Basic Approaches to Unraveling SSC in Single Endocrine Cells

The study of mechanisms of hormone release from ESCs was once a slow, arduous, and uncertain endeavor: it required the measurement of hormone release from many thousands of cells into the bulk solution, often after very unphysiological treatment of cells (e.g., total membrane permeabilization with detergent), using finicky and time-consuming radioimmunoassays often dependent on species-specific antibodies. Fortunately, over the past 25 years, several new tools have made possible the real-time, single cell study of hormone exocytosis, with nearly the ease, speed, and accuracy previously attainable only at synapses, where both presynaptic and postsynaptic synapses are sometimes assayable by simultaneously recording current and voltage (e.g., Ref. 28). The maintenance of long-term exocytosis requires endocytosis, or the retrieval of granule membrane from the surface, so as to preserve cell surface area, volume, and excitability as well as to recycle some granule components, although we shall not deal with this directly in this review.

The exocytosis of a secretory granule consists of two sequential processes: the fusion of the granule membrane with the plasma membrane producing an increase in the surface area of the plasma membrane followed by the discharge of the packet of to-be-secreted substance contained within the granule. Conceptually, exocytosis can be tracked in real time in several different ways (see Fig. 2A). The increase in the surface area of the plasma membrane can be detected electrically as an increase in plasma membrane capacitance (C_m), where C_m is the amount of charge needed to change the membrane potential by a fixed amount and, hence, is directly proportional to the surface area to be charged. Measurement of a stepwise increase in C_m of magnitude predictable from the surface area of a secretory granule is highly suggestive of the fusion of the membrane of the granule with the plasma membrane. In exocrine or endocrine cells, the fusion of a 300–nm-diameter granule with the plasma membrane often yields an electrically detectable (2–3 fF) step in C_m (e.g., Refs. 26, 33, and 36). The packet-like (or quantal) release of the thousands of molecules of hormone contained within a granule [or quantal release event (QRE)] can be detected electrochemically as the synchronized oxidization of C-OH to C=O in the cases of epinephrine, norepinephrine, dopamine, or serotonin or the reduction of a disulfide S-S bond in the case of insulin (8, 54, 55). Here, the tip of a small-diameter (1–10 μm) carbon fiber electrode, positioned directly over the release site, is held at a fixed potential to achieve maximal oxidation/reduction, a technique known as amperometry. In some ESCs, even a single action potential (AP) simultaneously evokes a small step in C_m and one or more accompanying QRE(s). Also, the approach and fusion of secretory granules with the plasma membrane can be detected optically using total internal reflectance microscopy to collect light emission from a region of a cell near the plasma membrane, after the loading of secretory granules with a membrane-permeant fluorescent dye or transfection of the cell with DNA for a fluorescent granule cargo (30, 31, 49). Combining amperometry and/or optical tracking with capacitance tracking provides a way of insuring that the fusion event actually represents the exocytosis of a transmitter-filled secretory granule rather than the fusion of a random endosome with the plasma membrane. Singly or in combination, these techniques can be used to measure stable (34) or cyclical (31) patterns of exocytosis over tens of minutes.

Open in a separate window

Fig. 2.

A primer on exocytosis. A: techniques for the stimulation and real-time monitoring of single cell exocytosis in endocrine cells. The monitoring methods shown are 1) patch-clamp electrophysiology to dissect tiny membrane capacitance (C_m) changes signifying granule membrane insertion into the plasma membrane; 2) optical fluorescence imaging (hν_emit) of the contents of the granule to visualize the approach of the secretory granule and the subsequent pore formation resulting in the release of the granule contents; and 3) electrochemistry, namely, the amperometric detection of quantal release of oxidizable/reducible molecules within the granule. All three approaches allow the tracking of fusion pore formation and expansion. Prolonged amperometrically measured pore flickers, often called amperometric “feet,” which stand alone and do not result in subsequent spikes, may constitute a type of rapid opening/closing of the fusion pore (56). Stimulation methods include the following: application of agonist X, resulting in cell depolarization (pathway 1a) or the receptor-mediated release of Ca²⁺ or other second messengers into the cytoplasm (pathway 1b); introduction into the cell cytoplasm of secretagogue intermediates by diffusion from a pipette (pathway 1c) or membrane permeation (pathway 1d); direct depolarization of the patch-clamped cell (pathway 2); and exposure of the cell to a flash of light (hν_excit), which can liberate Ca²⁺ or another second messenger compound (e.g., IP₃) from a photolysis-sensitive caging compound (X) previously injected into the cytoplasm. QRE, quantal release event; I_amp, current measured by amperometry. B: sequence showing the presumed docking (1), initial fusion pore formation with the transient release of some free hormone (from 1 to 2), and subsequent fusion pore expansion (3) with the release of the remaining free hormone as well as the matrix core and matrix-trapped hormone(s). Pore flickers likely represent fast transitions between 1 and 2. SNARE, receptor for N-ethylmaleimide-sensitive attachment protein.

Single cell exocytosis can be stimulated in several ways. The application of a natural or pharmacological stimulus can result in cell depolarization (Fig. 2A, pathway 1a) or receptor-mediated release of Ca²⁺ or other second messengers into the cytoplasm (Fig. 2A, pathway 1b) (e.g., Ref. 55). These actions can be mimicked by second messengers introduced into the cytoplasm by diffusion from the pipette (Fig. 2A, pathway 1c) (e.g., Refs. 37 and 41–43) or membrane permeation (Fig. 2A, pathway 1d). Alternatively, direct depolarization of the patch-clamped ESC can produce measurable Ca²⁺ entry-dependent exocytosis (Fig. 2A, pathway 2) (35, 36). Finally, exposure of the cell to a flash of light can liberate Ca²⁺ or another second messenger compounds [e.g., inositol (1,4,5)-trisphosphate (IP₃)] from a photolysis-sensitive caging compound previously introduced into the cytoplasm, whereupon the second messenger then widely distributes within the cytoplasm (Fig. 2A, pathway 3) (9, 51–53).

The processes of membrane fusion and release of granule contents are very complex. Once routinely referred to as “a miracle,” the basic steps have been slowly, although not totally, unraveled (see Fig. 2B). The fusion process itself involves at least the following steps (see Ref. 20 for an overview). Initially, the secretory granule is docked and primed at the plasma membrane by the interaction of sticky receptor for N-ethylmaleimide-sensitive attachment (SNARE) proteins, present on both membranes, which allows the close approach of the leaflets of the opposite bilayers. Later, the fusion pore is formed, which might be lipidic (i.e., formed by the contact of bilayer leaflets) or might be based on protein scaffolds (i.e., formed by the intramembranous domains of SNARE proteins), and provides the pathway of exit for granule contents. When exocytosis is triggered by a rise in cytosolic [Ca²⁺], the fusion pore may be catalyzed by the rearrangement of the intramembranous domain of a Ca²⁺ sensor protein, synaptotagmin. The fusion pore is a dynamic structure: it starts narrow, flickers open, closes several times to allow the exchange of granule Ca²⁺ for extracellular Na⁺, the influx of water, and a trickle of release of soluble contents, and then finally expands fully, coincident with granule swelling, to allow most of the granule contents to diffuse or ooze out (e.g., Ref. 56). In cells where the granule content is a tightly packed gel or matrix, which must expand and decondense before the dissociation of the hormone, the release of hormone into the extracellular space may lag behind a stable increase in C_m by up to seconds (30). Exocytosis is followed by endocytosis, where complex cytoskeletal machineries, consisting of actin, dynamin, myosin, and myosin light chain kinase, among other components, retrieve the granule membrane.

SSC in a “Model” Endocrine Cell: Evidence for Distinct Pools of Granules in the Catecholamine-Releasing ACC

The ACC, the most neuron-like of endocrine cells, has long been the ESC of choice for the study of SSC. As a neural crest-derived, cholinergically innervated, electrically excitable cell, long considered analogous to postganglionic sympathetic neurons, the ACC is triggered by ACh to fire APs (13) and secrete catecholamines by exocytosis (36). As shown in Fig. 3, left, early combined patch-clamp/exocytosis studies have shown that the binding of a cholinergic agent such as carbamylcholine (CCH) to a nicotinic ACh receptor channel (nAChR; step 1) produces membrane depolarization (step 2), resulting in electrical activity (step 3), opening voltage-activated Ca²⁺ entry (step 4), a rise of local cytosolic Ca²⁺ (step 5), and ultimately Ca²⁺ entry-induced exocytosis of catecholamines (step 6). Use of the C_m assay to compare the rates and overall quantities of exocytosis evoked by depolarization-induced Ca²⁺ entry versus flash photolysis-induced liberation of Ca²⁺ suggested that during depolarization-induced exocytosis, local cytosolic [Ca²⁺] rose as high as many micromolars in a small spatial domain, being maximal near the mouths of a cluster of Ca²⁺ channels but extending toward nearby docked secretory granules (9). Hence, the depolarization-induced, Ca²⁺ entry-dependent mode of SSC appeared to resemble that occurring in nerve terminals, where smaller-diameter vesicles are clustered at membrane hot spots known as active zones. However, it quickly became clear that chromaffin cells were quite distinct from nerve terminals: CCH can trigger chromaffin cells to secrete catecholamines in the depolarization-independent mode of exocytosis even after nAChRs are desensitized (55). As shown in Fig. 3, right, the binding of CCH to a G protein-coupled muscarinic ACh receptor (mAChR; step 1) produces the activation of PLC (step 2), resulting in the release of IP₃ from the plasma membrane (step 3). The binding of IP₃ to a Ca²⁺-release channel in the Ca²⁺-storing regions of the endoplasmic reticulum (ER) releases Ca²⁺ into the bulk of the cytoplasm (Fig. 3, right, step 4). Although only raising local cytosolic [Ca²⁺] near granules into the range of several micromolars (40) (Fig. 3, right, step 5), this internal Ca²⁺ release evokes major exocytosis (Fig. 3, right, step 6). Hence, the release of ACh from innervating neurons results in subsequent chromaffin cell exocytosis via two separate receptors that are linked with two separate effector systems that ultimately raise cytosolic Ca²⁺ in the vicinity of secretory granules over a wide range of Ca²⁺ concentrations. While both modes of exocytosis are Ca²⁺ dependent, the mAChR-evoked sequence is depolarization independent but internal Ca²⁺ release dependent.

Open in a separate window

Fig. 3.

Two modes of stimulation of adrenal medullary chromaffin cells (ACCs) resulting in vigorous quantal secretion of catecholamines measured by amperometry (I_amp). Left: initial application of the cholinergic agonist carbamylcholine (CCH) evokes the depolarization-induced, Ca²⁺ entry-dependent quantal release of catecholamine (as measured amperometrically, top). In the schematic immediately below, CCH activates nicotonic ACh receptors (nAChRs; step 1), producing membrane depolarization (step 2), trains of APs based on the opening of voltage-dependent Na⁺ and Ca²⁺ channels (steps 3 and 4), Ca²⁺ entry (step 5), and Ca²⁺ entry-dependent exocytosis [QREs of norepinephrine (NE) and epinephrine (E); step 6]. V_m, membrane potential. Right: after nAChRs have desensitized, subsequent larger applications of CCH result in a more slowly developing barrage of QREs by a depolarization-independent, Ca²⁺-release-dependent process. Binding of CCH to G protein-coupled muscarinic ACh receptors (mAChRs; step 1) activates PLC (step 2), releasing IP₃ (step 3), which triggers Ca²⁺ release from ER stores (step 4) and ultimately results in exocytosis (step 5). Each amperometric spike (or QRE) consists of the release of 0.5–2 million molecules of epinephrine or norepinephrine. Ca(L), L-type Ca²⁺ channels.

This brings us face to face with a paradox. How do both modes of SSC ultimately discharge chromaffin granules at high frequency, although at different kinetics of onset, even though each mode would be expected to evoke very different spatiotemporal profiles of cytosolic Ca²⁺ increases? Ca²⁺-releasing regions of ER are likely to be situated up to a micrometer from the granules and, even when fully emptied of their stores, raise “near-granule” cytosolic [Ca²⁺] to only a few micromolars, whereas AP-provoked Ca²⁺ entry raises cytosolic [Ca²⁺] in the vicinity of Ca²⁺ channels to tens of micromolars. Are the granules and Ca²⁺ channels less well colocalized in chromaffin cells than in nerve terminals? Are there two distinct sets of granules with different localizations vis-a-vis Ca²⁺ channels and with two very different Ca²⁺ sensitivities?

Consideration of carefully performed experiments using flash photolysis of caged Ca²⁺ as the secretory trigger has suggested a model consisting of two distinct pools of secretory granules individually or jointly activated by distinct stimulus patterns and giving rise to distinct exocytotic responses (55). This model has been examined in other ESCs as well (33, 52). As in neurons, there is a small immediately releasable pool of granules (IRP) with low Ca²⁺ affinity situated near or docked at clusters of voltage-dependent Ca²⁺ entry channels (9, 19). Granules in the IRP would likely be activated by individual APs, which transiently raise local Ca²⁺ into the range of many micromolars over narrow spatial domains near the open Ca²⁺ channels. However, these local domains of greatly enhanced [Ca²⁺] collapse rather quickly and, at most, extend a distance of hardly more than tens of nanometers from the mouths of open Ca²⁺ channels. That is, Ca²⁺ is rapidly buffered by mobile and fixed buffers, extruded across the plasma membrane or sequestered into Ca²⁺ stores (see the general model in Fig. 4A and the stimulus-response paradigm in Fig. 4B, top). Hence, in response to a single AP, release from the IRP begins within milliseconds and produces a nearly synchronous exocytosis, indicated by a step in C_m and coincident amperometrically detectable QRE(s) (not shown here). However, more prominently than in most neurons, in ACCs there is a much larger pool of granules of higher Ca²⁺ affinity, the so-called highly Ca²⁺-sensitive pool (HCSP), located farther from clusters of voltage-dependent Ca²⁺ entry channels (53). This pool is activated by repetitive or prolonged depolarization, producing long-term Ca²⁺ entry, saturating local cytoplasmic Ca²⁺ buffers, and allowing Ca²⁺ to diffuse over a longer distance, and equilibrate, at low concentrations, over a portion of the cytoplasm, especially its outermost, immediately submembrane ring. Ultimately, this produces delayed asynchronous exocytosis lasting up to several seconds after the depolarization has ceased.

Open in a separate window

Fig. 4.

Hypothesis for how two pools of secretory granules located at different distances from Ca²⁺ entry channels and having distinct Ca²⁺ sensitivities can account for a wide range of stimulus-secretion coupling (SSC) in ACCs. A: model of two pools. The first is an immediately releasable pool (IRP) with low Ca²⁺ affinity (depicted as a shallow Ca²⁺ threshold) closely colocalized with (or docked at) plasma membrane Ca²⁺ channels through which voltage-dependent Ca²⁺ current [I_Ca(v)] enters. The second is a highly Ca²⁺ sensitive pool (HCSP) with much higher Ca²⁺ affinity (depicted as a deep Ca²⁺ threshold) located at some distance from both Ca²⁺ channels and ER Ca²⁺ stores. Ca²⁺ diffusing through the cytoplasm is shown as encountering and binding to mobile and fixed Ca²⁺ buffers (denoted as <Ca²⁺>) and as being extruded across the plasma membrane or sequestered back into the ER, both by Ca²⁺ pumps. [Ca²⁺]_i, intracellular [Ca²⁺]. B: three patterns of stimulation resulting in distinct exocytotic responses reflected as C_m increases [stimulus → C_m response patterns (left) and associated with distinct spatiotemporal profiles of free cytosolic [Ca²⁺] (Ca²⁺ profiles, right)]. Top, an AP with brief underlying I_Ca(v), physiologically evoked by a brief pulse of ACh activating nAChRs, leads to a brief microdomain-like rise in local cytosolic Ca²⁺, exceeding the threshold for the activation of neighboring granules of the IRP; it rapidly results in a stepwise synchronous increase in C_m (C_m sync), representing the exocytosis of one or more granules. Middle, a train of APs, physiologically evoked by a longer pulse of ACh repeatedly activating nAChRs, results in the repeated establishment and waning of microdomain rises in local Ca²⁺ profiles that widen with each successive AP (compare AP1 with AP3). By the conclusion of the third AP, the Ca²⁺ profile is sufficiently wide that the threshold for triggering release from HCSP is exceeded. The combined release of granules from IRP and HCSP is evidenced as C_m synchronous steps representing the exocytosis with each AP of one or more granules as soon as these events can be recorded and smaller asynchronous C_m steps (C_m async) occurring after the end of the train and representing the exocytosis of single granules. Bottom, a longer pulse of ACh, activating mAChRs, results in the sustained release of Ca²⁺ from stores that saturates local Ca²⁺ buffers and results in wider Ca²⁺ diffusion producing a diffuse global rise in cytosolic Ca²⁺ sufficient to exceed the threshold for triggering granules from HCSP. The ultimate result is asynchronous C_m steps lasting many seconds.

It is likely that both the IRP and HCSP should be activated by a train of APs (see the stimulus-response paradigm in Fig. 4B, middle). As with single widely spaced APs, individual depolarizations in the train should transiently raise “near-channel” Ca²⁺ domains, activate granules in the IRP, and produce a synchronous step of C_m during each depolarization. However, additionally, the clustering of the APs should result in a sufficient buildup of Ca²⁺ and give rise to a slow asynchronous component of C_m lasting for up to several seconds after the train has ended (see the stimulus-response paradigm in Fig. 4B, middle) (32, 33, 50, 51). Recently, using genetic knockout techniques, the two granule pools with differing Ca²⁺ affinities have been found to have appropriately distinct isoforms of the Ca²⁺ sensor synaptotagmin (46).

Evidence has suggested that the mAChR-induced Ca²⁺ release by ER stores raises cytosolic [Ca²⁺] at discrete regions near the plasma membrane to values in the range of 1 μM to several micromolars. In our model, this increase should be sufficient to activate member granules of the HCSP and account for mAChR-induced exocytosis with the prolonged application of ACh (see the stimulus-response paradigm in Fig. 4B, bottom). In addition, pharmacologically, the slow sustained exocytosis evoked by α-latrotoxin, a neurotoxin forming long-lived Ca²⁺-permeable plasma membrane channels and producing a continuous rise in global cytosolic [Ca²⁺], should also result from the activation of the HCSP (27).

In addition to nAChR- and mACh-activated pathways evoking release individually, it is likely that the two pathways work in concert under a wide variety of patterns of cell stimulation to increase the efficiency or extent of stimulus-evoked exocytosis, thus priming the cell for subsequent exocytosis in a manner analogous to the short-term facilitation or rapid recovery from post-use depression seen at many synapses. There is evidence that low-level stimulation via the mAChR or nAChR, or low-amplitude depolarizations, which in themselves do not provoke enough of a rise in cytosolic Ca²⁺ to trigger release, nonetheless increase both the synchronous and asynchronous release seen after a train of APs. Here, the IRP and HCSP may be increased directly, by the residual Ca²⁺ left in the wake of plasma membrane Ca²⁺ entry and/or Ca²⁺ release from cytoplasmic stores, or indirectly, by the activation of PKC by residual Ca²⁺ or diacylglycerol generated by PLC (47).

Two complicating features merit comment. First, as chromaffin cells contain a variety of Ca²⁺ channels, it is likely that some types of Ca²⁺ channels are more closely colocalized with granules than others, and their preferential opening would be more likely to trigger release from the IRP. In addition, it is likely that some types of Ca²⁺ channels may be differentially modulated by neurotransmitters (1). Second, as chromaffin granules contain neuropeptides as well as catecholamines, and the peptides are preferentially released by intense activity patterns, it is not unreasonable to anticipate that intense activity predisposes to wider pore expansions and differentially favors the discharge of peptide contents, perhaps by altering the tug of the cytoskeleton on the evolving fusion pore or on granule membrane endocytosis (10).

SSC in Many Endocrine Cells: Variations on the Chromaffin Cell Theme

Conceptually, SSC in a wide variety of endocrine cells may be viewed as specialized variations on the chromaffin cell's theme of the mixed mode of exocytosis triggering. Here, we provide some key examples.

Cells may adapt the depolarization-induced, Ca²⁺ entry-dependent mode of exocytosis by exchanging the nAChR for another stimulus-receptor channel that provides the initial depolarization.

A prime example of this is the magnocellular neurosecretory neuron (MNN; see Fig. 5A, 1). MNNs, with cell bodies in the supraoptic nucleus and modified nerve terminals in the posterior pituitary, respond directly to increases in serum osmolality as small as 1%. Antidiuretic hormone (ADH) secretion by these cells results in the insertion of water channels, known as aquaporins, into the apical membrane of principal cells of the renal tubule collecting duct, hence completing the pathway for water reabsorption across the collecting duct and then into the capillaries that traverse the hypertonic renal medulla.

Open in a separate window

Fig. 5.

Modular approach to SSC in a variety of key endocrine cells. A: adaptation of depolarization-induced, Ca²⁺ entry-dependent SSC by osmosensitive magnocellular neurosecretory cells of the supraoptic nucleus (SON; 1); volume (stretch)-sensitive atrial myocytes (2); and K⁺-sensitive zona glomerulosa cells of the adrenal cortex (3). ADH, antidiuretic hormone; ANP, atrial natriuretic peptide; T-T, T-tubule; Ca(T), T-type Ca²⁺ channel. B: adaptation of internal Ca²⁺-release dependent SSC by pituitary gonadotrope cells. GnRH, gonadotrophin-releasing hormone; LH, luteinizing hormone; FSH, follicle-stimulating hormone. C: novel external Ca²⁺-inhibited, internal Ca²⁺-independent SSC displayed by chief cells of the parathyroid glands (paths A and B and steps 1–6 are defined in the text). PTH, parathyroid hormone; AC, adenylyl cyclase; CaSR, Ca²⁺-sensing receptor; β-Ad, β-adrenergic receptor.

The cell bodies of MNNs, situated in a region of the hypothalamus locally devoid of the blood-brain barrier, serve as tiny osmoreceptors/stretch receptors (5). In response to increases in serum osmolality, the plasma membrane deforms (or crinkles) slightly (Fig. 5A, 1, step 1), opening specialized nonselective cation channels (NSCCs) usually deactivated at resting cell volume (Fig. 5A, 1, step 2); this depolarizes the cell body and increases its rate of intrinsic AP firing. Worth noting is that the NSCCs, here of the transient receptor potential, vanilloid-type variety, are situated on a region of membrane underlain by an actin-based cytoskeleton. Membrane deformation either activates the channel directly or sets off a mechanosensitive enzyme that does this indirectly. APs propagating into the modified nerve terminal (Fig. 5A, 1, step 3) increase Ca²⁺ entry via voltage-activated Ca²⁺ channels ( Fig. 5A, 1, step 4), thereby stimulating the exocytosis of ADH (Fig. 5A, 1, step 5). Interestingly, although the nerve terminals have large Ca²⁺ currents, significant exocytosis of ADH is “tuned” to the firing of trains of APs, suggesting that the initial Ca²⁺ entry primes the pool of granules available for release as well as immediately triggering release (24).

Inversely, a decrease in ambient osmolality, which causes mild swelling of the cell body, hyperpolarizes it by opening stretch-activated K⁺ channels, thereby decreasing the intrinsic AP frequency. Additionally, these cells respond to decreases in intravascular volume of >10% via a polysynaptic vasomotor command pathway. The release of the neurotransmitter glutamate, from the final neuron of this pathway, results in the opening of transmitter-operated NSCCs.

SSC in the Most Clinically Explored Endocrine Cells, the β- and α-Cells of the Pancreatic Islet of Langerhans: Complexity, Ability to Be Modulated, and Possible Ability to Be Replaced

Pancreatic islets cells as metabolic sensor cells.

With the worldwide epidemic of type II DM, in large part due to increased caloric intake and decreased daily energy expenditure in both developed and developing nations, the normal function and decline of function of endocrine cells of the pancreatic islets of Langerhans has become a public health issue (4). Interspersed in a huge branching exocrine gland, these small islands, of ~1,000 ESCs, regulate metabolite uptake, storage, and release by depot store tissue such as hepatocytes, adipocytes, and skeletal myocytes (44). Sixty to sixty-five percent of islet cells are insulin-producing β-cells. Easily identified by their large diameter, autofluorescence, and distinctive (almost cuboidal) crystals in their secretory granules, β-cells are triggered to secrete by an increase in serum glucose. Twenty to twenty-five percent of islet cells are glucagon-producing α-cells; they are largely triggered to secrete by a decrease in serum glucose. The combination of β-cells with α-cells forms a “push-pull system” with respect to glucose metabolism and general energy homeostasis. The remaining 15% of islet cells are δ-cells or pancreatic polypeptide (PP) cells, which secrete the modulatory peptides somatostatin or PP, which inhibit SSC of both α-cells and β-cells. In its simplest form, β-cell insulin secretion, induced by high serum [glucose], stimulates glucose transport into, while inhibiting its release from, storage cells (hepatocytes and muscle), thereby rapidly reducing circulating glucose levels. In contrast, α-cell glucagon, induced by low serum [glucose], stimulates glycogenolysis in hepatocytes and muscle, thereby releasing glucose and increasing its increasing circulating serum concentration. In DM, glucose-induced insulin secretion is either reduced because β-cells have developed a reduced sensitivity to glucose as a result of their accumulation of lipids (in the case of type II DM) or is abolished because β-cells have been destroyed by an immunological attack (in the case of type I DM) (4). In type I DM, while insulin must be administered exogenously to maintain a semblance of glucose homeostasis, this treatment is often tricky because α-cells may no longer secrete glucagon in response to low glucose, thereby prolonging the recovery from episodes of hypoglycemia (low plasma [glucose]).

What makes α-, β-, and δ-cells of the islet cells so special is that they are all metabolic sensor cells. However, rather than sensing glucose using a specific GPCR linked to an effector cascade, these cells import glucose, aerobically metabolize it, and then use the resultant changes in cytosolic [ATP] and [ADP] to control their respective secretion. In addition, although individual islet cells secrete in response to the appropriate stimuli when in isolation, these cells secrete most potently when they are in clusters containing representatives of each cell type. In the latter case, β-cells are electrically coupled, while α-, β-, and δ-cells communicate via local paracrine secretory interactions. Glucagon enhances stimulus-induced insulin secretion, whereas somatostatin inhibits it. However, insulin, glucagon, and somatostatin all inhibit glucagon secretion.

An overall consensus scheme for SSC in β-cells (e.g., Ref. 32) is shown in Fig. 6A. A rise in serum glucose from a baseline level of 3 mM to stimulatory levels of 5-6 mM enhances glucose importation via a moderate-affinity transporter [glucose transporter (GLUT)] as well as phosphorylation by glucokinase (Fig. 6A, step 1) to permit its intracellular metabolism. Glycolysis and mitochondrial oxidation of 3-carbon glucose fragments results in a small rise in cytosolic ATP and a fall in cytosolic ADP (Fig. 6A, step 2) sufficient to close down ATP-inhibited (and ADP-disinhibited) K⁺ channels, known as K_ATP channels (Fig. 6A, step 3). K_ATP channels consist of an inner ring [a quartet of inward rectifier K⁺ channel subunits (Kir6.2) forming the pore] surrounded by an outer ring of sulfonylurea receptors (SUR; type 1) (2, 37). The binding of ATP to the Kir6.2 core closes, whereas the binding of MgADP to the SUR 1 outer ring opens, the pore in K_ATP channels; the binding of sulfonylurea to SUR 1 closes the pore in K_ATP channels (see Fig. 6C). As K_ATP channels constitute the bulk of the resting K⁺ permeability, their closure, against a background of tonically open NSCCs, depolarizes the membrane (Fig. 6A, step 4), thereby sequentially activating voltage-dependent Na⁺, Ca²⁺, and K⁺ channels (Fig. 6A, step 5), which underlie complex electrical activity. The opening of high-voltage-activated (HVA) Ca²⁺ channels at membrane potentials positive to −45 mV supports the rapid entry of Ca²⁺ (Fig. 6A, step 6) and Ca²⁺-dependent exocytosis of the contents of insulin granules (Fig. 6A, step 7) in a manner reminiscent of chromaffin cells but with subtle differences. To get all the insulin out, it is likely that the entire crystalline contents of the granule must ooze out. In addition, under some conditions, the release may be pulsatile, over periods of tens of seconds, rather than continuous (31).

Open in a separate window

Fig. 6.

A and B: SSC in pancreatic islet β-cells (A) and α-cells (B), both prominent metabolic sensors cells. Both cells use closure of ATP-sensitive K⁺ (K_ATP) channels as a means of coupling a rise in cytosolic [ATP]/[ADP] to cell depolarization and regulation of voltage-dependent Ca²⁺ entry resulting in secretory granule exocytosis. In the case of β-cells, quiescent at resting levels of glucose, closure of K_ATP channels provides the depolarization to open high-voltage-actived (HVA) Ca²⁺ channels. In contrast, in the case of α-cells, already depolarized and exhibiting Ca²⁺ entry-dependent exocytosis at resting levels of glucose, further depolarization inactivates Na⁺ and T-type Ca²⁺ channels, reducing the amplitude of AP underlain by the latter and curtailing exocytosis dependent on Ca²⁺ entry through HVA Ca²⁺ channels (steps 1–6 in both A and B are defined in the text). C: cartoon of gating of K_ATP channels, which consist of an outer ring of four sulfonylurea type 1 receptors (SUR 1), each composed of 17 transmembrane segments, surrounding an inner ring of four Kir6.2 subunits. The binding of ATP to Kir6.2 or sulfonylurea (S) to SUR 1 closes the channel, whereas the binding of MgADP to SUR 1 opens the channel.

Currently, much attention is being focused on mutations of the K_ATP channel as the loci of origin of two neonatal diseases: neonatal diabetes (ND) and its converse, congenital hyperinsulinemia of infancy (CHI) (2, 37). In CHI, K_ATP channels show loss of function; in the case of ND, there is gain of function. In CHI, SUR 1 receptors are often poorly synthesized, assembled, or trafficked to the membrane or else have lost the ability to appropriately respond to MgADP. With functional K_ATP channels absent or largely closed, the depolarizing effect of NSCCs is poorly opposed, and the β-cell remains chronically depolarized and stimulated to secrete insulin, resulting in systemic hypoglycemia. In the case of ND, the best-explored mutations of Kir6.2 show reduced K_ATP channel closure in response to MgATP, presumably leading to inability of the β-cell to depolarize in response to glucose. It is likely that defective K_ATP channels are heteromultimers containing different ratios of wild-type to defective subunits; these ratios may even change with the progression of the disease. Interestingly, both diseases are associated with neurological deficits. In CHI, this is likely due to cytotoxicity produced by repeated episodes of hypoglycemia. In ND, which is often associated with delayed motor and cognitive development and epilepsy, this may be due to reduced excitability of central inhibitory (GABAergic) neurons, skeletal muscle, and terminals of motor neurons, where K_ATP channels make a sizeable contribution to resting membrane permeability.

While α-cells appear to be the inverse of β-cells, in that they cease to secrete as [glucose] rises, in detail their basic working parts are quite similar to those of β-cells… with some interesting twists (see Ref. 14). Although α-cells exhibit similar K_ATP channels as found in β-cells, even at low [glucose] (2-3 mM), α-cells metabolize sufficient glucose to close most of the channels. Having voltage-activated Na⁺ channels and low voltage-activated (T-type) Ca²⁺ channels, in the face of largely closed K_ATP channels, α-cells fire “spontaneous” APs, which trigger Ca²⁺ entry through HVA Ca²⁺ channels and the exocytosis of glucagon-containing granules. As shown in Fig. 6B, an increase of extracellular [glucose] above 4-5 mM closes down the remaining K_ATP channels (steps 1–3) and depolarizes the α-cells into a voltage range (−40 to −35 mV) where Na⁺ and T-type Ca²⁺ channels are largely inactivated (steps 4 and 5) and the AP amplitude is decreased. This reduces the probability that an AP will open HVA Ca²⁺ channels, hence reducing Ca²⁺ entry (Fig. 6B, step 6) and inhibiting glucagon secretion (Fig. 6B, step 7). In addition, α-cells of some species have a high density of GABA receptor channels that carry Cl⁻ current. As neighboring β-cells corelease GABA with insulin, increases in [glucose] that stimulate exocytosis from β-cells may help fix the membrane potential of the α-cell near the Cl⁻ equilibrium potential (approximately −30 mV), thereby also reducing electrical activity of, and glucagon release from, α-cells.

The biphasic time course of insulin secretion and how gut incretins affect it.

β-Cells display two features that are being increasingly appreciated to be very significant clinically (4). In response to prolonged glucose stimulation, β-cells secrete insulin in two phases, a spike (or first) phase of insulin secretion (FPIS) lasting several minutes and a dome (or second) phase of insulin secretion (SPIS) lasting for the remainder of the duration of the glucose elevation. In addition, β-cells secrete insulin more efficiently in response to an oral versus intravenous glucose load, suggesting that the digestion of food primes insulin secretion by evoking the secretion of incretins, which are gastrointestinal tract-associated hormones that enhance glucose-induced insulin secretion at meal time.

Figure 7A, which shows the time course of changes in venous [glucose] and [insulin] after a meal, demonstrates the phenomenon of biphasic insulin secretion. We now know that FPIS is critical for stopping glucose release by the liver and muscle and reducing free fatty acid release from labile visceroabdomenal adipose stores. In contrast, SPIS is critical for stimulating the uptake of newly absorbed glucose into muscle and peripheral fat. In type II DM (which is associated with overweight/obesity), β-cells, filling with fat, display a decreased sensitivity to an initial rise in [glucose] induced by a meal, thereby resulting in markedly reduced FPIS. However, in response to persistently high [glucose] after a meal, β-cells display much enhanced and prolonged SPIS, especially as fat-stuffed myocytes and hepatocytes display decreased glucose uptake in response to the released insulin. However, over time, β-cells (pushed to synthesize and secrete large quantities of insulin) hypertrophy, increasing in cell size and number, suffer damaged from overwork, secrete less robustly, and may even die. Clinically, phasic insulin secretion is rarely examined directly by formal glucose tolerance testing after a meal. Hence, many type II DM patients are first diagnosed as diabetic only after SPIS is greatly reduced and blood [glucose] remains persistently elevated for long periods after a meal.

Open in a separate window

Fig. 7.

Biphasic insulin secretion: its timing, physiological significance, and possible underlying cellular mechanisms. A: in vivo timing of the first (FPIS) and second phase of insulin secretion (SPIS) in normal individuals and those with early type II diabetes mellitus (DM II). B: models for the cell biological/molecular bases of biphasic insulin secretion based on the concept of two pools of insulin secretory granules. These pools include one immediately available for release and another requiring one of the following: refilling of membrane-docked granules from a cytoplasmic pool, via movement along the cytoskeleton (i); a distinct glucose-derived metabolic stimulus to recruit granules into the second pool (ii); a second variety of voltage-activated Ca²⁺ channel, providing an alternative source of Ca²⁺ entry to help refill vesicles of the initial pool (iii); a set of granules docked by different SNAREs at distinct fusion sites (iv); and granules from an HCSP triggered by plateau depolarization (PD) rather than cluster of AP activity (v).

Biphasic insulin secretion is also seen in isolated islets and in individual β-cells, where it has been attributed to insulin release from two different pools of granules: 1) an immediately available but easy-to-deplete pool and 2) a late-to-release but rapidly refillable backup pool (see Fig. 7B). Shortly after learning to isolate and study islets in vitro, the islet pioneer Paul Lacy obtained evidence that SPIS could be inhibited by cytoskeleton-disrupting agents (24) and postulated that the backup pool consisted of granules transported by motors on microtubules from their site of manufacture in the ER/Golgi complex to their site of exocytosis at the plasma membrane (Fig. 7B, i).

Although the mechanism(s) underlying biphasic insulin secretion remain elusive, several newer hypotheses, originating from single cell studies of exocytosis, merit consideration. As shown in Fig. 7B, ii, there is evidence that glucose may generate intracellular signals other than the increased [ATP]/decreased [ADP] that closes K_ATP channels and triggers initial insulin secretion. These signals may help enhance SPIS. Alternatively, there is evidence from gene knockout studies in mice that the activation of R-type (Cav2.3) voltage-dependent Ca²⁺ channels, distinct from the L- or N-type channels usually described, recruits granules into a backup pool and is responsible for SPIS (21) (see Fig. 7B, iii). Also, from another knockout mouse model, there is optical evidence for differential docking of granules at distinct sites, with granules involved in FPIS docked at sites rich in the SNARE protein syntaxin A, whereas those involved in SPIS docked at syntaxin A-poor sites (39) (see Fig. 7B, iv). Finally, there is evidence for differential docking of granules of varying Ca²⁺ sensitivity vis-à-vis Ca²⁺ channels. While a body of evidence from mouse β-cells has suggested that the IRP underlies FPIS (3), emerging evidence from other species is consistent with the HCSP, discharged during prolonged depolarization in several species, underlying much of SPIS (Fig. 7B, v). Unlike mouse β-cells, which display persisitent bursts of electrical activity on exposure to secretagogue concentrations of glucose, single canine β-cells display biphasic electrical activity temporally and pharmacologically correlated with the two phases of insulin secretion (34). Trains of APs, ideal to activate the IRP, occur during FPIS; the sustained phase of plateau depolarization, ideal to activate the HCSP, occurs during the transition to SPIS. Both IRP and HCSP are enhanced by concentrations of glucose that trigger biphasic secretion, perhaps by its ability to increase transport or docking of granules, as Lacy suggested 35 years ago.

Let us return to the issue of incretins and their ability to enhance biphasic insulin secretion. The two most likely incretins are ACh, which is released, as a reflex by the vagus nerve, upon stretching of the gut, and glucagon-like intestinal peptide (GLP)-1, which is released by enterochromaffin-like cells of the gut in response to newly digested nutrients. Both work through GPCRs and ultimately exert a pair of similar effects, although through different second messengers (15, 16). Both enhance the closure of K_ATP channels initiated by glucose metabolism, likely by increasing their phosphorylation; both enhance depolarization-exocytosis coupling, likely by stimulating the recruitment of granules into both the IRP and HCSP (see Fig. 8). Specifically, ACh, via its mAChR, stimulates the activity of CaMK II (12). GLP-1, via its GPCR, activates adenylate cyclase, increasing cytosolic [cAMP], thereby stimulating either classical PKA or the cAMP-regulated guanidine nucleotide exchange factor known as Epac (18). GLP-1 has received the bulk of the attention for several reasons: its effect is more targeted to β-cells than ubiquitously acting ACh, its secretion appears to be blunted in type II DM, and, additionally, it may reduce apoptosis in β-cells exposed to high [glucose]. Efforts to enhance circulating [GLP-1] has entered mainstream type II DM therapy as a key to normalizing insulin secretion and potentially delaying the progression of β-cell damage. Two classes of drugs now in development/early use are 1) GPL-1 mimetics, or long-acting recombinant GLP-1 analogs; and 2) GLP-1 enhancers, or inhibitors of the dipeptide peptidases that breakdown circulating GLP-1. The advantages of these drugs over earlier insulin secretion enhancers, such as sulfonylureas (which bind to and close K_ATP channels even in the absence of glucose), are both predictable and novel (17). As GLP-1 mimetics/enhancers modulate glucose-induced K_ATP channel closure rather than bypassing it, they are less likely to cause hypoglycemia. As the latter agents work to recruit granules into the readily releasable pools, they are less likely to cause pool exhaustion. As these agents appear to extend the life of β-cells, they may retard the intrinsic progression of the disease. Finally, as GLP-1 mimetics/enhancers may act in the central nervous system to suppress appetite, they may ultimately help bring about a better caloric balance. The critical and as-yet-unresolved issue is whether GLP-1 receptors themselves become severely downregulated in type II DM.

Open in a separate window

Fig. 8.

Cellular basis for a role of incretins in enhancing biphasic insulin secretion. A: outlines of how glucagon-like peptide (GLP)-1 and ACh are released by gut enterochromaffin-like cells (ECLs) and the vagus nerve reflex, respectively (each likely activated by some combination of gut stretch, increase in luminal [glucose], and osmolality). Each of these secretions enhances glucose-induced closure of K_ATP channels and accelerates insulin granule recruitment into release-ready pools. CNS, central nervous system. B: traces obtained from canine islet β-cells showing the effects of GLP-1 on stimulus-depolarization coupling (1) and depolarization-exocytosis coupling (2). GLP-1 speeds cell depolarization and the initiation of a moderate frequency of AP activity after a transition from 3 to 15 mM extracellular glucose (1). GLP-1 also enhances the exocytotic response (C_m increases) in response to a train of five 200-ms depolarizations from −70 mV to +10 mV applied at 1 Hz without increasing Ca²⁺ currents. Note that GLP-1 enhances the slow continuous C_m increases seen after the third and fourth depolarizations, at high gain seen to consist of small asynch C_m steps, likely indicating greater recruitment and utilization of the HCSP, as well as enhancing the synchronous C_m step increases seen after the first and second depolarizations, likely indicating greater recruitment and utilization of the IRP. [S. Misler and D. W. Barnett, unpublished observation.]

Returning to our theme of SSC, evidence from clonal cell lines has suggested that GLP-1-secreting enterochromaffin-like cells, sparsely distributed endocrine cells of the gut mucosa, respond to glucose not by metabolizing it but by transporting it. Na⁺-glucose cotransport provides both a source of direct electrogenic depolarization and glucose entry for the closure of K_ATP channels. Depolarization sets off electrical activity and Ca²⁺ entry-dependent GLP-1 granule exocytosis (50). Perhaps gut stretch and the increasing osmolality occurring with progressive digestion also activate stretch-modulated channels in these cells and serve as backup sources of depolarization.

Implications for Translating Basic Physiology to the Clinics

In this review, we have demonstrated that fundamental mechanisms of SSC are shared by a variety of ESCs with widely differing sensory apparati and secretory products. Cells appear to adapt basic modular elements of design to achieve specific functions. Additionally, we have demonstrated how understanding the specifics of these mechanisms has contributed to the development of new pharmacotherapies for several wide-ranging clinical problems, the two prominent examples being calcimimetics, used to treat the osteodystrophy of chronic kidney disease, and GLP-1 mimetics/enhancers, used to treat type II DM. Furthermore, in single-gene defect diseases of ESCs, such as ND and CHI, it is likely that an intricate understanding of mechanisms of SSC should help to uniquely target therapy (e.g., sulfonylurea therapy in ND).

It is intriguing to speculate that the basic understanding of SSC in endocrine cells might bring to fruition a “holy Grail search” that has motivated endocrinology since its beginnings: cell replacement therapy for failing hormone-secreting organs. The latest incarnation of this approach is islet transplantation for brittle type I DM patients, for whom insulin replacement therapy, even by infusion pumps and coupled with frequent blood sugar checks to close the feedback loop, still results in frequent episodes of debilitating hypoglycemia. Unfortunately, even in this carefully chosen population, the current approaches for intrahepatic transplantation of islets have largely transitory therapeutic effects while carrying the attendant risks of immunosuppression and perhaps even reawakening the inciting autoimmune response (45). Against the background of our increased understanding of SSC, the challenge remains for the identification and controlled proliferation of endocrine stem cells or the genetic engineering of cells to “turn on” of key elements of the secretory apparatus expressed in SSCs. Realizing the similarities of basic modular processes in SSC, a general advance in the guidance of development or the genetic engineering of any one type of secretory cell may be widely applicable to several others.

In conclusion, the study and understanding of the cell physiology and biophysics of SSC in endocrine cells, which in the past has been a side avenue of experimental endocrinology, may have an illustrious future with wide-ranging therapeutic potential.