Cholecystokinin

Margery C. Beinfeld

Since its discovery in the brain, anatomical, physiological and pharmacological research on cholecystokinin (CCK) and its possible involvement in neurological and psychiatric disease has continued unabated. In just the last three years, over 1400 papers have been published on CCK. Much of this excitement has to do with CCK's colocalization with dopamine and its ability to modulate mesolimbic, mesocortical and nigral striatal dopaminergic neurotransmission. The presence of abundant CCK in the cortex and hippocampus colocalized with GABA and the CCK projection from prefrontal cortex to striatum has additional implications for a possible role in anxiety, attention deficit disorder, and in the negative symptoms and cognitive deficits of schizophrenia. This review will attempt to integrate this large body of knowledge (while providing only selective references) and point out future directions of investigation.

The discovery of the hormone which would eventually be called CCK dates to the beginning of the 20th century and the experiments of Bayliss and Starling and culminated with the sequencing of CCK 33 by Mutt and Jorpes in the 1960s (106). CCK was found to share the same 5 carboxyl-terminal amino acids with gastrin, a peptide isolated from stomach in 1964 (56). Gastrin has a much more limited distribution than CCK; it is present mainly in endocrine cells in the stomach (93). Gastrin and CCK arise from separate prohormones, although the many similarities between them suggest that they arose from a common precursor early in evolution.

The confusion over the chemical nature of the CCK-like material in brain started when Vanderhaegen and co-workers, using a gastrin antiserum that cross-reacted strongly with CCK, detected abundant CCK/gastrin-like material in the brains of several vertebrate species, including man (162).

Based on its chromatographic behavior, biological activity, and amino acid sequence analysis (49), it was soon clear that this material in brain was mainly the carboxyl-terminal amidated peptide CCK 8. The sequences of CCK 8, gastrin, and caerulein (a CCK-like peptide from the skin of the frog Hyla caerulea) are shown in Table 1).

Our knowledge of the structure of pre-pro-CCK was greatly enhanced by the cloning of the cDNA for the rat sequence (44). This was followed, in rapid succession, by the cloning of the porcine and human cDNAs. The human CCK gene is found on the third chromosome, while the gastrin gene is on the 17th (89). The transcription unit spans 7 kb and is interrupted by two introns. The promoter region is within 144 bases 5' to the transcription start site (43).

CCK has a long evolutionary history going back to Hydra, and it is thought that CCK and gastrin probably evolved from a common ancestral proto CCK-gastrin prohormone. This similarity, which is very high in the biologically active peptide, also extends to the structure of the prohormone and the fact that multiple biologically active amidated peptides with variable amino terminal extensions are made and released. The sequence of CCK 8 is highly conserved across mammalian species.

Rat pre-pro-CCK is shown schematically in Figure 1) with the known cleavage sites (inferred from products isolated and sequenced from brain and intestine). In the lower portion of the figure, a model of the temporal cleavages which produce CCK 8 is presented.

The cloning of the rat CCK gene has revealed that the promoter contains a number of regulatory elements all located within 100 bp of the TATA box including a putative basic helix-loop-helix leucine zipper element, an SP1 element, and a combined cyclic AMP and TPA response element (61,110). Reporter gene studies indicate about a six-fold increase in expression by addition of forskolin and about a two-fold increase with TPA (tetradecanoylphorbol 13-acetate).

The cAMP element appears to be active, as treatment of most CCK-secreting endocrine cells in culture with forskolin increased CCK mRNA expression (90). Treatment of some endocrine tumor cells with phorbol esters also increases their CCK mRNA content (104). CCK mRNA levels in rat brain are altered by estrogens (medial preoptic area, amygdala, bed nucleus stria terminalis) [97], caffeine (striatum) [143], N-methyl-D-aspartate [NMDA] (striatum) [47], opiates (spinal cord and hypothalamus) [46] and cocaine and benzotript (striatum) [48]. Kainic acid-induced seizures increased CCK mRNA and peptide levels in the cortex and hippocampus (114). Repeated electroconvulsive shock increased the number of neurons which were positive for CCK and substance P mRNA in the Edinger-Westphal nucleus (85). This may be related to the antidepressive and analgesic effects of electroconvulsive therapy. Lesioning the medial forebrain bundle (142) and the sciatic nerve increases CCK mRNA expression in striatum and dorsal root ganglia, respectively (172). Acute administration of the anxiogenic benzodiazepin inverse agonist FG 7142 (N methyl-b-carboline-3 carboxamide) increases CCK mRNA levels in basolateral amygdala and the CA3 pyramidal cell layer of the hippocampus, both areas implicated in anxiety (121).

CCK displays an unusually high degree of tissue heterogeneity, and this appears to be species-dependent. The major tissue difference is in processing of pro-CCK in brain and gut.

The major carboxyl-terminal amidated peptide in brain is CCK 8. Some CCK 4 has also been detected in rat brain (124). In gut, larger forms like CCK 22, 33, 39, and 58 are more abundant than CCK 8. The major forms of CCK in plasma appear to reflect the forms present in the gut. Larger forms like CCK 33, 39, and 58 have also been identified in the brains of some mammals, as have their amino-terminal peptides minus CCK 8. Small amounts of carboxyl-terminally extended peptides have been identified in brain, predominately 33 and 8. These forms are much more abundant in the gut and parallel the amidated forms in size.

Plasma levels of CCK are low, and their measurement by RIA is complicated by the cross-reactivity of many of the CCK antisera with circulating gastrin. A sensitive and specific bioassay has been developed which has been used successfully to document changes in plasma CCK levels under normal and pathological situations (84).

CCK is very abundant in the brain, more so than in the gut. To date, the only neuropeptide which is more abundant than CCK is neuropeptide Y (NPY).

CCK levels are very high (> 4 ng CCK/mg protein) in cerebral cortex, caudate-putamen, hippocampus, and amygdala, while thalamus, hypothalamus and olfactory bulb are lower (1–2 ng/mg protein). The pons, medulla and spinal cord have even lower levels (<1 ng/mg protein), while CCK is barely detectable in the cerebellum (11).

The anatomy of central CCKergic projections is very complex and has been the subject of intense investigation since its discovery. The ability to visualize CCK mRNA has provided much additional information about the location of CCK cell bodies.

Many brain regions like the cortex and hippocampus contain both CCK-positive interneurons and a mixture of afferent and/or efferent cells and terminals. Certain regions like the striatum, nucleus accumbens and olfactory tubercle have abundant CCK terminals, but few CCK-positive cells.

CCK cells are abundant in three sexually dimorphic nuclei in the rat forebrain (the central part of the medial preoptic nucleus, the encapsulated part of the bed nucleus of the stria terminalis, and the posterodorsal part of the media nucleus of the amygdala) [148]. CCK mRNA levels are altered by estrogen in these areas (97).

A number of major CCKergic projections have been identified in rat brain.

Nigral-striatal and cortico-striatal: The striatum contains very few CCK-positive cells, although the level of immunoreactive CCK peptides is among the highest in the brain. Lesion studies indicated that the bulk of this CCK originates outside the striatum. Although CCK is present in some dopamine cells in the substantia nigra which project to the striatum, these cells provide only a small proportion of its CCK. The degree of co-localization of CCK and dopamine in the nigra in humans is much lower than in rats (116). Significant co-localization of CCK and dopamine has been detected in the brains of medicated schizophrenic patients (137,138), suggesting that notable individual variations or alterations in disease states are possible. Subsequent experiments have indicated that a large amount of the immunoreactive CCK in the striatum originates in cerebral cortical cells (95,105). Much of this CCK comes from the prefrontal cortex. This is also a GABAergic projection, as CCK is colocalized with GABA in the cortex (62). Some of the CCK-containing cells in the nigra also project to the amygdala.

Mesolimbic and mesocortical: CCK cells in the ventral tegmental area (VTA), some of which also contain dopamine (66) and neurotensin project to specific subdivisions of the nucleus accumbens, olfactory tubercle, and pre-frontal cortex (176). The degree of co-localization of CCK and dopamine in the primate mesocortical projection is much lower than in rats (111). Recent studies demonstrate that dopaminergic (and presumably CCKergic) cells in the VTA innervate the basal forebrain cholinergic neurons which supply acetylcholine to the cortex and have been implicated in the pathophysiology of Alzheimer's disease (53).

Thalamo-cortical and thalamo-striatal: The thalamus has moderate levels of CCK immunoreactivity, but relatively few peptide-positive cells are visualized. With in situ hybridization, many more neurons are identified which contain CCK transcripts, from 100% in the medial geniculate or ventral lateral nucleus to none in the ventrolateral geniculate. A high percentage of the cells (which are not GABAergic) project topographically to sites in the cortex and striatum (24). Some of the CCK neurons which project to cortex are reactive to vasoactive intestinal peptide (VIP) [112]. There is also a major corticothalamic CCK projection (146).

Ascending pathways: CCK neurons project from the dorsal medial nucleus tractus solitarius (nts) and the outer rim of the area postrema to the parabrachial nucleus (63). From the parabrachial nucleus, an additional set of CCK neurons project to the ventromedial nucleus of the hypothalamus (177). Other CCK neurons in the nts project to the nucleus accumbens (169). CCK neurons in the raphe which do not contain 5-HT project to the forebrain (161). CCK is abundant in amacrine cells in the retinas of some species (80) and also has been found in retinal ganglion cells (21). CCK cells in the inferior olive project to the inferior colliculus, while CCK cells in the superior colliculus project to dorsal lateral geniculate, and CCK neurons in the dorsal lateral geniculate project to visual cortex. CCK neurons in the trigeminal ganglion project to the pial arteries and innervate the iris (12).

Hippocampal-septal: Lesions of the fornix caused a significant decrease in CCK levels in the lateral septum, bed nucleus of the stria terminalis, mammillary bodies, anteroventral nucleus of the thalamus, and the subiculum (59). These results suggest that there is a significant efferent CCKergic projection from the hippocampus. The bed nucleus of the stria terminalis also receives projections from the amygdala via the stria terminalis.

Olfactory intrabulbar associational system: CCK is the transmitter of the tufted cells in the olfactory bulb which project to the internal plexiform layer of the ipsilateral bulb (87).

Magnocellular/parvocellular hypothalamic-hypophyseal: CCK-containing neurons in the paraventricular and supraoptic nuclei, some of which contain oxytocin and some CRF (96), project to the median eminence and posterior lobe of the pituitary (117).

Descending projections to spinal cord: CCK neurons in the periaqueductal gray, Edinger-Westphal nucleus, and ventral medulla project to the spinal cord (150). Spinal cord: The use of mRNA hybridization revealed small to medium cells in layer II, III and X and motoneurons in layer IX of cervical, thoracic, and lumbo-sacral spinal cord (141), as well as in motor trigeminal and hypoglossal nuclei (32). CCK levels in spinal motoneurons are increased by dorsal rhizotomy (159).

CCK has been found to be colocalized with a number of "classical transmitters" and other neuropeptides with a distinct distribution. This information is summarized in Table 2.

Significant progress has been made in the elucidation of the mechanism and enzymology of pro-CCK cleavage using endocrine tumor cells which process pro-CCK to CCK 8 (8). CCK biosynthesis takes place in the context of the regulated secretory pathway. Drugs which disrupt the Golgi (Brefeldin) or alter acidification of secretory granules (ammonium chloride) drastically reduce or completely eliminate the pool of processed CCK available for secretion (Beinfeld, unpublished observations).

After insertion into the endoplasmic reticulum (ER) and removal of the signal peptide, pro-CCK assumes its correct, three-dimensional structure. From there it is probably not further modified until it reaches the trans-Golgi. Pro-CCK has no consensus sequence(s) for N- or O-linked glycosylation, and there is no indication that it is glycosylated.

In the trans-Golgi, three out of four of the tyrosine residues of pro-CCK are sulfated by a specific membrane-bound tyrosine sulfotransferase (109). The sulfation of one of the tyrosine residues (in CCK 8) is important for its biological activity at CCK A receptors. Recent evidence suggests that sulfation of the tyrosines in pro-CCK allows for correct sorting and/or processing of pro-CCK in an endocrine cell in culture (9). This is probably also true for gastrin.

In the trans-Golgi network, sulfated pro-CCK is sorted into secretory granules with other material destined for the regulated secretory pathway. The structural features of pro-CCK (or any proneuropeptide) that allow it to be recognized as secretory material are unknown. Attention has focused on the amino terminal domains of these proteins, and it has been proposed that hydrophobic, amphipathic a-helical domains may represent sorting signals (78). A membrane-bound form of carboxypeptidase E has been proposed as a sorting receptor for neuropeptide prohormones (31).

A serine in the carboxyl terminal extension of progastrin is known to be phosphorylated by casein kinase II (163). The carboxyl terminal extension of pro-CCK contains a similar serine which may be phosphorylated, although this has not been examined directly. The phosphorylation of this serine may play a role in regulating the processing of gastrin, because decreased phosphorylation of canine gastrin was associated with decreased conversion of glycine extended forms to amidated gastrin (164).

Based on the products isolated from brain and endocrine cells in culture and some other observations on both CCK and gastrin processing, the most probable model of the processing of pro-CCK to CCK 8 is depicted in Figure 1 and involves the following steps: 1) signal peptide cleavage (by signalase in ER membrane); 2) cleavage of pro-CCK at a single arginine site to release carboxyl terminal extended CCK 8; 3) removal of the carboxyl-terminal extension by cleavage at an ArgArg site; 4) the possible action of a carboxypeptidase E activity to remove extra arginines on the carboxyl terminal to produce CCK 8 Gly; 5) conversion of CCK 8 Gly to CCK 8 amide by the amidating enzyme. The amino terminal of pro-CCK is further processed, particularly at the monobasic site, which would yield CCK 58. Whether this cleavage occurs before or after the removal of carboxyl-terminal extended 8 is unknown.

The processing of pro-CCK is somewhat unusual because most of the cleavage sites occur at single Arg residues. This is not true for pro-gastrin, although many other features of the prohormone structure have been retained in evolution.

The enzyme(s) responsible for the monobasic cleavages in vivo has not been definitively identified. Three enzymes have been identified which will cleave CCK 33 to produce CCK 8. 1) A serine protease called CCK 8 generating enzyme has been extensively purified from rat brain synaptosomes (166). This enzyme will also cleave CCK 33 at the single lysine residue, generating CCK 22. It has a very broad pH optimum, with a maximum at 8, and is highly specific, requiring the sulfated tyrosine or a suitably charged residue in the same position in CCK 8 for cleavage. The cloning of this enzyme is in progress. 2) Yeast aspartyl protease 3 (YAP3) cleaves a number of substrates and has the ability to cleave both single basic and dibasic sites. Like the serine protease from rat brain, YAP3 will cleave CCK 33 at a Lys-Asp bond to produce CCK 22 (26). 3) Recombinant PC2 will produce CCK 8 but not CCK 22 (167). From these observations, it is clear that there is nothing "special" about the single arginine cleavage of CCK 33 to produce CCK 8, as two enzymes with known abilities to cleave dibasic sites can also cleave at these single arginine or lysine sites.

It is likely that the subtilisin-like proteases PC1 (145,152) and perhaps PC2 (153) are involved in the processing of pro-CCK. They are widely distributed in the brain and are present in a number of endocrine cells (175) which express CCK mRNA and correctly process pro-CCK to amidated CCK 8 (8). The fact that the AtT20 anterior pituitary cell line (which makes PC1 but not PC2) has the ability to correctly process pro-CCK to amidated CCK 8 (when stably transfected with the CCK cDNA) suggests that PC2 is not required for CCK 8 processing. Experiments in which the effect of expression of anti PC1 and PC2 mRNA on pro-CCK processing in stably transfected RIN5F (rat insulinoma cell) and STC-1 (mouse intestinal cell) confirm a role of PC1 in producing CCK 8. These cells also make and secrete some amidated CCK 22 in addition to CCK 8. Inhibition of PC2 expression causes a comparative depletion of CCK 22 while sparing CCK 8. That inhibition of PC1 or PC2 expression can shift the ratio of CCK 22 to CCK 8 in these cells suggests that CCK 22 and CCK 8 arise by different pathways, and they are not interconverted. It also suggests that the key to tissue-specific differential processing lies in tissue differences in peptidase expression or activity.

The mammalian equivalent of YAP3 has not yet been identified, but an antibody generated against YAP3 stains neurons in peptide-rich areas of the rat brain, including CCK-positive cells in the cortex and hippocampus (27).

The turnover of CCK 8 in rat brain in vivo has been measured by intraventricular injection of labeled amino acids or sulfate, followed by HPLC analysis of the labeled products. The peak of incorporation of label into CCK 8 occurs after about 4 hours. The half-life of turnover of labeled CCK 8 in the brains of intact rats is about 16 hours (94).

The enzyme(s) responsible for the degradation of CCK 8 after it is released has not been definitely identified, but a likely candidate has recently been isolated and cloned (136). There is no evidence that de-sulfation is the route of degradation, as little non-sulfated CCK 8 has been isolated from brain. Early experiments indicated that enkephalinase (E.C. 3.4.11.7) might be involved in the degradation of both enkephalin and CCK (45). This is a logical candidate, as it has been localized to pre- and post-synaptic membranes (7!popup(ch56ref7)). More recent studies have indicated that either a thiol (92) or a serine protease (135) in concert with an aminopeptidase (99) are responsible. The serine peptidase has been studied in detail (136) and is a membrane bound isoform of tripeptidylpeptidase II (EC 3.4.14.10). It cleaves CCK 8 fairly specifically between the Met and Gly and the Met and Asp bonds. A specific "peptoid" (meaning peptide-like but not containing natural amino acids) inhibitor with good bioavailability and ability to penetrate the brain has been developed with a Ki in vitro of 9 nM. This inhibitor protected CCK 8 released from slices and potentiated CCK effects on gastric emptying and the satiety effect of endogenously released CCK (136).

Another route of CCK inactivation may involve uptake. Tritiated propionyl CCK 8 was specifically taken up unaltered by cortical synaptosomes by an energy-dependent transport process with a Km of 10.7 nm (101).

The release of CCK is relatively easy to study because of its abundance. Most studies have focused on release of CCK from brain slices in vitro. Some studies used synaptosomes and, more recently, some push-pull and microdialysis.

Like other neuropeptides, the release of CCK can be elicited with potassium stimulation, veratridine, or an electrical field. Unlike the catecholamines, CCK requires a relatively strong stimulus for release, 40–60 mM potassium, compared with 25 mM for dopamine (58). Even in the presence of 60 mM potassium, CCK release can be modulated up or down by pharmacological agents (3-5).

The fractional release of CCK from brain slices varies from about 2% in the cortex and hippocampus to less than 1% in the caudate-putamen. The lower fractional release from caudate-putamen has been attributed to the tonic inhibition of an unknown substance that is released by a calcium-dependent mechanism, along with CCK, from a number of brain regions, but whose effect is most apparent in the caudate-putamen (58).

Studies of the effect of different pharmacological agents on the release of CCK from synaptosomes, brain slices or intact rats indicated that the modulation of CCK release was region-specific. Most of the studies found that dopamine, acting through D1 receptors, stimulated CCK release from the caudate, cortex and hippocampus (22). Numerous studies convincingly demonstrate that GABA inhibits CCK release from cortex (including human synaptosomes) (126) and spinal cord, while excitatory amino acids increase it (173). Morphine decreased CCK release from the hypothalamus (98). Serotonin increased CCK release from cortical and nucleus accumbens synaptosomes (118), while 5-HT antagonists decreased veratridine-induced CCK release from rats in vivo (127).

Psychotomimetic drugs like phencyclidine (PCP) inhibit CCK release from striatal slices (5!popup(ch56ref5)). A recent study showed that infusion of CCK 8 into the rostral nucleus accumbens antagonized EEG and behavioral effects induced by PCP, supporting a role for CCK as an endogenous antipsychotic agent (120). Pre-incubation for 40 minutes with lithium increased the subsequent potassium-evoked CCK release (57).

A few studies have evaluated the effect of alteration of intracellular mediators on CCK release. Incubation with phorbol esters causes a 2–3 fold increase in potassium-evoked release from cortex, hippocampus, and caudate-putamen (4). Incubation with forskolin + IBMX (which activates adenylate cyclase) also produced a small increase in potassium-evoked CCK release from cerebral cortical, as well as caudate slices (10). A similar result was seen with primary cortical cells in culture.

A vast literature exists on the pharmacology and physiology of CCK. By the early 1980s, it was clear that two major sub-types of CCK receptors existed. Both of these have equally high affinity for CCK 8 sulfate, while they differ substantially for unsulfated CCK peptides, gastrin and amidated peptides shorter than CCK 7, like CCK 4 and CCK 5 (also called pentagastrin). The CCKA subtype, typically found in the pancreas, is relatively specific for sulfated CCK 8; unsulfated CCK 8, CCK 4 and gastrin are 2–3 orders of magnitude less potent. For the CCKB subtype, the difference in potency between sulfated CCK 8, unsulfated CCK 8, gastrin and CCK 4 is about one order of magnitude. The CCKB subtype appeared to be identical to the gastrin receptor. In April 1992, in the same issue of the Proceedings of the National Academy of Science, the cloning of the rat pancreatic CCKA receptor (170) and the canine parietal gastrin receptor (79) were reported. Both appear to be classical seven-domain, membrane-spanning receptors which are homologous to the ;=+gb-adrenergic receptor. The CCKA and B receptors are 48% identical to each other and code for a protein of about 450 amino acids. They contain potential sites for N-linked glycosylation and serine phosphorylation.

Subsequently, it was found that the gastrin receptor in the stomach was the same as the CCKB receptor found in brain. The CCKB receptor is also found in the pancreas, gall bladder, and bowel. Unlike other neurotransmitter receptor systems (e.g., acetylcholine, serotonin and somatostatin), the CCK system, with only two receptor subtypes, appears to be a model of simplicity.

The CCKB receptor expressed in COS-7 cells responds to the addition of CCK by mobilizing internal calcium, suggesting that CCK's effect is mediated by the activation of phospholipase C (79). This mechanism has been previously established in the pancreas. Similar results have been observed in cultured striatal neurons, also using fura-2 and calcium imaging (103).

The distribution of the CCKA and CCKB subtypes in brain is species-specific. In general, the CCKB receptor predominates in the brain, although significant populations of CCKA receptors are also present, and a number of physiological effects of CCK in the brain are mediated by CCKA receptors. Involvement of CCK subtypes in specific behaviors is frequently species-dependent. CCKA receptors have been shown to be most abundant in the nucleus tractus solitarius, dorsal motor nucleus of the vagus, area postrema, interpeduncular nucleus, supraoptic, paraventricular, dorsomedial, infundibular, submammilary nuclei and mammillary bodies of the hypothalamus, nucleus accumbens, VTA, substantia nigra, caudate and ventral pallidum by receptor autoradiography in the rat (25,65,68.) CCKB receptors, which are much more widely distributed and overlap with CCKA receptors, are abundant in the cortex, caudate, brain stem and amygdala (25,68).

The development of a number of selective, potent antagonists, some of which penetrate the blood-brain barrier, has greatly enhanced our understanding of the pharmacology and physiology of CCK. The discovery of asperlicin (29) [a potent, non-peptide, benzodiazepine-like metabolite of Aspergillus alliaceus] set the stage for the new round of discovery. These antagonists have been derived from cyclic nucleotides, amino acids, CCK and gastrin, benzodiazepines, quinazolinone, and diphenylpryzolidinone derivatives (122). The affinities of some of these antagonists for CCKA and B receptors are summarized in Table 3.

Porcine CCK 33, isolated by Viktor Mutt, was the first material available for studies of the biological activity of CCK. Amidated sulfated CCK 7 is the smallest CCK peptide with full biological activity, while larger amidated, sulfated peptides like CCK 33 are also fully active. The frog skin peptide caerulein is as active as CCK 8 and may be somewhat more degradation resistant because of its amino terminal pyroglutamate residue. The chemical synthesis of sulfated CCK 12 was reported by Ondetti and co-workers at Squibb (115) in 1970. They provided synthetic CCK 8 to investigators for years until it became commercially available in the early 1980s. The synthesis of CCK is complicated by the difficulty and low efficiency of the addition of the sulfate group to the tyrosine after conventional peptide synthesis with t-boc (t-butyloxycarbonyl) amino acids. Recent improvements in peptide synthesis, in which the sulfated tyrosine is incorporated into the peptide backbone as an FMOC (9-fluorenyl-methoxycarbonyl) amino acid, has allowed the synthesis of human and porcine CCK 33 (119) and canine CCK 58.

CCK 8 sulfate and caerulein are the most commonly used non-selective CCK agonists. A number of synthetic agonists specific for the CCKB receptor have been synthesized. BC 264 (Table 3) is the most potent and widely used.

The pharmacology and physiology of CCK have been studied in detail. The biological actions of CCK have been reviewed recently (35). This review will deal solely with the physiology and pharmacology of CCK in the brain. CCK is well established as a gastrointestinal hormone, but determination of the precise role of CCK in the brain has been more elusive. A large and rapidly expanding body of literature supports a neurotransmitter role for CCK. Some the animal studies have yielded conflicting results, but differences in the behavioral measures and the emotional or motivational state of the animal may be partly responsible. It is important to remember that CCK release requires a stronger stimulus than dopamine release, so it may be more pronounced when animals are stressed or activated.

Major questions remain about the site(s) of action of CCK agonists and antagonists after intraperitoneal, intravenous, or subcutaneous administration. It is unlikely that peripherally administered CCK 8 is entering the brain in significant quantities (36): the penetration of some CCK 4 or CCK 5 is more likely. The possibility that small amounts of CCK 8 could be entering the brain in areas where the blood-brain barrier is leaky cannot be excluded. However, it is clear from many studies that peripherally administered CCK activates specific neuronal populations in areas where the blood-brain barrier is not leaky, such as the midbrain (151) or the hippocampus (38). Peripheral administration of CCK is known to activate ascending pathways which use the vagus, the nts, area postrema, leading to the hypothalamus. Some of the antagonists, on the other hand, do penetrate the blood-brain barrier fairly well. The CCKA antagonist devazepine penetrates with an uptake index about one-half of that of diazepam (123). The CCKB antagonist CL-988, given subcutaneously or orally, blocks behavioral effects of centrally administered CCK (69).

A large body of evidence indicates that CCK is an excitatory neurotransmitter or neuromodulator, usually activating specific cells directly. In some cases, the action of CCK can be blocked by antagonists of other neurotransmitter systems like DA, 5-HT , GABA or endogenous opiates, suggesting that CCK is releasing other agents which are excitatory or, in some cases, inhibitory (2). CCK excites neurons in the periaqueductal gray (86), dorsal raphe (15), nucleus accumbens (174), hippocampus (18,38,71), ventral medial hypothalamus (15), substantia nigra (130), thalamus (33), and spinal cord (72). It depolarizes oxytocin-containing neurons in the supraoptic nucleus of the hypothalamus (88) and serotonin-containing neurons in the dorsal raphe nucleus.

CCK can excite specific neuronal pathways. Two recent examples of this are outlined below.

Thalamic reticular nucleus: The thalamic reticular nucleus (nrt) has dual innervation from corticothalamic and thalamocortical relay cells, both of which contain CCK (24). Neuronal activity in these circuits affects sensory processing and the level of behavioral arousal, while abnormal activity may underlie generalized absence epilepsy (33). CCK evokes prolonged spike discharge in nrt neurons which is associated with an increase in input resistance by suppressing a potassium conductance (33).

Visual system: CCK cells which originate in the superior colliculus innervate the dorsal lateral geniculate body (dlgb). Iontophoretically applied CCK, acting through CCKA receptors, alters the bursting pattern of dlgb neurons which are activated by light, which suggests that they may be involved in control of retino-cortical transmission (39).

The ionic mechanism by which CCK excites neurons is generally either a decrease in potassium conductance or an increase in a non-selective cation current (14). In some preparations, such as the nodose ganglion, CCK can produce a fast depolarization, followed by a slow response caused by blockade of a potassium current. A recent study demonstrated that in acutely dissociated dopaminergic substantia nigra neurons which are excited by CCK, this response is mediated by Gaq and Ga11 (171).

CCK increases excitatory amino acid release in the hippocampus (100). Most studies indicate that CCK increases GABA release (128) while it decreases dopamine release (54).

The role of CCK in modulating opiate analgesia has been reviewed recently (157). In rats, CCK acting on CCKB receptors reduces m-opiate-mediated analgesia. In primates, CCK acting through CCKA receptors has similar actions in the spinal cord. CCK may be acting presynaptically to antagonize m-opiate action by opposing m-opiate-mediated suppression of the rise in internal calcium concentration produced by depolarization of c-fiber terminals. This results in a decreased release of glutamate and neurokinins presynaptically. CCK may also be opposing opiate action by a similar mechanism postsynaptically. CCKB antagonists increase the potency of spinal morphine and tend to reverse morphine tolerance (50). Further, changes in the levels of CCK or CCK release may participate in alterations of the effectiveness of morphine in inflammation or neuropathic pain models. CCK antagonists could thus be used to enhance morphine analgesia. The use of CCKB antagonists paired with enkephalin catabolism inhibitors in the management of pain and drug addiction was reviewed recently (134).

CCK released from the duodenum after a meal is thought to be a physiological regulator of food intake (147). This satiety-inducing effect of CCK was first reported in 1973, before the discovery of CCK in the brain (55). Peripheral CCK appears to act additively with other hormones like bombesin, somatostatin, calcitonin, and glucagon and with absorbed nutrients to produce termination of a meal, followed by the inter-meal interval. The precise mechanism by which CCK elicits satiety is not known, but decreased gastric emptying, production of hyperglycemia, and antagonism of the opiate feeding system are just a few possibilities. This "peripheral" action of CCK is thought to be mediated by CCKA-type receptors located in the stomach, as sectioning of the gastric vagus prevents the satiety effect of peripherally administered CCK (154). Studies of the role of CCK in satiety have been confounded by significant species and experimental paradigm differences. Studies of the effect of 5-HT antagonists on CCK-induced satiety suggest that central 5-HT receptors are involved in CCK-induced satiety (156)).

CCK release in the hypothalamus is increased after feeding in both rats and primates (139), and central administration of CCK causes satiety (140). The fat/fat mouse (which is obese, diabetic and sterile), whose primary genetic defect is loss of carboxypeptidase E (107), has about 20% of the whole brain CCK and about 60% of the duodenal CCK of its heterozygous littermates. (Cain, Wang, and Beinfeld, in preparation). This partial loss of CCK may contribute to the maturity-onset obesity of this mouse strain.

In several well controlled studies, CCK (at doses which produce the same circulating levels of CCK observed after a meal and which produce no adverse side effects) decreased food intake in experimental animals (reviewed in reference 35) and in humans (77). It is likely that CCK is acting both centrally and peripherally in a complex fashion with other factors like NPY, insulin, glucagon, serotonin, melanocortin and leptin in the regulation of feeding.

The corticostriatal pathway from the prefrontal cortex to the striatum is a major pathway implicated in attention deficit disorder (ADD) and in the negative symptoms and cognitive deficits of schizophrenia. CCK is very abundant in the cortex and striatum and is probably a major transmitter in this pathway. In a recent study, unmedicated schizophrenic patients had significantly fewer D1 receptors (113). Since CCK release is strongly influenced by D1 receptor activation (22), the CCK release in these patients may be significantly decreased. The spontaneously hypertensive rat (SHR), which is considered a possible model for ADD, has CCK levels which are lower relative to normotensive controls rats. In addition, the ability of CCK to facilitate release of dopamine from the nucleus accumbens was reduced in SHR (76). Dopamine agonists (e.g., apomorphine) which increase CCK release are known to improve cognitive function in rat and primate models. In a double-blind, cross-over study of normal men, caerulein improved selective attention tasks (144).

Feeding mice following a training session has been shown to increase memory (51). This effect is mimicked by injection of CCK 8 and is blocked by cutting the vagus (52) and by administration of a CCKA antagonist. CCKB agonists prolong LTP in the CA1 region of the hippocampus (6). CCKA agonists enhance olfactory recognition (a model of learning in rats) while CCKB agonists disrupt it (83).

The discovery that CCK and dopamine are colocalized in the mesolimbic and mesocortical systems has inspired many studies of the interaction of these two transmitters. Numerous studies have examined their ability to alter each other's release, receptor binding and pharmacology. Although some of the results are conflicting and reveal that the pharmacology of CCK in these regions probably involves both CCKA and B receptors, the overall picture that has emerged is that CCK is an endogenous antagonist of dopamine. In the substantia nigra and VTA, CCK by itself is excitatory (67,151), but it potentiates the inhibitory action of dopamine or apomorphine (67,151). In the nucleus accumbens, CCK increased the firing rate of neurons activated by glutamate but antagonized the inhibitory actions of dopamine (174). CCK altered D2 dopamine receptor binding and antagonized the effect of dopamine electrophysiologically (174) and behaviorally (34,81).

The midbrain neurons which project to the NAC and contain both CCK and dopamine have a distinct topology. They terminate mainly in the medial and posterior NAC (also called the shell), while the anterior and lateral NAC (also called the core) contain mainly non-colocalized terminals (176). Some CCK neurons containing dopamine project to the caudate-putamen (which is also considered to be the core), but the bulk of the CCK neurons innervating the caudate-putamen originate mainly in the cerebral cortex and do not contain dopamine (95).

A number of studies have indicated that this anatomical difference has functional consequences. The interactions between CCK and dopamine are different in the anterior, lateral nucleus, compared with the posterior, medial NAC. CCK increased dopamine release from the posterior NAC through CCKA receptors; in the anterior NAC, CCK (acting through CCKB receptors) inhibited dopamine release (91). Most studies indicated that dopamine release from the caudate was inhibited by CCK (75), although the opposite has also been reported (54,168). CCK acting through CCKA receptors in posterior NAC, potentiated dopamine-induced hyperlocomotion, while in the anterior NAC, CCKB receptors inhibit it (34). CCK potentiates reward (intracranial self-stimulation) when injected into posterior NAC, while it inhibits in the anterior NAC (160). CCK 8, acting through type A receptors in the posterior NAC, decreased the number of open-arm entries in an elevated plus-maze test (a measure of anxiety); in the anterior NAC, it had no effect on this behavior. CCK increased dopamine-induced adenylate cyclase activity in posterior NAC, while it inhibited it in the anterior NAC. Medial NAC neurons were excited by CCK, while lateral NAC neurons were not. CCK also has a different effect on dopamine metabolism in the core vs. shell region of the nucleus accumbens (82).

Like neuroleptics, CCK agonists and antagonists alter the number of spontaneously active dopamine neurons in the A9 and A10 regions (129). CCK antagonists reverse the effect of chronic clozapine and haloperidol treatment on the number of spontaneously active cells in these regions, with the greatest activity on the A10 neurons (102). Long-term haloperidol treatment decreases levels of CCK (125) and increases CCK and dopamine D2 receptor binding in the nucleus accumbens (41).

CCK potentiates amphetamine-conditioned place preference (an animal model of drug abuse potential), while CCK antagonists decrease morphine-conditioned place preferences. CCKB antagonists display antidepressant effects in the forced swim test in mice which were blocked by D1 and D2 antagonists but were potentiated by the dopamine uptake inhibitor nomifensine (64).

A possible mediator of CCK/dopamine interaction is the phosphorylation/dephosphorylation of DARPP-32 (dopamine- and cAMP-dependent 32 kda phosphoprotein). Striatal medium spiny neurons are enriched in DARPP-32, and they are excited by glutamate acting through NMDA receptors. Dopamine acting through D1 receptors increases cAMP-dependent kinase and increases the phosphorylation of DARPP-32, converting it to a potent protein phosphatase I inhibitor. Excitation by glutamate increases calcium influx, which stimulates the calcium/calmodulin-dependent protein phosphatase calcineurin to dephosphorylate and inactive DARPP-32. CCK, like glutamate, decreases DARPP-32 phosphorylation in slices of rat striatum. This effect is blocked by CCKB and NMDA antagonists, implying CCK is stimulating the release of an excitatory transmitter working through NMDA receptors (155).

Much of the CCK (including the cortex and hippocampus) in the brain is colocalized with GABA (62). CCK is known to stimulate GABA release (128) while GABA inhibits CCK release (126). The observation that the excitation of hippocampal neurons by CCK was antagonized by benzodiazepines revealed the possibility of involvement of CCK in anxiety (18). Some of the best CCK antagonists are modified benzodiazepine derivatives, a fact that supports this observation. The CCK-benzodiazepine interaction appears to be indirect, since flurazepam is a weak (KD 12.5 mM) antagonist at CCKB receptors in the VMH or dorsal raphe (15). Anecdotal evidence of panic-like symptoms following human administration of pentagastrin (132) and evidence that administration to sheep caused panic-like symptoms led to further investigations. A number of well controlled human studies have shown that administration of CCK 4 or CCK 5 to panic attack patients or normal volunteers (42) evoked panic-like attacks which were blocked by CCKB antagonists (20) and the benzodiazepine anxiolytic lorazepam. The benzodiazepine antagonist flumazenil did not block the anxiogenic action of CCK 4 (19). Panic attack patients were more sensitive than normal volunteers to the effect of CCK 4, and preadministration of the CCKB receptor antagonist was shown to inhibit the effect of CCK 4.

Further studies in rodents and primates generally supported the observation that CCK 4 given peripherally induced panic-like attacks, while CCKB antagonists are anxiolytic (149). CCKB antagonists were anxiolytic in four rodent models of anxiety: the black-white call box, the elevated plus-maze, the conditioned suppression of drinking, and the acoustic startle response. The ability of CCK antagonists to work in these different models is highly variable and appears to depend on the motivational state of the animals (30,40). CCK may be acting with 5-HT and other transmitters in these models. Further, CCK antagonists decrease the anxiogenic activity of benzodiazepin receptor inverse agonists, and diazepam withdrawal is blocked by CCK antagonists. Diazepam inhibits stress and yohimbine-induced release of CCK in the frontal cortex of freely moving rats (108).

From these results, it is clear that CCK 4 given peripherally induces panic attack symptoms which are blocked by CCK antagonists. This suggested that CCK antagonists might prove useful as anxiolytic drugs. However, initial human studies on endogenous anxiety did not confirm this. One clinical trial of the CCKB antagonist CI-988 found no beneficial effect, but the presence of a significant treatment-by-center interaction and variable placebo response rate made the interpretation of the study results more difficult (1). Another study, with L-354,260, did not find a significant effect (37).

The ability of CCKB antagonists to decrease anxiety in animal models is variable. Concerns about the ability of these agents to gain access to their sites of action have not been resolved. Further concerns about possible species differences in CCK receptor properties between animals and humans may make it difficult to extrapolate animal studies to humans.

CCK causes secretion of both vasopressin and oxytocin from isolated neural lobes of the rat (16) and is known to stimulate oxytocin secretion in intact rats (23). Hypothalamic CRF and pituitary ACTH secretion are stimulated in intact rats by CCK acting through A receptors (73). CCK has been reported to stimulate ACTH release (133) and to act additively with vasopressin and CRF in stimulating ACTH release from isolated rat pituitary and mouse pituitary tumor cells in culture.

CCK at submicromolar concentrations, acting through CCKB receptors, protects neuronal cells in culture from glutamate or NMDA cytotoxicity (74). It has been suggested that this effect of CCK is caused by an inhibition of nitric oxide formation (158).

Caerulein had no beneficial effects in patients with Parkinson's disease, although CCKB antagonists, used in addition to l-dopa, improved neurological performance in MPTP-treated monkeys (17). Caerulein was beneficial in reducing voluntary movements in chorea caused by several neurological diseases (60). In a single patient with palatal myoclonus following Behcet's disease, involuntary movement was improved by caerulein (70).

Uncontrolled studies of the effects of caerulein on schizophrenic patients were somewhat promising, but controlled studies later found that results with caerulein were not significantly different from placebo. Animal studies suggest that CCKB antagonists might be more beneficial than the agonists.

In the last few years, progress toward the elucidation of details of the biosynthesis, physiology and pharmacology of CCK, some of which has been made by groups that are new to the field. The evidence that CCK plays a role in the neurochemical balance of the nervous system is well established. Elucidation of the precise nature of this role and determination of whether alterations in CCK function contribute to neurological and psychiatric disease are the next challenges. There is ample rationale from animal studies to begin and continue human testing of the new generation of CCK agonists and antagonists and inhibitors of CCK degradation.

This work was supported in part by NIH grants NS18667 and NS 31602

published 2000