that define the excitability of these neurons. Under the influence of BDNF and its two receptors trkB and p75 they form so-called CREB-proteins which determine the number of membranous glutamate receptors during the process of long-term potentiation (LTP). At low cortisol concentrations MR are primarily active and these downregulate their own synthesis. MR also promote the synthesis of BDNF receptors (trkB). At high cortisol concentrations GR are mainly active, and these also downregulate their own production. Under the influence of GR, BDNF-receptors (p75) are produced. BDNF stimulates via its high-affinity trkB receptors the CREB gene, while it inhibits the CREB gene via its low-affinity BDNF receptors (p75). CREB-proteins lead to LTP and a durable alteration in the number of membranous glutamate AMPA receptors. In this way glutamatergic transmission is subject to modulation by cortisol and BDNF, and this can be stabilized over the long-term by LTP. (C) Ventromedial hypothalamus. Ventromedial hypothalamus-(VMH)-neurons stimulate the CRH-neurons in the paraventricular nucleus (PVN) and with that both the sympathetic nervous system and ACTH-release from the pituitary. These excitatory VMH neurons also mediate GABAergic output to the lateral hypothalamus. Limbic neurons stimulate VMH neurons. These excitatory VMH neurons are also subject to a dual feedback-control: At high brain-glucose concentrations, ATP-sensitive potassium channels on presynaptic GABAergic neurons are closed, which as a result release GABA and act inhibitorily on the excitatory VMH neurons. At high leptin concentrations the same ATP-sensitive potassium channels on the presynaptic GABAergic neurons are opened so that the neurons release less GABA and a stimulatory effect on the excitatory VMH neurons results. a-MSH from the ARC amplifies the stimulatory effect of leptin on the excitatory VMH neurons, while NPY from the ARC decreases the leptin effect. The neurons of the VMH measure the difference between the peripheral (leptin) and central (brain-glucose) feedback signals and generate the VMH output from this result. (D) Lateral hypothalamus. The glucose-sensitive neurons of the lateral hypothalamus release orexigenic peptides and in so doing stimulate food intake. The orexigenic neurons are stimulated by glutamatergic neurons and inhibited by GABAergic VMH neurons. They are subject to feedback-inhibition by brain glucose. With high brain glucose concentrations their sodium/potassium ATPase is activated so that these neurons become hyperpolarized and stop releasing orexigens. In addition, neuropeptides from the ARC also exert a modulatory influence. In energetic homeostasis, however, the stimulatory influence of NPY and the inhibitory influence of a-MSH are at a state of balance. (E) The blood–brain barrier and the cell membranes of muscle/adipose tissue. Glucose is transported by glucose transporter 1 (GLUT1) across the blood–brain barrier. Neuronal glutamate is taken up by the astrocytes and stimulates glucose uptake across the blood– brain barrier. Glucose is transported by the insulin-sensitive GLUT4 across the membranes of muscle and fat cells. The sympathetic nervous system regulates glucose uptake into muscle and adipose tissue by inhibiting pancreatic b-cells, thereby limiting the insulin-receptor (IR) mediated glucose uptake into peripheral tissues. Both neuronal glutamate release and activation of the sympathetic nervous system lead to an allocation of glucose to the brain, whereby both processes restrict peripheral glucose uptake. (F) Arcuate nucleus. The glucose-responsive neurons of the arcuate nucleus produce POMC and NPY. Both neuronal populations project into the VMH and the LH. In these two core regions the POMC neurons release a-MSH and the NPY neurons NPY. In the ARC, leptin binds to the leptin receptor (LR), thereby stimulating the POMC-neurons, and inhibiting the NPY-neurons. In addition, leptin directly accesses the VMH. At very high leptin concentrations the ATP-sensitive potassium channels which are localized on both ARC-neuron types are opened so that these neurons become hyperpolarized and deactivated.

158

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

Fig. 4 (continued )

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

Fig. 4 (continued )

159

160

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

the hypothalamus-pituitary – adrenal system [114 – 116] (model b2 – c2). Hippocampal stimulation or microinjections have been shown to activate both VMH neurons [117] and PVN neurons [118 – 121] (see Fig. 4c). A subsequent rise was observed in the hormones ACTH, corticosterone, epinephrine, norepinephrine [121 – 123] and in the substrates blood glucose and lactate [122,124]. We want to emphasize that although these structures and functions are understood as potential candidates they do not represent the absolute and final mechanisms. In summary, glutamatergic cortical signals can activate limbic structures that enhance hypothalamic neuronal activity, and in so doing request and mobilize circulating fuels. A minor reduction in ATP opens the KATP channels in the presynaptic VMH-neurons so that the VMH neurons are not just stimulated by the limbic system, but they are also locally disinhibited. These hypothalamic KATP channels belong to the network of hierarchically organized ATP sensors that maintain glucose homeostasis [125 – 128]. The VMH governs glucose allocation by limiting peripheral glucose uptake (model c2 – e2). Local or systemic glucoprivation increases signal output from the VMH [129, 130]. VMH neurons project towards the PVN, where they can stimulate CRH and vasopressin release [131,132]. At the same time, GABA- and BDNF-containing neurons are activated in the VMH [133,134] which project to the LH and inhibit appetite. Both a release of ACTH from the pituitary and cortisol from the adrenal gland as well as a stimulation of the sympathetic nervous system coincide with the stimulation of PVN neurons. The sympathetic nervous system innervates the pancreatic b cells [66,135] where the stimulation of a2-adrenergic receptors suppresses the secretion of insulin [68,69], and antagonizes insulin effects on muscle and adipose tissue [70]. As a result, less glucose transporter 4 (GLUT4) is translocated onto the cell membranes of muscle and adipose tissue (Fig. 4e). The sympathetic nervous system also innervates the musculature where it can open KATP channels, and this also decreases the insulin-mediated glucose uptake [136,137]. Since GLUT4 is the main glucose transporter for these peripheral tissues, a decreased peripheral glucose uptake is seen during fasting. Glucose is therefore guided past the peripheral tissues and is instead available for uptake by the brain [138,139]. Again, each particular mechanism identified here can fulfil its role in the model, although the whole theory is not necessarily bound to any specific one. In conclusion, a raised glutamatergic tone in the cerebral cortex activates allocation of glucose to the brain by promoting GLUT1-mediated glucose transport across the blood – brain barrier and restricting GLUT4-mediated glucose transport into muscle and adipose tissue. During a 5-day fasting period, a latent brain ATP deficit therefore activates glucose allocation to the brain. How does ketone body formation on the one hand and a reduction in body mass on the other alter the burdening of the regulatory system?

During the first day of fasting, serum leptin concentration lowers drastically due to the reduced allocation of glucose to adipose and muscular tissue [140]. During this phase the liver forms ketone bodies from free fatty acids. Since the brain can metabolize these ketone bodies as an alternative substrate, the loading of the regulatory system for glucose allocation is alleviated (Fig. 5b). The decrease in body mass on the other hand occurs far more slowly than does the leptin reduction. At the end of the fasting period Olaf had lost approx. 4 kg in mass. His brain at the end of the fasting period therefore had to compete with a reduced glucose-utilizing organ mass. This entails a certain disburdening for the regulatory system which is already burdened by a threatening state of ATP deficiency in the brain. In Fig. 5b the characteristics for low body mass and additional ketone formation are graphically illustrated, and they lie below the normal characteristic. The body mass M is included into the calculation of this characteristic according to Eq. (3). This means that with a small body mass M; less glucose allocation suffices in order to supply the brain with adequate energy for the same intake of foods. On the 5th day of fasting the brain’s energy supply remains critical, but the competition between brain and periphery has shifted somewhat to the benefit of the brain. 2.2.1.2. Replenishment. During fasting, the glutamate command signal acts on the appetite controlling LH (Fig. 4d) (model a2 – d2). It is beyond question that on the 5th day of fasting Olaf in case 1 suffered an extreme feeling of hunger. Upon intake of food a large amount of glucose is then available for allocation to the brain and the periphery. After a few minutes the raised substrate supply results in more glucose reaching the brain across the blood – brain barrier. The brain ATP increases further during food intake until the setpoint is achieved for ATP (model a3 – a2). The KATP channels on the GABAergic cortical neurons are closed and the GABA-ergic tone increases. Glutamatergic and GABA-ergic activity return to a state of equilibrium, and the ‘glutamate command’ signal normalizes. The energy supply to the brain returns to an optimal state. Only in the next step is the allocation of glucose to the periphery regulated. The KATP channels in the VMH and the pancreatic b cells play key roles during the replenishment phase for the periphery (model c3 – c2). Immediately after the start of food intake, the brain is already optimally supplied again while the peripheral energy depots are still depleted. The brain ATP is normal, while the leptin concentration and the body mass are still noticeably reduced. How does this pattern of events influence the functional state of the KATP channels in the VMH? Leptin can open KATP channels directly and within minutes (model c1 – c2). However, circulating Leptin is subject to only slow changes over time. It is transported via a specific transport mechanism across the blood – brain barrier. Leptin binds to its leptin receptor in the VMH [141, 142] which is localized on presynaptic GABAergic neurons [143] (Fig. 4c). Mike Ashford’s research group from

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

161

Fig. 5. (A) Balance and Setpoint. The food intake required for brain nourishment (ordinate) depends on glucose allocation to the brain (abscissa). The reciprocal relationship describes how much food intake is necessary in order to achieve an adequate energy supply to the brain with a given glucose allocation to the brain. If there is adequate energy supply of the brain, the effects of high- and low-affinity ATP-sensitive potassium channels are in equilibrium. Glucose allocation is described as the ratio between the glucose uptake rate of the brain and the glucose uptake rate of the muscle and adipose tissue. The higher the allocation,

162

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

the University of Aberdeen, United Kingdom, succeeded in determining the key facts surrounding this process. The leptin receptor activated in the VMH opens the cells own KATP channel via an intracellular signal cascade (via phosphoinositide-3-kinase) [25]. The KATP channel in the VMH has a dual input: an inhibitory input for ATP and a stimulatory input for leptin. This KATP-sensor with a dual input measures the difference between a signal representing brain ATP and one representing the peripherally stored energy. From a regulatory-theoretical viewpoint, one can expect such a dual input for a sensor controlling allocation to two organs. The principle of allocation control can be described briefly as follows: if the signal for brain energy status dominates (i.e. ATP is sufficiently available), the hypothalamic KATP channels are closed and glucose is assigned to the periphery; if the signal for peripheral energy status (leptin) dominates, the hypothalamic KATP channels are opened and glucose is assigned to the brain.

Immediately after starting food intake the inhibitory signal of brain ATP dominates over the stimulatory leptin signal and the KATP channels are closed. Closed KATP channels in the VMH lead to a decrease in noradrenergic tone (that stimulates the sympatho-adrenal system) and via vagal stimulation lead to a secretion of insulin from the beta cells [135,144]. In parallel to that the LH is disinhibited, and these neurons stimulate insulin secretion via direct efferents to the beta cells [145]. Insulin enables glucose allocation to the muscle and adipose tissue. During the replenishment phase, leptin increases in the blood and increasingly opens the KATP channels in the VMH. If the KATP channels are opened, the tone of the sympatho-adrenal system increases and allocation of glucose to the periphery is curbed (Fig. 4c). The energy needs of the brain and the entire organism can be balanced over the long-term with the aid of the brain’s hypothalamic allocation regulator. The extent to

R the more glucose is assigned to the brain. With very high glucose allocation the glucose is assigned almost exclusively to the brain so that food intake almost exclusively covers the energy needs of the brain (dashed line). With very low glucose allocation the greater proportion of the glucose is assigned to the musculature and adipose tissue so that a larger food intake is necessary in order to ensure that the brain is supplied. The LHPA system determines glucose allocation to the brain and strives to achieve a balance state (setpoint) under the influence of the corticosteroid feedback loop. The setpoint of the LHPA system (vertical straight line) is determined by the MR-GR balance. The intersection point of the reciprocal function and the vertical straight line characterizes the setpoint for homeostasis: here, the brain ATP concentration is sufficient and the LHPA system is at a state of balance. (B) Loading of the regulatory system. The reciprocal function describes the forward signal path of the regulatory system. With a given glucose-allocation to the brain the required food intake is controlled according to the reciprocal function. The feedback signal-path (diagonal straight line) describes the reverse effect of the food intake on glucose allocation. The more food that is consumed, the more glucose enters into the musculature and adipose tissue so that the leptin concentration increases. Leptin stimulates the allocation centre (VMH) in the hypothalamus. The feedback signal path therefore describes the association whereby an increased food intake leads to increased glucose allocation to the brain.The lower reciprocal function describes the forward signal path when load on the regulatory system is alleviated. With a reduction in body mass (M), the food intake required for brain energy supply is smaller (compare equation 3). Correspondingly, alternative substrates that provide a portion of the brains energy supply, such as ketones, lead to a disburdening of the regulatory system. The numbering describes the chronological sequence of events provided in the first case study (ranger training): 1. 2. 3. 4. 5.

Increased glucose allocation to the brain due to stress from fasting. Disburdening of the regulatory system through ketogenesis and body mass reduction. Replenishment of the stores due to glucose allocation to the muscle and adipose tissue. Normalization of body mass. Return of the LHPA system to a state of balance (setpoint).

(C) Feedback amplification. The leptin signal can be modulated by neuropeptides from the ARC, i.e. a-MSH and NPY. In this way the feedback function is complemented by the feedback signal l(leptin). The way in which the feedback signal l (leptin) is generated in the ARC is illustrated in Fig. 3c. The modulated feedback signal [Leptin þ l(Leptin)] at high leptin concentrations produces an amplified stimulation of the sympatho-adrenal system and as such an increased glucose allocation to the brain. At low leptin concentrations it attenuates the sympatho-adrenal system and with that the glucose allocation to the brain. At extremely high leptin concentrations the amplifier mechanism is ineffective ½lðleptinÞ ¼ 0 and leptin only exerts its own, unmodulated effect on glucose allocation. (D) Setpoint-change. Glucose allocation to the brain is determined by the LHPA system. The LHPA-system is influenced by the MR–GR balance which determines the system’s balance state (setpoint). Extreme loading of the LHPA system can lead to a pathological change in the MR-GR balance. If the setpoint of the LHPA-system is chronically too low, glucose allocation to the brain also decreases. Food intake, necessary for an adequate brain supply, becomes greater as a result. With a lowered setpoint the brain supply is therefore maintained more by intake than by allocation of nutrients. (E) Development of obesity. Obesity can develop in three stages: (1) A setpoint-reduction of the LHPA-system leads to an inadequate glucose allocation to the brain (with increased allocation to muscle and adipose tissue) and makes it necessary to increase the uptake of nutrients. (2) Consequently, a loading of the brain-supplying regulatory system is due to the continuous increase in body mass. (3) Under the feedback influence of leptin the LHPA system is stimulated in the burdened regulatory system and glucose allocation to the brain is increased. Therefore the LHPA-system, whose setpoint is reduced but whose load is increased, shows an apparently ‘normal’ activity. The ‘normal’ activity of the LHPA-system does not suffice, however, to supply the brain with energy so well that a reduction in food intake can be enabled. [The deep red areas describe situations in which body mass has a tendency to increase, while the yellow areas describe situations where it tends to decrease.] (f) Development of a type 2 diabetes mellitus. The development of type 2 diabetes mellitus can occur in 3 stages: (1) A setpointreduction of the LHPA system. (2) A loading of the regulatory system due to increase in body mass. (3) An amplified feedback-effect of leptin from melanocortins. Leptin and melanocortins both stimulate the LHPA system. Individuals of a certain genetic background with an especially strong melanocortin system can amplify the leptin-effect on the sympatho-adrenal system in such a way that the burdened LHPA system is hyperactive. Such an overactivity of the LHPA-system leads to an increased glucose allocation to the brain, a rise in blood glucose, and in extreme cases to glucosuria. The loss of glucose via the kidneys represents an additional load on the regulatory system. An amplified glucose allocation to the brain may lead not only to a type 2 diabetes mellitus, but also during the further course of disease to a loss in body mass (yellow area).

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

which the limbic system also represents a center that integrates the feedback-signals brain ATP and leptin is not yet clear. There are KATP channels [146] and leptin receptors [142,147] in the limbic system that also play a role in memory formation and long-term potentiation [148 – 150]. These findings indicate that the limbic system plays a supraordinate role over the hypothalamus in balancing the energy needs of the brain and the entire organism (compare chapter ‘sleep and the consolidation of setpoints’). In case 1 a homeostasis gradually appeared over the ensuing three weeks during which both the brain/ periphery energy balance and the LHPA system returned to a state of calm. Fig. 5b summarizes the five different constellations that appeared in case 1 during fasting and replenishment of nutritional stores: 1. At the beginning of fasting, glucose allocation to the brain was activated because of a threatening state of cerebral energy deficiency. 2. After 2 days the brain is supplied in part by ketone bodies. During the further course of fasting the body mass decreased. Both ketones and the reduced body mass led to a disburdening of the regulatory system. With falling leptin concentrations the allocation of glucose to the brain was somewhat reduced. 3. Immediately after the start of the replenishment phase the brain supply returned to an optimal state so that allocation of glucose to the musculature was restored. 4. During the replenishment phase the body mass renormalized. The increasing leptin concentrations activated glucose allocation to the brain more and more. 5. The LHPA system returned to its resting state with an equilibrated MR-GR balance. Thus, the allocation of glucose to the brain and the muscle/adipose tissue had returned to a state of equilibrium. 2.2.1.3. Economics of the alternative brain-specific substrates. The brain’s ‘strategy’ of efficiently using lactate or ketones for its energy needs may shed new light on the pathogenesis of type 2 diabetes (see Chapter 3.3.4. diabetes mellitus). Lactate and ketones pass the blood –brain barrier with the help of specific monocarboxylate transporters, can be directly metabolized by neurons as energy substrates [151,152], and can protect the brain against hypoglycemia [153 – 157]. Lactate arises in the muscle tissue from the glycolytic, anaerobic breakdown of glucose. Ketones are formed in the liver. The liver utilizes free fatty acids for ketogenesis that are lipolytically released from visceral central adipose tissue and reach the liver via the portal vein. Contrastingly, only a small proportion of the free fatty acids circulating in the general circulation (arising from peripheral adipose tissue) gain access to the liver, and to all intents and purposes these are not available for ketogenesis. The free fatty acids arising from peripheral adipose tissue are used as substrates for cardiac and skeletal musculature.

163

The central adipose tissue is an energy depot for the brain, while the peripheral adipose tissue is an energy depot for the musculature. The brain can optimize its energy supply by maintaining lactate and ketones during phases of food supply and deficit. In order to safeguard its energy needs during critical nutritional states over particularly long periods it can utilize central adipose tissue stores (Fig. 6). Case 2: The natives of the pacific island of Nauru often undertook long canoe journeys from one island to the next before the era of European colonization. Numerous reports have suggested that such canoe rides sometimes lasted several weeks and, contrary to expectations, were frequently survived [19]. How can an organism maintain its brain supply over extremely long periods of undernutrition? There is a neuronal network in the hypothalamus that reinforces the effect of leptin during nutritional surplus and curbs its effects during undernutrition [12]. This reinforcement of the leptin effect works as a kind of servomechanism. With food surplus it leads to an increase in sympatho-adrenal activity which is associated with an increased release of lactate from the musculature. During this phase, when peripheral stores are replenished, the brain employs glucose and lactate. In this process it leaves the glucose, spared by the lactate, for retention in central adipose tissue. With prolonged fasting, free fatty acids are released from central adipose tissue and converted to ketones in the liver; during this phase of fasting the brain utilizes glucose and ketones. The metabolization of ketones in the brain allows a reduction of sympatho-adrenal tone

Fig. 6. Economics of alternative brain-specific substrates. An amplification mechanism of the leptin-feedback effect allows at times of nutrient intake a particularly strong activation of the sympatho-adrenal system that promotes lactate release from the muscle. As a result, lactate instead of glucose can cover a portion of the brain’s energy. The glucose spared in this way can be stored instead under the influence of insulin and cortisol in central adipose tissue stores. With long-term fasting, free fatty acids from central adipose tissue can be employed for ketogenesis in the liver. As a result, ketones instead of glucose can cover a portion of the brain’s energy needs. Therefore less glucose-allocation to the brain is required, and load on the sympatho-adrenal system is alleviated, i.e. it can operate in ‘save mode’ and can maintain brain supplies economically over a long period of time.

164

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

and consequently a reduction in glucose allocation to the brain. Thus, the utilization of ketones as an alternative substrate for the brain allows a ‘saving mode’ to be entered. Leptin amplification is provided by the melanocortin system in the hypothalamic ARC (Fig. 4f). Under the stimulating influence of leptin a neuronal population of the ARC synthesizes the precursor molecule pro-opiomelanocortin (POMC), from which the melanocortin a-MSH is cleaved [12]. The axons of these POMC neurons innervate the VMH, where a-MSH is released and stimulates noradrenergic neurons via its melanocortin (MC4) receptor (Fig. 4c). This signal pathway runs parallel to the abovedescribed primary leptin signal which directly acts on the KATP channels in the VMH (Fig. 4c) (model d1– c2). One can distinguish a primary leptin signal that sends information directly from the periphery to the VMH (allocation), from an amplifying secondary leptin signal, which only acts on the VMH with the aid of a-MSH after conversion in the POMC neuron. In this way, a-MSH amplifies the leptin effect on CRH (which promotes glucose allocation to the brain) and thereby attenuates food intake [158]. The POMC neuron also projects with its axons directly into the appetite-controlling LH (model d1– d2). Here a-MSH acts inhibitorily via its MC4 receptor on the appetite-inducing orexigenic neurons. POMC neurons are glucose-responsive, i.e. they possess KATP channels that are closed by ATP and opened by leptin or insulin [26,159]. With cerebral ATP deficiency or peripheral leptin surplus the KATP channels are opened and the POMC neurons are inactivated. This means that the leptin-amplifying melanocortin system is only fully functional with a balanced cerebral ATP content and a normal body mass. Leptin amplification can also be accomplished through a second signal pathway emanating from the ARC. Here, leptin inhibits a neuronal population that synthesizes neuropeptide Y (NPY), a powerful appetite stimulator (Fig. 4f) [160]. The NPY neurons project with their axons also into the VMH, where unlike a-MSH they act inhibitorily, and they project to the LH, where unlike aMSH they act in a stimulatory manner [12]. Fig. 3c provides an overview of the activity of POMC and NPY neurons depending on leptin concentration. At low leptin concentrations, e.g. after long-term fasting, the NPY neurons are activated, with moderate leptin concentrations the POMC neurons are active, while with high leptin concentrations both the POMC and NPY neurons are inactivated. From Fig. 3c it should also be noted that with moderate leptin concentrations mainly melanocortin (a-MSH) is released by the ARC while with low leptin concentrations NPY is mostly secreted. In conclusion, the melanocortin system is activated with nutrient surplus (with moderate leptin concentrations); as such the leptin effect on glucose allocation in the VMH is amplified. In contrast, glucose allocation in the VMH is reduced during fasting (with low leptin concentrations). Fig. 2c shows that the melanocortin system is able to amplify glucose allocation and decrease the intake of foods.

While the body energy stores are being replenished the melanocortin system is able to amplify the sympathoadrenal activity [161,162]. Catecholamines and cortisol favor the increased build up of lactate in the musculature [163 –165]. The brain can employ lactate directly as an alternative substrate [151 –157]. A certain proportion of the glucose is therefore available for other tasks: to be stored in visceral central adipose tissue. Cortisol also favors the storage of glucose in exactly this fat compartment. One such cortisol effect with central fat accumulation is known to clinicians in the form of Cushing’s disease. The melanocortin system can also increase allocation to the brain. Under the influence of melanocortin, the brain not only receives an additional energy supply from lactate—as described above—but also an additional share of glucose. The increased allocation under these conditions enables the organism to take in less food. In the same way an appetite suppressing effect of a-MSH is possible without the brain supply being endangered. The melanocortin system increases the role of allocation and decreases that of food intake (Fig. 2c). Under the influence of the melanocortin system a somewhat reduced intake of foods occurs while store replenishment is taking place, so that less energy is stored in peripheral adipose tissue. In conclusion, the melanocortin system amplifies the leptin effect; it mobilizes lactate and glucose for the brain and in this way promotes central adipose tissue growth while maintaining adequate supply to the brain. In times of prolonged fasting these accumulated central fat depots can ensure a long-lasting supply of ketone bodies to the brain. From a biomechanical standpoint the storage of energy in certain strongholds, i.e. in the core of the body, is particularly favorable. Peripherally stored masses can considerably restrict physical mobility (larger inertial momentum), whereby a mass at the body’s center of gravity (smallest inertial momentum) [166] permits an almost unrestricted mobility. The central (visceral) adipose tissue produces less leptin than peripheral fat [167], and therefore allows more energy uptake. Currently it is not known whether the natives of Oceania described in case 2 have a supranormally effective melanocortin system. James Neel in his ‘thrifty gene hypothesis’ [168] postulated that a combination of genes might exist that allows an especially efficient utilization of nutrients and improves the survival chance of their carriers in situations of long-term starvation. Our above-described views on how an effective melanocortin system can allow a long-lasting brain supply allows us to assume that the natives of Oceania do indeed carry genes that allow particularly effective hypothalamic leptin amplification. Fig. 5c summarizes the behavior of a brain equipped with a leptin amplification system in the ARC: It responds to an increased intake of foods with an especially effective loading of its stress system that coincides with an increased glucose allocation to the brain. It responds to undernutrition

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

with a highly effective restriction of its stress systems, enabled by an economically effective utilization of ketones. 2.2.1.4. The ‘primacy of the brain’. What mechanisms underlie the primacy of the brain regarding energy supply? Two independent mechanisms are available to the brain for measuring cerebral ATP content: the KATP channels and the Naþ/Kþ-ATPase. As described above, the KATP channels act as sensors for the regulation of glucose allocation. The Naþ/Kþ-ATPase is a molecular structure that is ATPsensitive and which can activate the appetite-stimulating orexigenic neurons in the LH [169,170]. The neurons in the VMH (KATP channels) and the LH (Naþ/Kþ ATPase) differ fundamentally. With ATP deficiency the VMH neurons hyperpolarize, while the LH neurons depolarize. Correspondingly, the VMH neurons are also known as ‘glucoseresponsive’, while those in the LH are designated as ‘glucose-sensitive’ [171,172]. With cerebral ATP deficiency the hyperpolarization of presynaptic GABAergic neurons in the VMH allows glucose allocation to the brain. Contrastingly, depolarization of glucose-deprived LH neurons leads to the release of orexigenic neuropeptides (e.g. orexins; Fig. 4d) [173–175] (model d3–d2). In the life-threatening situation of a cerebral energy deficit, also known as neuroglucopenia, both the allocation mechanisms as well as hunger are activated independently of one another. One such situation occurs for example with insulin-induced hypoglycemia which is frequently seen in individuals with type-1 diabetes. The brain has yet another mechanism to guarantee its primacy regarding its energy supply. If a cerebral energy deficit or a peripheral energy surplus occurs (both of which are critical constellations regarding the competition for energy resources), signals from the periphery are only conveyed in a restricted way to the regulatory centers of the brain. There is a signal-filter in the ARC that controls whether the leptin signal should be relayed. Although the leptin-signal can reach the VMH at any time unhindered, the melanocortin signal from the ARC is transmitted only under certain conditions: i.e. with sufficient brain ATP and at physiological leptin concentrations. With cerebral ATP deficiency or leptin excess the postsynaptic KATP channels of the ARC neurons are opened [25,26,159] so that these neurons are hyperpolarized and stop releasing melanocortin (a-MSH) at their synapses. In obese individuals, who have high leptin concentrations, the stimulatory effect of leptin on the melanocortin system has been lost as a rule. This phenomenon is often described as ‘leptin resistance’. The melanocortin system suppresses appetite and food intake and can therefore theoretically induce a serious conflict in a life threatening neuroglucopenia. Since in neuroglucopenia the brain is dependent on all possibilities for obtaining energy, i.e. allocation and food intake, it deactivates the melanocortin system in such crisis situations. This neuroprotective mechanism permits a ravenous hunger sensation during hypoglycemia, even in situations where an excessive peripheral body mass provides a ‘satiety’ signal to

165

the brain. It has indeed been shown that hyperphagia during hypoglycemia is not mediated via signals from the ARC [176]. The leptin resistance of the melanocortin system is therefore a necessary condition to maintain the ‘primacy of the brain’ regarding energy supply. 2.2.2. Psychological stress Case 3: Sascha, a 22-year-old architectural student, undertakes his first parachute jump. In the parachuteschool beginners are strapped for a so-called ‘tandem jump’ to the front of an experienced trainer. At 10 a.m. the plane ascends to a height of 3500 m. During the 15 min ascent and particularly during the seconds before the jump Sascha appears somewhat excited. He senses severe palpitations, shaking, and inner unrest. The tandem jumpers experience 45 s of freefall before the parachute opens and 5 min coasting under the opened parachute. A safe landing follows. The psychological stress is resolved immediately. At 1.00 p.m. during the subsequent communal lunch with the other beginners, Sascha expresses euphorically that he wants to learn to parachute at any cost. During the evening he falls exhausted but satisfied into a deep, restful sleep. What effects does psychological stress have on glucose metabolism? Sascha was exposed to a strong psychological stressor and put into a state of great tension, excitement, and attentiveness. Many new experiences had to be dealt with, selected, and judged. In this state, cortical excitatory activities, especially of glutamatergic neuronal populations were dominant. The GABAergic neuronal activity in this situation took the back seat. As described above, the glutamatergic neuron population provides the glutamate command signal. The cortical balance between glutamatergic and GABAergic activity prevailing during exposure to psychological stress resembles that observed during metabolic stress such as neuroglucopenia. Glucose transport across the blood – brain barrier is increased. Cortical glutamate via receptors in the limbic system stimulates projections activating the LHPA system [73]. Activation of the LHPA system coincides with stimulation of the sympathetic nervous system (leading to symptoms such as ‘palpitation’) and with the release of hormones from the hypothalamus-pituitary system [177,178]. As already shown in the chapter ‘malnutrition’, activation of the stress systems promotes glucose allocation to the brain and reduces the uptake of glucose into muscle and adipose tissue. In conditions of psychological stress the brain is adequately supplied with energy in this way. The degree of activation of the LHPA system is indicated by blood cortisol concentrations. With a parachute jump, as in the above described example, serum cortisol climbs from 15 to 25 mg/dl [179]. Catecholamine, cortisol and insulin play a role in allocating glucose. Cortisol and insulin also

166

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

feedback to the brain [98,180]. The rise in cortisol indicates how strongly the organism reacts to stressor exposure. Cortisol easily crosses the blood – brain barrier as a lipophilic hormone. Sites of cortisol feedback appear to include the limbic system, the hypothalamus, the pituitary, and the adrenals. There is a debate as to whether the locus of limbic feedback action can be traced by lesion studies to the hippocampal formation [181 – 183]. Such controversial findings suggest that other brain regions, e.g. the amygdala, are also essentially involved. The limbic neurons belong to a complex information processing system, and they can act in concert via triple (excitatory, inhibitory, and disinhibitory) descending pathways to exert corticosteroid feedback. In our opinion, it is the information stored within these limbic neuron populations, in terms of emotional or declarative memories, that optimises the energy request signal. The energy request is only a crude—but necessarily—anticipatory estimate of the energy need, i.e. the amount of energy that will actually be utilized. An inadequate energy request might be harmful to the brain. An excessive energy request mediated by the LHPA system may be harmful to the body (catabolism). It is therefore adaptive to match the energy request closely with the energy need. The limbic system modulates the glutamate command signal on the basis of previous experiences. Learning modifies limbic signal transduction in three ways: fear conditioning enhances, contextual learning of familiar content attenuates, and contextual learning of aversive content potentiates [184]. The triple descending projections described by Swanson which consist of excitatory, inhibitory, and disinhibitory components precisely reflect this functionality [113]. Accordingly, the limbic system can use memories to make the most economic energy request. Cortisol binds to MR and GR in the limbic system and to GR in the hypothalamus. At high cortisol concentrations it is primarily GR –GR homodimers and MR – GR that form in the pyramidal cells of the limbic system. These GR-dimers with their feedback effects control the excitability of the limbic neurons that determine activity of the LHPA system (model b3– b2). With long-lasting stressors, cortisol inhibits the mRNA of those peptides and receptors that convey or amplify the glutamate command signal from the limbic system to the body periphery: in the PVN CRH [185] and vasopressin [186], in the ARC POMC [187], in the pituitary CRH receptors [188] and POMC [189], in the adrenal gland melanocortin (MC2) receptors [190], and on glucoseresponsive cells high-affinity ATP-sensitive potassium channels [191]. However, as long as the psychological stress persists, the LHPA system continues to be stimulated by widespread cortical glutamatergic input. After a safe parachute landing the psychological stressors are no longer present. Cortical excitation normalizes, as does the stimulatory input to the LHPA system. Now the LHPA system is subject to the dominant inhibitory feedback influence of the GR-dimers in particular. Under the inhibitory feedback influence, Sascha’s serum cortisol concentrations

fall off according to their half-life. VMH neurons that inhibit the LH [84,134] decline in activity. The LH in turn releases appetite-stimulating orexigens. In addition, cortisol suppresses CRH in the PVN which also has an appetite suppressing effect aside from its function as a releasing hormone [192]. Because of this Sascha is able to eat food again after his successful landing. Interestingly, the ingestion of glucose may additionally suppress the LHPA system activity [193,194]. The stores are replenished, as described in detail in the chapter ‘malnutrition’. In the evening, serum cortisol concentrations reduced continuously. The lower the cortisol is, the less binding there is to GR in the limbic system, and MR– MR homodimers predominate. MR – MR homodimers increase excitability of the limbic neurons that control the LHPA system (model b1– b2). Under the influence of CRH and ACTH, cortisol is released from the adrenals sufficiently to prevent further decline. Before sleeping the cortisol concentrations reach a nadir of about 3 mg/dl. At this time-point the adrenal cortisol production is approximately as large as the hepatic cortisol elimination so that cortisol concentrations stabilize. During slow wave sleep in the first half of the night the MR-MR homodimers are mainly active at the low cortisol concentrations that then prevail. As we shall explain in the following chapter, a consolidation of memory and metabolic setpoints can occur under the influence of MR – MR homodimers during the slow wave sleep phase. 2.3. Sleep and the consolidation of setpoints In addition to its general restorative function, sleep also serves to consolidate memory. The definition of memory refers in this case to a fundamental function of controlled biological systems whereby a memory is formed for the purpose of consolidating and stabilizing setpoints. During the wake phase setpoints are newly acquired (learned) as part of an adaptive interaction between the organism and its environment; existing (inborn as well as learned) setpoints are subject to destabilizing influences (stressors). Unlike the wake phase, sleep represents a phase during which the consolidation of memory is optimized. During a period of minimum reactivity towards environmental stimuli, sleep allows the reorientation of an inner homeostasis through the stabilization of setpoints and with that a lasting ‘strategic’ adaptation to stressors. Glutamate receptors are necessary for conveying the glutamate command signal to the body periphery [73]. The cerebral cortex communicates with the limbic system via a glutamate-mediated synaptic transmission. There are two main types of ionotrophic glutamate receptors: the AMPA receptor involved in basal synaptic transmission, and the NMDA receptor that specifically modulates a form of synaptic plasticity, i.e. so-called long-term potentiation (LTP) [195]. LTP is thought of as the basic neuronal process underlying the formation of memory. Activation of the NMDA receptors increases the calcium concentration

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

in the postsynaptic neuron and thereby activates a range of intracellular biochemical signal pathways. These intracellular signals modify both the number and the activity of the postsynaptic AMPA receptors with lasting effect. In LTP, AMPA receptors are recruited in greater numbers to the cell membrane so that synaptic signal transmission is increased [196,197]. In the counteracting process, i.e. long-term depression (LTD), AMPA receptors are endocytosed so that synaptic transmission is reduced [196] (see Fig. 4b). 2.3.1. Stressors and the limbic system How do limbic neurons find the optimal conditions for LTP? As an answer we propose: 1. With the assistance of high-affinity and low-affinity BDNF receptors and 2. with the assistance of high-affinity and low-affinity brain corticosteroid receptors. A major breakthrough was the identification of the neurotrophin BDNF by Yves Barde and Hans Thoenen at the Max-Planck-Institute in Martinsried/Munich [198]. BDNF and other neurotrophins, such as the formerly known nerve growth factor, are trophic proteins necessary for the differentiation and survival of neurons. In addition they modulate synaptic transmission and plasticity [199]. Unlike their slow effects on neuronal survival or differentiation (h or days), modulation of synaptic transmission is much faster (s to min). It has now become apparent that BDNF exerts its effects on LTP and LTD via two different membrane receptors: the high-affinity trkB receptor and the low-affinity p75 receptor [200,201]. Like most growth factor receptors the trkB receptor displays a cytoplasmic domain with a tyrosine kinase. After binding BDNF, tyrosine kinase mediates autophosphorylation of the trkB receptor as the first step of intracellular signal transduction. The p75-receptor on the other hand lacks the intracellular tyrosine kinase domain. That means that the p75-receptor competes with the trkB-receptor to prevent trkB-induced activation of cellular processes. The neurotrophin receptors form three kinds of dimers: trkB – trkB homodimers, trkB – p75 heterodimers and p75 – p75 homodimers [202 – 204]. The receptor dimers bind in each case one BDNF molecule. Consequently trkB – trkB homodimers are predominantly active at low BDNF concentration, while at high BDNF concentrations p75-containing dimers are. The complete effect of BDNF is mediated in this case via the homodimer trkB –trkB, through mutual phosphorylation of the two cytoplasmic tyrosine kinase residues. The p75-containing dimers do not transmit the signal because of their lacking cytoplasmic domains. Opposing (e.g. antiapoptotic and proapoptotic) effects of trkB and p75 receptors have indeed been demonstrated [203]. BDNF mediates the induction of LTP [199,205] and attenuates LTD [206] via its trkB receptor. Autophosphorylation of the trkB receptor activates a cascade of kinases leading to

167

gene activation and protein synthesis, with this effect fading within a few hours. Alcino Silva and Eric Kandel of Columbia University in New York showed that the ‘cAMP responsive element binding’ (CREB) gene produces exactly the proteins that appear to be essential for neuronal plasticity and longterm memory [207,208]. CREB is a so-called ‘early’ gene, and the CREB protein a transcription factor [209]. With LTP the CREB protein controls production of effector proteins that stabilize synaptic modification and help to form new synapses: amongst these include structural proteins for new synapses, receptors for the cell membrane, as well as kinases and proteins that maintain vital functions of the neuron. Genes like CREB activate a genetic chain reaction by activating diverse ‘later’ genes. After a few hours the synthesis of glutamatergic AMPA receptors intensifies. The number of AMPA receptors at the cell membrane stabilizes over the long-term. In summary, BDNF mediates its effect on LTP as follows: 1. BDNF binds with high affinity to the trkB receptor and low affinity to the p75 receptor. 2. BDNF-bound trkB and p75 receptors form three types of dimers: trkB-trkB, trkB-p75, and p75-p75. 3. The trkB-trkB homodimers mediate the BDNF effects on LTP while the p75 containing dimers prevent these. This interaction of a ligand with two receptors also reminds us of a general principle which we have already illustrated with the brain corticosteroid receptors (MR and GR) and the high- and low-affinity ATP-sensitive potassium channels. A bell-shaped dose-effect-relationship for BDNF on the production of CREB-protein results from the behavior of BDNF and its two receptors (Fig. 7).

Fig. 7. Optimal conditions for long-term potentiation. Induction and maintenance of long-term potentiation (LTP) involves induction of CREB proteins. CREB protein expression depends on BDNF concentration, with a bell-shaped dose-effect relationship. BDNF-(trkB) receptors must be present in sufficient numbers for the complete BDNF effect. BDNF(trkB) receptor synthesis depends on cortisol, again with a bell-shaped dose-effect relationship. After a stressful event, LTP induced during wakefulness is maintained during slow wave sleep only if both cortisol and BDNF lie within optimal ranges. In this way the effects of BDNF (as an indicator for the scale of the stressor effect) and cortisol (as an indicator for the scale of the stress reactions) are integrated.

168

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

2.3.2. Stress reactions and the limbic system Cortisol also modulates synaptic transmission and plasticity. The effects of cortisol in the limbic system are influenced by the interaction between MR and GR. Originally, various groups found that cortisol-bound MR [210–213] and GR [214–218] were subject to ‘autoregulation’, i.e. they both suppressed their own production. Bruce McEwen and colleagues showed that MR activates LTP while GR attenuates it [219–221]. Extending these findings, Ronald de Kloet and coworkers showed how cortisol modulated LTP when they confirmed that corticosteroids influence the production of BDNF receptors [222]. Expression of trkB– mRNA was highest with low corticosteroid concentrations, while at high corticosteroid concentrations it was barely evident and not different from the vehicle injected controls. This dose-effect relationship supported the idea that predominant MR activation is critical for the production of trkB receptors. Recently it was shown that GR activation enhances the production of p75 receptors [223]. Together, these findings lead us to conclude a bell-shaped dose-effect curve for corticosteroids on trkB receptor production (Fig. 7). BDNF and cortisol jointly regulate LTP. Fig. 7 shows that both BDNF and cortisol have bell-shaped dose-effect curves regarding corresponding gene products: cortisol on the BDNF-(trkB) receptor, BDNF on the CREB protein. The effect on CREB gene transcription is maximal if BDNF and cortisol are ‘balanced’, i.e. within an optimal concentration range. Cortisol concentrations are optimal during the nadir of the circadian cortisol profile, allowing for maximum numbers of trkB receptors to be formed. The number of trkB receptors is the limiting factor for the effect of BDNF on the CREB gene. Thus, LTP and associated induction of immediate early genes as the basis of memory formation, occur only if both BDNF and cortisol are within their optimal concentration range. BDNF is released upon cortical excitation together with glutamate. For this reason BDNF can be considered as a general indicator of metabolic or psychological stress. Complementarily, cortisol is an indicator for the magnitude of an organism’s stress response. In healthy individuals, BDNF and cortisol responses with stress coincide, while pathological states may selectively impair the cortisol response. Because of the tight association between BDNF and cortisol in their effect on LTP, it might be speculated in this context that the magnitude of a ‘stressor’ might be balanced by the ‘stress reaction’. This means that storage of information in memory would be favored especially if a state of balanced cortical activity and balanced stress reaction would prevail after stressor exposure. 2.3.3. Memory formation during sleep Slow wave sleep, which is dominant during the early part of nocturnal sleep, has been proposed to enhance the formation of declarative types of memory [224,225]. These are memories for facts and episodes (including the survival strategies taken by experimental rats to find food in a maze)

the storage of which relies essentially on the functions of the hippocampal formation. It is assumed that the memory enhancing effect of sleep is generated by a ‘replay’ of the memory representation newly acquired in a learning phase during prior wakefulness [226]. Thus, a certain spatiotemporal pattern of excitation within limbic neuron networks associated with a particular behavior solution to a problem (i.e. a stressor) during wakefulness becomes reactivated during slow wave sleep. This eventually results in strengthened synaptic efficacy and enduring memory for this behavior, induced by the process of LTP and the expression of immediate early genes discussed above. A success of this replay for enhancing memory can be predicted from the balanced cortisol concentration achieved during this period. If cortisol is too high during the early period of sleep replay activation would fail to enhance present memory [227], and the respective behavior would not be associated with a return to homeostasis. If specific behavioral programs are activated during the stress of the wake phase day and the organism returns to homeostasis (MR – GR balance) during the evening, these programs are consolidated during slow wave sleep. This strategy appears beneficial towards the organism’s survival. It would be just as beneficial if programs whose activation is not associated with a return to homeostasis were to be deleted. Both excessively high and low cortisol during the evening hours have unfavorable effects on slow wave sleep and potentially disrupt the consolidation of memories [227 – 229]. Low but balanced cortisol concentrations, such as those established during slow wave sleep periods of the early night provide conditions optimal for the consolidation of memories. Here we present the view that BDNF and cortisol furthermore regulate the consolidation of metabolic setpoints. If the LHPA system returns to its setpoint at the beginning of the night with balanced cortisol concentrations, MR and GR are in equilibrium. In this constellation, MR– MR homodimers are present in sufficient numbers to allow an adequate number of trkB receptors to be produced. There are hints that BDNF concentrations in limbic regions such as hippocampus, are maintained during sleep and thus remain in an optimal range [230]. Under these conditions BDNF binds to a sufficient number of trkB receptors and allows for optimal expression of LTP and CREB, required for the presumed stabilization of glucose setpoints in limbichypothalamic networks. In this way a stable and effective communication between the brain cortex and the body periphery is present which ensures the conveyance of the glutamate command signal.

3. Pathological glucose regulation The limbic system plays a decisive role in the communication between the cerebral cortex and the body periphery. Limbic neurons modulate and mediate the cortical glutamate

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

command signal to the muscle and adipose tissue. They do not just control the activity of the LHPA system, but in fact they also have the capacity to store information on a longterm basis with the help of the molecular mechanisms involved in long-term potentiation (LTP). With these properties, limbic neurons are in principle able to ‘memorize’ experiences and using these to persistently change the setpoint of the LHPA system. One such LHPA system setpoint alteration entails considerable long-term consequences for glucose allocation to the brain and the body periphery. The MR–GR balance can be seen as an indicator for the setpoint of the LHPA system. This MR-GR balance is genetically programmed, but can be altered by experiences in the postnatal period [231–233] as well as in later life [65]. A unique traumatic experience can lead for example to persistent modifications in the LHPA system. Israel Liberzon and coworkers of the Mental Health Research institute in Ann Arbor were able to show that an amplified negative cortisol feedback effect, as observed with post-traumatic stress disorder, can be induced by application of a single prolonged stress paradigm [234]. The research group employed different stressors in animal experiments during a single extended session, and then allowed a stress-free period to pass before the MR–GR balance was examined post mortem one week later. A week after stressor exposure, a reduction in the MRGR ratio was revealed in the hippocampus; these changes also persisted for 14 days after the stressor exposure. It is certainly remarkable that a single stress-intensive episode is able to lead to such long lasting, possibly even permanent modifications in the MR–GR balance. Fig. 5d shows the consequences that result from an alteration in the MR – GR balance: the pathologic balance setpoint, at which both the energy supply of the brain and the LHPA system are balanced, results in a lower glucose allocation and a higher food intake. In such a constellation the brain increases the intake of foods and in so doing safeguards its energy supply. Thus, a change in the MR – GR balance leads in the long-term to a change in body mass. What diseases arise when the setpoint of the LHPA system is altered and glucose allocation to the brain changes as a consequence? Considering this question we shall now discuss the origin of hypoglycemia unawareness, anorexia nervosa, obesity and type-2 diabetes mellitus. 3.1. Hypoglycemia unawareness (type 1 diabetes mellitus) Case 4: Walter, a 50 year-old plumber, has suffered from type 1 diabetes for 25 years. He attends a company function where the best-working employees are being honored. Before he leaves home he injects his usual dose of insulin. The dinner is served about one hour too late. Walter notices that he has difficulties going to the podium and conducting his thanksgiving speech. He goes back to his table, wobbles on his feet, and falls to the floor

169

unconscious. He is brought immediately to the hospital by ambulance. Upon arrival his plasma-glucose value is 25 mg/dl. He had already suffered a hypoglycemic coma a year previously. Asymptomatic hypoglycemia represents a major problem in the modern treatment of diabetes mellitus. Approximately a third of people with type-1 diabetes mellitus suffer severe hypoglycemia with comas, without noting any earlier warning symptoms such as shaking, sweating or palpitations. The inability to sense such warning symptoms or the absence of such symptoms before the development of a severe neuroglucopenia is known as hypoglycemia unawareness [178]. So-called ‘glucose counterregulation’ is the physiological mechanism that normally prevents hypoglycemia very effectively and corrects it wherever necessary [235]. This counter-regulation is based mostly on the activation of the LHPA system. This activation includes the stimulation of the sympatho-adrenal system with the associated increase in serum catecholamines and cortisol. In addition, insulin concentrations fall in line with the counterregulation [236]. Individuals with type-1 diabetes mellitus, who must inject insulin, can not lower their serum insulin concentrations at short notice through suppression of b cells. Catecholamines must compensate for this lacking insulin reaction. There is broad agreement that severe hypoglycemia arises from an interaction between absolute and relative insulin excess on the one hand and an impaired counter-regulation on the other. Patients at risk to severe hypoglycemia have more likely type-1 than type-2 diabetes mellitus, a longer diabetes duration, a low concentration of glycosylated hemoglobin, an autonomous neuropathy, or have already suffered episodes of hypoglycemic coma [237,238]. Recurrent hypoglycemia plays a key role in the origin of disrupted hypoglycemia awareness. During a hypoglycemic episode the counter-regulatory hormones increase strongly, while during recurrences these reactions are quite clearly reduced [239]. With mild, short hypoglycemia, such an attenuation is barely evident [240]. With moderate hypoglycemia, the attenuation lasts for up to 5 days [241], whereas amongst patients with recurring hypoglycemic coma the counter-regulation is decreased on a long-term basis [242]. Two research groups have studied whether the infusion of a high cortisol dose results in a reduced increase in the counter-regulatory hormones with hypoglycemia on the following day [243,244]. Only Stephen Davis’s group could confirm this presumed effect [243]. They supported their first results by verifying in animal experiments that intra-cerebroventricular infusion of cortisol also decreases the hormonal hypoglycemia response on the following day [245]. These findings support the notion that episodic cortisol excess plays a key role in the pathogenesis of hypoglycemia unawareness. The role of the limbic MR– GR balance in the origin of disrupted hypoglycemia awareness has not yet been

170

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

examined. An initial study on the MR–GR balance suggested that the ratio of MR and GR in the hippocampus is reduced in acute hypoglycemia [246]. Both a BDNF and a glucocorticoid excess occur with severe hypoglycemia [247,248]. Our model of brain ATP regulation predicts that the coincidence of BDNF and glucocorticoid excess in the limbic system induces an LTD. Furthermore, it also predicts that the number of AMPA receptors on the limbic neurons that control the LHPA system is also reduced. A hypocortisolism then results on the day after the hypoglycemic episode which can be demonstrated by the reduced morning cortisol concentrations [239]. Because of the autoregulation of the brain corticosteroid receptors, the ratio of MR to GR decreases and the circadian cortisol profile stabilizes to a lower concentration. Although the essential steps of this sequence of events have been demonstrated by various individual experiments elsewhere (see chapter 2), it still remains to be checked whether a single episode of a hypoglycemic coma can lead to a lasting setpoint reduction of the LHPA system in humans. With our current state of knowledge it can be hypothesized that hypoglycemic comas, similar to the ‘single prolonged stress paradigm’ employed by Liberzon, can lead to persistent alterations in the limbic MR–GR balance. If one assumes a setpoint reduction of the LHPA system after one or several hypoglycemic comas, glucose allocation to the brain can be considered to be inadequate with type 1 diabetes and as a result the sympatho-adrenally mediated alarm symptoms are less marked. In such individuals, the brain with its falling ATP content is dependent on an immediate intake of foods. Unlike the sympatho-adrenally mediated alarm symptoms, hunger is not attenuated with recurrent hypoglycemia [83]. With rapidly progressing neuroglucopenia there is often insufficient time to allow the correction of an energy deficiency in the brain through food intake. With mild neuroglucopenia, presynaptic low-affinity ATP-sensitive potassium channels on inhibitory cortical neurons are opened, which causes an increase in food intake and glucose allocation to the brain. With type-1 diabetes with disrupted detection only an inadequate time period is available for food intake. Because of this a stronger neuroglucopenia can develop. With critically low ATPconcentrations in the brain the high-affinity ATP-sensitive potassium channels are also opened (model a1 –a2). The result of this is that neurons over the entire neocortex become hyperpolarized and deactivate [55]. This resting state of the neocortex can be understood as a neuroprotective mechanism, whereby neurons employ their last remaining energy merely for structural maintenance. Susumu Seino and Nabuya Inagaki at the Akita and Chiba Universities in Japan showed recently that KATP channels localized in the substantia nigra can also support neuroprotection at critical brain ATP concentrations [35]. The research groups showed that KATP channels activate a nigral protection mechanism against generalized seizures at critical brain ATP-concentrations. In conclusion, two

neuroprotective mechanisms appear to be mobilized at critical brain ATP concentrations leading to a hypoglycemic coma, i.e. one involving cortical, and the other involving nigral, KATP channels. From this viewpoint it would appear that the syndrome of hypoglycemia unawareness is based on a ‘learned’ setpoint reduction of the LHPA system and an inadequate glucose allocation to the brain. As a result of episodic, metabolic insults, the brain then competes inadequately with the body periphery for energy resources and does not care for itself sufficiently. 3.2. Anorexia nervosa A long-term, regular loading with mid-level stressors can lead to a completely different reaction pattern of the organism compared to a single short-term overwhelming stressor load. While in the previous chapter we showed that hypoglycemic comas can lead to a hyporesponsiveness of the LHPA system, we shall show in the current chapter that a chronic stress can instead lead to a hyperresponsiveness of the LHPA system. Case 5: The 15 year-old Anna lives in permanent stress. Both parents are alcohol-dependent and get very violent when they are drunk. Anna and her younger brother have already been hit very frequently. Despite this burdensome situation she is very ambitious and has set goals for her life. She has good grades at school, and every afternoon she goes to the sport’s club where she trains for 2 h. During a summer camp Anna intensifies her training and diets. Her weight falls from 54 kg at 1.65 m (BMI of 16.5 kg/m2) to 45 kg. When she comes back to school her friends are shocked by her appearance, but Anna feels lively and active, sleeps well, does more sport and continues her fasting. Chronic exposure to various psychological stressors, as described with Anna, leads to a loading of the LHPA system [249,250]. Even a relatively short-lasting stimulation of the LHPA system has effects on glucose allocation: the allocation of glucose to the brain is increased while that to the body periphery is decreased. The activated allocationcontrolling VMH is hierarchically superior to the appetite adjusting LH and inhibits it (forward signal pathway). In addition, both VMH and LH are subject to inhibition by brain ATP (feedback signal pathway) independently of one another. Since the VMH is subject to a lasting stimulatory influence of the limbic system, food intake is curbed so far that the brain ATP concentration is balanced (high- and lowaffinity ATP-sensitive potassium channels in balance). As a result, a smaller amount of glucose is available for glucose allocation, and glucose uptake is reduced in the muscular and adipose tissue. Thus, body mass decreases over the long term. Overall, there is a decrease in body mass brought about by a centrally defined ratio between brain-glucose uptake and peripheral glucose uptake.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

The model of brain ATP regulation predicts that under chronic stressor exposure there is not just a loading of the LHPA system, but there is in fact also a limbic amplification of the glutamate command signal. A chronic stimulation of the LHPA system typically leads to hypercortisolism. In animal experiments it has been shown that corticosterone delivery to the amygdala increases CRH mRNA in the central amygdaloid nucleus, anxiety-like behaviour, and hypothalamo-pituitary-adrenal responses to behavioral stress [251,252]. In hypercortisolism, the number of GR decreases due to autoregulation of the brain corticosteroid receptors. Conversely, the number of MR increases. If the hypercortisolism is only mild, sleep patterns remain unchanged for the most part. During slow wave sleep there is an increased influence of MR leading to LTP. The number of AMPA receptors on the limbic neurons that control the LHPA system increases. This prediction is consistent with the observation that an early postnatal stress exposure due to maternal deprivation leads to a persisting hyper-responsiveness of the LHPA system [233]. If the LHPA system is stimulated chronically, this can lead to an increase in the system setpoint. If one assumes a setpoint increase of the LHPA system with chronic psychological stress, glucose allocation to the brain was increased while that to the body periphery was decreased in Anna’s case. As such the strength of the glutamate command signal determines the low body mass. If the glutamate command signal of the brain is very large or is amplified by the limbic system, a loss in body mass critical to survival results. With chronic fasting, ketone body formation occurs in the liver. These ketone bodies can be metabolized by the brain in place of glucose. If ketone body formation in progressive anorexia nervosa is activated, a ketone-induced inhibition of the sympatho-adrenal system in the hypothalamus occurs during this late stage of disease [153]. As a result, glucose allocation to the brain decreases. Up until now no effect of ketone bodies on hunger has been confirmed [253,254]. Ketones allow that at least a proportion of the available glucose is assigned to the musculature and adipose tissue. In the progression of anorexia nervosa, ketones therefore assume an important function for supplying energy to the brain and in so doing enable the body mass to be maintained and stabilized. From this new way of looking at limbic, long-term changes, anorexia nervosa is based on a ‘learned’ setpoint increase of the LHPA system with inadequate glucose allocation to the muscle and adipose tissue. As a result of this chronic stress, the brain competes excessively with the body periphery for energy resources. 3.3. Obesity A short-term load with a traumatizing stressor can cause a setpoint reduction of the LHPA system [255,256]. Such a setpoint change entails considerable metabolic consequences.

171

We shall now explain how secondary metabolic consequences can burden the LHPA system. Case 6: The now 23 year-old Clara grew up in prosperous and well-protected circumstances. After her schoolleaving examinations she traveled with her boyfriend who was an enthusiastic surfer to the Atlantic coast. After a wind accident her boyfriend became a quadriplegic and died after a 3 month treatment in hospital. Clara never left his side during this time. After the death of her boyfriend she refused to speak. She starts her law studies, but withdraws socially, avoiding friendships, sporting activities and travel. After a year she becomes increasingly and continuously depressed, and is very sensitive and feels easily offended. She sleeps regularly for 12 h, eats a lot and has already gained 20 kg in weight (current weight 74 kg at 1.65 m, BMI of 27.2 kg/m2). Her fasting plasma glucose (102 mg/dl) and cholesterol (180 mg/dl) are normal. After the traumatic loss of her partner, Clara developed an atypical depression. Patients with atypical depression are lethargic, tired, hyperphagic, hypersomnolent, and avoid social contact. Unlike the endocrine modifications with typical depression, patients with atypical depression show a less active LHPA system, characterized by CRH deficiency and hypocortisolism [255]. Dissociative behavior is also associated with low cortisol [257]. In its clinical presentation and neuroendocrine status atypical depression shows many common features with ‘posttraumatic stress disorder’ [258]. The neuroendocrine modifications resemble those caused by Liberzon in the ‘single prolonged stress’ paradigm (see above) [234]. The findings with atypical depression are consistent with a decreased MR to GR ratio and can therefore be considered to represent a setpoint reduction of the LHPA system. If one assumes a setpoint reduction of the LHPA system after a single traumatic experience, it can be understood that glucose allocation to the brain is inadequate amongst such traumatized individuals. Unlike the situation with progressing hypoglycemia whereby only a few minutes are available for taking food, patients with a setpoint alteration after a traumatic experience have sufficient time to ensure supply to the brain by increasing their food intake (see Fig. 5d). The combination of increased glucose allocation to muscle and adipose tissue results over the long term in a marked increase in peripheral body mass. In the Canadian Twin Study it could indeed be shown that low plasma cortisol is a predictor for a substantial increase in body weight [259]. A setpoint elevation of the LHPA-system, in contrast, results in opposite effects on body mass. Experiments in mice with inactivated GR in the nervous system (i.e. increased MR/GR-ratio and elevated setpoint) showed increased activity of the LHPA-system and decreased peripheral fat accumulation [260]. A setpoint-reduction of the LHPA system may result from a wide spectrum of conditions and disorders that

172

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

disrupt or alter the ‘glutamate command’ signal and the MR – GR balance. These conditions may include extreme stress situations (neuroglucopenia, traumatization, etc.), starvation, exercise, or infectious diseases. Also ‘endocrine disrupting chemicals’ [261], i.e. environmental agents that alter the endocrine system, may displace the limbic MR – GR balance and in so doing influence the mass of the body. Disturbances of the LHPA may also be located at the hypothalamic level. As predicted by the model, VMH and PVN lesions induce obesity [87,262]. Obesity also arises from an attenuation of the stimulatory input from the periphery (e.g. fat-derived leptin or muscle-derived interleukin-6) into these hypothalamic nuclei [6,263 – 266]. Similarly, the stimulatory input into these hypothalamic regions is reduced in melanocortin (MC4) receptor deficiency. In our model, a-MSH and NPY act as leptin-amplifying and leptin-attenuating signals, respectively. Selective disruption of the a-MSH pathway at the MC4-receptor level would produce a predominance of NPY neurons and their attenuating action on the leptin signal. Hypothalamic stimulatory input would be attenuated and, as expected, obesity does indeed develop in MC4 deficient mice [267]. Such disruptions may be localized in the neocortex, the limbic system, or the hypothalamus. Higher brain structures and functions have therefore found a new role in the pathogenesis of obesity. With increasing body weight ðMÞ; the brain ATP regulatory system is burdened secondarily (Fig. 5b and e). A larger peripheral adipose tissue mass is then able to absorb more glucose, representing a larger competitor to the brain for energy resources. The organism can respond to this metabolic load by increasing either food intake or allocation. The feedback effect of leptin mediates an activation of glucose allocation to the brain (see the feedback characteristic in Fig. 5b and e). The organism therefore reacts to the body weight increase by indirectly increasing the burden on the LHPA system. In conclusion, the development of obesity occurs in two stages: (1) an initial setpoint reduction and (2) a compensatory feedback-response of the LHPA system. In other words, the brain in the first phase competes too little for energy resources with the body periphery. In the second phase it sees a growing, more forceful competitor in the periphery, so that ultimately it has to compete even more with the body periphery for energy. 3.4. Type 2 diabetes mellitus A setpoint change of the LHPA system can lead to the development of obesity. The way in which the organism reacts to this metabolic load depends greatly on its genetic background [268]. Case 7: Pedro, a 40 year-old Mexican factory worker, had suffered a severe accident at work 10 years previously. He was in a coma for five days, was operated on repeatedly for

various fractures, recovered only sluggishly, and resumed work only a year later. Before the accident Pedro was healthily and athletically built. In the time after his hospitalization, however, he puts on 15 kg in weight, mainly around the abdomen. For the last four weeks Pedro is feeling weak, has resorted to drinking large amounts of water, and loses 2 kg. After he finally seeks hospital treatment for angina pectoris the doctors diagnose diabetes mellitus (blood glucose 350 mg/dl) as well as arterial hypertension (200/100 mmHg). Diabetes mellitus occurred frequently in Pedro’s family. Pedro suffered a physical trauma. Similar to the psychological trauma suffered by Clara, a serious physical condition can lead to a setpoint alteration in the LHPA system. Common to both situations is the coincidence of an insult and cortisol excess, which can lead to a decreased tone of the LHPA system. Similar to Clara, Pedro in the subsequent period developed obesity. However, the further course of disease differed fundamentally. The genetic background of people with a MexicanAmerican origin differs from those of a European origin [19]. The prevalence of type 2 diabetes mellitus and the metabolic syndrome (hyperglycemia, hyperlipoproteinemia, hypertension and abdominal obesity) is increased amongst individuals of a Mexican-American origin [269 –271]. In the chapter ‘economics of the alternative brain-specific substrates’ we discussed that some groups with genetic peculiarities (Nauruans, Pima Indians) can survive particularly long periods of starvation. We presume that this adaptation is made possible by a leptin amplifier mechanism. The melanocortin system fulfills the criteria for one such leptin amplifier mechanism. If obesity develops amongst people with specific genetic background allowing leptin amplification, it can be expected that such individuals react to obesity in a special way. People with a leptin amplifier mechanism react to the increase in serum leptin by increasing glucose allocation to the brain (see Fig. 5f). In an impressive experiment, Akira Mizuno and coworkers of the Takushima University in Japan were able to demonstrate the leptin-mediated feedback-effect on glucose allocation. The researchers treated vagotomized animals with leptin and examined their insulin production under glucose loading. Leptin suppressed insulin secretion almost completely. The team then carried out the same experiment with animals that were also sympathectomized; here, a renormalized insulin secretion was shown. The experiment showed that leptin, via the hypothalamus and its sympathetic efferents, can suppress insulin secretion and with that glucose allocation to muscle and adipose tissue. That means in reverse that leptin concentrations increasing with body mass can promote glucose allocation to the brain. It was recently shown that the melanocortin system can amplify this leptin feedback effect even further; centrally applied melanocortin agonists inhibit insulin secretion [272]. Individuals with a marked

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

leptin amplification mechanism therefore react to increasing leptin with a particularly strong activation of glucose allocation to the brain. How does the body mass change with increased glucose allocation to the brain? An increased glucose allocation to the brain coincides with an increased sympatho-adrenally mediated lactate formation in the muscle [163 – 165]. Additional glucose and lactate for the brain allows the body to reduce food intake while still guaranteeing the energy supply of the brain. A decrease in appetite and food intake results in less energy reaching the muscle and adipose tissues. An inhibited glucose uptake into the peripheral tissues results in an increased blood glucose concentration. If the blood glucose concentrations exceed the renal threshold, glucosuria occurs. In this way significant amounts of glucose can be eliminated (at 250 mg/dl blood glucose there is an approx. 10 g/l glucosuria). Such a glucose loss represents an additional burden to the regulatory system. As such an increased allocation to the brain coincides with both appetite inhibition and glucosuria, and both can lead to a negative energy balance. A population-based study from Arizona has indeed shown that amongst patients not taking insulin after a type 2 diabetes mellitus is diagnosed, a longlasting decrease in weight is observed [273]. In conclusion, a type 2 diabetes mellitus can arise in two phases: (1) an initial setpoint-reduction and (2) an increased feedback-mediated regulatory response of the LHPA system. In the first phase the brain competes too little with the body periphery so that obesity arises. With certain genetic predispositions the brain reacts in a second phase particularly strongly to the increase in body weight and restricts the further growth of its peripheral competitor. Interestingly, there is a certain analogy between the development of type 2 diabetes mellitus and anorexia nervosa. With anorexia nervosa, ketone bodies alleviate the regulatory system as additional energy carriers. The ketone bodies prevent the body mass from falling below a certain critical mass. During the development of diabetes mellitus the organism loses energy through glucosuria (elimination of energy-carriers). The ‘flowing through’ (in Greek ‘diabeinein’) of energy prevents the body exceeding a certain critical mass. From this viewpoint a type 2 diabetes mellitus actually serves to safeguard the primacy of the brain in competing for energy resources.

4. Conclusions There are four basic tenets of the theory: First, the brain prioritizes adjustment of its own ATP concentration. High and low affinity ATP-sensitive potassium channels regulate the brain’s self-allocation of glucose. Second, brain ATP regulation is a learning control process. The self-allocation process is under short-term and long-term feedback control by cortisol and it’s high and low affinity receptors. Third,

173

the brain uses limbic plasticity to stabilize the selfallocation process in the long run. After a stressful experience, such a newly achieved consolidation and stabilization of the self-allocation process may allow the brain to reappraise its energy demand strategies. Fourth, alterations in glucose allocation can be compensated by changes in food intake or by utilization of alternate substrates such as lactate or ketones. Alternative strategies of the brain to safeguard its energy needs can result in diseases such as obesity or type 2 diabetes mellitus. These four basic tenets summarize the essentials of the new theory: refutal of any of these tenets should lead to a major questioning of the validity of the theory. From these basic tenets we built various ramifications and in each case proposed molecular mechanisms to explain them. The theory itself is not bound to one particular mechanism or another to explain these secondary aspects. For example, we have proposed BDNF and its receptors as one mechanism underlying plasticity, but we cannot rule out that another factor was more involved. If this were true, however, it would not actually jeopardize the theory. The presented theory also includes a newly discovered general principle of how two receptors can originate a setpoint in biological systems: High and low affinity receptors bind to a ligand and after so doing they generate a signal that results in an increase or decrease of ligand concentrations, respectively. This ‘principle of balance’ describes the molecular mechanisms of setpoint signal generation which is essential for defining the ideal behavior of a biological feedback control system. Unlike traditional models, the paradigm presented by us fundamentally changes the hierarchy of the regulated parameters and places the brain’s energy needs at the highest level. The size of the adipose tissue or muscle mass therefore becomes a secondary regulatory goal. According to these concepts, obesity and type 2 diabetes represent brain diseases with defects in neuroendocrine functions. Attempts to prevent or treat obesity can be successful only when these basic aspects are taken into account.

References [1] Attwell D, Laughlin SB. An energy budget for signaling in the grey matter of the brain. J Cereb Blood Flow Metab 2001;21(10): 1133– 45. [2] Pardridge WM. Receptor-mediated peptide transport through the blood–brain barrier. Endocr Rev 1986;7(3):314 –30. [3] Gjedde A. Blood-barrier glucose transfer. In: Bradbury WB, editor. Physiology and pharmacology of the blood– brain barrier. Berlin: Springer; 1992. p. 65–115. [4] DuPont RL. The selfish brain: learning from addiction. Center City, Minnesota: Hazelden; 1997. [5] Kennedy GC. The role of depot fat in the hypothalamic control of food intake in the rat. Proc R Soc Lond Ser 1953;140:578–92. [6] Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse obese gene and its human homologue. Nature 1994;372(6505):425–32.

174

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180

[7] Woods SC, Stein LJ, McKay Jr. LD, Porte D. Suppression of food intake by intravenous nutrients and insulin in the baboon. Am J Physiol 1984;247(2 Pt 2):R393–R401. [8] Schwartz MW, Bergman RN, Kahn SE, Taborsky Jr GJ, Fisher LD, Sipols AJ, Woods SC, Steil GM, Porte Jr. D. Evidence for entry of plasma insulin into cerebrospinal fluid through an intermediate compartment in dogs. Quantitative aspects and implications for transport. J Clin Invest 1991;88(4):1272–81. [9] Igel M, Becker W, Herberg L, Joost HG. Hyperleptinemia, leptin resistance, and polymorphic leptin receptor in the New Zealand obese mouse. Endocrinology 1997;138(10):4234–9. [10] Clement K, Vaisse C, Lahlou N, Cabrol S, Pelloux V, Cassuto D, Gourmelen M, Dina C, Chambaz J, Lacorte JM, Basdevant A, Bougneres P, Lebouc Y, Froguel P, Guy-Grand B. A mutation in the human leptin receptor gene causes obesity and pituitary dysfunction. Nature 1998;392(6674):398– 401. [11] El Haschimi K, Pierroz DD, Hileman SM, Bjorbaek C, Flier JS. Two defects contribute to hypothalamic leptin resistance in mice with diet-induced obesity. J Clin Invest 2000;105(12):1827–32. [12] Schwartz MW, Woods SC, Porte Jr. D, Seeley RJ, Baskin DG. Central nervous system control of food intake. Nature 2000; 404(6778):661–71. [13] Bouchard C, Tremblay A, Despres JP, Nadeau A, Lupien PJ, Moorjani S, Theriault G, Kim SY. Overfeeding in identical twins: 5year postoverfeeding results. Metabolism 1996;45(8):1042–50. [14] Brownell KD, Rodin J. Medical, metabolic, and psychological effects of weight cycling. Arch Intern Med 1994;154(12):1325–30. [15] UKPDS Study Group, Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998;352(9131):837–53. [16] Kopelman PG, Hitman GA. Diabetes. Exploding type II. Lancet 1998;352(4):SIV5. [17] Friedman JM. A war on obesity, not the obese. Science 2003; 299(5608):856–8. [18] Kopelman PG. Obesity as a medical problem. Nature 2000; 404(6778):635–43. [19] Diamond J. The double puzzle of diabetes. Nature 2003;423(6940): 599–602. [20] Seeley RJ, Woods SC. Monitoring of stored and available fuel by the CNS: implications for obesity. Nat Rev Neurosci 2003;4(11):901– 9. [21] Dallman MF, Pecoraro N, Akana SF, la Fleur SE, Gomez F, Houshyar H, Bell ME, Bhatnagar S, Laugero KD, Manalo S. Chronic stress and obesity: a new view of comfort food. Proc Natl Acad Sci USA 2003;100(20):11696–701. [22] Berthoud HR. Multiple neural systems controlling food intake and body weight. Neurosci Biobehav Rev 2002;26(4):393–428. [23] Peters A, Schweiger U, Fruhwald-Schultes B, Born J, Fehm HL. The neuroendocrine control of glucose allocation. Exp Clin Endocrinol Diabetes 2002;110(5):199 –211. [24] DiStefano JJ, Stubberud AR, Williams IJ. Theory and problems of feedback and control systems. New York: McGraw-Hill; 1997. [25] Spanswick D, Smith MA, Groppi VE, Logan SD, Ashford ML. Leptin inhibits hypothalamic neurons by activation of ATP-sensitive potassium channels. Nature 1997;390(6659):521– 5. [26] Spanswick D, Smith MA, Mirshamsi S, Routh VH, Ashford ML. Insulin activates ATP-sensitive K þ channels in hypothalamic neurons of lean, but not obese rats. Nat Neurosci 2000;3(8):757–8. [27] Einstein A. In: Calaprice A, editor. The expanded quotable Einstein. New Jersey: Princeton University Press; 2000. [28] Aguilar-Bryan L, Bryan J. Molecular biology of adenosine triphosphate-sensitive potassium channels. Endocr Rev 1999;20(2): 101–35. [29] Ashcroft FM. Adenosine 50 -triphosphate-sensitive potassium channels. Annu Rev Neurosci 1988;11:97–118. [30] Lazdunski M. ATP-sensitive potassium channels: an overview. J Cardiovasc Pharmacol 1994;24(4):S1–S5.

[31] Aguilar-Bryan L, Nichols CG, Wechsler SW, Clement JP, Boyd III AE, Gonzalez G, Herrera-Sosa H, Nguy K, Bryan J, Nelson DA. Cloning of the beta cell high-affinity sulfonylurea receptor: a regulator of insulin secretion. Science 1995;268(5209):423–6. [32] Inagaki N, Gonoi T, Clement JP, Namba N, Inazawa J, Gonzalez G, Aguilar-Bryan L, Seino S, Bryan J. Reconstitution of IKATP: an inward rectifier subunit plus the sulfonylurea receptor. Science 1995; 270(5239):1166–70. [33] Garcia dA, Franke H, Pissarek M, Nieber K, Illes P. Neuroprotection by ATP-dependent potassium channels in rat neocortical brain slices during hypoxia. Neurosci Lett 1999;273(1):13–16. [34] Heurteaux C, Bertaina V, Widmann C, Lazdunski M. K þ channel openers prevent global ischemia-induced expression of c-fos, c-jun, heat shock protein, and amyloid beta-protein precursor genes and neuronal death in rat hippocampus. Proc Natl Acad Sci USA 1993; 90(20):9431–5. [35] Yamada K, Ji JJ, Yuan H, Miki T, Sato S, Horimoto N, Shimizu T, Seino S, Inagaki N. Protective role of ATP-sensitive potassium channels in hypoxia-induced generalized seizure. Science 2001; 292(5521):1543–6. [36] Babenko AP, Gonzalez G, Bryan J. Two regions of sulfonylurea receptor specify the spontaneous bursting and ATP inhibition of KATP channel isoforms. J Biol Chem 1999;274(17):11587–92. [37] Matsuo M, Tanabe K, Kioka N, Amachi T, Ueda K. Different binding properties and affinities for ATP and ADP among sulfonylurea receptor subtypes, SUR1, SUR2A, and SUR2B. J Biol Chem 2000;275(37):28757–63. [38] Song DK, Ashcroft FM. ATP modulation of ATP-sensitive potassium channel ATP sensitivity varies with the type of SUR subunit. J Biol Chem 2001;276(10):7143–9. [39] Inagaki N, Gonoi T, Clement JP, Wang CZ, Aguilar-Bryan L, Bryan J, Seino S. A family of sulfonylurea receptors determines the pharmacological properties of ATP-sensitive K þ channels. Neuron 1996;16(5):1011 –7. [40] Karschin C, Ecke C, Ashcroft FM, Karschin A. Overlapping distribution of K(ATP) channel-forming Kir6.2 subunit and the sulfonylurea receptor SUR1 in rodent brain. FEBS Lett 1997;401(1): 59– 64. [41] Lee K, Dixon AK, Rowe IC, Ashford ML, Richardson PJ. The high-affinity sulphonylurea receptor regulates KATP channels in nerve terminals of the rat motor cortex. J Neurochem 1996;66(6): 2562–71. [42] Hernandez-Sanchez C, Basile AS, Fedorova I, Arima H, Stannard B, Fernandez AM, Ito Y, LeRoith D. Mice transgenically overexpressing sulfonylurea receptor 1 in forebrain resist seizure induction and excitotoxic neuron death. Proc Natl Acad Sci USA 2001;98(6):3549 –54. [43] Ohno-Shosaku T, Yamamoto C. Identification of an ATP-sensitive K þ channel in rat cultured cortical neurons. Pflugers Arch 1992; 422(3):260–6. [44] Lee K, Dixon AK, Rowe IC, Ashford ML, Richardson PJ. Direct demonstration of sulphonylurea-sensitive KATP channels on nerve terminals of the rat motor cortex. Br J Pharmacol 1995;115(3): 385 –7. [45] Dunn-Meynell AA, Rawson NE, Levin BE. Distribution and phenotype of neurons containing the ATP-sensitive K þ channel in rat brain. Brain Res 1998;814(1-2):41–54. [46] Holemans S, Javoy-Agid F, Agid Y, De Paermentier F, Laterre EC, Maloteaux JM. Sulfonylurea binding sites in normal human brain and in Parkinson’s disease, progressive supranuclear palsy and Huntington’s disease. Brain Res 1994;642(1–2):327–33. [47] Freiman TM, Kukolja J, Heinemeyer J, Eckhardt K, Aranda H, Rominger A, Dooley DJ, Zentner J, Feuerstein TJ. Modulation of K þ -evoked [3H]-noradrenaline release from rat and human brain slices by gabapentin: involvement of KATP channels. Naunyn Schmiedebergs Arch Pharmacol 2001;363(5):537 –42.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 [48] Luhmann HJ, Heinemann U. Hypoxia-induced functional alterations in adult rat neocortex. J Neurophysiol 1992;67(4):798–811. [49] Ohno-Shosaku T, Sawada S, Yamamoto C. ATP-sensitive K þ channel activators suppress the GABAergic inhibitory transmission by acting on both presynaptic and postsynaptic sites in rat cultured hippocampal neurons. Neurosci Lett 1993;159(1–2):139–42. [50] Amoroso S, Schmid-Antomarchi H, Fosset M, Lazdunski M. Glucose, sulfonylureas, and neurotransmitter release: role of ATPsensitive K þ channels. Science 1990;247(4944):852–4. [51] Ye GL, Leung CK, Yung WH. Pre-synaptic effect of the ATPsensitive potassium channel opener diazoxide on rat substantia nigra pars reticulata neurons. Brain Res 1997;753(1):1 –7. [52] Watts AE, Hicks GA, Henderson G. Putative pre- and postsynaptic ATP-sensitive potassium channels in the rat substantia nigra in vitro. J Neurosci 1995;15(4):3065–74. [53] During MJ, Leone P, Davis KE, Kerr D, Sherwin RS. Glucose modulates rat substantia nigra GABA release in vivo via ATPsensitive potassium channels. J Clin Invest 1995;95(5):2403 –8. [54] Jiang C, Haddad GG. Modulation of K þ channels by intracellular ATP in human neocortical neurons. J Neurophysiol 1997;77(1): 93–102. [55] Mobbs CV, Kow LM, Yang XJ. Brain glucose-sensing mechanisms: ubiquitous silencing by aglycemia vs. hypothalamic neuroendocrine responses. Am J Physiol Endocrinol Metab 2001;281(4):E649–54. [56] Magistretti PJ, Pellerin L, Rothman DL, Shulman RG. Energy on demand. Science 1999;283(5401):496–7. [57] Chatton JY, Pellerin L, Magistretti PJ. GABA uptake into astrocytes is not associated with significant metabolic cost: Implications for brain imaging of inhibitory transmission. Proc Natl Acad Sci USA 2003;12456–61. [58] Bergles DE, Jahr CE. Synaptic activation of glutamate transporters in hippocampal astrocytes. Neuron 1997;19(6):1297–308. [59] Pellerin L, Magistretti PJ. Glutamate uptake stimulates Na þ ,K þ ATPase activity in astrocytes via activation of a distinct subunit highly sensitive to ouabain. J Neurochem 1997;69(5):2132–7. [60] Pellerin L, Magistretti PJ. Glutamate uptake into astrocytes stimulates aerobic glycolysis: a mechanism coupling neuronal activity to glucose utilization. Proc Natl Acad Sci USA 1994; 91(22):10625–9. [61] Takahashi S, Driscoll BF, Law MJ, Sokoloff L. Role of sodium and potassium ions in regulation of glucose metabolism in cultured astroglia. Proc Natl Acad Sci USA 1995;92(10):4616– 20. [62] Bittar PG, Charnay Y, Pellerin L, Bouras C, Magistretti PJ. Selective distribution of lactate dehydrogenase isoenzymes in neurons and astrocytes of human brain. J Cereb Blood Flow Metab 1996;16(6): 1079–89. [63] Schurr A, Miller JJ, Payne RS, Rigor BM. An increase in lactate output by brain tissue serves to meet the energy needs of glutamateactivated neurons. J Neurosci 1999;19(1):34–9. [64] Ainscow EK, Mirshamsi S, Tang T, Ashford ML, Rutter GA. Dynamic imaging of free cytosolic ATP concentration during fuel sensing by rat hypothalamic neurones: evidence for ATP-independent control of ATP-sensitive K(þ ) channels. J Physiol 2002;544(Pt 2):429– 45. [65] Lopez JF, Chalmers DT, Little KY, Watson SJ, A.E. Bennett Research Award. Regulation of serotonin1A, glucocorticoid, and mineralocorticoid receptor in rat and human hippocampus: implications for the neurobiology of depression. Biol Psychiatry 1998; 43(8):547–73. [66] Jansen AS, Hoffman JL, Loewy AD. CNS sites involved in sympathetic and parasympathetic control of the pancreas: a viral tracing study. Brain Res 1997;766(1 –2):29–38. [67] Dunning BE, Ahren B, Veith RC, Taborsky Jr. GJ. Nonadrenergic sympathetic neural influences on basal pancreatic hormone secretion. Am J Physiol 1988;255(6 Pt 1):E785–92. [68] Chapman JC, McClenaghan NH, Cosgrove KE, Hashmi MN, Shepherd RM, Giesberts AN, White SJ, Ammala C, Flatt PR,

[69]

[70]

[71] [72]

[73]

[74]

[75]

[76] [77]

[78]

[79]

[80]

[81]

[82]

[83]

[84]

[85]

[86] [87]

[88]

175

Dunne MJ. ATP-sensitive potassium channels and efaroxan-induced insulin release in the electrofusion-derived BRIN-BD11 beta-cell line. Diabetes 1999;48(12):2349–57. Chan SL, Dunne MJ, Stillings MR, Morgan NG. The alpha 2adrenoceptor antagonist efaroxan modulates K þ ATP channels in insulin-secreting cells. Eur J Pharmacol 1991;204(1):41–8. Baron AD, Wallace P, Olefsky JM. In vivo regulation of non-insulinmediated and insulin-mediated glucose uptake by epinephrine. J Clin Endocrinol Metab 1987;64(5):889–95. Deibert DC, DeFronzo RA. Epinephrine-induced insulin resistance in man. J Clin Invest 1980;65(3):717–21. Lembo G, Capaldo B, Rendina V, Iaccarino G, Napoli R, Guida R, Trimarco B, Sacca L. Acute noradrenergic activation induces insulin resistance in human skeletal muscle. Am J Physiol 1994;266(2 Pt 1): E242–7. Molina PE, Abumrad NN. Contribution of excitatory amino acids to hypoglycemic counter-regulation. Brain Res 2001;899(1– 2): 201– 8. Molina PE, Tepper PG, Yousef KA, Abumrad NN, Lang CH. Central NMDA enhances hepatic glucose output and non-insulin-mediated glucose uptake by a nonadrenergic mechanism. Brain Res 1994; 634(1):41–8. Yousef KA, Tepper PG, Molina PE, Abumrad NN, Lang CH. Differential control of glucoregulatory hormone response and glucose metabolism by NMDA and kainate. Brain Res 1994; 634(1):131 –40. Anand BK, Brobeck JR. Hypothalamic control of food intake in rats and cats. Yale J Biol Med 1951;24:123–46. Watts AG. Understanding the neural control of ingestive behaviors: helping to separate cause from effect with dehydration-associated anorexia. Horm Behav 2000;37(4):261–83. Stanley BG, Willett VL, Donias HW, Ha LH, Spears LC. The lateral hypothalamus: a primary site mediating excitatory amino acidelicited eating. Brain Res 1993;630(1–2):41–9. Stanley BG, Ha LH, Spears LC, Dee MG. Lateral hypothalamic injections of glutamate, kainic acid, D ,L -alpha- amino-3-hydroxy5-methyl-isoxazole propionic acid or N-methyl-D -aspartic acid rapidly elicit intense transient eating in rats. Brain Res 1993; 613(1):88–95. Petrovich GD, Setlow B, Holland PC, Gallagher M. Amygdalohypothalamic circuit allows learned cues to override satiety and promote eating. J Neurosci 2002;22(19):8748–53. Petrovich GD, Gallagher M. Amygdala subsystems and control of feeding behavior by learned cues. Ann N Y Acad Sci 2003;985: 251– 62. Borg MA, Borg WP, Tamborlane WV, Brines ML, Shulman GI, Sherwin RS. Chronic hypoglycemia and diabetes impair counterregulation induced by localized 2-deoxy-glucose perfusion of the ventromedial hypothalamus in rats. Diabetes 1999;48(3): 584– 7. Schultes B, Oltmanns KM, Kern W, Fehm HL, Born J, Peters A. Modulation of hunger by plasma glucose and metformin. J Clin Endocrinol Metab 2003;88(3):1133–41. Oomura Y, Ono T, Ooyama H. Inhibitory action of the amygdala on the lateral hypothalamic area in rats. Nature 1970;228(276): 1108– 10. Oomura Y, Kimura K, Ooyama H, Maeo T, Iki M, Kuniyoshi N. Reciprocal activities of the ventromedial and lateral hypothalamic area of cats. Science 1964;143:484 –5. Bray GA. Reciprocal relation between the sympathetic nervous system and food intake. Brain Res Bull 1991;27(3–4):517–20. Sakaguchi T, Arase K, Bray GA. Sympathetic activity and food intake of rats with ventromedial hypothalamic lesions. Int J Obes 1988;12(4):285–91. Cai XJ, Widdowson PS, Harrold J, Wilson S, Buckingham RE, Arch JR, Tadayyon M, Clapham JC, Wilding J, Williams G. Hypothalamic

176

[89]

[90] [91]

[92]

[93]

[94]

[95]

[96]

[97]

[98]

[99] [100]

[101]

[102]

[103]

[104]

[105]

[106]

[107]

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 orexin expression: modulation by blood glucose and feeding. Diabetes 1999;48(11):2132–7. McEwen BS, Weiss JM, Schwartz LS. Selective retention of corticosterone by limbic structures in rat brain. Nature 1968; 220(170):911–2. Pfaff DW, Silva MT, Weiss JM. Telemetered recording of hormone effects on hippocampal neurons. Science 1971;172(981):394–5. Patel PD, Lopez JF, Lyons DM, Burke S, Wallace M, Schatzberg AF. Glucocorticoid and mineralocorticoid receptor mRNA expression in squirrel monkey brain. J Psychiatr Res 2000;34(6):383–92. Roozendaal B, Griffith QK, Buranday J, De Quervain DJ, McGaugh JL. The hippocampus mediates glucocorticoid-induced impairment of spatial memory retrieval: dependence on the basolateral amygdala. Proc Natl Acad Sci USA 2003;100(3):1328– 33. Roozendaal B, Phillips RG, Power AE, Brooke SM, Sapolsky RM, McGaugh JL. Memory retrieval impairment induced by hippocampal CA3 lesions is blocked by adrenocortical suppression. Nat Neurosci 2001;4(12):1169–71. Liu W, Wang J, Sauter NK, Pearce D. Steroid receptor heterodimerization demonstrated in vitro and in vivo. Proc Natl Acad Sci USA 1995;92(26):12480–4. Ou XM, Storring JM, Kushwaha N, Albert PR. Heterodimerization of mineralocorticoid and glucocorticoid receptors at a novel negative response element of the 5-HT1A receptor gene. J Biol Chem 2001; 276(17):14299–307. Trapp T, Rupprecht R, Castren M, Reul JM, Holsboer F. Heterodimerization between mineralocorticoid and glucocorticoid receptor: a new principle of glucocorticoid action in the CNS. Neuron 1994;13(6):1457 –62. Fehm HL, Voigt KH, Lang RE, Beinert KE, Kummer GW, Pfeiffer EF, Paradoxical ACTH. response to glucocorticoids in Cushing’s disease. N Engl J Med 1977;297(17):904–7. de Kloet ER, Vreugdenhil E, Oitzl MS, Joels M. Brain corticosteroid receptor balance in health and disease. Endocr Rev 1998;19(3): 269–301. Joels M. Corticosteroid actions in the hippocampus. J Neuroendocrinol 2001;13(8):657 –69. van Haarst AD, Oitzl MS, de Kloet ER. Facilitation of feedback inhibition through blockade of glucocorticoid receptors in the hippocampus. Neurochem Res 1997;22(11):1323–8. Campion J, Maestro B, Mata F, Davila N, Carranza MC, Calle C. Inhibition by aldosterone of insulin receptor mRNA levels and insulin binding in U-937 human promonocytic cells. J Steroid Biochem Mol Biol 1999;70(4-6):211–8. Campion J, Maestro B, Molero S, Davila N, Carranza MC, Calle C. Aldosterone impairs insulin responsiveness in U-937 human promonocytic cells via the downregulation of its own receptor. Cell Biochem Funct 2002;20(3):237–45. Joels M, de Kloet ER. Corticosteroid actions on amino acidmediated transmission in rat CA1 hippocampal cells. J Neurosci 1993;13(9):4082– 90. Wellhoener P, Fruehwald-Schultes B, Kern W, Dantz D, Kerner W, Born J, Fehm HL, Peters A. Glucose metabolism rather than insulin is a main determinant of leptin secretion in humans. J Clin Endocrinol Metab 2000;85(3):1267–71. Mueller WM, Gregoire FM, Stanhope KL, Mobbs CV, Mizuno TM, Warden CH, Stern JS, Havel PJ. Evidence that glucose metabolism regulates leptin secretion from cultured rat adipocytes. Endocrinology 1998;139(2):551–8. Vannucci SJ, Gibbs EM, Simpson IA. Glucose utilization and glucose transporter proteins GLUT-1 and GLUT-3 in brains of diabetic ðdb=dbÞ mice. Am J Physiol 1997;272(2 Pt 1):E267–74. Satoh N, Ogawa Y, Katsuura G, Numata Y, Tsuji T, Hayase M, Ebihara K, Masuzaki H, Hosoda K, Yoshimasa Y, Nakao K. Sympathetic activation of leptin via the ventromedial hypothalamus: leptin-induced increase in catecholamine secretion. Diabetes 1999; 48(9):1787 –93.

[108] Tang-Christensen M, Havel PJ, Jacobs RR, Larsen PJ, Cameron JL. Central administration of leptin inhibits food intake and activates the sympathetic nervous system in rhesus macaques. J Clin Endocrinol Metab 1999;84(2):711–7. [109] Haynes WG, Morgan DA, Walsh SA, Mark AL, Sivitz WI. Receptor-mediated regional sympathetic nerve activation by leptin. J Clin Invest 1997;100(2):270 –8. [110] Opstad PK. Androgenic hormones during prolonged physical stress, sleep, and energy deficiency. J Clin Endocrinol Metab 1992;74(5): 1176–83. [111] Cahill Jr. GF. Starvation in man. N Engl J Med 1970;282(12): 668 –75. [112] Maher F, Simpson IA. Modulation of expression of glucose transporters GLUT3 and GLUT1 by potassium and N-methyl-D aspartate in cultured cerebellar granule neurons. Mol Cell Neurosci 1994;5(4):369 –75. [113] Swanson LW. Cerebral hemisphere regulation of motivated behavior. Brain Res 2000;886(1–2):113– 64. [114] Herman JP, Figueiredo H, Mueller NK, Ulrich-Lai Y, Ostrander MM, Choi DC, Cullinan WE. Central mechanisms of stress integration: hierarchical circuitry controlling hypothalamo-pituitary-adrenocortical responsiveness. Front Neuroendocrinol 2003; 24(3):151–80. [115] Feldman S, Conforti N, Weidenfeld J. Limbic pathways and hypothalamic neurotransmitters mediating adrenocortical responses to neural stimuli. Neurosci Biobehav Rev 1995;19(2):235 –40. [116] Jacobson L, Sapolsky R. The role of the hippocampus in feedback regulation of the hypothalamic-pituitary – adrenocortical axis. Endocr Rev 1991;12(2):118–34. [117] Iguchi A, Uemura K, Miura H, Ishiguro T, Nonogaki K, Tamagawa T, Goshima K, Sakamoto N. Mechanism of intrahippocampal neostigmine-induced hyperglycemia in fed rats. Neuroendocrinology 1992;55(1):44– 50. [118] Uemura K, Iguchi A, Yatomi A, Miura H, Honmura A, Yanase M, Sakamoto N. Involvement of the hippocampus in central nervous system-mediated glucoregulation in rats. Endocrinology 1989; 124(5):2449–55. [119] Iguchi A, Yatomi A, Gotoh M, Matsunaga H, Uemura K, Miura H, Satake T, Tamagawa T, Sakamoto N. Neostigmine-induced hyperglycemia is mediated by central muscarinic receptor in fed rats. Brain Res 1990;507(2):295–300. [120] Iguchi A, Kunoh Y, Gotoh M, Miura H, Uemura K, Tamagawa T, Sakamoto N. Relative contribution of nervous system and hormones to CNS-mediated hyperglycemia is determined by the neurochemical specificity in the brain. Physiol Behav 1991;50(5):1019–25. [121] Zhu W, Umegaki H, Yoshimura J, Tamaya N, Suzuki Y, Miura H, Iguchi A. The elevation of plasma adrenocorticotrophic hormone and expression of c-Fos in hypothalamic paraventricular nucleus by microinjection of neostigmine into the hippocampus in rats: comparison with acute stress responses. Brain Res 2001;892(2): 391 –5. [122] Iguchi A, Uemura K, Miura H, Ishiguro T, Nonogaki K, Tamagawa T, Goshima K, Sakamoto N. Mechanism of intrahippocampal neostigmine-induced hyperglycemia in fed rats. Neuroendocrinology 1992;55(1):44– 50. [123] Feldman S, Saphier D, Conforti N. Hypothalamic afferent connections mediating adrenocortical responses that follow hippocampal stimulation. Exp Neurol 1987;98(1):103–9. [124] Watanabe G, Ishiguro T, Miura H, Uemura K, Hiyoshi Y, Ozawa K, Nonogaki K, Tamagawa T, Sakamoto N, Iguchi A. CNS regulation of blood lactate concentration in anesthetized rats. Life Sci 1994; 54(20):1491–9. [125] Miki T, Liss B, Minami K, Shiuchi T, Saraya A, Kashima Y, Horiuchi M, Ashcroft F, Minokoshi Y, Roeper J, Seino S. ATPsensitive K þ channels in the hypothalamus are essential for the maintenance of glucose homeostasis. Nat Neurosci 2001;4(5): 507 –12.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 [126] Levin BE, Dunn-Meynell AA, Routh VH. Brain glucosensing and the K(ATP) channel. Nat Neurosci 2001;4(5):459– 60. [127] Yang X, Kow LM, Funabashi T, Mobbs CV. Hypothalamic glucose sensor. Diabetes 1999;48:1763–72. [128] Lee K, Dixon AK, Richardson PJ, Pinnock RD. Glucose-receptive neurones in the rat ventromedial hypothalamus express KATP channels composed of Kir6.1 and SUR1 subunits. J Physiol 1999; 515(Pt 2):439– 52. [129] Beverly JL, De Vries MG, Bouman SD, Arseneau LM. Noradrenergic and GABAergic systems in the medial hypothalamus are activated during hypoglycemia. Am J Physiol Regul Integr Comp Physiol 2001;280(2):R563–9. [130] De Vries MG, Lawson MA, Beverly JL. Dissociation of hypothalamic noradrenergic activity and sympathoadrenal responses to recurrent hypoglycemia. Am J Physiol Regul Integr Comp Physiol 2004;. [131] Fehm HL, Voigt KH, Lang RE, Pfeiffer EF. Effects of neurotransmitters on the release of corticotropin releasing hormone (CRH) by rat hypothalamic tissue in vitro. Exp Brain Res 1980;39(2): 229–34. [132] Cole RL, Sawchenko PE. Neurotransmitter regulation of cellular activation and neuropeptide gene expression in the paraventricular nucleus of the hypothalamus. J Neurosci 2002;22(3):959–69. [133] Beverly JL, De Vries MG, Beverly MF, Arseneau LM. Norepinephrine mediates glucoprivic-induced increase in GABA in the ventromedial hypothalamus of rats. Am J Physiol Regul Integr Comp Physiol 2000;279(3):R990–6. [134] Xu B, Goulding EH, Zang K, Cepoi D, Cone RD, Jones KR, Tecott LH, Reichardt LF. Brain-derived neurotrophic factor regulates energy balance downstream of melanocortin-4 receptor. Nat Neurosci 2003;6(7):736–42. [135] Gerendai I, Halasz B. Central nervous system structures connected with the endocrine glands. Findings obtained with the viral transneuronal tracing technique. Exp Clin Endocrinol Diabetes 2000;108(6):389–95. [136] Chutkow WA, Samuel V, Hansen PA, Pu J, Valdivia CR, Makielski JC, Burant CF. Disruption of Sur2-containing K(ATP) channels enhances insulin- stimulated glucose uptake in skeletal muscle. Proc Natl Acad Sci USA 2001;98(20):11760– 4. [137] Wasada T, Yano T, Ohta M, Yui N, Iwamoto Y. ATP-Sensitive potassium channels modulate glucose transport in cultured human skeletal muscle cells. Endocr J 2001;48(3):369–75. [138] Capaldo B, Napoli R, Guida R, Di Bonito P, Antoniello S, Auletta M, Pardo F, Rendina V, Sacca L. Forearm muscle insulin resistance during hypoglycemia: role of adrenergic mechanisms and hypoglycemia per se. Am J Physiol 1995;268(2 Pt 1):E248–54. [139] Shamoon H, Friedman S, Canton C, Zacharowicz L, Hu M, Rossetti L. Increased epinephrine and skeletal muscle responses to hypoglycemia in non-insulin-dependent diabetes mellitus. J Clin Invest 1994;93(6):2562–71. [140] Ahima RS, Prabakaran D, Mantzoros C, Qu D, Lowell B, Maratos Flier E, Flier JS. Role of leptin in the neuroendocrine response to fasting. Nature 1996;382(6588):250–2. [141] Elias CF, Kelly JF, Lee CE, Ahima RS, Drucker DJ, Saper CB, Elmquist JK. Chemical characterization of leptin-activated neurons in the rat brain. J Comp Neurol 2000;423(2):261– 81. [142] Mercer JG, Hoggard N, Williams LM, Lawrence CB, Hannah LT, Trayhurn P. Localization of leptin receptor mRNA and the long form splice variant (Ob-Rb) in mouse hypothalamus and adjacent brain regions by in situ hybridization. FEBS Lett 1996;387(2 –3):113–6. [143] Song Z, Levin BE, McArdle JJ, Bakhos N, Routh VH. Convergence of pre- and postsynaptic influences on glucosensing neurons in the ventromedial hypothalamic nucleus. Diabetes 2001;50(12): 2673–81. [144] Havel Jr PJ, Taborsky GJ. The contribution of the autonomic nervous system to changes of glucagon and insulin secretion during hypoglycemic stress. Endocr Rev 1989;10(3):332– 50.

177

[145] Curry DL, Safarik RH, Joy RM, Stern JS. Direct neural effect of lateral hypothalamic stimulation on insulin secretion by pancreases of normal and obese rats. Horm Metab Res 1990;22(3):129 –35. [146] Zawar C, Plant TD, Schirra C, Konnerth A, Neumcke B. Cell-type specific expression of ATP-sensitive potassium channels in the rat hippocampus. J Physiol 1999;514(Pt 2):327–41. [147] Treherne JM, Ashford ML. The regional distribution of sulphonylurea binding sites in rat brain. Neuroscience 1991;40(2):523–31. [148] Ghelardini C, Galeotti N, Bartolini A. Influence of potassium channel modulators on cognitive processes in mice. Br J Pharmacol 1998;123(6):1079–84. [149] Rashidy-Pour A. ATP-sensitive potassium channels mediate the effects of a peripheral injection of glucose on memory storage in an inhibitory avoidance task. Behav Brain Res 2001;126(1–2):43–8. [150] Shanley LJ, Irving AJ, Harvey J. Leptin enhances NMDA receptor function and modulates hippocampal synaptic plasticity. J Neurosci 2001;21(24):RC186. [151] Pellerin L. Lactate as a pivotal element in neuron-glia metabolic cooperation. Neurochem Int 2003;43(4–5):331– 8. [152] Abi-Saab WM, Maggs DG, Jones T, Jacob R, Srihari V, Thompson J, Kerr D, Leone P, Krystal JH, Spencer DD, During MJ, Sherwin RS. Striking differences in glucose and lactate levels between brain extracellular fluid and plasma in conscious human subjects: effects of hyperglycemia and hypoglycemia. J Cereb Blood Flow Metab 2002; 22(3):271–9. [153] Veneman T, Mitrakou A, Mokan M, Cryer PE, Gerich J. Effect of hyperketonemia and hyperlacticacidemia on symptoms, cognitive dysfunction, and counterregulatory hormone responses during hypoglycemia in normal humans. Diabetes 1994;43(11):1311–7. [154] King P, Parkin H, Macdonald IA, Barber C, Tattersall RB. The effect of intravenous lactate on cerebral function during hypoglycaemia. Diabet Med 1997;14(1):19–28. [155] Maran A, Cranston I, Lomas J, Macdonald I, Amiel SA. Protection by lactate of cerebral function during hypoglycaemia. Lancet 1994; 343(8888):16–20. [156] Maran A, Crepaldi C, Trupiani S, Lucca T, Jori E, Macdonald IA, Tiengo A, Avogaro A, Del Prato S. Brain function rescue effect of lactate following hypoglycaemia is not an adaptation process in both normal and type I diabetic subjects. Diabetologia 2000;43(6): 733– 41. [157] Sakurai T, Yang B, Takata T, Yokono K. Synaptic adaptation to repeated hypoglycemia depends on the utilization of monocarboxylates in Guinea pig hippocampal slices. Diabetes 2002;51(2):430 –8. [158] Lu XY, Barsh GS, Akil H, Watson SJ. Interaction between alphamelanocyte-stimulating hormone and corticotropin-releasing hormone in the regulation of feeding and hypothalamo-pituitary– adrenal responses. J Neurosci 2003;23(21):7863–72. [159] Ibrahim N, Bosch MA, Smart JL, Qiu J, Rubinstein M, Ronnekleiv OK, Low MJ, Kelly MJ. Hypothalamic proopiomelanocortin neurons are glucose responsive and express K(ATP) channels. Endocrinology 2003;144(4):1331–40. [160] Baskin DG, Breininger JF, Schwartz MW. Leptin receptor mRNA identifies a subpopulation of neuropeptide Y neurons activated by fasting in rat hypothalamus. Diabetes 1999;48(4):828–33. [161] Dunbar JC, Lu H. Leptin-induced increase in sympathetic nervous and cardiovascular tone is mediated by proopiomelanocortin (POMC) products. Brain Res Bull 1999;50(3):215–21. [162] Satoh N, Ogawa Y, Katsuura G, Numata Y, Masuzaki H, Yoshimasa Y, Nakao K. Satiety effect and sympathetic activation of leptin are mediated by hypothalamic melanocortin system. Neurosci Lett. 1998;249(2–3):107–10. [163] Waldhausl WK, Gasic S, Bratusch-Marrain P, Komjati M, Korn A. Effect of stress hormones on splanchnic substrate and insulin disposal after glucose ingestion in healthy humans. Diabetes 1987; 36(2):127–35. [164] Goldstein RE, Abumrad NN, Lacy DB, Wasserman DH, Cherrington AD. Effects of an acute increase in epinephrine and cortisol on

178

[165]

[166]

[167]

[168] [169]

[170]

[171]

[172] [173]

[174]

[175]

[176]

[177]

[178] [179]

[180]

[181]

[182]

[183]

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 carbohydrate metabolism during insulin deficiency. Diabetes 1995; 44(6):672– 81. Connolly CC, Ivy RE, Adkins-Marshall BA, Dobbins RL, Neal DW, Williams PE, Cherrington AD. Counterregulation by epinephrine and glucagon during insulin-induced hypoglycemia in the conscious dog. Diabetes Res Clin Pract 1996;31(1–3):45–56. Peters A, Preuschoft H. External biomechanics of leaping in Tarsius and its morphological and kinematic consequences. In: Niemitz C, editor. Biology of Tarsius. Stuttgart: G. Fischer; 1984. p. 227 –57. Montague CT, Prins JB, Sanders L, Digby JE, O’Rahilly S. Depotand sex-specific differences in human leptin mRNA expression: implications for the control of regional fat distribution. Diabetes 1997;46(3):342–7. Neel JV. Diabetes mellitus: a thrifty genotype rendered detrimental by progress? Am J Hum Genet 1962;14:353–62. Oomura Y, Ono T, Ooyama H, Wayner MJ. Glucose and osmosensitive neurones of the rat hypothalamus. Nature 1969; 222(190):282–4. Oomura Y, Ooyama H, Sugimori M, Nakamura T, Yamada Y. Glucose inhibition of the glucose-sensitive neurone in the rat lateral hypothalamus. Nature 1974;247(439):284–6. Levin BE, Dunn-Meynell A, Routh VH. Brain glucose sensing and body energy homoestasis: role in obesity and diabetes. Am J Physiol 1999;276:R1223– 31. Routh VH. Glucose-sensing neurons: are they physiologically relevant? Physiol Behav 2002;76(3):403– 13. Moriguchi T, Sakurai T, Nambu T, Yanagisawa M, Goto K. Neurons containing orexin in the lateral hypothalamic area of the adult rat brain are activated by insulin-induced acute hypoglycemia. Neurosci Lett 1999;264(1–3):101–4. de Lecea L, Kilduff TS, Peyron C, Gao X, Foye PE, Danielson PE, Fukuhara C, Battenberg EL, Gautvik VT, Bartlett FS, Frankel WN, van den Pol AN, Bloom FE, Gautvik KM, Sutcliffe JG. The hypocretins: hypothalamus-specific peptides with neuroexcitatory activity. Proc Natl Acad Sci USA 1998;95(1):322–7. Cai XJ, Evans ML, Lister CA, Leslie RA, Arch JR, Wilson S, Williams G. Hypoglycemia activates orexin neurons and selectively increases hypothalamic orexin-B levels: responses inhibited by feeding and possibly mediated by the nucleus of the solitary tract. Diabetes 2001;50(1):105 –12. Dryden S, Pickavance L, Henderson L, Williams G. Hyperphagia induced by hypoglycemia in rats is independent of leptin and hypothalamic neuropeptide Y (NPY). Peptides 1998;19(9):1549– 55. Deary IJ, Hepburn DA, MacLeod KM, Frier BM. Partitioning the symptoms of hypoglycaemia using multi-sample confirmatory factor analysis. Diabetologia 1993;36:771–7. Gerich JE, Mokan M, Veneman T, Korytkowski M, Mitrakou A. Hypoglycemia unawareness. Endocr Rev 1991;12:356 –71. Richter SD, Schurmeyer TH, Schedlowski M, Hadicke A, Tewes U, Schmidt RE, Wagner TO. Time kinetics of the endocrine response to acute psychological stress. J Clin Endocrinol Metab 1996;81(5): 1956– 60. Schwartz MW, Figlewicz DP, Baskin DG, Woods S, Porte DJ. Insulin in the brain: A hormonal regulator of energy balance. Endocr Rev 1992;13:387–414. Feldman S, Weidenfeld J. The dorsal hippocampus modifies the negative feedback effect of glucocorticoids on the adrenocortical and median eminence CRF-41 responses to photic stimulation. Brain Res 1993;614(1–2):227–32. Bradbury MJ, Strack AM, Dallman MF. Lesions of the hippocampal efferent pathway (fimbria-fornix) do not alter sensitivity of adrenocorticotropin to feedback inhibition by corticosterone in rats. Neuroendocrinology 1993;58(4):396–407. Tuvnes FA, Steffenach HA, Murison R, Moser MB, Moser EI. Selective hippocampal lesions do not increase adrenocortical activity. J Neurosci 2003;23(10):4345–54.

[184] LeDoux JE. Emotion circuits in the brain. Annu Rev Neurosci 2000; 23:155–84. [185] Makino S, Gold PW, Schulkin J. Corticosterone effects on corticotropin-releasing hormone mRNA in the central nucleus of the amygdala and the parvocellular region of the paraventricular nucleus of the hypothalamus. Brain Res 1994;640(1–2):105–12. [186] Erkut ZA, Pool C, Swaab DF. Glucocorticoids suppress corticotropinreleasing hormone and vasopressin expression in human hypothalamic neurons. J Clin Endocrinol Metab 1998;83(6):2066 –73. [187] Beaulieu S, Gagne B, Barden N. Glucocorticoid regulation of proopiomelanocortin messenger ribonucleic acid content of rat hypothalamus. Mol Endocrinol 1988;2(8):727–31. [188] Kageyama K, Li C, Vale WW. Corticotropin-releasing factor receptor type 2 messenger ribonucleic acid in rat pituitary: localization and regulation by immune challenge, restraint stress, and glucocorticoids. Endocrinology 2003;144(4):1524–32. [189] Birnberg NC, Lissitzky JC, Hinman M, Herbert E. Glucocorticoids regulate proopiomelanocortin gene expression in vivo at the levels of transcription and secretion. Proc Natl Acad Sci USA 1983;80(22): 6982–6. [190] Leavitt MG, Aberdeen GW, Burch MG, Albrecht ED, Pepe GJ. Inhibition of fetal adrenal adrenocorticotropin receptor messenger ribonucleic acid expression by betamethasone administration to the baboon fetus in late gestation. Endocrinology 1997;138(7): 2705–12. [191] Hernandez-Sanchez C, Ito Y, Ferrer J, Reitman M, LeRoith D. Characterization of the mouse sulfonylurea receptor 1 promoter and its regulation. J Biol Chem 1999;274(26):18261–70. [192] Cullen MJ, Ling N, Foster AC, Pelleymounter MA. Urocortin, corticotropin releasing factor-2 receptors and energy balance. Endocrinology 2001;142(3):992–9. [193] Laugero KD, Bell ME, Bhatnagar S, Soriano L, Dallman MF. Sucrose ingestion normalizes central expression of corticotropinreleasing-factor messenger ribonucleic acid and energy balance in adrenalectomized rats: a glucocorticoid-metabolic-brain axis? Endocrinology 2001;142(7):2796–804. [194] Bell ME, Bhargava A, Soriano L, Laugero K, Akana SF, Dallman MF. Sucrose intake and corticosterone interact with cold to modulate ingestive behaviour, energy balance, autonomic outflow and neuroendocrine responses during chronic stress. J Neuroendocrinol 2002;14(4):330 –42. [195] Bliss TV, Collingridge GL. A synaptic model of memory: longterm potentiation in the hippocampus. Nature 1993;361(6407): 31 –9. [196] Malinow R, Malenka RC. AMPA receptor trafficking and synaptic plasticity. Annu Rev Neurosci 2002;25:103 –26. [197] Takahashi T, Svoboda K, Malinow R. Experience strengthening transmission by driving AMPA receptors into synapses. Science 2003;299(5612):1585–8. [198] Barde YA, Edgar D, Thoenen H. Purification of a new neurotrophic factor from mammalian brain. EMBO J 1982;1(5):549 –53. [199] Minichiello L, Calella AM, Medina DL, Bonhoeffer T, Klein R, Korte M. Mechanism of TrkB-mediated hippocampal long-term potentiation. Neuron 2002;36(1):121 –37. [200] Chao MV, Hempstead BL. p75 and Trk: a two-receptor system. Trends Neurosci 1995;18(7):321– 6. [201] Huang EJ, Reichardt LF. Neurotrophins: roles in neuronal development and function. Annu Rev Neurosci 2001;24:677 –736. [202] Dechant G. Molecular interactions between neurotrophin receptors. Cell Tissue Res 2001;305(2):229–38. [203] Bibel M, Hoppe E, Barde YA. Biochemical and functional interactions between the neurotrophin receptors trk and p75NTR. EMBO J 1999;18(3):616–22. [204] Ross GM, Shamovsky IL, Lawrance G, Solc M, Dostaler SM, Weaver DF, Riopelle RJ. Reciprocal modulation of TrkA and p75NTR affinity states is mediated by direct receptor interactions. Eur J Neurosci 1998;10(3):890–8.

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 [205] Kovalchuk Y, Hanse E, Kafitz KW, Konnerth A. Postsynaptic induction of BDNF-mediated long-term potentiation. Science 2002; 295(5560):1729–34. [206] Ikegaya Y, Ishizaka Y, Matsuki N. BDNF attenuates hippocampal LTD via activation of phospholipase C: implications for a vertical shift in the frequency-response curve of synaptic plasticity. Eur J Neurosci 2002;16(1):145–8. [207] Frey U, Huang YY, Kandel ER. Effects of cAMP simulate a late stage of LTP in hippocampal CA1 neurons. Science 1993;260(5114): 1661–4. [208] Bourtchuladze R, Frenguelli B, Blendy J, Cioffi D, Schutz G, Silva AJ. Deficient long-term memory in mice with a targeted mutation of the cAMP-responsive element-binding protein. Cell 1994;79(1): 59–68. [209] Silva AJ, Kogan JH, Frankland PW, Kida S. CREB and memory. Annu Rev Neurosci 1998;21:127–48. [210] Chao HM, Ma LY, McEwen BS, Sakai RR. Regulation of glucocorticoid receptor and mineralocorticoid receptor messenger ribonucleic acids by selective agonists in the rat hippocampus. Endocrinology 1998;139(4):1810–4. [211] Luttge WG, Rupp ME, Davda MM. Aldosterone-stimulated downregulation of both type I and type II adrenocorticosteroid receptors in mouse brain is mediated via type I receptors. Endocrinology 1989; 125(2):817– 24. [212] Kalman BA, Spencer RL. Rapid corticosteroid-dependent regulation of mineralocorticoid receptor protein expression in rat brain. Endocrinology 2002;143(11):4184–95. [213] Hansson AC, Fuxe K. Biphasic autoregulation of mineralocorticoid receptor mRNA in the medial septal nucleus by aldosterone. Neuroendocrinology 2002;75(6):358 –66. [214] Reul JM, van den Bosch FR, de Kloet ER. Differential response of type I and type II corticosteroid receptors to changes in plasma steroid level and circadian rhythmicity. Neuroendocrinology 1987; 45(5):407–12. [215] Reul JM, Pearce PT, Funder JW, Krozowski ZS. Type I and type II corticosteroid receptor gene expression in the rat: effect of adrenalectomy and dexamethasone administration. Mol Endocrinol 1989;3(10):1674–80. [216] Erdeljan P, MacDonald JF, Matthews SG. Glucocorticoids and serotonin alter glucocorticoid receptor (GR) but not mineralocorticoid receptor (MR) mRNA levels in fetal mouse hippocampal neurons, in vitro. Brain Res 2001;896(1 –2):130–6. [217] Peiffer A, Lapointe B, Barden N. Hormonal regulation of type II glucocorticoid receptor messenger ribonucleic acid in rat brain. Endocrinology 1991;129(4):2166–74. [218] Kalinyak JE, Dorin RI, Hoffman AR, Perlman AJ. Tissue-specific regulation of glucocorticoid receptor mRNA by dexamethasone. J Biol Chem 1987;262(22):10441–4. [219] Pavlides C, Watanabe Y, Magarinos AM, McEwen BS. Opposing roles of type I and type II adrenal steroid receptors in hippocampal long-term potentiation. Neuroscience 1995;68(2):387–94. [220] Pavlides C, Ogawa S, Kimura A, McEwen BS. Role of adrenal steroid mineralocorticoid and glucocorticoid receptors in long-term potentiation in the CA1 field of hippocampal slices. Brain Res 1996; 738(2):229– 35. [221] Pavlides C, McEwen BS. Effects of mineralocorticoid and glucocorticoid receptors on long-term potentiation in the CA3 hippocampal field. Brain Res. 1999;851(1–2):204–14. [222] Schaaf MJ, Hoetelmans RW, de Kloet ER, Vreugdenhil E. Corticosterone regulates expression of BDNF and trkB but not NT-3 and trkC mRNA in the rat hippocampus. J Neurosci Res 1997; 48(4):334–41. [223] Shi B, Mocchetti I. Dexamethasone induces TrkA and p75NTR immunoreactivity in the cerebral cortex and hippocampus. Exp Neurol 2000;162(2):257–67. [224] Born J, Gais S. Roles of early and late noctirnal sleep for the consolidation of human memories. In: Maquet P, Smith C, Stickgold

[225]

[226] [227]

[228]

[229]

[230]

[231]

[232]

[233]

[234]

[235]

[236]

[237]

[238]

[239]

[240]

[241]

[242] [243]

179

R, editors. Sleep and brain plasticity. Oxford: Oxford University Press; 2003. p. 65–86. McNaughton BL, Barnes CA. Off-line reprocessing of recent memory and its role in memory consolidation: a progress report. In: Maquet P, Smith C, Stickgold R, editors. Sleep and brain plasticity. Oxford: Oxford University Press; 2003. p. 225–46. Wilson MA, McNaughton BL. Reactivation of hippocampal ensemble memories during sleep. Science 1994;265(5172):676–9. Plihal W, Pietrowsky R, Born J. Dexamethasone blocks sleep induced improvement of declarative memory. Psychoneuroendocrinology 1999;24(3):313– 31. Born J, DeKloet ER, Wenz H, Kern W, Fehm HL. Glucoand antimineralocorticoid effects on human sleep: a role of central corticosteroid receptors. Am J Physiol 1991;260(2 Pt 1): E183–8. Plihal W, Born J. Memory consolidation in human sleep depends on inhibition of glucocorticoid release. Neuroreport 1999;10(13): 2741– 7. Taishi P, Sanchez C, Wang Y, Fang J, Harding JW, Krueger JM. Conditions that affect sleep alter the expression of molecules associated with synaptic plasticity. Am J Physiol Regul Integr Comp Physiol 2001;281(3):R839–45. Francis D, Diorio J, Liu D, Meaney MJ. Nongenomic transmission across generations of maternal behavior and stress responses in the rat. Science 1999;286(5442):1155–8. Liu D, Diorio J, Tannenbaum B, Caldji C, Francis D, Freedman A, Sharma S, Pearson D, Plotsky PM, Meaney MJ. Maternal care, hippocampal glucocorticoid receptors, and hypothalamic-pituitary– adrenal responses to stress. Science 1997;277(5332):1659–62. Ladd CO, Huot RL, Thrivikraman KV, Nemeroff CB, Meaney MJ, Plotsky PM. Long-term behavioral and neuroendocrine adaptations to adverse early experience. Prog Brain Res 2000;122:81–103. Liberzon I, Lopez JF, Flagel SB, Vazquez DM, Young EA. Differential regulation of hippocampal glucocorticoid receptors mRNA and fast feedback: relevance to post-traumatic stress disorder. J Neuroendocrinol 1999;11(1):11–17. Gerich JE, Davis J, Lorenzi M, Rizza RA, Bohannon N, Karan J, Lewis S, Kaplan R, Schultz T, Cryer PE. Hormonal mechanism of recovery from insilin-induced hypoglycemia in man. Am J Physiol 1979;236:E380–5. Fruehwald-Schultes B, Kern W, Born J, Fehm HL, Peters A. Comparison of the inhibitory effect of insulin and hypoglycemia on insulin secretion in humans. Metabolism 2000;49(7):950–3. Gold AE, Frier BM, MacLeod KM, Deary IJ. A structural equation model for predictors of severe hypoglycaemia in patients with insulin-dependent diabetes mellitus. Diabet Med 1997;14(4): 309– 15. The DCCT Research Group, Epidemiology of severe hypoglycemia in the diabetes control and complications trial. Am J Med 1991;90: 450– 9. Fruehwald-Schultes B, Kern W, Deininger E, Wellhoener P, Kerner W, Born J, Fehm HL, Peters A. Protective effect of insulin against hypoglycemia-associated counterregulatory failure. J Clin Endocrinol Metab 1999;84(5):1551 –7. Peters A, Rohloff F, Kerner W. Preserved counterregulatory hormone release and symptoms after short term hypoglycemic episodes in normal men. J Clin Endocrinol Metab 1995;80: 2894– 8. George E, Harris N, Bedford C, Macdonald IA, Hardisty CA, Heller SR. Prolonged but partial impairment of the hypoglycaemic physiological response following short-term hypoglycaemia in normal subjects. Diabetologia 1995;38:1183–90. Widom B, Simonson DC. Intermittent hypoglycemia impairs glucose counterregulation. Diabetes 1992;41:1597–602. Davis SN, Shavers C, Costa F, Mosqueda Garcia R. Role of cortisol in the pathogenesis of deficient counterregulation after

180

[244]

[245]

[246]

[247]

[248]

[249]

[250]

[251]

[252]

[253]

[254]

[255]

[256]

[257]

[258]

A. Peters et al. / Neuroscience and Biobehavioral Reviews 28 (2004) 143–180 antecedent hypoglycemia in normal humans. J Clin Invest 1996; 98(3):680– 91. Raju B, McGregor VP, Cryer PE. Cortisol elevations comparable to those that occur during hypoglycemia do not cause hypoglycemiaassociated autonomic failure. Diabetes 2003;52(8):2083 –9. Sandoval DA, Ping L, Neill AR, Morrey S, Davis SN. Cortisol acts through central mechanisms to blunt counterregulatory responses to hypoglycemia in conscious rats. Diabetes 2003;52:2198–204. Chan O, Chan S, Inouye K, Shum K, Matthews SG, Vranic M. Diabetes impairs hypothalamo-pituitary-adrenal (HPA) responses to hypoglycemia, and insulin treatment normalizes HPA but not epinephrine responses. Diabetes 2002;51(6):1681–9. Lindvall O, Ernfors P, Bengzon J, Kokaia Z, Smith ML, Siesjo BK, Persson H. Differential regulation of mRNAs for nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3 in the adult rat brain following cerebral ischemia and hypoglycemic coma. Proc Natl Acad Sci USA 1992;89(2):648–52. Tkacs NC, Dunn-Meynell AA, Levin BE. Presumed apoptosis and reduced arcuate nucleus neuropeptide Y and pro- opiomelanocortin mRNA in non-coma hypoglycemia. Diabetes 2000;49(5): 820–6. Duclos M, Corcuff JB, Roger P, Tabarin A. The dexamethasonesuppressed corticotrophin-releasing hormone stimulation test in anorexia nervosa. Clin Endocrinol (Oxf) 1999;51(6):725–31. Putignano P, Dubini A, Toja P, Invitti C, Bonfanti S, Redaelli G, Zappulli D, Cavagnini F. Salivary cortisol measurement in normalweight, obese and anorexic women: comparison with plasma cortisol. Eur J Endocrinol 2001;145(2):165–71. Shepard JD, Barron KW, Myers DA. Stereotaxic localization of corticosterone to the amygdala enhances hypothalamo-pituitary– adrenal responses to behavioral stress. Brain Res 2003;963(1–2): 203–13. Shepard JD, Barron KW, Myers DA. Corticosterone delivery to the amygdala increases corticotropin-releasing factor mRNA in the central amygdaloid nucleus and anxiety-like behavior. Brain Res 2000;861(2):288– 95. Davis JD, Wirtshafter D, Asin KE, Brief D. Sustained intracerebroventricular infusion of brain fuels reduces body weight and food intake in rats. Science 1981;212(4490):81– 3. Sun M, Martin RJ, Edwards GL. ICV beta-hydroxybutyrate: effects on food intake, body composition, and body weight in rats. Physiol Behav 1997;61(3):433 –6. Gold PW, Chrousos GP. Organization of the stress system and its dysregulation in melancholic and atypical depression: high vs low CRH/NE states. Mol Psychiatry 2002;7(3):254–75. Grossman R, Yehuda R, New A, Schmeidler J, Silverman J, Mitropoulou V, Sta MN, Golier J, Siever L. Dexamethasone suppression test findings in subjects with personality disorders: associations with posttraumatic stress disorder and major depression. Am J Psychiatry 2003;160(7):1291–8. Morgan III CA, Wang S, Rasmusson A, Hazlett G, Anderson G, Charney DS. Relationship among plasma cortisol, catecholamines, neuropeptide Y, and human performance during exposure to uncontrollable stress. Psychosom Med 2001;63(3):412 –22. Yehuda R. Post-traumatic stress disorder. N Engl J Med 2002; 346(2):108 –14.

[259] Pritchard J, Despres JP, Gagnon J, Tchernof A, Nadeau A, Tremblay A, Bouchard C. Plasma adrenal, gonadal, and conjugated steroids before and after long-term overfeeding in identical twins. J Clin Endocrinol Metab 1998;83(9):3277–84. [260] Kellendonk C, Eiden S, Kretz O, Schutz G, Schmidt I, Tronche F, Simon E. Inactivation of the GR in the nervous system affects energy accumulation. Endocrinology 2002;143(6):2333–40. [261] McLachlan JA. Environmental signaling: what embryos and evolution teach us about endocrine disrupting chemicals. Endocr Rev 2001;22(3):319–41. [262] Sakaguchi T, Bray GA, Eddlestone G. Sympathetic activity following paraventricular or ventromedial hypothalamic lesions in rats. Brain Res Bull 1988;20(4):461–5. [263] Pedersen BK, Steensberg A, Schjerling P. Muscle-derived interleukin-6: possible biological effects. J Physiol 2001;536(Pt 2): 329 –37. [264] Bethin KE, Vogt SK, Muglia LJ. Interleukin-6 is an essential, corticotropin-releasing hormone-independent stimulator of the adrenal axis during immune system activation. Proc Natl Acad Sci USA 2000;97(16):9317–22. [265] Farooqi IS, Jebb S, Langmack G, Lawrence E, Cheetham C, Prentice AM, Hughes IA, McCamish MA, O’Rahilly S. Effects of recombinant leptin therapy in a child with congenital leptin deficiency. N Engl J Med 1999;341:879 –84. [266] Wallenius V, Wallenius K, Ahren B, Rudling M, Carlsten H, Dickson SL, Ohlsson C, Jansson JO. Interleukin-6-deficient mice develop mature-onset obesity. Nat Med 2002;8(1):75–9. [267] Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q, Berkemeier LR, Gu W, Kesterson RA, Boston BA, Cone RD, Smith FJ, Campfield LA, Burn P, Lee F. Targeted disruption of the melanocortin4 receptor results in obesity in mice. Cell 1997;88(1):131 –41. [268] Bouchard C, Tremblay A, Despres JP, Nadeau A, Lupien PJ, Theriault G, Dussault J, Moorjani S, Pinault S, Fournier G. The response to long-term overfeeding in identical twins. N Engl J Med 1990;322(21):1477–82. [269] Ford ES, Giles WH, Dietz WH. Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. JAMA 2002;287(3): 356 –9. [270] Meigs JB, Wilson PW, Nathan DM, D’Agostino Sr. RB, Williams K, Haffner SM. Prevalence and characteristics of the metabolic syndrome in the San Antonio Heart and Framingham Offspring Studies. Diabetes 2003;52(8):2160– 7. [271] Aguilar-Salinas CA, Velazquez MO, Gomez-Perez FJ, Gonzalez CA, Esqueda AL, Molina CV, Rull-Rodrigo JA, Tapia CR. Characteristics of patients with type 2 diabetes in Mexico: Results from a large population-based nationwide survey. Diabetes Care 2003;26(7):2021 –6. [272] Fan W, Dinulescu DM, Butler AA, Zhou J, Marks DL, Cone RD. The central melanocortin system can directly regulate serum insulin levels. Endocrinology 2000;141(9):3072–9. [273] Looker HC, Knowler WC, Hanson RL. Changes in BMI and weight before and after the development of Type 2 diabetes. Diabetes Care 2001;24(11):1917–22.