- Email: [email protected]
The controls of eating: a shift from nutritional homeostasis to behavioral neuroscience
INGESTIVE BEHAVIOR AND OBESITY
The Controls of Eating: A Shift from Nutritional Homeostasis to Behavioral Neuroscience Gerard P. Smith, MD From the Bourne Laboratory, New York-Presbyterian Hospital, Westchester Division, White Plains, New York, USA The experimental analysis of the controls of eating has undergone a paradigmatic shift in the past decade. Instead of seeing meals as a problem of how intake serves metabolism and nutritional homeostasis, meals are now seen as a problem in behavioral neuroscience. The major developments underlying this significant change are behavioral, neurological, and theoretical. Behavioral analysis has shown that a central pattern generator in the caudal brainstem organizes eating movements and that the size of a liquid meal is determined by the number and size of clusters of licking. Neurologic analysis has shown that eating is under orosensory positive-feedback control and postingestive, preabsorptive, negative-feedback control. These feedback controls are activated by food ingested during a meal. The sensory information of the feedbacks is carried by afferent fibers that project to the caudal brainstem. The new theory is based on the fact that the feedback controls are stimulated by food acting directly on mucosal receptors along the gastrointestinal tract, from the mouth to the end of the small intestine. Thus, they are referred to as direct controls, and the caudal brainstem is sufficient for organizing their action. All other controls, such as metabolic, rhythmic, and ecologic, that do not contact the mucosal receptors are indirect controls. Indirect controls act by modulating the potency of the central effects of the direct controls, and they require the forebrain and its reciprocal connections with the caudal brainstem for their control of eating and meal size. Nutrition 2000;16:814 – 820. ©Elsevier Science Inc. 2000 Key words: nutritional homeostasis, caudal brainstem, orosensory positive-feedback control, postingestive, preabsorptive, negative-feedback control, mucosal receptors, direct and indirect controls
INTRODUCTION Food intake, energy expenditure, and energy storage determine nutritional homeostasis. After Richter’s classic experiments in rats,1 it became clear that intake in all mammals under a variety of conditions is episodic, not continuous. The episodes of eating are called meals. Thus, total food intake is determined by the size of each meal that occurs during the period of interest. A large literature developed that described changes in meal size or number in response to changes in diet, deprivation, diurnal and ovarian rhythms, pregnancy and lactation, operant contingencies, experience, brain lesions, social stimuli, neurotransmitters, neuromodulators, and drugs. This literature was descriptive and served primarily to provide a measurement of energy intake over time. Attempts to explain the changes in meal pattern at a mechanistic level failed repeatedly. For example, Le Magnen and Tallon2 found a correlation between meal size and intermeal interval, but this correlation was not robust because it was not seen by other investigators under similar or different conditions.3 Some success has been achieved more recently in identifying endogenous changes that occur just before the initiation of meals. These are a small decrease of plasma glucose,4 a decrease of metabolism,5 and an increase in liver temperature.6 These endog enous changes may be the result of learning,7 and they are not
necessary for meal initiation because the presentation of food or stimuli (temporal, sensory, or social) previously associated with food initiate a meal at times when the endogenous changes are not present.8 There are two important points about these adequate stimuli for initiating eating: First, the mechanisms that mediate the correlation between stimuli and meal initiation are not known. Second, none of these stimuli control meal size. The second point is fundamental because it means that the controls of meal size are different from the controls of meal number. Thus, the duration of eating and the amount eaten during a meal is determined by mechanisms that maintain eating once eating has begun.
MAINTENANCE OF EATING When the maintenance of eating was seen as an experimental problem, it shifted investigation from the meal as an episode of energy intake to the meal as a form of behavior. This was the beginning of what has become a major change in the analytic approach to meals. Instead of seeing meals as a problem of how intake serves metabolism and nutritional homeostasis, meals are now seen as a problem in behavioral neuroscience.9 The rest of this article reviews the major developments underlying this paradigm shift, the recent insights it has led to, and the new experiments it suggests.
The work was supported by NIH grant MH 40010. Correspondence to: Gerard P. Smith, MD, Bourne Laboratory, New YorkPresbyterian Hospital, Westchester Division, 21 Bloomingdale Road, White Plains, NY 10605, USA. E-mail: [email protected] Date accepted: July 7, 2000. Nutrition 16:814 – 820, 2000 ©Elsevier Science Inc., 2000. Printed in the United States. All rights reserved.
SHAM FEEDING AND THE MAINTENANCE OF EATING The application of the sham-feeding technique to the rat in the 1960s10 and early 1970s11,12 proved influential in the analysis of the maintenance of eating. Two important observations were made 0899-9007/00/$20.00 PII S0899-9007(00)00457-3
Nutrition Volume 16, Number 10, 2000 in this early work. First, sham-fed meals were always larger then real-fed meals. In fact, after overnight deprivation, eating continued for hours.12 This demonstrated that postingestive stimulation terminated eating. Second, the size of a sham-fed meal was dependent on diet, duration of food deprivation,12,13 and experience with the diet.11 This demonstrated that, in the absence of postingestive stimulation, orosensory stimulation determined the rate and duration of eating. Although the opposing functions of orosensory and postingestive stimulation had been suggested repeatedly, these experiments brought the problem under experimental control. Because food stimuli contacted oral receptors before they contacted postingestive receptors, orosensory stimulation occurred before postingestive stimulation. Thus, it was clear that the rate and duration of eating and, thus, the size of a meal are determined by the dynamic integration in the brain of the neural information provided by the orosensory and postingestive stimulation produced by the food being eaten. Because both types of stimuli are produced by the oral movements of eating and both influence the probability of the maintenance of eating, orosensory and postingestive stimuli function as feedbacks in the control of eating: orosensory stimuli provide positive feedback* and postingestive stimuli provide negative feedback.14 Davis et al. grasped the implications of these feedback relations, mathematized them using control theory, and presented a quantitative model for the control of meal size.15,16 This was new, it represented a developmental milestone in the field, and it was the direct result of using sham feeding to study the problem of the maintenance of eating. The model was based on experiments with carbohydrate solutions in which the rate of licking was measured. The model proposed 1) that the initial rate of licking measured the gustatory stimulation produced by the carbohydrate solution and 2) that the slope of the function describing the subsequent rate of licking until the meal ended described the central comparison of the relative potencies of the positive feedback produced by gustatory stimulation and the negative feedback produced by the postingestive stimulation. In addition to providing a better framework for the investigation of meal size, the model made specific and testable predictions. One was that removal of postingestive negative feedback should result in a rate of eating that was determined solely by the potency of the positive feedback provided by gustatory stimulation. Thus, when gustatory stimulation was fixed by using the same carbohydrate solution in real feeding and the first sham-feeding test, the model predicted that the increase in meal size during the first sham-feeding test should be maximal. When this prediction was tested, the results did not fulfill it. In fact, this discrepancy between prediction and experimental result was first noticed by Davis and Campbell in 197311 and subsequently reported by Young et al.12 and Mook et al.17 The failure to confirm could have been due to adaptation to the novelty of sham feeding or to the gradual extinction of a learned, inhibitory, orosensory control of eating. Davis and Smith18 tested these alternatives and found that the progressive increase of intake after the first sham-feeding test was from the extinction of a
*The use of positive feedback in the control of eating has been objected to by some investigators on the grounds of the well-known fact that, when positive feedback acts alone, the system runs out of control. But real feeding is controlled by a system that has peripheral positive and negative feedbacks and a central mechanism for comparing their potencies.55 Indeed, control by a system that has positive and negative feedbacks is ubiquitous at all levels of biological analysis from genomic transcription, enzymatic action, hormonal effects, neural control of movements, and behavioral interactions. See the recent study by Shearman et al.56 for a relevant example at the molecular level in the function of the suprachiasmatic nucleus in the control of diurnal rhythms.
Controls of Eating
815
conditioned inhibitory control that functioned during the first 8 min of eating. In contrast, the increase of intake during the first sham-feeding test was from increased intake after the first 8 min. Because this delayed increase was maximal in the first shamfeeding test, it was unconditioned. These experiments not only identified the intervals in which conditioned and unconditioned negative feedbacks occurred, they demonstrated that the defect in the model was from the lack of a conditioned component of negative feedback.19 Conditioned negative feedback is based on the association between orosensory and postingestive stimulation during real feeding.20 This association can be formed within one or two meals. The unconditioned stimuli have begun to be characterized: gastric volume and an unidentified postpyloric stimulus are effective,21 but increased plasma glucose is not.22 Davis and his colleagues pursued the role of conditioned negative feedback in a series of recent experiments that changed the concentration of milk or sucrose. They used repetitive shamfeeding tests to identify the extinction of the conditioned inhibition. The major result was that conditioned negative feedback occurred with high concentrations, but not with low concentrations.23,24 This observation suggests that postingestive stimulation must be relatively strong to serve as an unconditioned stumulus for the formation of conditioned, orosensory negative feedback. What is the function of this concentration-dependent conditioned negative feedback? Davis et al.23 proposed that it decreases the rate of gastric emptying of concentrated solutions into the small intestine because, by slowing the rate of ingestion, it slows the rate of increase of gastric volume, a major stimulant of gastric emptying. This in turn diminishes or avoids the aversive consequences of the rapid entry of concentrated solutions into the small intestine, such as occur in the dumping syndrome.
THE MICROSTRUCTURE OF THE MAINTENANCE OF EATING The demonstration that the maintenance of eating was under peripheral positive- and negative-feedback controls naturally led to the question of how those controls affected the maintenance of eating. To answer this question, one had to measure the pattern of eating, i.e., its microstructure, and its rate during a meal. By the 1980s, the use of lickometers had shown that liquid nutrients were eaten in bouts of licks separated by pauses of non-licking. Although the duration of bouts and the pauses between them differed, the rate of licking was fixed at a frequency of five to seven per second.25 The fact that the rate of licking was fixed had important consequences for the analysis of microstructure. First, it indicated that licking was the output of a central neural network that had the functional characteristics of a central pattern generator (cpg). Second, it meant that the effects of both positive and negative feedback would be expressed temporally in how long a bout of licking lasted and how long a pause between bouts lasted.26 Thus, the pattern of licking recorded by a lickometer during the maintenance of eating was the record of when the cpg was on and off. Positive feedback turned the cpg on and negative feedback turned it off.27 Their relative potency at any moment during a meal is reflected by the number of bouts of licking at that time. This interpretation of the number of bouts of licking has been confirmed with carbohydrates, dextrins, and oils. It is a significant penetration into the problem of the maintenance of eating because it is a precise and sensitive measure of the neural control of an oral movement involved in eating. This opened up the problem to neurologic analysis.
THE NEUROLOGY OF THE MAINTENANCE OF EATING The hindbrain contains the neurons of the cpg and the neurons that receive positive feedback from the mouth and negative feedback
816
Smith
from the stomach and small intestine. The neurons of the cpg are in a relatively large area of the medial reticular formation28; their premotor neurons extend as far forward as the substantia nigra.29 The final motor neurons for oral movements are in the trigeminal, facial, and hypoglossal nuclei. The hindbrain also contains vagal motor neurons that project widely to abdominal viscera for the control of gut movements, the release of pancreatic hormones, and hepatic metabolism. The neurons that receive positive and negative feedback from peripheral, preabsorptive stimulation are in the nucleus of the solitary tract (NTS). The rostral and intermediate NTS receive positive feedback from orosensory stimuli conducted over afferent fibers in the fifth, seventh, ninth, and tenth cranial nerves. The intermediate, subpostremal, and caudal NTS receive negative feedback from the stomach and small intestine conducted by afferent fibers of the abdominal vagus nerves and probably from afferent fibers of the splanchnic nerves that synapse in the spinal cord and then ascend to the NTS. The midportion of the hindbrain is capped by the area postrema, one of the circumventricular organs that is permeable to many circulating substances. Intrinsic neurons and extrinsic afferent terminals in the area postrema transduce humoral stimuli into neural information for the control of eating. For example, the area postrema is necessary for lithium chloride to function as an unconditioned stimulus in the formation of a conditioned taste aversion. The presence in the hindbrain of premotor and motor neurons, a cpg, and second-order neurons that receive peripheral sensory information from functional feedback loops that stimulate and inhibit oromotor movements is very similar to the functional arrangement of the spinal cord for limb movement: the spinal cord has cpg, sensory, and motor components. There are three other similarities. First, the afferent nerves enter the hindbrain dorsal to the site where efferent nerves leave the hindbrain. This, of course, is the classic arrangement in the spinal cord. Second, the hindbrain and spinal cord receive afferent inputs from neural regions cephalad to them that provide important information for the normal control of oromotor and limb movements, respectively. Third, when the hindbrain or spinal cord are disconnected from the neural regions cephalad to them, oromotor or limb movements can be organized into eating and walking, respectively, but only when stimuli are applied directly to the relevant sensory surface, i.e., the mouth for eating and the soles of the feet for walking.9 That the hindbrain is the spinal cord for ingestion is deduced from analogous functional anatomy. Such an argument by analogy needs support from experiments. The critical evidence has come from investigation of the controls of eating in the chronic decerebrate rat.
THE CHRONIC DECEREBRATE RAT AND THE MAINTENANCE OF EATING The introduction of the chronic decerebrate rat into the investigation of the maintenance of eating and meal size in 1978 by Grill and Norgren30 was extremely important for the neurology of eating. It was more important than the previous discoveries of the disordered eating produced by ventromedial and lateral hypothalamic lesions because of the nature of the lesions. The hypothalamic lesions damaged cells and fibers to different degrees. The extent of the disconnections was never known. In contrast, the decerebrate lesion severed all of the reciprocal fibers that connected the forebrain and hindbrain. Thus, the disorders of eating after complete disconnection are simpler to interpret neurologically than after the partial disconnections produced by hypothalamic lesions. The most obvious disorder of eating in the chronic decerebrate
Nutrition Volume 16, Number 10, 2000 rat is that the rat never initiates eating unless liquid food is infused directly into its mouth. This is consistent with the caudal brainstem being the spinal cord for ingestion in that it responds only to contact. It also demonstrates that the forebrain must be connected to the hindbrain for eating to be initiated. This is not simply a deficit in the movements necessary to find food because the chronic decerebrate rat can locomote, climb, and groom itself.30 When milk or other liquid foods are infused into the mouth through chronic oral catheters, the decerebrate rat eats and swallows normally until it is satiated. At that point, the decerebrate rat lets the infused liquid drip out of its mouth. Thus, the decerebrate rat eats meals and the size of the meals changes with the liquid food infused. For example, decerebrate rats have the same intakeresponse function to a series of sucrose concentrations as intact rats. Decerebrate rats also increase intake of sucrose solutions when they sham feed31 and decrease intake after gastric preloads. Thus, the brainstem has the capacity to respond to positive and negative feedbacks and to integrate them into an output of the cpg that produces a meal that changes in size as a function of the relative potency of the peripheral feedbacks. These functions are consistent with the anatomic facts of the central projections of the afferent fibers mediating the positive and negative feedbacks into the hindbrain behind the disconnection and the presence of the cpg there. They also demonstrate what those anatomic facts did not: the disconnected brainstem has the neural complexity to perform the comparator function necessary to integrate positive and negative feedbacks into oromotor outputs that maintain eating for the appropriate time. A number of controls of eating, however, do not work in the chronic decerebrate rat,32 including the metabolic effect of food deprivation33 and the acquisition or expression of learned taste aversions.34 Because the decerebrate rat only eats when food contacts its oral receptors, it is clear that distal stimuli involved in the control of eating by foraging experience or contingencies35 or by social stimuli36 are also not effective. Furthermore, the importance of the forebrain for diurnal and ovarian rhythms that control eating makes it very likely that these controls of eating are not effective either. The neural interpretations of the capacities and the defects of the controls of eating in the chronic decerebrate rat are different. The neural networks of the disconnected brainstem are sufficient for the controls of eating that have been demonstrated. The neurologic interpretation of the controls that do not operate in the chronic decerebrate rat is that the disconnected brainstem is not sufficient to respond to them and some unidentified fibers of the large number passing back and forth between the forebrain and brainstem are necessary for them.37,38 When the interpretation of the neurology of eating in the chronic decerebrate rat was combined with the results of a separate analysis of the adequate stimuli of the positive and negative feedbacks from the periphery in intact rats, a new theory of the control of eating emerged.
DIRECT AND INDIRECT CONTROLS OF THE MAINTENANCE OF EATING The critical fact about the adequate stimuli of ingested food for the peripheral feedbacks is that all of them act preabsorptively.† It was well known, of course, that orosensory and gastric stimuli acted preabsorptively, but it was a new development when it was shown
†Preabsorptive in this context means that the stimulus acts before absorption into the blood or lymph. Only additional experiments can refine this usage to distinguish action at the surface of the mucosa (preabsorptive in the strict sense) from action within mucosal cells or in the extracellular space before entry into the blood or lymphatic vessels.
Nutrition Volume 16, Number 10, 2000
Controls of Eating
817
TABLE I. INDIRECT CONTROLS OF MEAL SIZE Categories*
Examples
Rhythmic Metabolic Thermal Conditioned Cognitive Ecologic
Diurnal, estrogen Chronic changes in insulin and fat mass Environmental and fever Preferences, aversions, and satiations Social and, in humans, cultural and esthetic Relative densities of predators and foods
* The list of categories is not mutually exclusive nor exhaustive; this is particularly true for conditioned, cognitive, and ecologic.
that food stimuli that entered the small intestine also acted only preabsorptively in controlling the size of the current meal.39‡ Although the population of preabsorptive receptors stimulated during a meal depends on the stimuli produced by the ingested food and its digestion products, some preabsorptive receptors must be stimulated during each meal. The stimulation of the receptors activates the mechanisms of the positive and negative feedbacks. The feedback information is carried over afferent fibers that project into the hindbrain, where the information can be integrated into a meal in the chronic decerebrate rat. These controls are the direct controls of meal size.40 The defining criterion is that the adequate stimuli act directly on preabsorptive receptors along the surface of the gastrointestinal tract, from the mouth through the small intestine. (The possibility that preabsorptive stimuli in the large intestine contribute to the direct controls of meal size under physiologic conditions has received little experimental attention.41) All other controls of meal size are categorized as indirect controls.40 They affect more than one meal, they are numerous and diverse (Table I), and they require normal neural connections between the forebrain and brainstem because none of them has been demonstrated to act in the chronic decerebrate rat. The functional neurology of the direct and indirect controls is summarized schematically in Figure 1. The functional neurology emphasizes the importance of the reciprocal connections of the forebrain and hindbrain for the effect of the indirect controls. The fact that some direct controls act during every meal means that all indirect controls must affect meal size by modulating the potency of the positive or negative feedbacks of the direct controls that are active during that meal. An example of such modulation was proposed by Figlewicz et al. in 1986. They suggested that insulin, a negative-feedback signal of fat mass (an indirect control), decreased food intake by increasing the inhibitory effect of peptide mechanisms of direct controls, such as cholecystokinin, that were released during a meal.42 I have expanded that idea to include all indirect controls and both types of feedback. The sites and neural mechanisms involved in the changes of processing of afferent information that changes the potency of the peripheral feedbacks is open to experiment. Recent results have
‡The demonstration that radioactive glucose can pass from the mouth to the brain57 and that infusions of glucose into the hepatic portal vein can decrease meal size39 have been proposed as evidence of postabsorptive sites of action of these stimuli in the control of meal size. Unfortunately, the demonstration of a phenomenon is not sufficient to prove that the phenomenon controls meal size under physiologic conditions. For example, there is no evidence that experimental blockade of the actions of absorbed glucose in the liver during a meal increases the size of that meal.
FIG. 1. The space between the forebrain (left) and the hindbrain (right) emphasizes the disconnection produced by the chronic decerebrate lesion at the upper brainstem. Functional analysis of the controls of eating in such rats showed that the direct controls of eating that are mediated by stimulation of orosensory and viscerosensory neurons by food are effective because these afferent neurons project to the hindbrain. The indirect controls that depend on distributed processing of diverse types of relevant information are not, however, because the efficacy of indirect controls depends on connections between the forebrain and the hindbrain. When the connections are intact in the normal rat, eating is the integrated action of indirect and direct controls of the central pattern generator in the hindbrain that organizes the oral movements of eating.
demonstrated such modulation can occur at the level of the NTS (neuropeptide Y and naloxone; for other possibilities, see Smith39).
ASSESSMENT OF THE THEORY OF DIRECT AND INDIRECT CONTROLS OF MEAL SIZE A good theory must be comprehensive, explicit, and heuristic. It must serve as an unambiguous framework in which the relationships about what is known can be determined and tested, and it must suggest new experiments that could require its revision or replacement. The theory of direct and indirect controls has these characteristics. It is the most comprehensive theory for the control of meal size that has ever been proposed. No previously suggested control, e.g., psychological, physiologic, or ecologic, long-term or short-term, glucostatic, aminostatic, lipostatic, neuroendocrine, or metabolic mechanism is excluded, but the functional relations among them and their effect and potency under various conditions can be assessed. The theory is more explicit than any previous theory. It provides an unambiguous criterion for categorizing direct and indirect controls. It identifies the peripheral positive and negative feedbacks produced by preabsorptive food stimuli as the crucial feature of the neural control of eating because they provide essential information for the action of all direct and indirect controls. It asserts that all indirect controls act by modulating the potency of direct controls and that the reciprocal connections between the forebrain and brainstem are necessary for this modulation. All these assertions are testable by current experimental techniques. The theory also provides a coherent framework for what is known about neurotransmitters, hormones, and neuropeptides. Peripheral molecules released during a meal can be assigned as mechanisms of positive or negative feedback (Table II). Some of the central amines and peptides can also be classified this way (Table III). The number of central and peripheral peptides and their interactions is large and growing (for recent reviews, see Smith44). However, many of the new gene products such as leptin, melanocortin, agouti-related peptide, orexin, cocaine and
818
Smith
Nutrition Volume 16, Number 10, 2000 TABLE II. MOLECULAR MECHANISMS OF DIRECT CONTROLS OF MEAL SIZE
Direct controls
Peripheral
Central
Orosensory
Gustatory and olfactory transducers
Gastric
CCK* at CCKA vagal mechanoreceptors, other mechanoreceptors, and bombesin-like peptides CCK* at CCKA receptors on vagal mechano- and chemoreceptors; glucagon*, amylin, enterostatin, apolipoprotein IV, and insulin released by contact with mucosal receptors or by the release of incretins by nutrient or digestive stimuli
Small intestinal
Dopamine* Opioids* Amino acids from gastric vagal afferent terminals in NTS; Serotonin* Serotonin*
* This molecule has been demonstrated to be a physiologic mechanism. The physiologic status of the other molecules is uncertain. CCK, cholecystokinin; NTS, nucleus of the solitary tract.
amphetamine-related transcript (CART), and previously known peptides such as corticotropin-releasing factor (CRF), galanin, and neuropeptide Y cannot be assigned to a specific feedback effect on
TABLE III. CANDIDATE CENTRAL MECHANISMS OF DIRECT AND INDIRECT CONTROLS OF MEAL SIZE* Increase intake Dopamine† Neuropeptide Y† Norepinephrine† Galanin Growth hormone-releasing hormone Orexins A and B (hypocretins 1 and 2) Dynorphin -Endorphin Agouti-related protein Melanin-concentrating hormone
Decrease intake Serotonin‡ Dopamine Norepinephrine Estrogen§ Leptin㛳
eating. Although all of them have been shown to decrease or increase food intake, we do not know how most of them produce this change.45 Only leptin has been shown to decrease meal size without changing meal number. This observation triggers the next question: How does leptin decrease meal size? The new theory provides a framework for answering that question because it reduces the possible explanations of a decrease in meal size to four combinations of changes in positive and negative feedbacks (Table IV). The experimental techniques for assessing the effect of a molecule on the feedbacks in rodents and humans are currently available.38 Furthermore, the potency of any effect on feedback can be quantitated by traditional curve-shift analysis.38,40 Such analysis in terms of feedback potency is necessary for determining the function of any molecule in the control of meal size. This in turn will facilitate the search for the fundamental biological meaning of a molecule that is defined by specifying its site(s) and function(s) in the central network that controls meal size.
Insulin¶ Corticotropin-releasing hormone Bombesin-like peptides Cholecystokinin Glucagon-like peptide-1 ␣-Melanocyte–stimulating hormone
* Although these amines and peptides have been shown to increase or decrease meal size under some conditions, their relative physiologic importance and their mediation of direct and indirect controls of meal size is uncertain except in the following cases. †Dopamine, neuropeptide Y, and norepinephrine mediate orosensory positive feedback of sweet taste and other foods. ‡ Central serotonin mediates the postingestive negative-feedback effect of small intestinal CCK.45 It is probably involved in numerous direct and indirect inhibitory controls. It is also possible that peripheral serotonin mediates inhibition of intake.46 § Estrogen mediates the ovarian-rhythmic indirect control by increasing the satiating potency of postingestive negative-feedback mechanisms, especially CCK.47 㛳,¶ Centrally administered leptin and insulin mediate the inhibitory indirect control of meal size exerted by adipose mass.48,49 At least part of insulin’s effect may be produced by increasing the satiating potency of CCK.42,50 Leptin’s action depends on the melanocortin system45; its synergism with CCK is not clear.51–54 Note that dopamine and norepinephrine increase or decrease intake depending on the site of injection, phase of the diurnal rhythm, etc. This underlines the fact that the meaning of any central molecule is determined by the function it has in the neural network in which it is a part. CCK, cholecystokinen.
CONCLUSION The shift from seeing the control of meal size as a problem in nutritional homeostasis to seeing it as a problem in behavioral neuroscience has a number of advantages. The first is generality.
TABLE IV. CHANGES IN POTENCY OF AFFERENT FEEDBACKS THAT DETERMINE CHANGES IN MEAL SIZE Afferent feedback*
Increase Increase Increase Increase Decrease Decrease Decrease Decrease
Positive
Negative
Increase Increase Increase No change Decrease Decrease Decrease No change
Decrease No change Smaller increase Decrease Increase No change Smaller decrease Increase
* The list of changes of afferent feedbacks is logically complete and unambiguous. Identification of the mechanisms of a specific change in potency of feedbacks is an experimental problem (see Table III for candidates).
Nutrition Volume 16, Number 10, 2000 How the central nervous system integrates nutritional processes into the controls of meal size retains its significance and experimental importance, but this is seen as a subset of the general problem of how any relevant information is integrated into the maintenance of appropriate oromotor movements. In this sense, the solution of any one of the problems should facilitate the solution of the others because all are using the same or similar systematic rules. The second advantage is that the shift has been the basis for a new theory that is more comprehensive, explicit, and heuristic than previous ones. The third advantage is that the framework of experimental analysis derived from the behavioralneuroscientific approach is specific and quantitative. The fourth advantage is that it provides an experimental framework in which to search for the meaning of molecules that change meal size. Fifth, and most important, all of the assertions of the theory are testable. This makes the theory open to refinement, revision, or replacement.
ACKNOWLEDGMENT The author thanks Ms. Laurel Torres for processing this manuscript and Dr. Nori Geary for constructive criticism and an important contribution to Table IV.
REFERENCES 1. Richter CP. A behavioristic study of the activity of the rat. Comp Psychol Monogr 1922;1(2) 2. LeMagnen J, Tallon S. La periodicite spontanee de la prise d’aliments ad libitum du rat blanc. J Physiol 1966;58:323 3. Collier G, Johnson DF, Mitchell C. The relation between meal size and the time between meals: effects of cage complexity and food cost. Physiol Behav 1999; 67:339 4. Campfield LA, Smith FJ. Systemic factors in the control of food intake. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiology of food and fluid intake. New York: Plenum Press, 1990:183 5. Even P, Coulaud H, Nicolaidis S. Integrated metabolic control of food intake after 2-deoxy-D-glucose and nicotinic acid injection. Am J Physiol 1988;255:R82 6. De Vries J, Strubbe JH, Wildering WC, Gorter JA, Prins AJ. Patterns of body temperature during feeding in rats under varying ambient temperatures. Physiol Behav 1993;53:229 7. Woods SC, Strubbe JH. The psychobiology of meals. Psychon Bull Rev 1994; 1:141 8. Weingarten HP. Conditioned cues elicit feeding in sated rats: a role for learning in meal initiation. Science 1983;220:431 9. Smith GP. The controls of eating: brain meanings of food stimuli. In: Mayer EA, Saper CB, eds. Progress in brain research, Vol. 122. The biological basis for mind– body interactions. New York, Elsevier Science BV, 2000:173 10. Mook DG. Oral and postingestional determinants of the intake of various solutions in rats with esophageal fistulas. J Comp Physiol Psychol 1963;56:645 11. Davis JD, Campbell CS. Peripheral control of meal size in the rat: effect of sham feeding on meal size and drinking rate. J Comp Physiol Psychol 1973;83:379 12. Young RC, Gibbs J, Antin J, Holt J, Smith GP. Absence of satiety during sham feeding in the rat. J Comp Physiol Psychol 1974;87:795 13. Kraly FS, Carty WJ, Smith GP. Effect of pregastric food stimuli on meal size and intermeal interval in the rat. Physiol Behav 1978;20:779 14. Campbell CS, Davis JD. Peripheral control of food intake; interaction between test diet and postingestive chemoreception. Physiol Behav 1974;12:377 15. Davis JD, Collins BJ, Levine MW. Peripheral control of drinking: gastrointestinal filling as a negative feedback signal: a theoretical and experimental analysis. J Comp Physiol Psychol 1975;89:985 16. Davis JD, Levine M. A model for the control of ingestion. Psychol Rev 1977; 89:379 17. Mook DG, Culberson R, Gelbart RJ, McDonald K. Oropharyngeal control of ingestion in rats: acquisition of sham-drinking patterns. Behav Neurosci 1983; 97:574 18. Davis JD, Smith GP. Learning to sham feed: behavioral adjustments to loss of physiological postingestional stimuli. Am J Physiol 1990;259:R1228 19. Davis JD. A model for the control of ingestion—20 years later. In: Fluharty SJ, Morrison AR, eds. Progress in psychobiology and physiological psychology. New York: Academic Press, 1998:127
Controls of Eating
819
20. Weingarten HP, Kulikovsky OT. Taste-to-postingestive consequence conditioning: is the rise in sham feeding with repeated experience a learning phenomenon? Physiol Behav 1989;45:471 21. Davis JD, Smith GP, Miesner J. Postpyloric stimuli are necessary for the normal control of meal size in real feeding and sham feeding rats. Am J Physiol 1993;265:R888 22. Gowans SE, Weingarten HP. Elevations of plasma glucose do not support taste-to-postingestive consequence learning. Am J Physiol 1991;261(6, Pt 2): R1409 23. Davis JD, Smith GP, Singh B, McCann DP. Increase in intake with sham feeding experience is concentration dependent. Am J Physiol 1999;277:R565 24. Davis JD, Smith GP, Singh B, McCann DP. The impact of milk-derived unconditioned and conditioned negative feedback on the microstructure of ingestive behavior. Physiol Behav 2000;70:1 25. Davis JD, Smith GP. Analysis of the microstructure of the rhythmic tongue movements of rats ingesting maltose and sucrose solutions. Behav Neurosci 1992;106:217 26. Davis JD. Some new developments in the understanding of orophyrangeal and postingestional controls of meal size. Nutrition 1999;15:32 27. Smith GP. Feeding. Control of eating. In: Adelman G, Smith BH, eds. Elsevier’s encyclopedia of neuroscience. New York: Elsevier Science, 1999:711 28. Travers JB, Dinardo LA, Kariminamazi H. Motor and premotor mechanisms of licking. Neurosci Biobehav Revs 1997;21:631 29. Fay RA, Norgren R. Identification of rat brainstem multisynaptic connections to the oral motor nuclei using pseudorabies virus. III. Lingual muscle motor systems. Brain Res Rev 1997;25:291 30. Grill HJ, Norgren R. Neurological tests and behavioral deficits in chronic thalamic and chronic decerebrate rats. Brain Res 1978;143:299 31. Grill HJ, Kaplan JM. Sham feeding in intact and chronic decerebrate rats. Am J Physiol 1992;262:R1070 32. Grill HJ, Kaplan JM. Caudal brainstem participates in the distributed neural control of feeding. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiology of food and fluid intake. New York: Plenum Press, 1990:125 33. Seeley RJ, Grill HJ, Kaplan JM. Neurological dissociation of gastrointestinal and metabolic contributions to meal size control. Behav Neurosci 1994;108:347 34. Grill HJ, Norgren R. Chronic decerebrate rats demonstrate satiation, but not bait shyness. Science 1978;201:267 35. Collier G, Johnson DG. The time window of feeding. Physiol Behav 1990;48:771 36. Galef BG Jr, Beck M. Diet selection and poison avoidance by mammals individually and in social groups. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiolgy of food and fluid intake. New York: Plenum Press, 1990:329 37. Kaplan JM, Seeley RJ, Grill HJ. Daily caloric intake in intact and chronic decerebrate rats. Behav Neurosci 1993;107:876 38. Smith GP. Control of food intake. In: Shils ME, Olson JA, Shike M, Ross AC, eds. Modern nutrition in health and disease. Baltimore: Williams & Wilkins, 1999:631 39. Greenberg D. Intestinal satiety. In: Smith GP, ed. Satiation: from gut to brain. New York: Oxford University Press, 1998:40 40. Smith GP. The direct and indirect controls of meal size. Neurosci Biobehav Rev 1996;20:41 41. Meyer JH, Hlinka M, Tabrizi Y, DiMaso N, Raybould HE. Chemical specificities and intestinal distributions of nutrient-driven satiety. Am J Physiol 1998;275: R1293 42. Figlewicz DP, Stein LJ, West D, Porte D Jr, Woods SC. Intracisternal insulin alters sensitivity to CCK-induced meal suppression in baboons. Am J Physiol 1986;250:R856 43. Smith GP. Introduction to the reviews on peptides and the control of food intake and body weight. Neuropeptides 1999;33:323 44. Schwartz MW, Woods SC, Porte D Jr, Seeley RJ, Baskin DG. Central nervous system control of food intake. Nature 2000;404:661 45. Poeschla B, Gibbs J, Simansky KJ, Smith GP. The 5-HT1A agonist 8-OH-DPAT attenuates the satiating action of cholecystokinin. Pharmacol Biochem Behav 1992;42:541 46. Simansky KJ. Serotonin and the structure of satiation. In: Smith GP, ed. The structure and mechanisms of satiation. New York: Oxford University Press, 1998:217 47. Eckel LA, Geary N. Endogenous cholecystokinin’s satiating action increases during estrus in female Long-Evans rats. Peptides 1999;20:451 48. Woods SC. Insulin and the brain: a mutual dependency. In: Fluharty S, Morrison AR, Sprague J, Stellar E, eds. Progress in psychobiology and physiological psychology, Vol. 16. New York: Academic Press, 1995:53
820
Smith
49. Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P. Recombinant mouse OB protein: evidence for a peripheral signal linking adiposity and central neural networks. Science 1995;269:546 50. Riedy CA, Chavez M, Figlewicz DP, Woods SC. Central insulin enhances sensitivity to cholecystokinin. Physiol Behav 1995;58:755 51. Matson CA, Reid DF, Cannon TA, Ritter RC. Cholecystokinin and leptin act synergistically to reduce body weight. Am J Physiol Regul Integr Comp Physiol 2000;278:R882 52. Emond M, Schwartz GJ, Ladenheim EE, Moran TH. Central leptin modulates behavioral and neural responsivity to CCK. Am J Physiol 1999;276:R1545 53. Barrachina MD, Martinez V, Wang L, Wei JY, Tache Y. Synergistic interaction
Nutrition Volume 16, Number 10, 2000
54.
55. 56. 57.
between leptin and cholecystokinin to reduce short-term food intake in lean mice. Proc Natl Acad Sci USA 1997;94:10455 Wildman HF, Chua S Jr, Leibel RL, Smith GP. Effects of leptin and cholecystokinin in rats with a null mutation of the leptin receptor Lepr(fak). Am J Physiol Regul Integr Comp Physiol 2000;278:R1518 McFarland DJ. Feedback mechanisms in animal behavior. New York: Academic Press, 1971 Shearman LP, Sriram S, Weaver DR, et al. Interacting molecular loops in the mammalian circadian clock. Science 2000;288:1013 Pilcher CWT, Jarman SP, Booth DA. The route of glucose to the brain from food in the mouth of the rat. J Comp Physiol Psychol 1974;87:56
The Controls of Eating: A Shift from Nutritional Homeostasis to Behavioral Neuroscience Gerard P. Smith, MD From the Bourne Laboratory, New York-Presbyterian Hospital, Westchester Division, White Plains, New York, USA The experimental analysis of the controls of eating has undergone a paradigmatic shift in the past decade. Instead of seeing meals as a problem of how intake serves metabolism and nutritional homeostasis, meals are now seen as a problem in behavioral neuroscience. The major developments underlying this significant change are behavioral, neurological, and theoretical. Behavioral analysis has shown that a central pattern generator in the caudal brainstem organizes eating movements and that the size of a liquid meal is determined by the number and size of clusters of licking. Neurologic analysis has shown that eating is under orosensory positive-feedback control and postingestive, preabsorptive, negative-feedback control. These feedback controls are activated by food ingested during a meal. The sensory information of the feedbacks is carried by afferent fibers that project to the caudal brainstem. The new theory is based on the fact that the feedback controls are stimulated by food acting directly on mucosal receptors along the gastrointestinal tract, from the mouth to the end of the small intestine. Thus, they are referred to as direct controls, and the caudal brainstem is sufficient for organizing their action. All other controls, such as metabolic, rhythmic, and ecologic, that do not contact the mucosal receptors are indirect controls. Indirect controls act by modulating the potency of the central effects of the direct controls, and they require the forebrain and its reciprocal connections with the caudal brainstem for their control of eating and meal size. Nutrition 2000;16:814 – 820. ©Elsevier Science Inc. 2000 Key words: nutritional homeostasis, caudal brainstem, orosensory positive-feedback control, postingestive, preabsorptive, negative-feedback control, mucosal receptors, direct and indirect controls
INTRODUCTION Food intake, energy expenditure, and energy storage determine nutritional homeostasis. After Richter’s classic experiments in rats,1 it became clear that intake in all mammals under a variety of conditions is episodic, not continuous. The episodes of eating are called meals. Thus, total food intake is determined by the size of each meal that occurs during the period of interest. A large literature developed that described changes in meal size or number in response to changes in diet, deprivation, diurnal and ovarian rhythms, pregnancy and lactation, operant contingencies, experience, brain lesions, social stimuli, neurotransmitters, neuromodulators, and drugs. This literature was descriptive and served primarily to provide a measurement of energy intake over time. Attempts to explain the changes in meal pattern at a mechanistic level failed repeatedly. For example, Le Magnen and Tallon2 found a correlation between meal size and intermeal interval, but this correlation was not robust because it was not seen by other investigators under similar or different conditions.3 Some success has been achieved more recently in identifying endogenous changes that occur just before the initiation of meals. These are a small decrease of plasma glucose,4 a decrease of metabolism,5 and an increase in liver temperature.6 These endog enous changes may be the result of learning,7 and they are not
necessary for meal initiation because the presentation of food or stimuli (temporal, sensory, or social) previously associated with food initiate a meal at times when the endogenous changes are not present.8 There are two important points about these adequate stimuli for initiating eating: First, the mechanisms that mediate the correlation between stimuli and meal initiation are not known. Second, none of these stimuli control meal size. The second point is fundamental because it means that the controls of meal size are different from the controls of meal number. Thus, the duration of eating and the amount eaten during a meal is determined by mechanisms that maintain eating once eating has begun.
MAINTENANCE OF EATING When the maintenance of eating was seen as an experimental problem, it shifted investigation from the meal as an episode of energy intake to the meal as a form of behavior. This was the beginning of what has become a major change in the analytic approach to meals. Instead of seeing meals as a problem of how intake serves metabolism and nutritional homeostasis, meals are now seen as a problem in behavioral neuroscience.9 The rest of this article reviews the major developments underlying this paradigm shift, the recent insights it has led to, and the new experiments it suggests.
The work was supported by NIH grant MH 40010. Correspondence to: Gerard P. Smith, MD, Bourne Laboratory, New YorkPresbyterian Hospital, Westchester Division, 21 Bloomingdale Road, White Plains, NY 10605, USA. E-mail: [email protected] Date accepted: July 7, 2000. Nutrition 16:814 – 820, 2000 ©Elsevier Science Inc., 2000. Printed in the United States. All rights reserved.
SHAM FEEDING AND THE MAINTENANCE OF EATING The application of the sham-feeding technique to the rat in the 1960s10 and early 1970s11,12 proved influential in the analysis of the maintenance of eating. Two important observations were made 0899-9007/00/$20.00 PII S0899-9007(00)00457-3
Nutrition Volume 16, Number 10, 2000 in this early work. First, sham-fed meals were always larger then real-fed meals. In fact, after overnight deprivation, eating continued for hours.12 This demonstrated that postingestive stimulation terminated eating. Second, the size of a sham-fed meal was dependent on diet, duration of food deprivation,12,13 and experience with the diet.11 This demonstrated that, in the absence of postingestive stimulation, orosensory stimulation determined the rate and duration of eating. Although the opposing functions of orosensory and postingestive stimulation had been suggested repeatedly, these experiments brought the problem under experimental control. Because food stimuli contacted oral receptors before they contacted postingestive receptors, orosensory stimulation occurred before postingestive stimulation. Thus, it was clear that the rate and duration of eating and, thus, the size of a meal are determined by the dynamic integration in the brain of the neural information provided by the orosensory and postingestive stimulation produced by the food being eaten. Because both types of stimuli are produced by the oral movements of eating and both influence the probability of the maintenance of eating, orosensory and postingestive stimuli function as feedbacks in the control of eating: orosensory stimuli provide positive feedback* and postingestive stimuli provide negative feedback.14 Davis et al. grasped the implications of these feedback relations, mathematized them using control theory, and presented a quantitative model for the control of meal size.15,16 This was new, it represented a developmental milestone in the field, and it was the direct result of using sham feeding to study the problem of the maintenance of eating. The model was based on experiments with carbohydrate solutions in which the rate of licking was measured. The model proposed 1) that the initial rate of licking measured the gustatory stimulation produced by the carbohydrate solution and 2) that the slope of the function describing the subsequent rate of licking until the meal ended described the central comparison of the relative potencies of the positive feedback produced by gustatory stimulation and the negative feedback produced by the postingestive stimulation. In addition to providing a better framework for the investigation of meal size, the model made specific and testable predictions. One was that removal of postingestive negative feedback should result in a rate of eating that was determined solely by the potency of the positive feedback provided by gustatory stimulation. Thus, when gustatory stimulation was fixed by using the same carbohydrate solution in real feeding and the first sham-feeding test, the model predicted that the increase in meal size during the first sham-feeding test should be maximal. When this prediction was tested, the results did not fulfill it. In fact, this discrepancy between prediction and experimental result was first noticed by Davis and Campbell in 197311 and subsequently reported by Young et al.12 and Mook et al.17 The failure to confirm could have been due to adaptation to the novelty of sham feeding or to the gradual extinction of a learned, inhibitory, orosensory control of eating. Davis and Smith18 tested these alternatives and found that the progressive increase of intake after the first sham-feeding test was from the extinction of a
*The use of positive feedback in the control of eating has been objected to by some investigators on the grounds of the well-known fact that, when positive feedback acts alone, the system runs out of control. But real feeding is controlled by a system that has peripheral positive and negative feedbacks and a central mechanism for comparing their potencies.55 Indeed, control by a system that has positive and negative feedbacks is ubiquitous at all levels of biological analysis from genomic transcription, enzymatic action, hormonal effects, neural control of movements, and behavioral interactions. See the recent study by Shearman et al.56 for a relevant example at the molecular level in the function of the suprachiasmatic nucleus in the control of diurnal rhythms.
Controls of Eating
815
conditioned inhibitory control that functioned during the first 8 min of eating. In contrast, the increase of intake during the first sham-feeding test was from increased intake after the first 8 min. Because this delayed increase was maximal in the first shamfeeding test, it was unconditioned. These experiments not only identified the intervals in which conditioned and unconditioned negative feedbacks occurred, they demonstrated that the defect in the model was from the lack of a conditioned component of negative feedback.19 Conditioned negative feedback is based on the association between orosensory and postingestive stimulation during real feeding.20 This association can be formed within one or two meals. The unconditioned stimuli have begun to be characterized: gastric volume and an unidentified postpyloric stimulus are effective,21 but increased plasma glucose is not.22 Davis and his colleagues pursued the role of conditioned negative feedback in a series of recent experiments that changed the concentration of milk or sucrose. They used repetitive shamfeeding tests to identify the extinction of the conditioned inhibition. The major result was that conditioned negative feedback occurred with high concentrations, but not with low concentrations.23,24 This observation suggests that postingestive stimulation must be relatively strong to serve as an unconditioned stumulus for the formation of conditioned, orosensory negative feedback. What is the function of this concentration-dependent conditioned negative feedback? Davis et al.23 proposed that it decreases the rate of gastric emptying of concentrated solutions into the small intestine because, by slowing the rate of ingestion, it slows the rate of increase of gastric volume, a major stimulant of gastric emptying. This in turn diminishes or avoids the aversive consequences of the rapid entry of concentrated solutions into the small intestine, such as occur in the dumping syndrome.
THE MICROSTRUCTURE OF THE MAINTENANCE OF EATING The demonstration that the maintenance of eating was under peripheral positive- and negative-feedback controls naturally led to the question of how those controls affected the maintenance of eating. To answer this question, one had to measure the pattern of eating, i.e., its microstructure, and its rate during a meal. By the 1980s, the use of lickometers had shown that liquid nutrients were eaten in bouts of licks separated by pauses of non-licking. Although the duration of bouts and the pauses between them differed, the rate of licking was fixed at a frequency of five to seven per second.25 The fact that the rate of licking was fixed had important consequences for the analysis of microstructure. First, it indicated that licking was the output of a central neural network that had the functional characteristics of a central pattern generator (cpg). Second, it meant that the effects of both positive and negative feedback would be expressed temporally in how long a bout of licking lasted and how long a pause between bouts lasted.26 Thus, the pattern of licking recorded by a lickometer during the maintenance of eating was the record of when the cpg was on and off. Positive feedback turned the cpg on and negative feedback turned it off.27 Their relative potency at any moment during a meal is reflected by the number of bouts of licking at that time. This interpretation of the number of bouts of licking has been confirmed with carbohydrates, dextrins, and oils. It is a significant penetration into the problem of the maintenance of eating because it is a precise and sensitive measure of the neural control of an oral movement involved in eating. This opened up the problem to neurologic analysis.
THE NEUROLOGY OF THE MAINTENANCE OF EATING The hindbrain contains the neurons of the cpg and the neurons that receive positive feedback from the mouth and negative feedback
816
Smith
from the stomach and small intestine. The neurons of the cpg are in a relatively large area of the medial reticular formation28; their premotor neurons extend as far forward as the substantia nigra.29 The final motor neurons for oral movements are in the trigeminal, facial, and hypoglossal nuclei. The hindbrain also contains vagal motor neurons that project widely to abdominal viscera for the control of gut movements, the release of pancreatic hormones, and hepatic metabolism. The neurons that receive positive and negative feedback from peripheral, preabsorptive stimulation are in the nucleus of the solitary tract (NTS). The rostral and intermediate NTS receive positive feedback from orosensory stimuli conducted over afferent fibers in the fifth, seventh, ninth, and tenth cranial nerves. The intermediate, subpostremal, and caudal NTS receive negative feedback from the stomach and small intestine conducted by afferent fibers of the abdominal vagus nerves and probably from afferent fibers of the splanchnic nerves that synapse in the spinal cord and then ascend to the NTS. The midportion of the hindbrain is capped by the area postrema, one of the circumventricular organs that is permeable to many circulating substances. Intrinsic neurons and extrinsic afferent terminals in the area postrema transduce humoral stimuli into neural information for the control of eating. For example, the area postrema is necessary for lithium chloride to function as an unconditioned stimulus in the formation of a conditioned taste aversion. The presence in the hindbrain of premotor and motor neurons, a cpg, and second-order neurons that receive peripheral sensory information from functional feedback loops that stimulate and inhibit oromotor movements is very similar to the functional arrangement of the spinal cord for limb movement: the spinal cord has cpg, sensory, and motor components. There are three other similarities. First, the afferent nerves enter the hindbrain dorsal to the site where efferent nerves leave the hindbrain. This, of course, is the classic arrangement in the spinal cord. Second, the hindbrain and spinal cord receive afferent inputs from neural regions cephalad to them that provide important information for the normal control of oromotor and limb movements, respectively. Third, when the hindbrain or spinal cord are disconnected from the neural regions cephalad to them, oromotor or limb movements can be organized into eating and walking, respectively, but only when stimuli are applied directly to the relevant sensory surface, i.e., the mouth for eating and the soles of the feet for walking.9 That the hindbrain is the spinal cord for ingestion is deduced from analogous functional anatomy. Such an argument by analogy needs support from experiments. The critical evidence has come from investigation of the controls of eating in the chronic decerebrate rat.
THE CHRONIC DECEREBRATE RAT AND THE MAINTENANCE OF EATING The introduction of the chronic decerebrate rat into the investigation of the maintenance of eating and meal size in 1978 by Grill and Norgren30 was extremely important for the neurology of eating. It was more important than the previous discoveries of the disordered eating produced by ventromedial and lateral hypothalamic lesions because of the nature of the lesions. The hypothalamic lesions damaged cells and fibers to different degrees. The extent of the disconnections was never known. In contrast, the decerebrate lesion severed all of the reciprocal fibers that connected the forebrain and hindbrain. Thus, the disorders of eating after complete disconnection are simpler to interpret neurologically than after the partial disconnections produced by hypothalamic lesions. The most obvious disorder of eating in the chronic decerebrate
Nutrition Volume 16, Number 10, 2000 rat is that the rat never initiates eating unless liquid food is infused directly into its mouth. This is consistent with the caudal brainstem being the spinal cord for ingestion in that it responds only to contact. It also demonstrates that the forebrain must be connected to the hindbrain for eating to be initiated. This is not simply a deficit in the movements necessary to find food because the chronic decerebrate rat can locomote, climb, and groom itself.30 When milk or other liquid foods are infused into the mouth through chronic oral catheters, the decerebrate rat eats and swallows normally until it is satiated. At that point, the decerebrate rat lets the infused liquid drip out of its mouth. Thus, the decerebrate rat eats meals and the size of the meals changes with the liquid food infused. For example, decerebrate rats have the same intakeresponse function to a series of sucrose concentrations as intact rats. Decerebrate rats also increase intake of sucrose solutions when they sham feed31 and decrease intake after gastric preloads. Thus, the brainstem has the capacity to respond to positive and negative feedbacks and to integrate them into an output of the cpg that produces a meal that changes in size as a function of the relative potency of the peripheral feedbacks. These functions are consistent with the anatomic facts of the central projections of the afferent fibers mediating the positive and negative feedbacks into the hindbrain behind the disconnection and the presence of the cpg there. They also demonstrate what those anatomic facts did not: the disconnected brainstem has the neural complexity to perform the comparator function necessary to integrate positive and negative feedbacks into oromotor outputs that maintain eating for the appropriate time. A number of controls of eating, however, do not work in the chronic decerebrate rat,32 including the metabolic effect of food deprivation33 and the acquisition or expression of learned taste aversions.34 Because the decerebrate rat only eats when food contacts its oral receptors, it is clear that distal stimuli involved in the control of eating by foraging experience or contingencies35 or by social stimuli36 are also not effective. Furthermore, the importance of the forebrain for diurnal and ovarian rhythms that control eating makes it very likely that these controls of eating are not effective either. The neural interpretations of the capacities and the defects of the controls of eating in the chronic decerebrate rat are different. The neural networks of the disconnected brainstem are sufficient for the controls of eating that have been demonstrated. The neurologic interpretation of the controls that do not operate in the chronic decerebrate rat is that the disconnected brainstem is not sufficient to respond to them and some unidentified fibers of the large number passing back and forth between the forebrain and brainstem are necessary for them.37,38 When the interpretation of the neurology of eating in the chronic decerebrate rat was combined with the results of a separate analysis of the adequate stimuli of the positive and negative feedbacks from the periphery in intact rats, a new theory of the control of eating emerged.
DIRECT AND INDIRECT CONTROLS OF THE MAINTENANCE OF EATING The critical fact about the adequate stimuli of ingested food for the peripheral feedbacks is that all of them act preabsorptively.† It was well known, of course, that orosensory and gastric stimuli acted preabsorptively, but it was a new development when it was shown
†Preabsorptive in this context means that the stimulus acts before absorption into the blood or lymph. Only additional experiments can refine this usage to distinguish action at the surface of the mucosa (preabsorptive in the strict sense) from action within mucosal cells or in the extracellular space before entry into the blood or lymphatic vessels.
Nutrition Volume 16, Number 10, 2000
Controls of Eating
817
TABLE I. INDIRECT CONTROLS OF MEAL SIZE Categories*
Examples
Rhythmic Metabolic Thermal Conditioned Cognitive Ecologic
Diurnal, estrogen Chronic changes in insulin and fat mass Environmental and fever Preferences, aversions, and satiations Social and, in humans, cultural and esthetic Relative densities of predators and foods
* The list of categories is not mutually exclusive nor exhaustive; this is particularly true for conditioned, cognitive, and ecologic.
that food stimuli that entered the small intestine also acted only preabsorptively in controlling the size of the current meal.39‡ Although the population of preabsorptive receptors stimulated during a meal depends on the stimuli produced by the ingested food and its digestion products, some preabsorptive receptors must be stimulated during each meal. The stimulation of the receptors activates the mechanisms of the positive and negative feedbacks. The feedback information is carried over afferent fibers that project into the hindbrain, where the information can be integrated into a meal in the chronic decerebrate rat. These controls are the direct controls of meal size.40 The defining criterion is that the adequate stimuli act directly on preabsorptive receptors along the surface of the gastrointestinal tract, from the mouth through the small intestine. (The possibility that preabsorptive stimuli in the large intestine contribute to the direct controls of meal size under physiologic conditions has received little experimental attention.41) All other controls of meal size are categorized as indirect controls.40 They affect more than one meal, they are numerous and diverse (Table I), and they require normal neural connections between the forebrain and brainstem because none of them has been demonstrated to act in the chronic decerebrate rat. The functional neurology of the direct and indirect controls is summarized schematically in Figure 1. The functional neurology emphasizes the importance of the reciprocal connections of the forebrain and hindbrain for the effect of the indirect controls. The fact that some direct controls act during every meal means that all indirect controls must affect meal size by modulating the potency of the positive or negative feedbacks of the direct controls that are active during that meal. An example of such modulation was proposed by Figlewicz et al. in 1986. They suggested that insulin, a negative-feedback signal of fat mass (an indirect control), decreased food intake by increasing the inhibitory effect of peptide mechanisms of direct controls, such as cholecystokinin, that were released during a meal.42 I have expanded that idea to include all indirect controls and both types of feedback. The sites and neural mechanisms involved in the changes of processing of afferent information that changes the potency of the peripheral feedbacks is open to experiment. Recent results have
‡The demonstration that radioactive glucose can pass from the mouth to the brain57 and that infusions of glucose into the hepatic portal vein can decrease meal size39 have been proposed as evidence of postabsorptive sites of action of these stimuli in the control of meal size. Unfortunately, the demonstration of a phenomenon is not sufficient to prove that the phenomenon controls meal size under physiologic conditions. For example, there is no evidence that experimental blockade of the actions of absorbed glucose in the liver during a meal increases the size of that meal.
FIG. 1. The space between the forebrain (left) and the hindbrain (right) emphasizes the disconnection produced by the chronic decerebrate lesion at the upper brainstem. Functional analysis of the controls of eating in such rats showed that the direct controls of eating that are mediated by stimulation of orosensory and viscerosensory neurons by food are effective because these afferent neurons project to the hindbrain. The indirect controls that depend on distributed processing of diverse types of relevant information are not, however, because the efficacy of indirect controls depends on connections between the forebrain and the hindbrain. When the connections are intact in the normal rat, eating is the integrated action of indirect and direct controls of the central pattern generator in the hindbrain that organizes the oral movements of eating.
demonstrated such modulation can occur at the level of the NTS (neuropeptide Y and naloxone; for other possibilities, see Smith39).
ASSESSMENT OF THE THEORY OF DIRECT AND INDIRECT CONTROLS OF MEAL SIZE A good theory must be comprehensive, explicit, and heuristic. It must serve as an unambiguous framework in which the relationships about what is known can be determined and tested, and it must suggest new experiments that could require its revision or replacement. The theory of direct and indirect controls has these characteristics. It is the most comprehensive theory for the control of meal size that has ever been proposed. No previously suggested control, e.g., psychological, physiologic, or ecologic, long-term or short-term, glucostatic, aminostatic, lipostatic, neuroendocrine, or metabolic mechanism is excluded, but the functional relations among them and their effect and potency under various conditions can be assessed. The theory is more explicit than any previous theory. It provides an unambiguous criterion for categorizing direct and indirect controls. It identifies the peripheral positive and negative feedbacks produced by preabsorptive food stimuli as the crucial feature of the neural control of eating because they provide essential information for the action of all direct and indirect controls. It asserts that all indirect controls act by modulating the potency of direct controls and that the reciprocal connections between the forebrain and brainstem are necessary for this modulation. All these assertions are testable by current experimental techniques. The theory also provides a coherent framework for what is known about neurotransmitters, hormones, and neuropeptides. Peripheral molecules released during a meal can be assigned as mechanisms of positive or negative feedback (Table II). Some of the central amines and peptides can also be classified this way (Table III). The number of central and peripheral peptides and their interactions is large and growing (for recent reviews, see Smith44). However, many of the new gene products such as leptin, melanocortin, agouti-related peptide, orexin, cocaine and
818
Smith
Nutrition Volume 16, Number 10, 2000 TABLE II. MOLECULAR MECHANISMS OF DIRECT CONTROLS OF MEAL SIZE
Direct controls
Peripheral
Central
Orosensory
Gustatory and olfactory transducers
Gastric
CCK* at CCKA vagal mechanoreceptors, other mechanoreceptors, and bombesin-like peptides CCK* at CCKA receptors on vagal mechano- and chemoreceptors; glucagon*, amylin, enterostatin, apolipoprotein IV, and insulin released by contact with mucosal receptors or by the release of incretins by nutrient or digestive stimuli
Small intestinal
Dopamine* Opioids* Amino acids from gastric vagal afferent terminals in NTS; Serotonin* Serotonin*
* This molecule has been demonstrated to be a physiologic mechanism. The physiologic status of the other molecules is uncertain. CCK, cholecystokinin; NTS, nucleus of the solitary tract.
amphetamine-related transcript (CART), and previously known peptides such as corticotropin-releasing factor (CRF), galanin, and neuropeptide Y cannot be assigned to a specific feedback effect on
TABLE III. CANDIDATE CENTRAL MECHANISMS OF DIRECT AND INDIRECT CONTROLS OF MEAL SIZE* Increase intake Dopamine† Neuropeptide Y† Norepinephrine† Galanin Growth hormone-releasing hormone Orexins A and B (hypocretins 1 and 2) Dynorphin -Endorphin Agouti-related protein Melanin-concentrating hormone
Decrease intake Serotonin‡ Dopamine Norepinephrine Estrogen§ Leptin㛳
eating. Although all of them have been shown to decrease or increase food intake, we do not know how most of them produce this change.45 Only leptin has been shown to decrease meal size without changing meal number. This observation triggers the next question: How does leptin decrease meal size? The new theory provides a framework for answering that question because it reduces the possible explanations of a decrease in meal size to four combinations of changes in positive and negative feedbacks (Table IV). The experimental techniques for assessing the effect of a molecule on the feedbacks in rodents and humans are currently available.38 Furthermore, the potency of any effect on feedback can be quantitated by traditional curve-shift analysis.38,40 Such analysis in terms of feedback potency is necessary for determining the function of any molecule in the control of meal size. This in turn will facilitate the search for the fundamental biological meaning of a molecule that is defined by specifying its site(s) and function(s) in the central network that controls meal size.
Insulin¶ Corticotropin-releasing hormone Bombesin-like peptides Cholecystokinin Glucagon-like peptide-1 ␣-Melanocyte–stimulating hormone
* Although these amines and peptides have been shown to increase or decrease meal size under some conditions, their relative physiologic importance and their mediation of direct and indirect controls of meal size is uncertain except in the following cases. †Dopamine, neuropeptide Y, and norepinephrine mediate orosensory positive feedback of sweet taste and other foods. ‡ Central serotonin mediates the postingestive negative-feedback effect of small intestinal CCK.45 It is probably involved in numerous direct and indirect inhibitory controls. It is also possible that peripheral serotonin mediates inhibition of intake.46 § Estrogen mediates the ovarian-rhythmic indirect control by increasing the satiating potency of postingestive negative-feedback mechanisms, especially CCK.47 㛳,¶ Centrally administered leptin and insulin mediate the inhibitory indirect control of meal size exerted by adipose mass.48,49 At least part of insulin’s effect may be produced by increasing the satiating potency of CCK.42,50 Leptin’s action depends on the melanocortin system45; its synergism with CCK is not clear.51–54 Note that dopamine and norepinephrine increase or decrease intake depending on the site of injection, phase of the diurnal rhythm, etc. This underlines the fact that the meaning of any central molecule is determined by the function it has in the neural network in which it is a part. CCK, cholecystokinen.
CONCLUSION The shift from seeing the control of meal size as a problem in nutritional homeostasis to seeing it as a problem in behavioral neuroscience has a number of advantages. The first is generality.
TABLE IV. CHANGES IN POTENCY OF AFFERENT FEEDBACKS THAT DETERMINE CHANGES IN MEAL SIZE Afferent feedback*
Increase Increase Increase Increase Decrease Decrease Decrease Decrease
Positive
Negative
Increase Increase Increase No change Decrease Decrease Decrease No change
Decrease No change Smaller increase Decrease Increase No change Smaller decrease Increase
* The list of changes of afferent feedbacks is logically complete and unambiguous. Identification of the mechanisms of a specific change in potency of feedbacks is an experimental problem (see Table III for candidates).
Nutrition Volume 16, Number 10, 2000 How the central nervous system integrates nutritional processes into the controls of meal size retains its significance and experimental importance, but this is seen as a subset of the general problem of how any relevant information is integrated into the maintenance of appropriate oromotor movements. In this sense, the solution of any one of the problems should facilitate the solution of the others because all are using the same or similar systematic rules. The second advantage is that the shift has been the basis for a new theory that is more comprehensive, explicit, and heuristic than previous ones. The third advantage is that the framework of experimental analysis derived from the behavioralneuroscientific approach is specific and quantitative. The fourth advantage is that it provides an experimental framework in which to search for the meaning of molecules that change meal size. Fifth, and most important, all of the assertions of the theory are testable. This makes the theory open to refinement, revision, or replacement.
ACKNOWLEDGMENT The author thanks Ms. Laurel Torres for processing this manuscript and Dr. Nori Geary for constructive criticism and an important contribution to Table IV.
REFERENCES 1. Richter CP. A behavioristic study of the activity of the rat. Comp Psychol Monogr 1922;1(2) 2. LeMagnen J, Tallon S. La periodicite spontanee de la prise d’aliments ad libitum du rat blanc. J Physiol 1966;58:323 3. Collier G, Johnson DF, Mitchell C. The relation between meal size and the time between meals: effects of cage complexity and food cost. Physiol Behav 1999; 67:339 4. Campfield LA, Smith FJ. Systemic factors in the control of food intake. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiology of food and fluid intake. New York: Plenum Press, 1990:183 5. Even P, Coulaud H, Nicolaidis S. Integrated metabolic control of food intake after 2-deoxy-D-glucose and nicotinic acid injection. Am J Physiol 1988;255:R82 6. De Vries J, Strubbe JH, Wildering WC, Gorter JA, Prins AJ. Patterns of body temperature during feeding in rats under varying ambient temperatures. Physiol Behav 1993;53:229 7. Woods SC, Strubbe JH. The psychobiology of meals. Psychon Bull Rev 1994; 1:141 8. Weingarten HP. Conditioned cues elicit feeding in sated rats: a role for learning in meal initiation. Science 1983;220:431 9. Smith GP. The controls of eating: brain meanings of food stimuli. In: Mayer EA, Saper CB, eds. Progress in brain research, Vol. 122. The biological basis for mind– body interactions. New York, Elsevier Science BV, 2000:173 10. Mook DG. Oral and postingestional determinants of the intake of various solutions in rats with esophageal fistulas. J Comp Physiol Psychol 1963;56:645 11. Davis JD, Campbell CS. Peripheral control of meal size in the rat: effect of sham feeding on meal size and drinking rate. J Comp Physiol Psychol 1973;83:379 12. Young RC, Gibbs J, Antin J, Holt J, Smith GP. Absence of satiety during sham feeding in the rat. J Comp Physiol Psychol 1974;87:795 13. Kraly FS, Carty WJ, Smith GP. Effect of pregastric food stimuli on meal size and intermeal interval in the rat. Physiol Behav 1978;20:779 14. Campbell CS, Davis JD. Peripheral control of food intake; interaction between test diet and postingestive chemoreception. Physiol Behav 1974;12:377 15. Davis JD, Collins BJ, Levine MW. Peripheral control of drinking: gastrointestinal filling as a negative feedback signal: a theoretical and experimental analysis. J Comp Physiol Psychol 1975;89:985 16. Davis JD, Levine M. A model for the control of ingestion. Psychol Rev 1977; 89:379 17. Mook DG, Culberson R, Gelbart RJ, McDonald K. Oropharyngeal control of ingestion in rats: acquisition of sham-drinking patterns. Behav Neurosci 1983; 97:574 18. Davis JD, Smith GP. Learning to sham feed: behavioral adjustments to loss of physiological postingestional stimuli. Am J Physiol 1990;259:R1228 19. Davis JD. A model for the control of ingestion—20 years later. In: Fluharty SJ, Morrison AR, eds. Progress in psychobiology and physiological psychology. New York: Academic Press, 1998:127
Controls of Eating
819
20. Weingarten HP, Kulikovsky OT. Taste-to-postingestive consequence conditioning: is the rise in sham feeding with repeated experience a learning phenomenon? Physiol Behav 1989;45:471 21. Davis JD, Smith GP, Miesner J. Postpyloric stimuli are necessary for the normal control of meal size in real feeding and sham feeding rats. Am J Physiol 1993;265:R888 22. Gowans SE, Weingarten HP. Elevations of plasma glucose do not support taste-to-postingestive consequence learning. Am J Physiol 1991;261(6, Pt 2): R1409 23. Davis JD, Smith GP, Singh B, McCann DP. Increase in intake with sham feeding experience is concentration dependent. Am J Physiol 1999;277:R565 24. Davis JD, Smith GP, Singh B, McCann DP. The impact of milk-derived unconditioned and conditioned negative feedback on the microstructure of ingestive behavior. Physiol Behav 2000;70:1 25. Davis JD, Smith GP. Analysis of the microstructure of the rhythmic tongue movements of rats ingesting maltose and sucrose solutions. Behav Neurosci 1992;106:217 26. Davis JD. Some new developments in the understanding of orophyrangeal and postingestional controls of meal size. Nutrition 1999;15:32 27. Smith GP. Feeding. Control of eating. In: Adelman G, Smith BH, eds. Elsevier’s encyclopedia of neuroscience. New York: Elsevier Science, 1999:711 28. Travers JB, Dinardo LA, Kariminamazi H. Motor and premotor mechanisms of licking. Neurosci Biobehav Revs 1997;21:631 29. Fay RA, Norgren R. Identification of rat brainstem multisynaptic connections to the oral motor nuclei using pseudorabies virus. III. Lingual muscle motor systems. Brain Res Rev 1997;25:291 30. Grill HJ, Norgren R. Neurological tests and behavioral deficits in chronic thalamic and chronic decerebrate rats. Brain Res 1978;143:299 31. Grill HJ, Kaplan JM. Sham feeding in intact and chronic decerebrate rats. Am J Physiol 1992;262:R1070 32. Grill HJ, Kaplan JM. Caudal brainstem participates in the distributed neural control of feeding. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiology of food and fluid intake. New York: Plenum Press, 1990:125 33. Seeley RJ, Grill HJ, Kaplan JM. Neurological dissociation of gastrointestinal and metabolic contributions to meal size control. Behav Neurosci 1994;108:347 34. Grill HJ, Norgren R. Chronic decerebrate rats demonstrate satiation, but not bait shyness. Science 1978;201:267 35. Collier G, Johnson DG. The time window of feeding. Physiol Behav 1990;48:771 36. Galef BG Jr, Beck M. Diet selection and poison avoidance by mammals individually and in social groups. In: Stricker EM, ed. Handbook of behavioral neurobiology, Vol. 10. Neurobiolgy of food and fluid intake. New York: Plenum Press, 1990:329 37. Kaplan JM, Seeley RJ, Grill HJ. Daily caloric intake in intact and chronic decerebrate rats. Behav Neurosci 1993;107:876 38. Smith GP. Control of food intake. In: Shils ME, Olson JA, Shike M, Ross AC, eds. Modern nutrition in health and disease. Baltimore: Williams & Wilkins, 1999:631 39. Greenberg D. Intestinal satiety. In: Smith GP, ed. Satiation: from gut to brain. New York: Oxford University Press, 1998:40 40. Smith GP. The direct and indirect controls of meal size. Neurosci Biobehav Rev 1996;20:41 41. Meyer JH, Hlinka M, Tabrizi Y, DiMaso N, Raybould HE. Chemical specificities and intestinal distributions of nutrient-driven satiety. Am J Physiol 1998;275: R1293 42. Figlewicz DP, Stein LJ, West D, Porte D Jr, Woods SC. Intracisternal insulin alters sensitivity to CCK-induced meal suppression in baboons. Am J Physiol 1986;250:R856 43. Smith GP. Introduction to the reviews on peptides and the control of food intake and body weight. Neuropeptides 1999;33:323 44. Schwartz MW, Woods SC, Porte D Jr, Seeley RJ, Baskin DG. Central nervous system control of food intake. Nature 2000;404:661 45. Poeschla B, Gibbs J, Simansky KJ, Smith GP. The 5-HT1A agonist 8-OH-DPAT attenuates the satiating action of cholecystokinin. Pharmacol Biochem Behav 1992;42:541 46. Simansky KJ. Serotonin and the structure of satiation. In: Smith GP, ed. The structure and mechanisms of satiation. New York: Oxford University Press, 1998:217 47. Eckel LA, Geary N. Endogenous cholecystokinin’s satiating action increases during estrus in female Long-Evans rats. Peptides 1999;20:451 48. Woods SC. Insulin and the brain: a mutual dependency. In: Fluharty S, Morrison AR, Sprague J, Stellar E, eds. Progress in psychobiology and physiological psychology, Vol. 16. New York: Academic Press, 1995:53
820
Smith
49. Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P. Recombinant mouse OB protein: evidence for a peripheral signal linking adiposity and central neural networks. Science 1995;269:546 50. Riedy CA, Chavez M, Figlewicz DP, Woods SC. Central insulin enhances sensitivity to cholecystokinin. Physiol Behav 1995;58:755 51. Matson CA, Reid DF, Cannon TA, Ritter RC. Cholecystokinin and leptin act synergistically to reduce body weight. Am J Physiol Regul Integr Comp Physiol 2000;278:R882 52. Emond M, Schwartz GJ, Ladenheim EE, Moran TH. Central leptin modulates behavioral and neural responsivity to CCK. Am J Physiol 1999;276:R1545 53. Barrachina MD, Martinez V, Wang L, Wei JY, Tache Y. Synergistic interaction
Nutrition Volume 16, Number 10, 2000
54.
55. 56. 57.
between leptin and cholecystokinin to reduce short-term food intake in lean mice. Proc Natl Acad Sci USA 1997;94:10455 Wildman HF, Chua S Jr, Leibel RL, Smith GP. Effects of leptin and cholecystokinin in rats with a null mutation of the leptin receptor Lepr(fak). Am J Physiol Regul Integr Comp Physiol 2000;278:R1518 McFarland DJ. Feedback mechanisms in animal behavior. New York: Academic Press, 1971 Shearman LP, Sriram S, Weaver DR, et al. Interacting molecular loops in the mammalian circadian clock. Science 2000;288:1013 Pilcher CWT, Jarman SP, Booth DA. The route of glucose to the brain from food in the mouth of the rat. J Comp Physiol Psychol 1974;87:56