Recent accounts propose that (a) environmental food cues will tend to evoke eating until that behavior is inhibited by biological control mechanisms and (b) obesity may be more prevalent because these biological control mechanisms are failing (Berthoud, 2004b
; Prentice, 2005
)). What these control mechanisms might be, and why they fail are two questions fundamental to understanding, and ultimately controlling, continuing trends toward increased body weight and obesity in the human population. Much previous work aimed at addressing these questions has focused on hypothalamic control mechanisms and on identifying direct effects of changes in regulatory neuropeptides (e.g., leptin, CCK, ghrelin, etc.) and their receptors. By showing that damage to the hippocampus, a brain structure considered to be an important substrate for learning and memory, interferes with the control of food intake and body weight, the present findings encourage us to think about energy dysregulation, not solely as a deficit in some type of hypothalamic signaling system but, at least in part, as a disorder of higher-order learning and memory functioning (Davidson et al., 2007
; Davidson et al., 2005
There is wide agreement that learned cues can exert strong control over appetitive and consummatory behavior. This control depends, in part, on the formation of simple associations between food-related conditioned stimuli (CSs) and highly salient appetitive unconditioned stimuli (USs) that are produced as a consequence of eating (Berthoud, 2004a
; Davidson et al., 2005
; Davidson and Swithers, 2004
; Holland and Petrovich, 2005
; Sclafani, 1997
; Woods and Ramsay, 2000
). A food-related CS comes to promote the performance of appetitive and consummatory responses by exciting or activating a representation of its appetitive US in memory (Bouton and Moody, 2004
). This type of simple association formation does not appear to depend on the hippocampus as animals with the complete hippocampus removed are not impaired at learning that discrete CSs signal delivery of appetitive USs, or at solving simple discriminations where an event always signals reinforcement and another event is always nonreinforced (Benoit et al., 1999
; Han et al., 1995
; Squire, 1992
Although not necessary for the formation of simple associations, the hippocampus appears to be involved with the performance of certain higher-order learning and memory operations. Morrris (2006)
noted that several modern accounts converge on the idea that one function of the hippocampus is to solve problems that involve “predictable ambiguity”. These problems often require animals to learn that the relationship between an event and a particular outcome varies depending upon the presence or absence of other events or conditions. For example, animals with the hippocampus removed often show deficits in appetitive problems (e.g., extinction, discrimination-reversal, feature-negative discrimination, working memory) where performance depends on learning to refrain from responding to cues that are, under some conditions, reliable signals for reinforcement (Berger and Orr, 1983
; Chan et al., 2003
; Holland and Bouton, 1999
; Jarrard et al., 2004
). In these cases, it may be that hippocampal damage reduces the ability of animals to inhibit their appetitive behavior by impairing their ability to learn or remember when events will not be followed by reinforcing outcomes.
In the present experiment, damage to the complete hippocampus was not only accompanied by greater food intake and body weight gain, but also by increased appetitive responding in the conditioning apparatus, especially on trials under 24-hr food deprivation, and elevated behavioral activity in the home cage, especially in the vicinity of the food magazine, when the rats were fed ad libitum. Although heightened appetitive responding could be indicative of stronger simple excitatory appetitive conditioning, elevated appetitive performance can also be a consequence of impaired inhibitory learning. As discussed elsewhere (e.g., Benoit et al., 1999
) in Pavlovian conditioning, apparatus cues are both reinforced at the time of US delivery and nonreinforced during periods prior to presentation of the US. This could make apparatus cues ambiguous predictors of reinforcement. Intact rats could use handling or temporal cues to determine when to respond and to inhibit their responding to the apparatus cues. However, if removing the hippocampus interferes with the ability to use such contextual cues as signals that predict the nonreinforcement of apparatus cues, then weaker inhibition of responding to the apparatus cues would be expected. Furthermore, consistent with the training contingencies, if the expectation of receiving the sucrose pellet US was greater under 24-compared to 0-hr food deprivation, the effects of impaired inhibition would be more obvious when the rats were under the higher level of food deprivation—the outcome obtained in the present experiment.
Similarly, background cues in the home cage were presumably associated with a strong appetitive postingestive US when rats were hungry (e.g., prior to a meal) but not when they were food sated (e.g., after eating). Under these circumstances, the rats could use interoceptive cues produced by satiety to signal when food and food-related cues in the apparatus will not be followed by postingestive reinforcement. Increased activity on the part of CHip-lesioned rats, relative to controls, is consistent with the hypothesis that CHip lesions reduced the ability of satiety cues to signal the nonreinforcement of food cues, and thus to inhibit behaviors evoked by those cues. The finding that our CHip-lesioned rats spent significantly more time than controls on the side of the apparatus where food was delivered, indicates that some, if not all, of the increased homecage activity exhibited by CHip-lesioned rats was attributable to heightened appetitive behavior (Tracy et al., 2001
A general feature of this analysis is the assumption that the decision to eat or refrain from eating may involve higher-order or conditional learning processes that would help animals predict when food CSs are followed by an appetitive (pleasant or satisfying) postingestive US and when they are not (Davidson et al., 2007
; Davidson et al., 2005
). Given that survival depends on efficiently performing many behaviors (e.g., reproduction, defense, driving in rush hour traffic) in addition to procuring and consuming food, it would be highly adaptive if the ability of food CSs to excite memories of appetitive outcomes which promote food-seeking and eating responses was inhibited during times of food satiation. The present analysis is consistent with the idea that the performance of this adaptive function could depend on the hippocampus.
Could hippocampal dysfunction contribute to current global trends toward increased obesity in humans? Compared to the well-known and dramatic increases in food intake and body weight that accompany other types of experimental manipulations, such as lesioning the hypothalamus or genetic mutations (King, 2006
; Lindstrom, 2007
; Tschop and Heiman, 2001
), the effects of hippocampal lesions on food intake and weight gain that are reported here may seem modest. However, very few humans show dramatic increases in food intake and body weight like those shown by hypothalamic-lesioned or genetically-altered rodents. One could argue that the gradual increase in body weight seen in our rats makes them more similar to the current U.S. human population, which has exhibited about a 10% increase in body weight over the past 10 years (Lewis et al., 2000
Clear links between the function of the hypothalamus and recent increases in the incidence of obesity in the general population have not yet been identified. For example, there are relatively few cases of overweight or obese humans that can be attributed causally to hypothalamic pathologies or genetic mutations in hypothalamic signaling systems (Eikelis et al., 2007
; Pinkney et al., 2002
). Thus, although surgical, genetic, and other manipulations of the hypothalamus may have profound effects on energy regulation in laboratory settings, it is not yet clear how these manipulations are related to the reduced regulatory control that is occurring outside of the laboratory.
A potential link between the hippocampus and energy dysregulation in humans is suggested by evidence that dietary manipulations known to promote excessive food intake and body weight also disrupt hippocampal-dependent learning and memory processes. For example, Molteni et al., (Molteni et al., 2002
) reported that rats maintained for 60 days on a diet high in saturated fat and sucrose, showed impaired hippocampal-dependent spatial memory in a Morris water maze compared to rats maintained on normal (low-fat, high-carbohydrate) lab chow. Similarly, Kanoski et al (Kanoski et al., 2007
) found that giving rats 90-day ad libitum access to a diet high in saturated fat and dextrose had long-term detrimental effects on performance in Pavlovian conditioning tasks (reversal learning and extinction) that depend on the hippocampus or prefrontal cortex. These same rats did not exhibit performance deficits on a simple discrimination task that does not require an intact hippocampus or prefrontal cortex. Consistent with this general analysis, deficits in performance on hippocampal-dependent spatial learning problems are also observed in rat models of obesity (Matsushita et al., 2003
; Nomoto et al., 1999
; Winocur et al., 2005
Furthermore, Molteni (Molteni et al., 2002
) reported that spatial memory deficits by rats maintained on the high-fat diets, were accompanied by reduced levels of hippocampal brain-derived neurotrophic factor (BDNF). Kanoski et al (Kanoski et al., 2007
)) also found that BDNF was significantly reduced in the ventral hippocampus and medial prefrontal cortex, but not in the dorsal hippocampus, in rats that showed deficits in nonspatial reversal and extinction performance following maintenance on the high-fat diet. BDNF contributes to the survival, growth, and maintenance of many types of neurons (Allen and Dawbarn, 2006
; Nottebohm, 2004
) and is thought to contribute to hippocampal long-term potentiation (LTP) and neurogenesis ((Bramham and Messaoudi, 2005
; Lee et al., 2002
; Rossi et al., 2006
; Wibrand et al., 2006
). Both of these processes have been described as important mechanisms for hippocampal-dependent forms of learning and memory (Dalla et al., 2007
; Gruart et al., 2006
; Kitabatake et al., 2007
; Whitlock et al., 2006
). It may be that the ability of high-fat diets to promote increased food intake and body weight gain occurs as a consequence of interfering with the same hippocampal-dependent mechanisms that were disrupted by hippocampal lesions in our present experiment.
Obviously, hippocampal damage could also influence behavior by interfering with processes that do not involve learning and memory. In the present study, indirect calorimetry revealed that metabolic activity during the dark phase of the light-dark cycle was elevated for rats with CHip lesions compared to controls. It may be that this increased energy expenditure was a byproduct, at least in part, of the increased appetitive behavioral activity exhibited by rats with CHip lesions. However, heightened metabolism might have also been induced, in part, by the increased food intake on the part of the CHip-lesioned rats. Several studies have shown that metabolism increases, perhaps as a counterregulatory response, when animals are forced to consume calories in excess of their metabolic needs ((Balkan et al., 1993
; Harris et al., 2006
; Shibata and Bukowiecki, 1987
; Weyer et al., 2001
). It may be that increased metabolism is an effect of excess caloric intake that was difficult for rats with CHip lesions to control. However, in the present study increased metabolism was not enough to abolish weight gain on by rats with CHip lesions.
In addition, a relatively unexplored possibility is that the disturbances in energy regulation reported here involve a reduction in direct sensing by the hippocampus of nutrients or peripheral factors that regulate energy balance. As mentioned previously, the hippocampus expresses many of the same receptors (e.g., insulin, leptin, ghrelin and CCK) that are thought to be important for energy balance in the hypothalamus and brainstem. Thus, in our rats with hippocampal lesions, the sensing or relaying of this information may have been damaged contributing to increased food intake and/or body weight gain. Similarly, it is conceivable that intake of diets high in saturated fat could also interfere with this type of hippocampal functioning. An intriguing possibility is that selective genetic deletion of hippocampal nutrient or hormonal receptors might result in changes in energy balance as well. Consistent with this idea, Irani et al., (2007)
reported that intake of a high-fat diet is associated with reduced insulin binding in the hippocampal CA1 cell field of rats.
Rats with lesions confined to the hippocampal ventral pole also ate significantly more and gained significantly more weight relative to their controls. However, unlike rats with the complete hippocampus removed, VP lesions were not associated with significant increases in appetitive behavior, general activity, or metabolism. Furthermore, compared to rats with CHip lesions, weight gain for rats with VP lesions appeared to increase faster, relative to their controls, during the post-surgical recovery period and during the first 20 days after complete recovery from surgery. Despite exhibiting greater initial weight gain, the magnitude of the increase in food intake for VP-lesioned rats relative to controls appeared smaller than that observed for rats with CHip lesions. The finding that rats with VP lesions recovered from surgery more rapidly than CHip-lesioned rats could reflect that the debilitating effects VP surgery subsided more rapidly compared to the debilitation produced by much more extensive lesions of the complete hippocampus. These differences in recovery may have allowed the facilitating effects of VP lesions on intake and body weight gain to emerge more quickly compared to CHip lesions.
In the present study the damage produced by the CHip lesion encompassed all of the hippocampus including the ventral pole. The ventral pole lesion was relatively small by comparison (injection of IBO at 30 sites for the CHip lesion compared to 4 sites for the VP lesion). While the intent with the VP lesion was to remove all of the cells that comprise the ventral pole, it is possible that there was some sparing of the relevant cells in this group compared to the damage found in the CHip lesioned rats. Thus, differences in the effects of the two types of lesions on energy and body weight regulation can not be attributed solely to a common disruption of direct connections between the VP and the lateral hypothalamus. Further, It may be that the greater effect of the CHip lesion reflects interference with learned behavioral control processes in addition to those mediated by the hippocampal ventral pole-lateral hypothalamic circuit
On the other hand, it is possible that VP lesions interfered with the same learning and memory mechanisms as did CHip-lesions, but that the magnitude of this interference was smaller for VP-lesioned rats. It is difficult to evaluate the above possibilities since the effects of lesions confined to the VP on learning and memory, including occasion setting and similar hippocampal-dependent processes, have not yet been thoroughly studied.
It is also the case that rats with neurotoxic lesions of the medial prefrontal cortex did not differ from their controls with respect to any of the measures (e.g., intake, appetitive behavior, body weight gain, etc.) that were recorded in the present experiment. These rats required more time than rats with either CHip- or VP-lesions to achieve the criterion for post-operative recovery. However, it is not clear whether this effect was a consequence of greater general behavioral debilitation produced by mPFC lesions or weaker facilitation of eating and appetitive behavior, compared to CHip and VP lesions. The latter possibility seems likely based on the finding that during the post-recovery period, neither mean food intake nor weight gain for mPFC rats differed significantly relative to their controls.
Our findings that mPFC lesions had no significant effects on intake or weight gain is noteworthy for several reasons: first, the area of the mPFC that was lesioned in this experiment was the same area that showed reduced levels of BDNF following exposure to a maintenance diet high in saturated fat and dextrose (Kanoski et al., 2007
). Given that destruction of this area had little impact on energy regulation in the present study, this suggests that the excess intake and weight gain exhibited by rats maintained on the high fat + dextrose diet used in the study by Kanoski et al were not based on the effects of that diet on functions performed by the medial prefrontal cortex or by neural circuits that include this area of the brain. Second, the lack of effects of mPFC lesions on intake and body weight gain that we observed is consistent with another report that rats with lesions of the medial prefrontal cortex, albeit at a site slightly (but perhaps importantly-see below) ventral to the site of the mPFC lesions used in the present experiment, showed no differences in home cage food intake or in body weight relative to controls ((Petrovich and Gallagher, 2007
However, previous studies have shown that rats with lesions that include the ventral mPFC exhibited less “conditioned stimulus potentiated eating” when either discrete CSs or contextual cues that were trained to predict food when the rats were hungry, are presented when the rats are subsequently food sated (Petrovich and Gallagher, 2007
; Petrovich et al., 2007
). In the present study, rats with mPFC lesions did not differ significantly from controls with respect to their appetitive responding to contextual cues in the training apparatus, under either food deprived or nondeprived conditions. However, in addition to differences in exact location of medial prefrontal cortex damage, the present experiment also employed different food deprivation manipulations and training procedures compared to the earlier studies. It is possible that the different lesion effects reported in these experiments might be reconciled if rats were tested under more similar lesion or training conditions. In any event, the results of the present study provide no compelling evidence that energy and body weight regulation depends on the structural integrity of the medial prefrontal cortex or on any hippocampal-prefrontocortical neural pathway.
Previous research shows that the hypothalamus, especially the arcuate nucleus, contains receptors that are involved with the detection of a variety of neurohormonal hunger, satiety, and adiposity signals. The identification of these signals and their receptor sites has contributed much to our understanding of the control of food intake and body weight regulation. However, the question of how the detection of these cues is translated into adaptive behavioral outcomes has often been addressed by little more than an arrow in a diagram (e.g.,Berthoud, 2003
; Woods and Seeley, 2000
). The results of the present study suggest that to more fully understand the mechanisms that underlie energy and body weight regulation it may be necessary to describe how the operation of neurohormonal signaling systems that depend on the hypothalamus are integrated with higher-order learning and memory processes that depend on the hippocampus.
In the present study we found that destruction of the complete hippocampus in the rat is accompanied by increased food intake, body weight gain, appetitive behavior, metabolic, and general behavioral activity, whereas the effects of damaging the hippocampal ventral pole were limited to increased food intake and body weight gain. We suggested that the operation of higher-order, hippocampal-dependent learning and memory processes may underlie the ability of interoceptive satiety signals and perhaps other types of conditional cues to inhibit appetitive and consummatory responding evoked by food and food-related environmental stimuli. Within this model, damage to the hippocampus could therefore interfere with the inhibition of appetitive and eating behaviors. Thus, the question of “how” physiological satiety signals inhibit food intake and reduce body weight gain may be addressed, in part, with reference to learning and memory mechanisms that depend on the hippocampus. As others have suggested, improved understanding of the functional links between the neural controls of food and drug intake and the operation of higher-order learning and memory processes may be key to developing effective therapeutic interventions that can combat obesity (Berthoud, 2002
; Moran and Gao, 2006