Abstract (summary)

Translate
Full text
 TranslateFull text
 Turn on search term navigation
Headnote
Peptide signals regulating food intake and energy homeostasis1

Headnote
Abstract: The adiposity hormone leptin has been shown to decrease food intake and body weight by
acting on neuropeptide circuits in the hypothalamus. However, it is not clear how this primary
hypothalamic action of leptin is translated into a change in food intake. We hypothesize that the
behavioral effect of leptin ultimately involves the integration of neuronal responses in the forebrain
with those in the nucleus tractus solitarius in the caudal brainstem, where ingestive behavior signals
are received from the gastrointestinal system and the blood. One example is the peptide
cholecystokinin, which is released from the gut following ingestion of a meal and acts via vagal
afferent nerve fibers to activate medial nucleus tractus solitarius neurons and thereby decrease meal
size. While it is established that leptin acts in the arcuate nucleus in the hypothalamus to stimulate
anorexigenic neurons that inhibit food intake while simulataneously inhibiting orexigenic neurons that
increase food intake, the mechanisms linking these effects with regions of the caudal brainstem that
integrate cues related to meal termination are unclear. Based on an increasing body of supportive
data, we hypothesize that this integration involves a pathway comprising descending projections from
neurons from the paraventricular nucleus to neurons within the nucleus tractus solitarius that are
activated by meal-- related satiety factors. Leptin's anorexic effect comprises primarily decreased
meal size, and at subthreshold doses for eliciting an effect on food intake, leptin intensifies the satiety
response to circulating cholecystokinin. The location of neurons subserving the effects of
intracerebroventricular administration of leptin and intraperitoneal injection of cholecystokinin on food
intake has been identified by analysis of Fos expression. These studies reveal a distribution that
includes the paraventricular nucleus and regions within the caudal brainstem, with the medial nucleus
tractus solitarius having the most pronounced Fos expression in response to leptin and
cholecystokinin, and support the hypothesis that the long-term adiposity signal leptin and the short-
term satiety signal cholecystokinin act in concert to maintain body weight homeostasis.

Key words: brainstem, adiposity, satiety.

Headnote
Resume: Des travaux ont montre que l'hormone de l'obesite, leptine, diminue la prise alimentaire et le
poids corporel en agissant sur les neuropeptides de I'hypothalamus. Toutefois, on ne salt pas
exactement comment cette action hypothalamique primaire de la leptine peut modifier la prise
alimentaire. Nous emettons I'hypothese que l'effet comportemental de la leptine implique l'integration
des reponses neuronales dans le prosencephale et celles du noyau du faisceau solitaire (NFS) dans le
tronc cerebral caudal, ou les signaux du comportement alimentaire sont rebus du systeme gastro-
intestinal et du sang. Par exemple, la cholecystokine, qui est liberee de l'intestin apres l'ingestion d'un
repas, agit par l'intermediaire des fibres afferentes vagales pour activer les neurones NFS medians et
ainsi diminuer la prise alimentaire. Alors qu'il est etabli que la leptine agit dans le noyau arque de
l'hypothalamus pour stimuler les neurones anorexigenes qui inhibent la prise alimentaire, tandis
qu'elle inhibe les neurones orexigenes qui l'augmentent, les mecanismes reliant ces effets aux regions
du tronc cerebral caudal qui integrent les signaux associes A l'arret de la prise alimentaire ne sont pas
clairs. A l'appui d'un nombre croissant de donnees pertinentes, nous emettons l'hypothese que cette
integration met en cause une vote constitute de projections descendantes des neurones du noyau
paraventriculaire vers les neurones du noyau du faisceau solitaire qui sont actives par des facteurs de
satiete pendant le repas. L'effet anorexique de la leptine consiste principalement en une diminution de
la taille du repas et, aux doses infraliminaires declenchant un effet sur la prise alimentaire, la leptine
intensifie la reponse de satiete A la cholecystokinine circulante. Le site des neurones favorisant les
effets d'une administration intracer6broventriculaire de leptine et d'une injection intraperitoneale de
CCK sur la prise alimentaire a etc identifi6 par l'analyse de 1'expression de Fos. Ces examens revelent
un distribution qui inclut le noyau paraventriculaire et des regions du tronc cerebral caudal, le noyau
du faisceau solitaire median ayant la plus forte expression de Fos en reponse A la leptine et a la
cholecystokinine. Its confortent aussi I'hypothese que le signal d'adiposite de longue duree, leptine, et
le signal de satiete de courte duree, cholecystokinine, agissent de concert pour de maintenir
I'homeostase du poids corporel.

Mots cles : tronc cerebral, adiposite, satiete. [Traduit par la Redaction]

Introduction

It is an intuitive concept that body weight is influenced by the total amount of food consumed in many
individual meals over long periods of time. However, the physiological mechanisms that regulate the
amount of food consumed in a single meal, and how this regulation is functionally related to body
weight (e.g., adipose tissue mass), are only now becoming clear. It is known that the sizes of
individual meals are modulated by signals that are generated in response to eating itself as well as to
adiposity (adipose tissue mass), metabolism, energy expenditure, and a variety of neuroendocrine
influences. A large body of evidence has demonstrated that these signals can be conceptually
separated into two functional categories: short-term signals and long-term signals (Woods et al. 1998,
2000). Short-term signals are generated acutely on a meal-to-meal basis and their effects are short-
lived, primarily influencing the initiation or termination of a meal. Long-term signals are tonic
influences that are related to adiposity and they have a chronic effect on the amount of food
consumed over many meals. Ultimately, the total caloric intake of an individual is the result of the
interactions of long-term and short-term signals over extended periods of time. In the central nervous
system (CNS), these signals are integrated with autonomic and motor mechanisms that regulate
energy expenditure to maintain energy homeostasis via mechanisms involving both the hypothalamus
and the caudal brainstem (CBS). The focus of the present review is the hypothesis that neural circuits
between the hypothalamus neurons that detect adiposity signals and CBS neurons that respond to
meal-related signals are components of a physiological mechanism for the maintenance of body
weight. Since much of the early literature in this field has been reviewed elsewhere (Baskin et al.
1999c; Schwartz et al. 2000; Spiegelman and Flier 2001; Woods et al. 1998, 2000), the present
article emphasizes recent literature related to the above hypothesis.

Short-term signals

Two types of short-term signals related to the consumption of single meals have been proposed: those
that promote the initiation of eating and those that promote termination of eating. The strongest
candidate for a meal initiation signal is the recently discovered peptide ghrelin, which is secreted from
gastric mucosal glands and is proposed to stimulate meal initiation by acting on vagal afferents from
the intestine to the CBS (Cummings et al. 2001). However, it is not known if ghrelin is required for
normal meal onset or if its secretion or its effects on the CBS are influenced by adiposity or play a role
in the regulation of adipose tissue mass. Far more information is available for meal-related signals
that are generated in response to the presence of food in the stomach and intestine, which have the
net effect of causing cessation of feeding. These signals, loosely referred to as "satiety signals", arise
from the pancreas, liver, or gastrointestinal tract. Some of these signals are neurohormonal in nature
and may have access to brain sites within the CBS via the area postrema, where the blood-brain
barrier is patent, or they may cross the blood-brain barrier by specific transport mechanisms. They
also may activate afferent sensory relays (vagal, splanchnic, or spinal nerves) from
the gastrointestinalsystem to the CBS, where integration related to caloric content of the ingested
food takes place. Among these meal-related satiety signals are cholecystokinin (CCK), bombesin, and
the bombesin-related peptides neuromedin B and gastrin-releasing peptide as well as somatostatin,
amylin, enterostatin, glucagon, and more recently glucagon-like peptide 2 (GLP-2) (Woods et al.
2000; Tang et al. 2000). In addition to these peptides, gastric distension sends neural signals via the
vagus to the CBS to induce the termination of a meal (Woods et al. 2000).

Long-term signals

In contrast with short-term satiety signals, long-term "adiposity signals" operate over long durations
with accompanying changes in metabolic rate to maintain a regulated amount of body adiposity.
Adiposity signals are circulating hormones whose concentration in blood are directly proportional to
adipose tissue mass (Woods et al. 2000). These signals inform the brain about the status of adiposity
and changes in their circulating levels are proposed to be a key mechanism for the generation of
adaptive CNS and metabolic responses that restore body weight to a regulated level (Schwartz et al.
2000; Woods et al. 1998). In cases of body fat fluctuations from the norm, depending on the intensity
of the adiposity signal to key receptor sites within the brain, subsequent pathways can be activated
and (or) inhibited that are thought to regulate the amount of food eaten and energy expended to
bring body adiposity to its original state (Woods et al. 2000). The peptides hormones insulin and leptin
and the glucocorticoids have all been implicated as adiposity signals, yet the glucocorticoids are
thought to act through a mechanism different from that of insulin and leptin (Solano and Jacobsen
1999). The roles of insulin and leptin as adiposity signals will, therefore, be the emphasis in the
present article.

CNS site of action of insulin and leptin

In their roles as adiposity signals, insulin (from pancreatic islets) and leptin (from adipose tissue) are
proposed to function as metabolic feedback signals relating changes of body adiposity to
compensatory changes in food intake and energy homeostasis. The net effect of insulin and leptin
acting in the brain is to reduce food intake (i.e., promote anorexia) and stimulate sympathetic nervous
system activity and energy expenditure. The primary site of their action in the CNS is hypothesized to
be the arcuate nucleus (ARC) (Baskin et al. 1999c; Schwartz et al. 2000; Woods et al. 2000), a region
of the mediobasal hypothalamus that has reciprocal connections with multiple hypothalamic and CBS
nuclei that are linked to feeding behavior. While both leptin and insulin are proposed to function as
anorectic adiposity signals to the CNS, leptin is the subject of more recent investigations and is
emphasized here. However, it is likely that insulin's anorectic effects may involve CNS mechanisms
and neural circuits similar to those that mediate the anorectic effects of leptin. Further studies are
needed to examine whether insulin and leptin act on the same neurons within the ARC to affect food
intake and body weight, but it is known that both peptides exhibit decreases in the expression of an
anabolic peptide, neuropeptide Y (NPY), within the ARC (Baskin et al. 1999c; Schwartz et al. 2000;
Woods et al. 2000), which suggests that both may have similar or overlapping CNS mechanisms. Both
insulin and leptin are hypothesized to act via PI-- 3 kinase (Zhao et al. 2000). The work of Niswender
et al. (2001) indicates that the effect of leptin, administered intracerebroventricularly (ICV), is
mediated by PI-3 kinase and suggests along with the work of Spanswick et al. (2000) that both leptin
and insulin may have a similar action on hypothalamic neurons. More studies are needed to determine
whether the effects of both insulin and leptin are mediated through similar CNS circuits.

The importance of leptin in regulating adiposity has been demonstrated convincingly by studies on
experimental mammals that have defects in the ability to synthesize leptin or in leptin receptor
function. The phenotype of the ob/ob mice (unable to synthesize normal leptin) or in animals (db/db,
obese Zucker falfa, obese Koletsky rat) whose defects rest in the inability to synthesize functional
receptors for this hormone is characterized by a profound hyperphagia and obesity (Baskin et al.
1999c; Schwartz et al. 2000; Woods et al. 2000). Some of these mutations have also been identified
in humans. Humans with autosomal recessive mutations in the genes encoding leptin or leptin
receptors are morbidly obese, similar to their ob/ob or db/db counterparts in mice (Barsh et al. 2000;
Baskin et al. 2001). Taken together, these data are indicative of the importance of leptin in the
regulation of energy homeostasis. It has been well documented that leptin replacement into the brains
of either ob/ob or fasted animals (low endogenous leptin) reverses the hyperphagia and obesity
(Woods et al. 2000). Likewise, animals that have streptozotocin-induced insulin deficiency or have
insulin resistance are also hyperphagic. The insulin-deficient condition can be rescued following ICV
administration of low doses of insulin into the brain. ICV administration of insulin fails to reduce food
intake or body weight in the insulinresistant obese (fa/fa) but not lean Zucker rat (Ikeda et al. 1986)
due to an inability of insulin to correct the overexpression of NPY within the ARC of the obese but not
lean Zucker rats (Schwartz et al. 1991). In addition, both insulin and leptin have receptors within
brain areas linked to the control of body weight, and well-characterized transport systems exist for the
uptake of these peptides within the circulation, providing additional evidence that these peptides, of
peripheral origin, are able to act within areas of the CNS that control body
energy homeostasis (Baskin et al. 1999c; Schwartz et al. 2000; Woods et al. 2000).

To explain how changes of body adiposity are coupled to adaptive responses that promote
energyhomeostasis, we propose that the interaction of adiposity signals, such as leptin, with the
short-term satiety signals plays an important role by producing smaller meals and reduced net caloric
intake (Schwartz et al. 2000). According to this hypothesis (Baskin et al. 2001; Baskin et al. 1999c;
Emond et al. 1999; Matson et al. 2000; Schwartz et al. 2000; Wang et al. 2000; Woods et al. 1998,
2000), increased body weight results in increased leptin signaling to the hypothalamus, resulting in
reduced food intake and increased energy expenditure, which together promote weight loss. When
body weight has returned to a regulated level, leptin levels normalize, attenuating the activation of
leptin-sensitive brain circuits that affect meal size. Weight loss and metabolic conditions that are
accompanied by a state of relative leptin deficiency produce the opposite effect, resulting in activation
of CNS circuits that stimulate food intake and decrease energy expenditure, the net effect being
weight gain.

Thus, according to this model, satiety and cumulative food intake can potentially be adjusted over
many meals in proportion to the adipose tissue mass to maintain stable body weight. In order to do
this, the brain needs to receive information regarding adiposity signals (insulin/leptin) and nutrient
status and to integrate this information with shortterm satiety signals that determine the size of
individual meals. The CNS site of integration between the adiposity signals and short-term satiety
signals is not clear, but one likely region would be within the CBS.
CBS integrates short-term satiety signals

Meal-related satiety signals stemming from the presence of food within the gastrointestinal tract
communicate with areas in the CBS linked to feeding behavior (i.e., area postrema (AP), dorsal motor
nucleus of the vagus (DMV), nucleus tractus solitarius (NTS)), primarily via activation of afferent
nerve fibers in the vagus nerve. Humorally or neurally mediated satiety signals control meal size and
are relayed to the NTS, where these vagal fibers terminate. Removal of afferent sensory signals
increases meal size (Chavez et al. 1997). Several observations indicate the importance of endogenous
CCK, acting via the vagus nerve in the control of meal termination. For one, vagotomy and perivagal
application of capsaicin block the ability of peripheral injections of CCK to reduce food intake
(Schwartz 2000). Specific sets of medial NTS neurons are responsive to the presence of food in the
duodenum and stomach, to CCK (Raybould et al. 1985), and to infusion of nutrients into the intestine
(Zittel et al. 1994). Induction of Fos gene expression (a marker for neuronal activation) in neurons of
the NTS during feeding is significantly reduced in the presence of the CCK-A receptor antagonist
devazepide and also in OLETF rats, which lack the CCK-A receptor (Glatzle et al. 2001). Moreover,
gastric loads and CCK, in combination, stimulate the vagal afferent pathway in a synergistic manner
(Schwartz 2000). Although intraperitoneal administration of exogenous CCK potently decreases meal
size, this effect is compensated for during repeated administration of CCK such that the animals
initiate more meals and thereby maintain body weight (West et al. 1984; Moran 2000). Food intake is
not affected beyond the first day following chronic minipump infusion (Crawley and Beinfeld 1983) or
1 week of jugular vein infusion (Lukaszewski and Praissman 1988). Recent studies show that CCK
analogues that have greater bioavailability (Moran et al. 1992) or selectivity than CCK (Asin et al.
1992; Simmons et al. 1999) do not result in meal compensation over repeated administration.
However, these studies allowed only a fixed meal schedule such that compensation via increased meal
frequency over the remainder of the day would not have been possible (Moran 2000). This suggests
that meal-related cues (such as CCK) alone are insufficient to produce long-term changes in
energyhomeostasis under ad libitum feeding conditions.

Leptin enhances sensitivity to satiety signals

Leptin increases the efficacy of the satiety effects of peripheral CCK (Matson et al. 2000; Wang et al.
2000) and potentiates its ability to activate neurons in the NTS (Emond et al. 1999). Conversely,
leptin deficiency brought on by fasting decreased the satiating effect of CCK, suggesting that
physiological variation in leptin levels can modulate the response to meal-related satiety signals
(Baskin et al. 2001; Wildman et al. 2000). Recently, the effects of leptin on short-term food intake
were reported to be blocked by intraperitoneal administration of the CCK-A receptor antagonist
devazepide, a finding that suggests that in some situations, leptin's anorectic effects are dependent on
peripheral CCK action (Buyse et al. 2001). Interestingly, insulin has adiposity signaling effects on food
intake similar to those of leptin, and it also increases the satiety effects of peripheral CCK
administration (Figlewicz et al. 1986). Taken together, these findings support the hypothesis that
pathways from leptin-sensitive hypothalamic circuits must be connected with those in the CBS that are
responsible for adjusting food intake to meal-related satiety signals (Baskin et al. 2001; Schwartz et
al. 2000; Woods et al. 2000). Thus, the circuits mediating the effects of leptin and CCK likely converge
within the NTS to regulate meal size (and perhaps gastric emptying and other effectors of satiety
also). Therefore, future studies aimed at identification of pathways that link adiposity signaling in the
hypothalamus with hindbrain neurons that regulate meal size will be important steps toward
understanding the physiological mechanisms underlying the brain's regulation of body weight and how
they may be defective in disorders of food intake such as obesity and anorexia.
Bombesin and bombesin-related peptides in meal termination

Other short-term meal-related signals that induce satiety include the peptides bombesin, neuromedin
B, and gastrin-- releasing peptide as well as gastric distension and the presence of nutrients within the
duodenum; all act within CBS structures to decrease meal size and evidence suggests that their
effects may also be potentiated by leptin. When leptin is administered ICV to animals that have also
received intraperitoneal bombesin with both peptides given at doses that are subthreshold for an
effect on food intake, a significant suppression of food intake results (Ladenheim et al. 1999a). Both
central and peripheral administration of bombesin appear to act in regions of the CBS to elicit satiety
(Merali et al. 1999). Bombesin injected into the fourth ventricle inhibits feeding at doses 10-1000
times less than those that are effective following lateral ventricular injections (Ladenheim and Ritter
1988; Merali et al. 1999). Lesions of the AP and NTS abolish the suppression of feeding by fourth
ventricular bombesin and attenuate the suppression of feeding by peripheral bombesin (Ladenheim
and Ritter 1993), implicating these brain regions in bombesin's satiety-- inducing action.
Subdiaphragmatic vagotomies are ineffective in attenuating the anorectic effects of peripherally
administered bombesin, whereas capsaicin, which desensitizes primary sensory neurons (including
both vagal and spinal fibers), attenuates the bombesin-induced inhibition of food intake. Other studies
have examined whether bombesin may act through CCK, but the CCK-A antagonist devazepide was
found ineffective in blocking bombesin-induced inhibition of feeding at a dose that completely blocked
the suppression of gastric emptying by bombesin (Ladenheim et al. 1999b). This suggests that the
CCK-dependent capsaicin-sensitive fibers within the vagus are not required for bombesin's satiety
effects.

Gastric and intestinal mechanisms in meal termination

Recent data show that animals reduce intake when given a gastric preload of nutrients to much lesser
degree than a treatment involving the combination of gastric preload and ICV leptin at doses that are
ineffective without gastric distention. Similarly, Fos protein expression is stimulated within neurons of
the rostral, medial, and caudal NTS as well as the AP following the ingestion of food into the stomach.
However, this effect is much greater within the NTS when leptin (at a subthreshold dose) is combined
with food intake into the stomach (Emond et al. 2001). These results suggest that both gastric
distension and the presence of nutrients in the duodenum produce meal-related satiety signals that
are potentiated in the presence of leptin.

Ghrelin as a meal initiator

While more is known about the signals that cause a meal to be terminated, the recently discovered
peptide ghrelin, which is produced by X/A cells in oxyntic glands of the stomach (Date et al. 2000), is
the only hormone released from the gastrointestinal tract that is known to cause an animal to begin
eating. This effect of ghrelin can be reproduced by peripheral administration or ICV administration of
the hormone (Wren et al. 2000). Ghrelin secretion increases adaptively with prolonged fasting and
decreases following gastric infusions of nutrients (Tschop et al. 2000). The action of ghrelin is believed
to be mediated within the CNS, since effective ICV doses of ghrelin are 1000 times less than those
given subcutaneously (Tschop et al. 2000) and since ghrelin receptors are concentrated in brain areas
involved in food intake regulation (Willesen et al. 1999). Ghrelin given either peripherally or centrally
also increases gastric emptying (Masusda et al. 2000) and stimulates gastric acid secretion (Date et
al. 2001; Masuda et al. 2000). ICV administration of ghrelin inhibits vagal firing (Asakawa et al.
2001), stimulates Fos protein expression in ARC NPY and agouti-related protein (AGRP) neurons
(Nakazato et al. 2001), as well as NPY (Nakazato et al. 2001) and AGRP gene expression in these cells
(Shintani et al. 2001). Ghrelin gene expression is increased in ob/ob mice (leptin deficient) and
decreased with administration of leptin or IL-1B (Asakawa et al. 2001), and ghrelin also has been
shown to block the anorexia and weight loss induced by leptin (Shintani et al. 2001) and IL-1B
(Asakawa et al. 2001). Taken together, it appears that ghrelin and leptin are active at opposite ends
of a meal and act, in concert, to alter the release of neural substrates within key areas of the
hypothalamus (ARC) and CBS controlling food intake and energy expenditure.

Ghrelin appears to play a physiological role in meal initiation in humans, but the CNS mechanisms are
not clearly understood. Human plasma ghrelin levels nearly double shortly before each meal and fall to
trough levels within 1 h after food is consumed (Cummings et al. 2001). As evident with exogenous
CCK, ghrelin, given intraperitoneally, is ineffective following vagotomy (Asakawa et al. 2001). Because
many of the actions of exogenous ghrelin oppose those of CCK, it is possible that ghrelin might inhibit
neurons that are sensitive to CCK while stimulating a separate pool of CBS neurons responsive to meal
initiation cues. Ghrelininduced feeding has been abolished by NPY Y1/YS antagonists, anti-NPY
antibodies, anti-AGRP antibodies, as well as alpha-MSH, suggesting that these factors contribute to
the mechanism of ghrelin action (Nakazato et al. 2001). Further work will need to be done in order to
better elucidate specific subsets of neurons sensitive to meal initiation and meal termination cues and
how ghrelin-responsive CNS circuits are influenced by adiposity signals such as leptin.

ARC - paraventricular nucleus (PVN)/ARC lateral hypothalamic area (LHA) axis for adiposity signaling

It is widely agreed that stimulation of neuronal circuits originating in the ARC contributes to the
anorexic effects of leptin on food intake and body weight. Adjacent to the median eminence and third
ventricle, the ARC is strategically positioned such that it can potentially sample changes in leptin levels
in blood and cerebrospinal fluid. Monosodium glutamate (Tang et al. 1999) or gold thioglucose
(Bergen et al. 1998) destruction of the ARC prevents leptin-induced inhibition of food intake (Tang et
al. 1999) and can result in profound obesity in these mice (Bergen et al. 1998), suggesting that the
ARC is required for leptin's effects. Leptin activates the anorexia-promoting catabolic pathway
containing proopiomelanocortin (POMC) and cocaine- and amphetamine-related transcript (CART)
neurons (Mizuno et al. 1997; Schwartz et al. 1997) that project to the PVN and the LHA. Leptin
simultaneously inhibits an ARC anabolic (orexigenic) pathway that originates in the ARC, represented
by NPY/AGRP neuronal projections (Mizuno and Mobbs 1999; Schwartz et al. 1996b, 1998) to the
PVN, and the perifornical area (PFA)/LHA (Elias et al. 1998; Elmquist and Saper 1996; Schwartz et al.
2000; Wilson et al. 1999). Both the NPY/AGRP and POMC/CART neurons in the ARC express ObRb
(Baskin et al. 1999b; Elmquist et al. 1998; Fei et al. 1997; Mercer et al. 1996; Schwartz et al. 1996b),
the "long form" leptin receptor splice variant that is thought as the predominant signaling form of
leptin receptor in the brain. Leptin administration to leptin-deficient ob/ob mice downregulates ObRb
mRNA, whereas ObRb mRNA is upregulated in these mice under-leptin deficient conditions (Baskin et
al. 1998). Fasting increases binding of leptin to leptin receptors in the ARC (Baskin et al. 1999a).
Furthermore, leptin deficiency in ob/ob mice and fasting decreases ARC expression of the SOCS-3
protein (suppressor of cytokine signaling) (Bjorbaek et al. 1999), which is stimulated following leptin
action at its receptor. Exogenous leptin produces parallel changes in these anabolic/catabolic peptides
within the ARC in models of leptin deficiency with no change occurring in models with defective or
devoid leptin receptors (Mizuno et al. 1997; Schwartz et al. 1996a, 1996b). All of these observations
taken together suggest that modulation of leptin activity via ObRb within the ARC is important in
achieving energyhomeostasis.
The PVN is thought to be the most likely candidate for mediating downstream leptin signaling because
of anorectic peptides (coricotropin-releasing -hormone (CRH), oxytocin (OXY)) expressed in these
regions are known to project to CBS regions involved in the regulation of meal size and because PVN
lesions cause profound hyperphagia (Leibowitz et al. 1981). Conversely, it is well known that
electrolytic (Anand and Brobeck 1951) and neurotoxic lesions of the LHA (Stricker et al. 1978) elicit a
hypophagia and weight loss, suggesting that orexigenic responses subserved by LHA neurons are
important in body weight homeostasis. The presence of numerous neurons in the LHA that express
orexigenic peptides (orexins and melanin-concentrating hormone (MCH)) suggests that the LHA may
indeed be a leptin-sensitive anabolic signaling route to the CBS and spinal cord.

The next logical step in research in this field is to identify the neurons and connections that
interconnect leptin-- sensitive neurons in the PVN to neurons in the CBS that respond to meal-related
satiety signals (CCK). Research in this field has been heavily focused on the circuits involved with
leptin action within the hypothalamus, and these hypothalamic mechanisms appear to be well
established (Baskin et al. 1999c, 2001; Elias et al. 1999, 2000; Schwartz et al. 2000). However, it is
still not clear how the leptin-sensitive circuitry originating in the ARC is connected with CBS neurons
that integrate meal-related signals to determine meal size (McMinn et al. 2000a). The control of food
intake and body weight is ultimately regulated by the CBS, which receives a variety of meal-related
and metabolic information that it uses to regulate the amount of food consumed in individual single
meals. Evidence suggests that leptin is one of these signals and that leptin acting in the ARC
ultimately changes the efficacy of meal-related signals impinging on the CBS (especially NTS/DMV) to
influence meal size. The anatomical foundation for this mechanism, however, remains to be
determined.

PVN-CBS axis for adiposity signaling

CRH and OXY appear to be likely candidates to mediate at least some of the downstream leptin
interaction with short-term satiety signals that act within the NTS. Previous studies have identified
direct descending pathways containing CRH from the PVN to the NTS, and since decerebrate animals
(all connections between hypothalamus and CBS severed) are unable to respond to energy deficits (H.
Grill, personal communication), hypothalamus-NTS pathways appear to be required for
energy homeostasis. Furthermore, PVN lesions attenuate the satiety effects of CCK (Crawley and Kiss
1985) and cause hyperphagia and obesity (Leibowitz et al. 1981). CRH and OXY both act through
Gprotein-coupled receptors and both activate neuronal firing. The CRH family of peptides, including
CRH and urocortin II, are found within the PVN (Bittencourt et al. 1999; Reyes et al. 2001) and in
direct descending projections from the PVN to the NTS (CRH). Moreover, these peptides activate
circuits including the NTS (Benoit et al. 2000) and inhibit food intake following central injections.
Urocortin II is correlated with the presence of CRH2 in the NTS (Reyes et al. 2001) and inhibits food
intake following central injections (Reyes et al. 2001), with an efficacy following fourth ICV injection
(near the NTS) that is comparable with responses elicited by injection into the forebrain ventricles.
These findings suggest that CRH2 receptors within the CBS may be involved in food intake and body
weight homeostasis. Leptin, given ICV, has been reported to increase expression of CRH mRNA in the
PVN (Nishiyama et al. 1999; Schwartz et al. 1996b), whereas ICV administration of either a CRH
antagonist (Uehara et al. 1998) or a CRH antibody (Okamoto et al. 2001) attenuates the inhibition of
feeding mediated by ICV leptin administration. Other studies, however, show that conditions that alter
glucocorticoid levels (e.g., glucoprivation, starvation, or adrenalectomy) are accompanied by CRH
release or expression of CRH mRNA within the PVN (Arvaniti et al. 2001; Heiman et al. 1997; Huang
et al. 1998). These findings suggest that distinct subpopulations of CRH neurons within the PVN
respond differentially to leptin, depending on the state of energy balance, leptin, and glucocorticoids.
Taken together, these data are evidence for the involvement of CRH in the mediation of leptin-induced
suppression of food intake, although the mechanisms remain to be elucidated. It remains to be
determined whether CRH or urocortin II containing circuits that originate in the PVN make functional
connections with CBS neurons that show Fos activation by peripheral administration of CCK.

OXY and CCK in the control of food intake

Parvocellular OXY neurons in the PVN are also strategically situated to influence CCK signaling to the
NTS via the vagus. Previous studies have identified direct descending pathways containing OXY from
the PVN to the NTS (Rinaman 1998; Sawchenko and Swanson 1982), and OXY receptor mRNA is
found in the DMV (Verbalis 1999; Yoshimura et al. 1993). Moreover, OXY decreases food intake
following ICV injections (Olson et al. 1991a). Intraventricular administration of OXY produces a
medullary Fos activation pattern similar to peripheral CCK (Olson et al. 1993) and likewise suppresses
food intake (Arletti et al. 1989). OXY neurons in the PVN that are activated by exogenous CCK
innervate the DMV and the NTS (Rinaman et al. 1993), but it is unknown whether these projections in
fact are involved in food intake. OXY administration into the DMV or the NTS also decreases gastric
motility, an action similar to that of CCK (Rogers and Hermann 1992), while central OXY receptor
antagonism or PVN lesions increase gastric motility (Flanagan et al. 1992a, 1992b). Similarly, CCK's
inhibition of feeding is blunted by antagonism of central OXY receptors (Olson et al. 1991b). Together,
these data suggest that a combination of CRH and OXY may be critical in eliciting appropriate
modulation of meal-related signals at the level of the CBS.

One seeming contradiction of this proposed model lies in the hyperphagia that occurs during periods
of lactation, when OXY levels are quite pronounced. However, while it is known that the
neurohypophysis is the site of release of the OXY that is involved with stimulating the release of milk
during this time (Fruend-Mercier et al. 1988; Jiang and Wakerly 1995), it is unclear whether OXY is
also released centrally under these conditions. It is also known that during lactation hypothalamic
NPY/AGRP increases and POMC decreases, as well as serum leptin, which likely provides a mechanism
underlying the hyperphagia that occurs during this period (Brogan et al. 2000).

Synthetic model of adiposity and meal-- related signals

This model describes CRH and OXY as catabolic signals from the PVN to the CBS. Conditions of low
leptin (fast, weight loss) will be integrated at the level of the CBS with meal-related satiety cues to
increase individual meal size. According to the model, neurons in the CBS NTS/DMV that are
responsible for the timing of meal termination would have low excitatory input from CRH/OXY
pathways from the PVN when leptin levels are low as during a fast or following weight loss. Low leptin
levels also result in increased expression of NPY and AGRP in ARC neurons, with a concomitant release
of both peptides within the PVN. NPY (an inhibitory neuropeptide) receptor activation on OXY and CRH
neurons in the PVN is proposed to decrease the catabolic signaling en route to the CBS, thus
weakening the impact of satiety cues to the CBS. Another event occurring simultaneously that would
effectively prevent stimulation of the melanocortin catabolic pathways to the CBS would be an
increase in AGRP expression in the ARC and subsequent release in the PVN, thereby decreasing the
activity of alphaMSH on OXY and CRH neurons within the PVN. Together, these signals would decrease
the PVN-CBS catabolic pathways and result in less satiety and increased meal size. The net result of
the decreased effectiveness of satiety signals to the CBS would be less satiety and larger meals.
In conditions of normal or even elevated adiposity, the subsequent rise in plasma leptin is proposed to
result in stimulating the OXY and CRH catabolic circuits from the PVN to the CBS. As a result, leptin is
proposed to both inhibit NPY/AGRP expression in the ARC and release in the PVN while simultaneously
increasing PVN expression of CRH/OXY in the ARC and subsequent release in the CBS. Increased leptin
is proposed to also increase alpha-MSH release within the PVN, thereby decreasing AGRP expression
and increasing melanocortin signaling within the PVN-CBS circuit. The net result is proposed to lead to
enhanced sensitivity of CBS neurons responding to meal-related satiety cues (CCK), thereby causing
smaller meals and a more rapid onset of satiety.

It is also possible that the orexigenic and anorexigenic peptides have a direct action within the NTS,
thereby not requiring either the PVN or LHA as output projection sites en route to the CBS. The
hypothesis is that POMC/CART from the ARC act directly to stimulate cells responsive to meal
termination cues within the CBS and simultaneously inhibit cells related to meal initiation cues.
Similarly, immediately preceding a meal, NPY/AGRP could act directly within the CBS to activate cells
responsive to meal initiation and inhibit the activity of cells related to meal termination.
Neuroanatomical evidence reveals POMC expression in both the ARC and the NTS and also that alpha-
MSH (found in terminals and cell bodies) expression in the commissural NTS is comparable with the
concentrations found in the ARC (Palkovits et al. 1987). The melanocortin receptors (MC3/MC4) linked
to body weight and food intake are present in the dorsal vagal complex (DVC). In addition, behavioral
studies have shown that MC3/MC4 receptor agonists and antagonists elicit similar effects on food
intake and body weight following lateral and fourth ventricular injections (Grill et al. 1998), indicating
that it is possible that POMC of ARC and (or) NTS origin may stimulate populations of neurons within
the NTS that are activated by meal-related satiety signals. AGRP is also found in fibers and terminals
in both the median and lateral NTS (Bagnol et al. 1999; Broberger et al. 1998). Since NPY and AGRP
are colocalized within the ARC, AGRP found within the NTS, originating from the ARC, will also contain
NPY (Broberger et al. 1998). AGRP has been shown to induce Fos expression in the NTS, suggesting
that the NTS is, indeed, part of an AGRP-activated circuit (Hagan et al. 2001). Together, these results
argue in favor of a direct action of these orexigenic and anorexigenic peptides within the CBS. Further
studies examining whether cells activated in response to meal-related cues are in contact with fibers
containing AGRP/NPY and (or) POMC/CART, originating from the ARC, or are only in contact with
CRH/OXY, originating from the PVN, should determine the contribution of these ARC-NTS and PVN-
NTS pathways.

Recent evidence suggests that leptin may exert a direct catabolic action within the DVC, independent
of hypothalamic input, to decrease food intake and body weight, possibly involving modulation via
norepinephrine (NE), serotonin, NPY, and glucagon-like peptide 1 (GLP-1). Leptin injections directly
within the CBS decrease food intake and body weight, and there is no difference in food intake or body
weight whether leptin is given into the lateral or the fourth ventricle (Grill et al. 2000). Leptin
receptors are also found in all areas within the DVC (AP, NTS, DMV) (Grill et al. 2000) as well as raphe
nuclei within the brainstem (Finn et al. 2001). Although the identity of leptin-- sensitive neurons
within the CBS is unclear, recent studies indicate that ObRb is located within the DVC in all neurons of
the adrenergic C2 regions of the rostral NTS that contain tyrosine hydroxylase (TH) (Schmidt-H. et al.
2001). Neurons within the C2 region contain both NPY and TH, suggesting that leptin may influence
the activity of NPY within the CBS (Schmidt-H. et al. 2001). Electrophysiological studies indicate that
leptin reduced NPY overflow from brain slices of the medulla oblongata as well as the cerebral cortex
and hypothalamus (Lee and Morris 1998). Together, these results suggest that leptin's action in areas
within the CBS may involve both NE and NPY Within the raphe nuclei, the long form of the leptin
receptor has been identified on serotonergic neurons (Finn et al. 2001), suggesting serotonin to be
another potential mediator of leptin action in the CBS. GLP-1 has been found in neurons within the
CBS that contain the long form of the leptin receptor (Goldstone et al. 1997). GLP-1 neurons in the
NTS are also activated by leptin (Elias et al. 2000), suggesting that leptin may act, in part, through,
actions of GLP-1 to elicit changes in body weight.

Although the actions of short-term satiety cues such as CCK in some cases appear to be less effective
in leptin-- deficient states (fasting) (McMinn et al. 2000b; Stein et al. 1986), leptin does not appear to
be required or necessary in order for these cues to be effective. The obese Koletsky rat, which lacks
the long form of the leptin receptor, responds similarly to CCK as lean Koletsky rats, which express
the long form of the leptin receptor (Wildman et al. 2000). Finally, rats that have all connections
between the hypothalamus and CBS severed surgically, thereby eliminating the contributions of leptin-
sensitive neurons in the hypothalamus, show a significant satiety response to CCK (Grill and Smith
1988) or bombesin (Flynn 1992). These findings may indicate that leptin-sensitive pathways
descending from the hypothalamus to the CBS are not necessarily required for animals to respond to
meal-related stimuli.

Conclusions

We describe a model in which the net effect of leptin signaling through the PVN-CBS signaling pathway
is to modulate the size of individual meals so that total caloric intake is adjusted to counter the effects
of changes in peripheral metabolic conditions that are related to energy intake and energy storage in
adipose tissue. Leptin, and likely insulin, are important signals that the brain uses to recognize
changes in body adiposity, and leptin initiates the activity of critical hypothalamic circuits that transmit
information about metabolic status to the CBS, which determines the quantity of food ingested during
a meal. Peripheral or central defects along this adiposity signaling cascade would result in
inappropriate attenuation of downstream signaling in the PVN-- CBS pathway that produces satiety.
Although the LHA may play a role in decreasing meal size, the lack of available information on
catabolic peptides and their connections within the CBS makes it difficult to fit into this model of
adiposity signaling, regulation of meal size, and total food intake. The potential contribution of
connections between the orexin and MCH neurons in the PFA and LHA and downstream sites in the
CBS to the regulation of meal size is an open question that needs investigation. If this information
does become available at a later time, the relative contribution of pathways between the ARC, LHA,
PFA, and CBS can be incorporated into this model. Thus, this unified model provides a conceptual
means for conceiving how the brain responds to alterations in energy homeostasis and makes
appropriate changes in food intake to maintain body weight over long durations. It also gives us a
feasible model for use in studying the potential defects in neural adiposity signaling pathways that
could contribute to alterations in feeding and obesity in humans.