| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Unité dEndocrinologie et Métabolisme, University of Louvain Faculty of Medicine, B-1200 Brussels, Belgium
Correspondence: Address all correspondence and requests for reprints to: Dr. Patrick Gilon, Unité dEndocrinolgie et Métabolisme, UCL 55.30, Avenue Hippocrate 55, B-1200 Brussels, Belgium. E-mail: gilon{at}endo.ucl.ac.be
| Abstract |
|---|
|
|
|---|
I. Introduction
II. The Innervation of the Endocrine Pancreas
A. General anatomical considerations
B. The parasympathetic innervation
C. The sympathetic innervation
D. Sensory fibers
E. Other types of nerves
III. Physiological Role of the Parasympathetic Control of ß-Cells
A. Difficulties and pitfalls of in vivo studies
B. Physiological situations
C. Pathophysiological situations: hyperinsulinemia, obesity, and insulin resistance
IV. General Characteristics of Acetylcholine (ACh) Effects on Insulin Secretion in Vitro
V. Effects of ACh on ß-Cell Phospholipases
A. Activation of PLC
B. Activation of PLA2
C. Activation of PLD
VI. Effects of ACh on the Membrane Potential of ß-Cells
A. Dependence on the electrical resistance of the plasma membrane
B. Mechanisms of the depolarization
C. Paradoxical hyperpolarization by ACh
VII. Other Effects of ACh in Islet Cells
A. Effects on glucose metabolism
B. Effects on cyclic nucleotides
C. Effects on cytoplasmic pH
VIII. ACh Controls Free Cytosolic Ca2+ Concentration ([Ca2+]c) in ß-Cells
A. Mechanisms by which ACh increases [Ca2+]c
B. Mechanisms by which ACh decreases [Ca2+]c
IX. Mechanisms of the Stimulation of Insulin Secretion by ACh
A. The rise in [Ca2+]c by ACh triggers exocytosis
B. ACh increases the efficacy of Ca2+ on exocytosis
C. Delayed effects of ACh on insulin secretion
D. Muscarinic responses are often abnormal in insulin-secreting cell lines
X. Nature of the Muscarinic Receptor Activated by ACh
A. Pharmacological studies
B. Binding studies
C. Molecular identification of the receptor subtypes
D. One or several receptor subtypes for several transduction pathways?
XI. Summary and Conclusions
A. The physiological role of ACh
B. The mechanisms of action of ACh in ß-cells
| I. Introduction |
|---|
|
|
|---|
The complex neural control of hormone secretion by the endocrine pancreas has been the subject of other reviews (1, 2, 3). It will be addressed only briefly in our contribution, which focuses on the cholinergic control of the ß-cell function. After an overview of the in vivo data demonstrating the role of the parasympathetic system in the regulation of glycemia, we analyze and synthesize the in vitro experiments that have elucidated the cellular mechanisms by which ACh influences ß-cells. Particular attention is paid to the effects of ACh on phospholipid metabolism, membrane potential, free cytosolic Ca2+ concentration ([Ca2+]c), and insulin secretion. This article updates and extends other reviews on the subject (4, 5, 6, 7).
| II. The Innervation of the Endocrine Pancreas |
|---|
|
|
|---|
-cells,
somatostatin-secreting
-cells, and pancreatic polypeptide-secreting
cells (PP-cells). The endocrine pancreas is richly innervated, but the abundance and organization of this innervation are highly variable between species (8). Most of the nerve fibers enter the pancreas along the arteries (9, 10). Unmyelinated nerve fibers are found in the neighborhood of all islet cell types at the periphery and within the islet. At some distance from the islets, glial Schwann cells often form a thin sheet around nerve fibers on their travel toward and within the islet. In the vicinity of islet cells, however, it is not rare to see some nerve fibers lacking this glial protection and coming close to or ending blindly 2030 nm from the endocrine cells (8, 11, 12, 13, 14, 15, 16, 17). Well differentiated synapses with islet cells have rarely been observed (18, 19, 20). Interestingly, the innervation of the islet is very plastic, as suggested by the observation that islets transplanted in the portal vein of diabetic rats became reinnervated by hepatic nerves (21).
The autonomic innervation of the endocrine pancreas has several origins (for review, see Refs. 2 and 3). Classically, the autonomic nervous system uses two interconnected neurons to control effector functions and is divided into two systems, the sympathetic and the parasympathetic nervous systems, according to the location of the preganglionic cell bodies. However, there are indications suggesting that these two systems are not always independent of each other, but display anatomical interactions (22) or share similar neurotransmitters (23, 24, 25). The endocrine pancreas also receives other types of nerves, the anatomical origin and the function (motor efferent or sensory afferent) of which are not clearly known. These nerves are of peptidergic and nonpeptidergic nature (2, 3).
B. The parasympathetic innervation
The preganglionic fibers of the parasympathetic limb originate
from perikarya located in the dorsal motor nucleus of the vagus
(26, 27, 28, 29, 30, 31, 32, 33) and possibly also in the nucleus ambiguus
(26, 34, 35, 36, 37), which are both under the control of the
hypothalamus. They are organized in well separated branches traveling
within the vagus nerves (cranial nerve X), and through the hepatic,
gastric (31, 38), and possibly celiac branches of the
vagus (39), they reach intrapancreatic ganglia that are
dispersed in the exocrine tissue. These ganglia send unmyelinated
postganglionic fibers toward the islets (9, 10, 38, 40).
Preganglionic vagal fibers release ACh that binds to nicotinic
receptors on intraganglionic neurons. Postganglionic vagal fibers
release several neurotransmitters: ACh, VIP, gastrin-releasing peptide
(GRP), nitric oxide (NO), and pituitary adenylate cyclase-activating
polypeptide (PACAP) (3, 27, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51). Cholinergic terminals
are found in the neighborhood of all islet cell types at the periphery
and within the islet (50, 52, 53, 54, 55, 56). The importance of the
cholinergic innervation of the endocrine pancreas is attested by the
presence of a 10-fold higher activity of choline acetyltransferase and
acetylcholinesterase (the enzymes involved, respectively, in the
synthesis and the degradation of ACh) in the islets than in the
surrounding exocrine tissue (57). Cholinergic synapses
with endocrine cells have been observed in some species (58, 59).
Understanding the organization of the pancreatic innervation permits correct interpretation of some experiments using different cholinergic antagonists. The stimulation of insulin release occurring upon electrical stimulation of vagal nerves in the dog is abolished by both nicotinic and muscarinic antagonists (60). In the perfused rat pancreas, nicotine produces an increase of insulin secretion that is blocked by atropine (10). These observations can be explained by the presence of nicotinic receptors on pancreatic ganglia and nerves (61, 62, 63, 64) and muscarinic receptors on ß-cells (see Section X).
The overall effect of a parasympathetic stimulation is an increase of insulin secretion (see Section III). Because postganglionic fibers contain various neurotransmitters in addition to the classic neurotransmitter ACh, it is important to keep in mind that parasympathetic neurotransmission is the sum of various biological effects. VIP and PACAP stimulate insulin secretion by increasing cAMP levels (3). GRP and its amphibian homolog, bombesin (3), are also insulinotropic (3, 42, 65, 66, 67, 68). They act on the same family of receptors (69) and exert their action by two mechanisms, directly by stimulating ß-cells through the PLC-PKC pathway (3), and indirectly by activating intrapancreatic postganglionic nerves that stimulate insulin secretion (68). NO synthase has been detected in nerves in several organs wherein NO is considered a neurotransmitter (70, 71), and in pancreatic nitrergic nerves (45, 48, 49, 67). Various effects of NO on ß-cells have been reported (72, 73, 74, 75), but it is unclear whether NO is implicated in the parasympathetic modulation of insulin secretion.
The parasympathetic system also controls the secretion of the other
islet hormones. Vagal nerve stimulation increases glucagon (31, 41, 60, 76, 77, 78) and PP secretion (41). The effect
of vagal stimulation on
-cells is less clear, as it was reported to
stimulate (79) or inhibit somatostatin secretion
(78, 80). In vitro and in vivo
experiments using various cholinergic agents have shown that ACh
stimulates glucagon and PP secretion through atropine-sensitive
mechanisms (81, 82, 83, 84). The effects of cholinergic agonists
on in vitro somatostatin secretion are again controversial
(2, 80, 82, 85), although this might reflect species
differences.
C. The sympathetic innervation
The sympathetic innervation of the pancreas originates from
preganglionic perikarya located in the thoracic and upper lumbar
segments of the spinal cord (86). The myelinated axons of
these cells traverse the ventral roots to form the white communicating
rami of the thoracic and lumbar nerves that reach the paravertebral
sympathetic chain (87). Preganglionic fibers either
communicate with a nest of ganglion cells within the paravertebral
sympathetic chain or pass through the sympathetic chain, travel through
the splanchnic nerves, and reach the celiac (2, 3, 35, 86, 88) and mesenteric ganglia (86). Ganglia within the
paravertebral sympathetic chain, and the celiac and mesenteric ganglia,
give off postganglionic fibers that eventually reach the pancreas. The
existence of intrapancreatic sympathetic ganglia has also been reported
(25, 26, 37). The preganglionic fibers release ACh that
acts on nicotinic receptors on intraganglionic neurons, whereas the
postganglionic fibers release several neurotransmitters:
norepinephrine, galanin, and NPY (3, 51, 89, 90, 91). A rich
supply of adrenergic nerves in close proximity of the islet cells has
been observed in several mammalian species (53, 54, 55, 92).
The net physiological effect of splanchnic nerve stimulation is a
lowering of plasma insulin concentration (93, 94, 95, 96). This
effect is attributed to release of norepinephrine from nerve fibers
close to ß-cells and to elevation of catecholamine (epinephrine and
norepinephrine) plasma levels because of the stimulation of the adrenal
medulla. Catecholamines have long been known to inhibit insulin
secretion in vivo (1, 2, 3, 97) and in
vitro (1, 2, 3, 98, 99, 100, 101). Their action is mediated by
2-adrenoceptors (102), probably
of the
2a- and
2c-subtypes (103), which have
been identified in ß-cells by both pharmacological (104)
and molecular approaches (103, 105). Activation of
2-adrenoceptors interferes with the secretory
process through several mechanisms that are all prevented by pertussis
toxin treatment and are, thus, likely mediated by
G
i or G
o
(106): an inhibition of adenylate cyclase leading to a
lowering of ß-cell cAMP, an opening of K+
channels of small conductance leading to partial membrane
repolarization and decrease in Ca2+
influx, and a major inhibition of a late step of exocytosis
(106). Similar pathways are implicated in the inhibitory
action of galanin, which may cooperate with catecholamines to inhibit
insulin secretion in response to splanchnic nerve stimulation (3, 106). In contrast, an increase in plasma insulin can be
evoked by selective ß-adrenergic agonists, particularly of the
ß2-subtype (2, 95, 107, 108), that
activate adenylate cyclase and increase cAMP. However, these usually
have little effect on insulin secretion by isolated islets
(109). Moreover, the presence of
ß2-adrenoceptors in ß-cells remains
controversial (105, 110). It is also important to
emphasize that a number of pharmacological studies have been
misinterpreted because antagonists of adrenoceptors can influence
insulin secretion by acting on other targets, e.g., on
ATP-sensitive K+ (K+-ATP)
channels (111). The mechanisms by which NPY inhibits
insulin release are not clearly known and might involve a decrease in
cAMP levels (3).
The sympathetic nervous system exerts profound effects on the secretion
of the other islet hormones. Splanchnic nerve stimulation increases
glucagon secretion (93, 94, 95, 96, 112, 113), and epinephrine
stimulates glucagon secretion in vivo and in
vitro (84, 100, 114, 115). This effect results from
the activation of ß-adrenoceptors (101), probably of the
ß2-subtype (110), although one
report implicates
-adrenoceptors (96). It has been
shown that pancreatic
-cells express
1-,
2-, and
3-subtypes
(103). Splanchnic nerve stimulation decreases
somatostatin secretion (80, 96, 116), and
norepinephrine inhibits somatostatin release by isolated rat islets
(100). The results are less clear for PP secretion. Thus,
splanchnic nerve stimulation has been reported to increase (3, 113, 117) or inhibit PP secretion (2, 116).
Catecholamines stimulate PP release by isolated islets
(118).
Overall, the sympathetic nervous system serves to maintain or increase glycemia in various conditions of stress such as neuroglycopenia, hypovolemia, or physical exercise (3). Its pancreatic action not only involves inhibition of insulin secretion, but also stimulation of glucagon secretion (3, 119).
D. Sensory fibers
Calcitonin gene-related peptide (CGRP) and substance P (SP) are
thought to report sensory information in many systems
(120). CGRP (51, 121, 122)- and
SP-immunoreactive (51, 123, 124) nerve fibers have been
observed in both the exocrine and endocrine pancreas. Vanilloid
receptors, activated by heat, low pH, and various vanilloid agents
(such as capsaicin), are localized in sensory fibers and generally
report pain information (120). Neonatal treatment of mice
with capsaicin destroys the majority of capsaicin-sensitive neurons and
has often been used to identify sensory fibers (120). This
treatment was followed by a marked reduction of CGRP-immunoreactive
fibers in both the endocrine and exocrine pancreas (122)
and by a partial reduction in SP-immunoreactive fibers (36, 125).
It is thought that sensory afferents leave the pancreas along the sympathetic fibers within the splanchnic nerves and that the perikarya of the sensory fibers are present in dorsal root ganglia, mainly at the level of the lower thoracic segments of the spinal cord, transmitting noxious information to the central nervous system by synapsing on second-order neurons of the dorsal horn of the spinal cord (2, 86, 126, 127). The existence of such an anatomical route is supported by experiments of retrograde labeling (36, 86, 126, 128). It has been suggested that the pancreas is also innervated by sensory afferents that run within the vagus nerve, the perikarya of which are in the nodose ganglion and transmit information to the nucleus tractus solitarius (30, 35, 36, 129, 130).
There is no doubt that sensory nerve fibers report pain information associated with diseases of the exocrine tissue, such as pancreatic cancer and pancreatitis (127, 131), but there are no reports of sensations of pain associated with a destruction of the endocrine pancreas. However, it is possible that sensory fibers play a role in the control of insulin secretion. Thus, neonatal treatment of mice with capsaicin (to destroy these fibers) results in more glucose-stimulated insulin secretion than in nontreated mice, suggesting that sensory fibers exert a direct, tonic inhibition of insulin secretion (132). CGRP may inhibit insulin secretion through a direct action on the islets (121, 133), whereas both inhibitory (134) and stimulatory (124, 135) effects of SP have been reported. Indirect effects of capsaicin-sensitive fibers are also possible. Indeed, it has been reported that removal of endogenous sensory neuropeptides by deafferentation of capsaicin-sensitive sensory nerves improves glucose tolerance by increasing in vivo insulin sensitivity (136, 137).
E. Other types of nerves
Immunocytochemistry has revealed the presence of
neurotransmitters other than those described above in pancreatic
nerves: cholecystokinin (138), 5-hydroxytryptamine (5-HT
or serotonin) (139, 140), and methionine-enkephalin
(3, 51). These might also influence insulin secretion:
cholecystokinin stimulates insulin release by activating PLC and
PLA2 (138), but the effects of 5-HT
are controversial, as both inhibition (141, 142) and
stimulation (143) of insulin secretion have been reported.
Enkephalin also exerts variable effects depending on the concentration
used and the species studied (144, 145).
The pancreatic innervation presents other interesting features. The section of extrinsic pancreatic nerves has revealed that many of the intrinsic pancreatic neurons are independent of the integrity of the extrinsic nerves (146), suggesting that the pancreatic innervation might behave as an independent system. This is supported by the observation that intrapancreatic ganglia are interconnected with one another, as are enteric ganglia (37, 140). It has also been suggested that intrapancreatic ganglia are connected with the duodenal myenteric plexus by nerve fibers (50), suggesting the existence of an entero-pancreatic innervation. On the other hand, ganglia from the myenteric plexus of the stomach and duodenum send nerve fibers toward the pancreas (50, 139). Many of these nerves are immunoreactive for 5-HT. Whether these innervations play a physiological role in the regulation of hormone secretion by the endocrine pancreas remains to be investigated.
| III. Physiological Role of the Parasympathetic Control of ß-Cells |
|---|
|
|
|---|
The most important characteristic of the influence of ACh on insulin secretion is a tight dependence on the ambient glucose concentration. In vivo, electrical stimulation of the vagus nerve has little effect on the concentration of plasma insulin during hypoglycemia, but increases it more and more efficiently as the concentration of plasma glucose augments (41, 151, 152, 153, 154, 155). Similar observations have been made in vitro when the perfused pancreas (81, 156, 157, 158, 159, 160) or isolated islets (161, 162, 163, 164) were used to study insulin secretion directly (Fig. 1A). This behavior is typical of a potentiating agent. The mechanisms underlying this potentiation will be explained in detail in Section IX.
From here, it is important to bear in mind that the majority of in vitro experiments were conducted with rodent islets, and that the concentration dependence of glucose-induced insulin secretion is different in rodent and human islets. Indeed, the threshold glucose concentration and the half-maximal effective concentration for insulin secretion are, respectively, around 4 and 9 mM for human islets as compared with 7 and 15 mM for mouse islets (165).
A. Difficulties and pitfalls of in vivo studies
Species differences must be considered when interpreting the
effects of ACh on insulin secretion in vivo. As already
mentioned in Section II, vagal stimulation can release at
least five neurotransmitters (ACh, VIP, PACAP, GRP, and NO), the
relative contribution of which differs between species. In the dog
(49, 60, 76, 149, 166), rat (78), albino
mouse (167), and calf (41), vagal stimulation
of insulin secretion is mediated mainly or exclusively by muscarinic
receptors because it is largely or fully prevented by atropine. This is
not the case in the pig (113, 168) and in the cat
(10, 169), in which neurotransmitters other than ACh are
probably implicated.
A second difficulty is linked to the number of physiological events that are under parasympathetic control. Cholinergic agonists or antagonists, stimulation of the vagus nerve, and vagotomy induce multiple effects that can indirectly interfere with insulin secretion. Cholinergic agents influence the secretion of the other islet hormones (see Section II), but it is difficult to establish to what extent the observed changes in insulin secretion are influenced by paracrine or endocrine interactions. Such interactions depend on the organization of the microvascularization and the direction of blood flow, which are still a matter of debate (170, 171, 172). In addition, clear evidence for intraislet paracrine influences has not yet been reported (172). Importantly, the effects of ACh on insulin secretion are observed at glucose concentrations that are substimulating or stimulating for insulin release but inhibitory for glucagon release.
Several intestinal hormones, particularly glucose-dependent insulin-releasing peptide (GIP) and glucagon-like peptide-1, potently increase insulin secretion (173, 174, 175). GIP- and glucagon-like peptide-1-secreting cells possess muscarinic receptors (176, 177), and both the stimulation of the vagus nerve and cholinergic agonists stimulate their release (178).
ACh also stimulates gastric emptying (179), which may affect the rate of glucose absorption, change in glycemia, and hence, insulin secretion (180, 181). It has also been reported that the increase in islet blood flow produced by a rise in blood glucose is mediated by the central nervous system, which senses the changes in glycemia and sends signals to islet vessels through the vagus nerves (182).
B. Physiological situations
1. Role of the vagus nerve on glucose tolerance. In lean
animals or humans, basal insulin secretion is not affected or is only
slightly decreased by vagotomy or atropinization (9, 149, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196), which indicates that there is no significant tonic
stimulation of the ß-cells by the parasympathetic system in the
fasting state. In contrast, it is generally agreed that the vagus
nerves participate in the control of insulinemia during the periods of
feeding. The difficulty is to assess their contribution to the overall
insulin-secretory response. Thus, depending on the study, the tolerance
to a glucose load is unaffected or impaired by atropine or vagotomy,
whereas the associated insulin response is larger, similar, or smaller.
The results become more consistent when the insulinogenic index (
insulinemia/
glycemia) is calculated, and the mode of administration
of glucose (oral or iv) is taken into account (197, 198, 199).
Thus, when glucose is administered iv, the insulinogenic index is not
affected or is hardly modified by atropine or vagotomy (185, 187, 191, 193, 195, 197, 198, 199, 200, 201). In contrast, when glucose is given
orally, the insulinogenic index is significantly decreased by atropine
or vagotomy (184, 197, 199, 200, 202). In addition,
the rise in plasma insulin is delayed, which also contributes to the
glucose intolerance of vagotomized or atropine-treated rats (1, 199, 203). These results suggest that ACh potentiates the
insulin response to glucose after a glucose load, a conclusion that is
supported by experiments using animal models without parasympathetic
innervation of the ß-cells. After destruction of their ß-cells by
alloxan or streptozotocin, rats were transplanted with isolated islets.
The vagus nerves of the receivers were intact, but the transplanted
ß-cells were presumably denervated at the time of test
(203, 204, 205). Meal ingestion induced a glucose increase that
was larger in transplanted than in normal control rats, and that was
associated with a delayed insulin response (40, 206, 207).
This confirms that a direct parasympathetic innervation of ß-cells
improves glucose tolerance.
2. The vagus nerves transmit signals of several origins. When evaluating the physiological role of the muscarinic control of ß-cells, it is important to bear in mind that the vagus nerves are the parasympathetic effectors of signals that are all integrated in the brain but come from at least four sources: cephalic sensory organs including those of the oral cavity and the visual and olfactory systems, the gut, the liver, and the brain itself (208). The sequential activation of all these inputs will affect insulin secretion in a time-dependent manner upon meal ingestion.
a. Cephalic sensory organs and the gastrointestinal tract.
The preabsorptive insulin phase corresponds to the earliest plasma insulin
rise during the first minutes of food ingestion. It does not depend on
nutrient assimilation, as it occurs before the glycemia has increased
(1, 200, 206, 207, 209, 210, 211, 212, 213, 214, 215, 216) and is sometimes associated
with a transient hypoglycemia (28, 203, 217). The
amplitude of the preabsorptive insulin phase is highly variable from
one study to another, but it is consistently much smaller than the
postabsorptive insulin phase occurring when glycemia starts increasing.
It corresponds to a rise of approximately 20% (28, 216, 217) or more (1, 206) above basal insulinemia,
which may be an underestimation of the reality, because insulin is
measured in peripheral or heart blood and not directly in the portal
circulation. Insulin is indeed very rapidly degraded by the liver (50%
during the first passage of blood), and the amplitude of the changes in
insulinemia is much smaller in peripheral than in portal blood
(192, 218). The preabsorptive insulin response involves
both cholinergic and noncholinergic mechanisms (3, 216).
It can be subdivided into the cephalic phase and the enteric phase.
The cephalic phase does not even require ingestion of nutrients, as it can occur in response to oral saccharin or water intake in animals (1, 204), but not in humans (219). Its mechanisms involve stimulation of oropharyngeal receptors (210, 220, 221) and probably also conditioned visual and olfactory reflexes, because an early peak of insulin secretion can be observed in animals that simply see or smell food (28). A cephalic phase exists in humans (Refs. 28 and 222, 223, 224, 225 , but see Ref. 226), but is less easily conditioned than in animals (227, 228). Because no cephalic phase occurs after vagotomy, nor is this phase observed in diabetic animals transplanted with denervated islets, it is ascribed to a direct stimulation of ß-cells by both cholinergic and noncholinergic fibers of the vagus nerves (3, 203, 204, 205, 206). The sympathetic nervous system might also contribute to the cephalic phase in the dog by activating ß2-adrenoceptors (229).
The enteric phase has been much less extensively studied because of the difficulty in separating it from the preceding cephalic phase and the following postabsorptive phase, which rapidly causes an increase in glycemia (28, 203, 217, 230, 231). This phase has been observed after direct infusion of a meal into the stomach or the duodenum (176, 203, 232) and is sometimes reflected by a single preabsorptive insulinemia peak (203). Abolition of this phase by vagotomy and atropine implicates the vagus nerve (176, 217, 232), but it remains unclear whether the response is mediated indirectly by a vagally induced release of incretins (203), or more directly by a reflex involving gut glucoreceptors augmenting efferent activity of the pancreatic branch of the vagus nerve (233, 234). Glucoresponsive neurons equipped with K+-ATP channels similar to those of ß-cells have recently been identified in the myenteric plexus of the guinea pig ileum (235).
The role of the preabsorptive phases of insulin secretion was initially addressed by comparing the insulin and glucose responses to oral vs. iv glucose administrations (203, 236). However, it was later found that this type of comparison might be misleading because a lesser glucose tolerance after iv administration of the sugar could result from the lack of incretin effect rather than the lack of preabsorptive insulin secretion. The importance of the cephalic phase for glucose homeostasis was established by experiments showing that plasma glucose and insulin concentrations increased more after direct administration into the stomach than after oral intake of the same amount of nutrients (237, 238). These results were confirmed in a more recent study that compared the insulin and glucose responses to gastric glucose administration in humans allowed or not allowed to taste food (239). Prevention of cephalic phase during food intake diminished the glucose tolerance without changing insulin secretion during the 3-h period after the beginning of food intake. This glucose intolerance was attributed to differences in the kinetics of the changes in insulinemia, and possibly also to a larger glucagon secretion over the same period of time (239).
That the cephalic phase exerts a beneficial, long-lasting effect has also been elegantly demonstrated after oral glucose administration to insulin-deficient rats transplanted with denervated islets. The missing cephalic phase in these rats was mimicked by a small premeal iv injection of insulin. This restored early insulin peak did not affect subsequent plasma insulin levels during the period of glucose intake, but attenuated (without normalizing) the rise in plasma glucose levels (40).
With the exception of two reports (196, 240), all the above-described studies suggest that the timing of insulin secretion is important for optimal glucose homeostasis. By promoting anticipatory use of glucose by the liver, muscles, and adipose tissue, and by inhibiting glucose production by the liver, the preabsorptive insulin phases restrain the changes in glycemia and insulinemia within a narrow range. This may serve as a protection against overworking of the ß-cells.
b. Liver.
Like ß-cells, the liver receives efferent vagal
stimuli in response to an oral stimulus (220, 241).
Neurophysiological studies have also revealed that portal glucose
injection increases efferent vagus activity innervating the pancreas,
and it has been suggested that hepatic glucosensitive mechanisms may
affect pancreatic function by involving hepatic vagus afferents and
pancreatic vagus efferents (208). These results are
supported by some physiological experiments demonstrating that a rise
of the glucose concentration in the portal circulation to the liver
induces an increase of insulin secretion that is prevented by vagotomy
of the hepatic branch (242) and is mimicked by hepatic
vagal stimulation (243). However, other studies have
reported that a rise in insulinemia is mimicked by a section of
the hepatic branches of the vagus nerve, whereas a drop in insulinemia
is induced by electrical stimulation of this branch (241, 244). Therefore, it has been suggested that afferent fibers
exert a tonic inhibition in brainstem centers of an efferent vagal
branch innervating the pancreas (241, 244). This implies
that the afferent hepatic nerve activity is inversely related to the
portal glucose level, which is confirmed by neurophysiological data
(245, 246, 247). It is difficult to establish how glucose
homeostasis is influenced by these vagally mediated messages from the
liver to the endocrine pancreas.
c. Brain.
Several studies suggest that an increase in the
glucose concentration in the brain can increase vagal tone. Indeed,
injection of glucose in the carotid artery of rats, in an amount
insufficient to modify systemic plasma glucose concentration, induced a
rapid increase in insulin secretion that was abolished by vagotomy
(182). This effect likely involves glucoresponsive neurons
in the hypothalamus and the nucleus tractus solitarius (220, 248).
3. Is there a long-lasting vagal stimulus during the absorptive phase?Whereas it is clear that the parasympathetic system contributes to the preabsorptive insulin response, it is less obvious whether a parasympathetic stimulus of the endocrine pancreas persists during the meal. Because ACh is quickly degraded by cholinesterases in plasma, it is impossible to reliably measure the pancreatic ACh spillover as an index of parasympathetic neural activity of the pancreas (229). Measurements of plasma PP provide an alternative approach to evaluate the parasympathetic activity. Indeed, PP secretion is predominantly under vagal control because its secretion in vivo is nearly completely prevented by atropine or vagotomy (249, 250, 251). Meal ingestion induces a biphasic PP secretion characterized by a rapid first phase followed by a second sustained response. Both phases are prevented by atropine (229). The first phase has sometimes been correlated to the preabsorptive insulin response (250, 252). The presence of the second sustained phase supports the existence of a long-lasting cholinergic stimulus. Because cholinergic nerve fibers innervating PP and ß-cells likely have a common origin, it is highly plausible that ß-cells are also under the influence of a long-lasting vagal stimulus during meals. This suggestion is corroborated by the observation that atropine markedly suppressed insulin response to a meal (251). However, recent data obtained with mice lacking the M3 muscarinic receptor (the main muscarinic receptor on ß-cells, see Section X) do not clarify the issue. These mice do not show any signs of impaired glucose intolerance after oral or ip glucose administration (253), but it is unclear to which extent other factors that were observed in these mice, such as increased insulin sensitivity, hypoleptinemia, and hypophagia, contributed to glucose tolerance.
C. Pathophysiological situations: hyperinsulinemia, obesity, and
insulin resistance
The net effect of the central nervous system on insulin secretion
is the result of a balance between the influence of the inhibitory
sympathetic system and the stimulatory parasympathetic system. The
metabolic consequences of a dysregulation of this subtle balance have
been reviewed recently (254). Only the troubles associated
with an anomaly of the parasympathetic system will be briefly mentioned
here.
Two areas in the brain play a major role in the control of the efferent autonomic pathways, the ventromedial hypothalamic nuclei (VMH, also called the satiety center) and the lateral hypothalamic area (LHA, also called the feeding center) (241, 254). The VMH increases the activity of the sympathetic nervous system and decreases that of the parasympathetic nervous system, whereas opposite effects are produced by the LHA. The VMH and the LHA reciprocally inhibit each other.
Several animal models of hyperinsulinemia are characterized by a dysregulation of the sympathetic and parasympathetic pathways. A lesion of the VMH in the rat causes an exaggerated insulin response to an iv or intragastric glucose load. This hyperinsulinemia occurs rapidly, 10 min after the lesion (255), and is abolished by atropine or vagotomy (256, 257, 258, 259). In animal models of hyperinsulinemia associated with a defect in leptin signaling, such as the ob/ob mice (abnormal leptin) or fa/fa (Zucker) rats (abnormal leptin receptors), the earliest detectable alteration of insulin secretion is a hyperresponsiveness to glucose that occurs before the animals become hyperphagic (193, 260, 261, 262, 263, 264, 265). It is mediated by the vagus nerve, as it is abolished by atropine (194, 255, 260, 266) or vagotomy (193, 266). NPY is a potent physiological stimulator of feeding that is present at abnormally high levels in the hypothalamus of fa/fa rats and ob/ob mice. Rats that undergo a chronic intracerebroventricular infusion of NPY display basal and glucose-induced hyperinsulinemia that is prevented by vagotomy (267, 268). A common feature of all these animal models is a hyperinsulinemia that results from an excessive vagal cholinergic tone and an attenuation of the inhibitory sympathetic tone (254). The chronic influence of hyperglycemia on the autonomic system may also aggravate the syndrome (269).
An increased sensitivity of ß-cells to ACh might also contribute to hyperinsulinemia. This has been reported in ob/ob mice (262) and in genetically normal mice subjected to a high-fat diet (46, 270, 271). The hyperinsulinemia brought about by an enhanced cholinergic over sympathetic tone and/or an exaggerated sensitivity of ß-cells to ACh might be a compensatory mechanism for insulin resistance.
This fairly clear picture obtained in experimental animals cannot readily be extrapolated to human subjects. Although indirect evidence suggests that insulin secretion is more sensitive to cholinergic stimulation in insulin-resistant obese subjects than in lean subjects (272), it is widely agreed that atropine does not correct the hyperinsulinemia of obese subjects (195, 196). Moreover, the preabsorptive phase of insulin secretion was found to be enhanced (273), normal (224, 274, 275), or absent in obese subjects (276). In type 2 diabetes, the initial rise in insulin levels after a meal is often delayed or deficient (277, 278), but it is unknown whether an impaired preabsorptive vagal stimulus contributes to this defect.
| IV. General Characteristics of Acetylcholine (ACh) Effects on Insulin Secretion in Vitro |
|---|
|
|
|---|
|
10 µM) trigger a
rapid, transient peak of secretion that also requires the presence of a
high concentration of glucose (or other nutrients) (Refs.
162 and 289, 290, 291 and Fig. 1BThe mechanisms underlying these glucose, membrane potential, and Ca2+ dependencies of ACh-induced insulin secretion will be explained in the following paragraphs.
| V. Effects of ACh on ß-Cell Phospholipases |
|---|
|
|
|---|
,
and PLC-
), each containing several
subtypes (292, 293, 294). PLC-ß are activated by
heterotrimeric G proteins, whereas PLC-
are activated by tyrosine
kinases. The mechanisms of activation of PLC-
are unknown
(294). All three types hydrolyze phosphatidylinositol
(PI), PI 4-phosphate (PIP), and PI 4,5-bisphosphate
(PIP2) in a Ca2+-dependent
manner to produce DAG and inositol 1-phosphate (1 IP),
inositol 1,4-bisphosphate, and IP3, respectively. At low
[Ca2+]c,
PIP2 is the preferred substrate
(293). In some tissues, certain PLC isoforms can hydrolyze
plasmenylcholine, phosphatidylethanolamine, or phosphatidylcholine (PC)
(295, 296, 297, 298).
|
|
o protein (323). All other
studies, performed with RINm5F cells (324, 325), rat
islets (321, 326, 327), and ß-TC3 cells
(328), found the G protein coupled to the muscarinic
receptor to be pertussis and cholera toxin insensitive. It is thought
to belong to the Gq subfamily (328),
like the G protein that couples muscarinic receptors to PLC-ß in
other tissues. Activation of PLC-ß has been reported to be directly
mediated by the
-subunit of the Gq protein
(329). The Gq subfamily contains
several members (G
q,
G
11, G
14,
G
15, and G
16)
(330, 331) that seem to specifically interact with the
different PLC-ß subtypes (292). The nature of the
G
-subunit involved in the coupling of the
muscarinic receptors to PLC-ß in ß-cells is unknown. Phospholipid hydrolysis and inositol production is larger in ß-cells maintained in a Ca2+-containing medium than in a Ca2+-free medium (5, 162, 290, 303, 332), and are markedly reduced in insulin-secreting cells loaded with the Ca2+-chelator 1,2-bis(o-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA) (333). Two hypotheses have been put forward to explain this Ca2+ dependence. Because PLC is strictly Ca2+ dependent (293), the enhanced hydrolysis of phosphoinositides observed in cells bathed in a Ca2+-containing medium has sometimes been attributed to a direct activation of PLC by Ca2+ (300, 305, 307, 322, 332, 333, 334, 335, 336, 337, 338), independently from calmodulin (300, 337). This proposal was supported by the observations that high K+, which induces a large rise in [Ca2+]c, increased IP3 or total IPs levels (332, 339, 340) and accelerated the efflux of radioactivity from rat islets prelabeled with [3H]inositol (335). However, these results must be interpreted with caution. First, the effect of high K+ on IP3 levels is species dependent and larger in the rat than in the mouse (340), perhaps because of the expression of different PLC isoforms (317, 320, 341, 342). Second, even in the rat, phosphoinositide breakdown is much larger in response to carbachol than to high K+ (316, 332, 335, 339). Therefore, a second hypothesis suggests that the potentiation by Ca2+ of ACh-induced IP3 production results from a synergistic effect between Ca2+ and muscarinic activation (309, 316). As G proteins have been reported to enhance the Ca2+ sensitivity of PLC (343, 344, 345, 346), the Ca2+ requirement would be reduced to the resting [Ca2+]c levels. This would also explain how muscarinic agonists can elicit a significant phosphoinositide hydrolysis in cells bathed in a Ca2+-free medium. Finally, three studies have proposed that PLC activity can also be controlled by the membrane potential, independently from a change in [Ca2+]c. Depolarization in a medium supplemented with methoxyverapamil or in a Ca2+-free medium was reported to increase PLC activity in ob/ob islets (347) or in insulin-secreting ß-TC3 cells loaded with 1,2-bis(o-aminophenoxy)-ethane-N,N,N',N'-tetraacetic acid (BAPTA) (338). The opposite effect, a decrease in PLC activity, was measured in rat islets (339). Thus, the question remains unanswered.
2. Phosphoinositol and phosphoinositide metabolism.
Experiments with RINm5F cells or rat islets have shown that IP3 is very
rapidly transformed into inositol 1,3,4,5-tetrakisphosphate (307, 332, 348) by a kinase activated by
Ca2+-calmodulin (305, 349, 350), and
into inositol 1,4-bisphosphate by a phosphatase (305, 348, 351, 352) (Fig. 2
). Inositol 1,3,4,5-tetrakisphosphate can then
be degraded into inositol 1,3,4-trisphosphate. All inositol phosphate
isomers can be further metabolized through complex pathways (350, 353). In several studies in which the different isomers of
inositol phosphate were not separated (5, 290, 311, 313),
cholinergic agonists caused a monotonic increase of inositol
trisphosphate levels to a plateau that was reached within approximately
1 min and thereafter maintained with only a minor decline. When the two
major isomers of inositol trisphosphate, i.e., IP3 and
inositol 1,3,4-trisphosphate, were separated, very different time
courses of accumulation emerged (305, 311, 352). Indeed,
IP3 accumulation consisted in a burst, reaching a peak within the first
5 sec of stimulation, followed by a decrease and then a lower sustained
phase. By contrast, inositol 1,3,4-trisphosphate levels increased
slowly, reached a maximum after approximately 30 sec of stimulation,
and plateaued at that level thereafter. The biphasic increase in IP3
probably involves both a negative feedback effect of PKC on PLC
activity (see Section V.A.3) and a rapid degradation of IP3
(352, 354). Indeed, the two-step conversion of IP3 into
inositol 1,3,4-trisphosphate initiated by the
Ca2+-calmodulin-sensitive IP3 kinase was markedly
attenuated if the rise in
[Ca2+]c that IP3 produces
(see Section VIII.A.1) was abolished (305).
This suggests that the rise in
[Ca2+]c contributes to
the rapid degradation of IP3 and, therefore, also contributes to the
transient nature of its accumulation (305, 352).
Whereas it is well established that muscarinic stimulation of insulin secretion increases with the glucose concentration, it remains unclear whether glucose, per se, potentiates ACh-induced IP3 accumulation. The larger effect that carbachol produces in the presence of high glucose (164, 304, 316, 355) might well result from an additional Ca2+-dependent activation of PLC due to a greater increase in [Ca2+]c (see Section VIII.A.3). This interpretation is supported by the observation that glucose failed to enhance carbachol-induced accumulation of inositol trisphosphate in rat islets incubated in a Ca2+-free medium (339). Glucose itself and various intermediates of its metabolism were without effect on PLC activity in a cytosolic fraction of mouse islet homogenate (300). However, other reports have suggested that glucose metabolism might interact with phosphoinositol and phosphoinositide metabolism. Thus, glucose metabolites inhibited IP3 degradation (356, 357) and increased IP3 production (358) in rat islet and RINm5F cell homogenates, and directly stimulated Ca2+ release from intracellular Ca2+ stores of HIT-T15 cell homogenates (359). Moreover, glucose has been reported to stimulate the de novo synthesis of DAG, inositol phosphate, phosphoinositides, or polyphosphoinositides (4, 302, 305, 311, 358, 360, 361, 362, 363, 364, 365, 366). More experiments must be performed to evaluate the possible direct effects of glucose on ACh-induced IP3 accumulation.
In many tissues, PLC mainly hydrolyzes PIP2. In
pancreatic ß-cells, cholinergic agonists also stimulate PI hydrolysis
as shown by the rapid accumulation of 1 IP independently from the
other phosphoinositol intermediates in rat islets challenged with
carbachol (348) (Fig. 2
). In agreement with this
observation, PI was found to be a better substrate than
PIP2 for PLC from the cytosolic fraction of mouse
islet homogenates (300). Surprisingly,
carbachol-induced accumulation of 1 IP in rat islets is stimulated
by hyperpolarization and is inhibited by depolarization of the plasma
membrane (339). The underlying mechanisms are not known.
The predominance of PI hydrolysis over that of
PIP2 during prolonged stimulation
(348) implies that DAG can be formed independently of IP3
formation. However, the potential importance of this pathway for the
stimulation of insulin release has yet to be established.
At the same time cholinergic agonists hydrolyze phospholipids, they
also accelerate PI turnover. DAG can be resynthesized back to PI by the
following steps (Fig. 2
): 1) diacylglycerol kinase converts DAG into
phosphatidic acid (PA) at the expense of an ATP; and 2) PA then reacts
with CTP to form CMP-phosphatidate (CDP-DAG), which in turn reacts with
inositol to form PI (346, 367, 368, 369). The enzymes involved
in this cycle are present in rat islets (370, 371).
Because the cycle requires phosphorylated nucleotides, its turnover was
estimated by labeling islets with
32PO43-.
It was found that carbachol stimulates the labeling of PA and decreases
that of PIP and PIP2, which suggests that
cholinergic agonists accelerate PI turnover after PLC activation in rat
islets (5, 312, 360, 372, 373). This effect was strongly
inhibited in a Ca2+-free medium, which might
result from the Ca2+ dependence of PLC
(334, 372). Concomitantly, there is an enhanced flux from
PI
PIP
PIP2, which is necessary for
resynthesis of PIP2 (322). This
increased flux might result from a direct stimulation of the activity
of PI 4-kinase, the enzyme responsible for the synthesis of PIP
from PI (322). Because PIP kinase is inhibited by its
product, PIP2, hydrolysis of
PIP2 by PLC could also relieve this inhibition,
therefore stimulating the flux for PIP2
synthesis. This latter mechanism has been demonstrated in other cell
types (374), but not in ß-cells. Cholinergic agonists
not only accelerate PI turnover, they also stimulate de novo
synthesis of phospholipids, as deduced from the enhanced incorporation
of [3H]glycerol into DAG (375),
PA, and PI in rat islets (360). Again, this de
novo synthesis pathway is very much Ca2+
dependent (360).
Of all inositol species formed upon ACh stimulation, IP3 is the physiologically more important isomer. Its effects will be described later (see Section VI.A.1).
3. Diacylglycerol and PKC. DAG is liposoluble and remains in
the plasma membrane. It causes the translocation of its target, PKC,
from the cytosol to the membrane. This translocation also requires
Ca2+ and an acidic phospholipid, such as
phosphatidylserine (376, 377). Metabolism of DAG, either
by DAG lipase-catalyzed deacylation (which yields arachidonic acid) or
by DAG kinase-catalyzed phosphorylation (which yields PA), terminates
its action on PKC (299, 378) (Fig. 3
).
Cholinergic agonists produce two major species of DAG that are enriched
in either arachidonate (a polyunsaturated fatty acid) or palmitate (a
saturated fatty acid), and accumulate with different time courses in
ß-cells (Fig. 3
). The concentration of arachidonate-enriched DAG
increases quickly during the first seconds of stimulation, before
declining and remaining at a lower sustained level. This type of DAG
probably originates from PIP2 that mostly
contains arachidonate at the sn-2 position of the phospholipid
(302, 379). By contrast, the palmitate-enriched DAG
accumulates monotonically during the first minutes of stimulation
(302, 375). This species resembles that produced upon
glucose stimulation (365, 375), but its source upon ACh
stimulation is unknown. It might originate from PLD activation (see
Section V.C) and hydrolysis of PC, a phospholipid
enriched in palmitate in islets, from a synergistic effect with glucose
on de novo DAG synthesis, or from other undefined pathways
(302, 380). A biphasic increase in polyunsaturated DAG and
a delayed accumulation of saturated DAG have been documented in many
other cell types (381). The differential time course of
accumulation of the two DAG species might have an impact on PKC
activation. Indeed, polyunsaturated DAG is a much more potent PKC
activator than saturated DAG in HIT cells (365, 382) and
other cell types (379).
The PKC family comprises a number of phosphatidylserine-binding
isoforms that can be classified in four groups. The conventional
isoforms, or cPKC, are activated by Ca2+ and DAG
or phorbol esters; they include
, ßI, ßII, and
, of which
ßI and ßII refer to the two gene products resulting from
alternative splicing of the same PKCß gene. The novel isoforms, or
nPKC (
,
,
, and
), are unresponsive to
Ca2+ but are activated by DAG alone or phorbol
esters. The atypical isoforms, or aPKC (
, and
/
; PKC
is the
mouse homolog of human PKC
), are Ca2+
independent and do not bind DAG or phorbol esters. The PKCµ isoform
is also Ca2+ independent and is activated by
phorbol esters, but has a structure different from the other isoforms
(298, 381, 383, 384, 385, 386). The nature of the PKC isoforms
present in insulin-secreting cells remains controversial (reviewed in
Ref. 376). In pancreatic islets, the isoform
predominates, but one or several of the isoforms ß,
,
,
,
and
may also be present (376, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396). This does not
necessarily mean that all these isoforms are expressed in ß-cells
because approximately 2035% of the islet cells are non-ß-cells.
Insulin-secreting cell lines express one or several of the isoforms
, ß,
,
,
,
,
/
, and µ (382, 387, 392, 394, 397, 398, 399, 400).
Several studies demonstrate that cholinergic agonists induce the
translocation of PKC to membranes (163, 312, 365, 382, 401), but they do not establish which isoforms are translocated.
Because cholinergic agonists induce DAG accumulation (302, 339, 375), it seems reasonable to assume that they stimulate all
DAG-sensitive isoforms present in normal and tumoral insulin-secreting
cells. However, a recent study suggests that this might not be the
case. Carbachol was found to translocate the
, ß, and
isoforms
without affecting the
,
, µ, and
isoforms in RINm5F cells
(400). These data must be interpreted with caution because
they were obtained with an insulin-secreting cell line whose responses
to cholinergic agonists differ from those of normal ß-cells (see
Section IX.D).
PKC phosphorylate their substrates on serine and/or threonine residues.
It has been suggested that the targeting of PKC isoforms to particular
membranes is mediated by specific anchoring proteins including the
receptors for activated C kinases (384, 402), which might
explain why a PKC isoform is translocated either to the nucleus or the
plasma membrane. It is possible that the multiple phospholipid-derived
second messengers produced upon ACh stimulation activate different PKC
isoforms that, after being translocated to specific targets, activate
different pathways (403). Such a differential activation
of PKC isoforms has been reported upon glucose stimulation, with
PKC
and
PKC being translocated to the cell periphery and
PKC and
PKC being translocated to perinuclear sites (396).
Many proteins are phosphorylated by PKC in islets (for review, see Ref.
376), but their nature is largely unknown. One identified
target for PKC in ß-cells is the myristoylated alanine-rich C kinase
substrate (MARCKS) (404), a protein that binds actin and
Ca2+-calmodulin and that has been implicated in
cell movement and vesicle transport (405, 406). This
substrate is phosphorylated in response to carbachol (407, 408). Other PKC substrates might be the G proteins that are
associated with the
2-adrenoceptor and
uncouple from the receptor after phosphorylation. This mechanism might
explain how phorbol esters and carbachol reduce the ability of
adrenoceptors to inhibit glucose-induced insulin secretion from rat
islets (409, 410). In view of the importance of the
PKC-dependent pathway in the stimulation of insulin secretion by
ACh (see Section IX.B.1), identification of the targets of
PKC in ß-cells is an important question.
It has been suggested that PKC activation also exerts a negative feedback control on the signal transduction linked to PLC and activated by ACh. Indeed, stimulation of PKC inhibits the production of inositol phosphates induced by cholinergic agonists (290, 312, 322, 352, 393). This effect occurs within 10 min (perhaps within even less time) of stimulation with phorbol esters (352). It likely contributes to the biphasic time course of accumulation of IP3 and arachidonate-enriched DAG upon stimulation with ACh. This PKC-mediated negative feedback might result from the uncoupling of PLC from the ACh receptor (411), either by a direct phosphorylation of PLC by PKC (412) and/or Ca2+-calmodulin kinase (413), by a modification of the G protein coupling the ACh receptor to PLC (414, 415), or by phosphorylation of the receptor itself (416). A muscarinic receptor kinase has recently been identified that might fulfill this role leading to decreased PLC activity (315, 417). Alternatively, cholinergic stimulation could cause a down-regulation of muscarinic receptors via (312, 418) or independently (419) of PKC activation.
B. Activation of PLA2
PLA2 enzymes hydrolyze the sn-2 ester
linkages in phosphoglyceride molecules to release a lysophospholipid
and a free acid, such as arachidonate (Fig. 3
). They can hydrolyze
various substrates, such as PC, phosphatidylethanolamine,
phosphatidylserine, PI, PA, and plasmalogens (420).