I. INTRODUCTION

Top Previous Next References

Calcium is an essential second messenger (18, 24, 56). This ion mediates the major functions of exocrine epithelial cells. Most importantly, intracellular calcium regulates both fluid and exocytotic secretion. Exocrine epithelial cells display structural polarity, functional polarity, and polarity of calcium signals. Although the polarity of the cells can often be seen using conventional microscopy, the complex spatiotemporal organization of calcium signals was only revealed following relatively recent developments in technology. These findings led to a rapid expansion of the epithelial research field, and there is now a large group of epithelial cell types in which many different forms of polarized calcium signals have been described. Information on these signals has come from very many sources and, although there is an obvious similarity between such signals, comparison is not easily made when the data are so widely spread. This review brings together information about different types of polarized calcium signals, the mechanisms underlying them, and their functional consequences for secretory epithelial cells.

Section II of the review describes the techniques that have molded the polarized signaling research field. We highlight those technological advances that have been most important and discuss in detail some specific applications of that technology. We also outline the latest technical developments and try to assess their possible impact on the field.

Section III discusses the polarity of intracellular calcium signals. Secretagogues induce release of calcium from internal stores resulting in very specific patterns of calcium oscillations and/or waves. How these signals are generated and controlled in various types of secretory epithelial cells is reviewed. In particular, we consider the correlation between the distribution of calcium-handling mechanisms (pumps and channels) and the shape of the calcium signals.

Many of the mechanisms underlying intracellular calcium signals also have roles in the movement of substantial amounts of calcium from the interstitial fluid to the lumen of epithelial tissue. We discuss the different stages of this transcellular transport in section IV. At the end of this section, we consider the implications of the transport process for the pathology and physiology of exocrine glands.

Cells within epithelial tissue tend to be highly connected, forming multicellular secretory units. Calcium signals are able to pass between cells and across the tissue as a whole. These intercellular calcium waves are described in section V. We discuss the polarity of the waves and, importantly, we note the implications of intercellular communication for the sensitivity of the tissue to secretagogues.

In section VI, we move one step downstream of the calcium signaling cascade. Calmodulin is one of the major targets for calcium signals, and important work on the activation of calmodulin and calmodulin-dependent pathways has been carried out in secretory epithelial cells. The translocation of calmodulin and functional polarity of calmodulin signaling are considered.

	II. TECHNIQUES USED IN STUDIES OF POLARIZED CALCIUM AND CALMODULIN SIGNALING

Top Previous Next References

Before the advent of cellular calcium measurements with fluorescent probes, ⁴⁵Ca²⁺ was extensively used to investigate calcium fluxes. ⁴⁵Ca²⁺ measurements are not suitable for monitoring intracellular Ca²⁺ responses, particularly at the subcellular level. ⁴⁵Ca²⁺ measurements did provide, however, a considerable amount of useful information on calcium fluxes from exocrine glands or suspensions of isolated exocrine secretory cells (68, 205, 305, 307).

The correlation between ⁴⁵Ca²⁺ fluxes and the functional consequences of calcium signaling were assessed by monitoring of amylase secretion (307). Amylase secretion from isolated, intact pancreas has also been measured in other studies using various assays (140, 181, 213). Functional tests of secretion are also available for other exocrine tissues. For example, Canaff et al. (35) measured the level of bile flow to assess the effects of in situ stimulation of a calcium-sensing protein in hepatocytes.

Microelectrode techniques yielded important early information on calcium-dependent conductance changes in exocrine cells (230). The further development of electrophysiological techniques, including patch-clamp techniques (110), had a profound effect on epithelial biology and, in particular, on our understanding of calcium handling in secretory epithelial cells. These techniques have permitted detailed electrophysiological analysis of calcium-dependent ionic currents (13, 58, 77, 137, 170, 228). Also, the relationship between intracellular calcium level and activation of some channels (e.g., Ca²⁺-dependent Cl channel in pancreatic acinar cells) has permitted indirect monitoring of the calcium level by the whole cell, voltage-clamp technique (228).

Patch-clamp techniques have been extensively used to characterize the channels that are responsible for Ca²⁺ influx in nonexcitable cells (215). The functional consequences of calcium signals were also assessed using electrophysiological techniques. As vesicles fuse with the plasma membrane during exocytosis, the area, and hence the capacitance, of the membrane is increased. The link between exocytosis and calcium transients was characterized using capacitance measurements (124, 168, 169, 236). Measurements of the activity of Cl channels have provided important information about calcium-dependent activation and regulation of fluid secretion in exocrine cells (137, 228).

Electrophysiology has played an important role in gaining a general understanding of the mechanisms of calcium release. Detailed characterization of the inositol 1,4,5-trisphosphate (IP₃) receptors, channels that mediate calcium release from the endoplasmic reticulum (ER), and their calcium sensitivity was made in electrophysiological studies of the proteins in reconstituted lipid bilayers (20). Single-channel recordings on the surface of ER have also allowed characterization of release channels from internal stores (162).

Optical techniques developed during the last 15 years have allowed direct visualization of calcium signaling at the cellular and subcellular level. The optical probes used for calcium measurements fall into one of three categories: luminescent photoproteins (e.g., aequorin), fluorescent proteins (e.g., cameleons), and fluorescent dyes (e.g., fura 2). Each of the three approaches has advantages and drawbacks (for review, see Refs. 27, 276). It is the development of optical techniques that has been the basis of our understanding of polarized calcium and calmodulin signaling. Therefore, the remainder of this section focuses on such techniques.

The early direct intracellular Ca²⁺ measurements in secretory epithelia used luminescent proteins (67). The discovery of calcium oscillations was made in secretory epithelial cells (hepatocytes), also using a luminescent probe aequorin (309). Since then, however, the fluorescent dyes have become the standard probes used in secretory epithelial investigations. This has happened because secretory epithelial cells are relatively small and are difficult to microinject. The other reason is that most primary isolated secretory epithelial cells are difficult to transfect and maintain in culture. Microinjection and transfection are two of the most frequently used techniques for intracellular protein loading. These techniques of loading are difficult to implement on secretory epithelial cells.

Fluorescent dyes can be introduced into cells directly using microinjection or diffusion from a pipette in whole cell, patch-clamp configuration. Most commonly and simply, however, cells are loaded by incubation with the acetoxymethyl ester (AM) form of the dye, which is membrane permeant. Once inside the cell, the dye becomes trapped following cleavage of the AM group by the cell's own esterase enzymes (276). AM loading can, however, present some problems as dyes may become sequestered into intracellular compartments. Such compartmentalization can lead to difficulty in interpretation of the Ca²⁺ signals measured. However, as described later, the tendency of some dyes to specifically compartmentalize during AM loading can be used to the experimenter's advantage (118, 120).

Fluorescent dyes can be placed into three categories based on their spectral properties: 1) dual-wavelength excitation (e.g., fura 2), 2) dual emission (e.g., indo 1), and 3) single wavelength (e.g., fluo 3, CaGreen, fura red).

The dual-wavelength indicators display spectral shift (along the wavelength axis) upon calcium binding. The major advantage of dual-wavelength dyes is that they allow ratiometric analysis of fluorescence, which can minimize inconsistencies produced by dye loading, leakage, bleaching, and cell movement artifacts (154, 276).

An interesting property of fura 2 and indo 1 is that Mn²⁺ will quench their fluorescence. This has been put to good use in studies of store-operated calcium entry (32, 166, 262). The calcium channels in internal stores are also permeable to Mn²⁺, so measurement of the quench effect allowed Thomas and collaborators (237) to study opening of intracellular calcium channels.

Probes have been developed that can be used to measure Ca²⁺ in different cellular compartments, such as the ER and mitochondria. Some of the probes used for ER calcium measurements were originally designed as Mg²⁺ indicators: mag-fura 2, mag-indo 1, mag-fura red, and mag-fluo 4. They have low affinity for Ca²⁺ and should not usually react to calcium changes in the cytosol. However, in the calcium stores (most importantly the ER) where the calcium concentration is relatively high, these indicators, although designed as Mg²⁺ probes, behave as calcium indicators. There is also now a range of indicators with intermediate affinity to calcium and low affinity to magnesium (e.g., fura 2-FF, fluo 3-FF) which should be even more suitable for measurements of calcium in internal stores (118). Indicators can be loaded into ER by incubation with appropriate AM forms. Normally, to reduce cross-compartment signal interference, it is necessary to wash out the dye that has remained in the cytoplasm. Various approaches have been reported to achieve this including permeabilization (120, 301, 325) and dialysis via patch pipette in whole cell patch-clamp configuration (218).

Rhod 2-AM, by virtue of its positively charged nature (in the AM form), is compartmentalized into negatively charged organelles. This leads to the dye being trapped predominantly in the mitochondria after AM loading. By careful adjustment of loading protocols, it has proven possible to measure mitochondrial calcium transients using rhod 2 (69, 216).

Single cell measurements using fluorescent calcium probes require a fluorescent microscope with appropriate light detector (usually photomultiplier). The distribution of fluorescence in different regions of the cell can be imaged using fluorescent microscopes equipped with a charge-coupled device (CCD) camera (154). The increase of sensitivity of the cameras can be achieved by cooling or by using optical intensifiers. Deconvolution techniques can be applied to decrease contamination of images by out-of-focus fluorescence (256). Smaller and discrete calcium signals (for example, small cytosolic signals or signals in mitochondria) become difficult to assess accurately due to interference from so-called "out-of-focus" light. This is the light emitted from above and below the focal plane that adds an inherent noise that may mask the fluorescence signal coming from the structures in the focal plane. It is in this respect that confocal microscopy has distinct advantages over conventional imaging techniques. Confocal microscopy can increase vertical resolution (sometimes termed "Z" resolution) of fluorescent images more than 10 times. The principles of confocal microscopy are well described in several reviews (241, 270) and will not be discussed here. Most importantly, confocal microscopy allows higher signal-to-noise ratios of recordings of small subcellular Ca²⁺ signals and more accurate localization of organelles involved in calcium signaling. Most modern confocals are of the laser scanning variety; that is, the laser excitation light is scanned across the sample to produce the image. Recent technological developments in control of this scanning have dramatically increased the speed with which images can be acquired. This means that it is possible to follow events at the same rate as conventional imaging while maintaining the high spatial resolution (154). The improved control of the excitation light has also been effectively put to use in studies based around rapid and localized changes in excitation. These include fluorescence recovery after photobleaching (FRAP) experiments in which it is possible to monitor the movement of a fluorescent molecule into a region that has previously been photobleached by transient application of very strong excitation light. This allows insights into the speed of translocation of molecules and the connectivity of different regions of the cell. An extension of the same process is used for uncaging of caged compounds. These compounds are specialized derivatives of molecules (e.g., cell signaling messengers) that are attached to an ultraviolet (UV)-sensitive group which effectively "inactivates" the bioactive molecule. When the caged compound is exposed to UV light, the caging group is removed and the signaling molecule is free to complete its function. The most common of the caged compounds is caged Ca²⁺, which has been used to elevate Ca²⁺ inside the cell in a controlled manner that is independent of the usual calcium release pathways (1, 72). IP₃, cAMP, cADP-ribose, and nicotinic acid adenine dinucleotide phosphate (NAADP) are also available in caged form (112). The combination of carefully controlled UV exposure and simultaneous confocal monitoring of the effects of uncaging allows the experimenter to dissect intracellular signaling pathways and localize them within the cell.

Multiphoton (MP) microscopy is the latest innovation in imaging technology to have a significant impact on this field (64). MP uses long (infrared; IR) wavelength laser light to excite the fluorophore. Excitation is achieved when two (or more) photons hit the fluorophore and become absorbed at almost exactly the same time. The two-photon reaction depends nonlinearly on the intensity of excitation light; it occurs effectively only when the density of photons is very high. Therefore, excitation is only achieved in that very small volume where the laser is focused. Emitted light will only come from this region and so can be collected without a pinhole (since there is no out-of-focus light). Another important advantage of two-photon microscopy is that its depth of penetration is more than double that of confocal. Because IR excitation light is absorbed or scattered by tissues less than visible light, it is possible to scan areas >100 µm deep into the sample. This has been used in other tissues (for review, see Ref. 142) but could be particularly well-suited to secretory epithelia since it offers the possibility of studying intact secretory units or even tissue in situ (54). Principles, advantages, and limitations (photobleaching, phototoxicity) of two-photon microscopy have been considered in detail in a number of recent publications (141, 202) and are not discussed here.

As yet, there are few examples of this technique applied to signaling in secretory epithelia. However, those studies that have been published are significant and highlight the possibilities of the technique. The molecular structures of the cytoskeleton (microtubules/actin) and ER strands that underlie the polarity of pancreatic acinar cells were visualized by Fogarty et al. (82), and the importance of these structures to normal calcium signaling was demonstrated. Very recently, in the same cell type, Kasai's group has used MP microscopy to simultaneously visualize the exocytosis of individual granules and measure intracellular Ca²⁺ concentration ([Ca²⁺]_i). This led them to important insights into the link between different calcium signals and secretion and into to mechanism of secretion itself (200). Both these studies were dependent on the accurate detection of small fluorescent structures achieved by MP microscopy. Furthermore, some common calcium indicators (e.g., fura 2) appear to be well suited to two-photon microscopy (144). Recently developed calcium-sensitive fluorescent proteins, called cameleons, can also be used for two-photon calcium measurements (75).

It is noteworthy that there are many fluorescent dyes that can be used for labeling of cellular organelles. Fluorescent indicators are available for labeling ER, secretory vesicles and other acidic organelles, mitochondria, and nucleus (112). These dyes have proven useful for the localization of organelles in live cells and for correlation of the position of organelles and calcium signals. The accurate localization of (often small) organellar structures has relied on confocal detection of specialized dyes (9, 95, 216). The number of these dyes is now very large, most intracellular organelles can be targeted, and various wavelengths of excitation and emission light can be used.

Although more traditional microscopy methods are applicable, confocal microscopy has been particularly useful in the immunofluorescent analysis of protein distribution in fixed sections. Particularly important have been studies to determine the distribution patterns of the major proteins involved in calcium signaling pathways. The only limitations here are the antibodies available, their specificity, and the condition of the cells after the necessary permeabilization and fixing procedures. This approach has been widely used in many secretory epithelial cell types to localize calcium release channels (113, 148, 150, 191, 197, 278, 314, 318), calcium pumps (147, 150, 302), and other proteins involved in calcium signal transduction (35, 50, 59, 135).

An interesting extension of this is the use of bioactive fluorescently labeled compounds to target proteins in live cells. A recent example is the use of BIODIPY-conjugated ryanodine by Straub et al. (268) to localize ryanodine receptors in live pancreatic acinar cells. The advantages of this approach are that the cell disruption is minimal and the integrity of the cells can be determined by functional assays.

Fluorescent dyes that label lipid bilayers (e.g., FM1-43) were used to monitor granular dynamics while simultaneously measuring intracellular calcium using another dye (95). Optical measurement of exocytosis from exocrine epithelium was extended by Kasai's group (200) to multiphoton microscopy (described above). Another study of exocytosis in pancreatic acinar cells used nonfluorescent microscopy to visualize granule movements (34). This technique is based on the fact that the zymogen granules are very absorptive of visible light and thus appear dark on transmitted images. By continuously subtracting one bright-field image from the previous one, the authors of this study were able to see the changes in granule darkness which occur upon exocytosis. The technique seems to correlate well with fluorescent dye analysis of secretion and is applicable to any cell type with large, optically dense vesicles (254).

Particularly high resolution in studies of localization of specific organelles and molecules responsible for calcium signaling is produced by immunoelectron microscopy. This usually involves detection of electron-dense gold particles conjugated via antibodies to the protein under investigation. This technique can only be applied to fixed cells, but the spatial resolution is unrivalled. Khoo et al. (135) have used this technique to detect the presence of CD38 (an enzyme suggested to have a role in production of cADPr) to the inner nuclear membrane in hepatocytes. Also, IP₃ receptors (IP₃Rs) were localized to vesicles located in the apical region of submandibular gland duct cells using this technique (314).

A range of fluorescently labeled calmodulins that are suitable for live cell studies was recently developed (297, 298). 5-(4,6-Dichlorotriazinyl)-amino-fluorescein (DTAF)-calmodulin does not change its fluorescent intensity upon calcium binding, whereas 2-chloro-(-amino-Lys₇₅)-[6-(4-N,N-diethylamino)phenyl]-1,3,5-triazin-4-yl (TA)-calmodulin increases its fluorescence when it binds calcium. The development of TA-calmodulin gave experimenters the fascinating opportunity to monitor the calcium-calmodulin reaction in real time at a cellular and subcellular level (298, 327). Importantly, DTAF- and TA-calmodulins could be used simultaneously, due to differences in their optical properties. This allows the distinction between changes of fluorescence due to redistribution of the calmodulin and changes due to calcium binding (298). Studies by Craske et al. (59) used these two forms of calmodulin in pancreatic acinar cells. They were able to see translocation of DTAF-calmodulin upon stimulation with Ca²⁺-releasing agonist and monitor the interaction of Ca²⁺ and calmodulin using TA-calmodulin (59).

Persechini and colleagues (242) developed an alternative calcium-calmodulin (Ca-CaM) sensor based on fluorescent proteins (FPs) (242). This sensor used a short Ca-CaM binding protein to link two FP molecules (CFP and YFP), which allows the fluorescence resonance energy transfer (FRET) to occur between the two fluorophores. The spectral relationship between the two FPs meant that when Ca-CaM bound to the link protein, inducing a steric change, there was a change in the fluorescent emission of the molecule. This change was used to monitor activity and position of CaM (285).

The expression of targeted fluorescent (e.g., GFP) and bioluminescent (e.g., aequorin) proteins is being used more and more to investigate specific organelles or proteins within the cell (52, 323). As mentioned above, the expression of such proteins in secretory epithelia is not easy, and hence, there is little use of the approach in this field. There are, however, some studies in other cell types which tackle general phenomena, and it may be possible to extrapolate these results to problems in epithelial cells. For example, the interaction of ER and mitochondria described using targeted GFPs in HeLa cells (239) offers a tempting explanation for results found in pancreatic (216, 268, 291) and salivary acinar cells (325). Also, recent work using GFP constructs, which highlighted the role of gap junctions in the propagation of calcium signals (214), may be very relevant to intercellular signaling in secretory glands.

	III. POLARITY OF INTRACELLULAR CALCIUM SIGNALS

Top Previous Next References

Oscillations of cytoplasmic Ca²⁺ were first directly demonstrated in hepatocytes using the bioluminescent protein aequorin (309). Since then, calcium oscillations have been recorded in most exocrine cell types studied. The calcium oscillations seen with aequorin in hepatocytes have subsequently been confirmed using fluorescent calcium dyes (243, 244). Several different types of calcium oscillations are triggered by vasopressin, phenylephrine, angiotensin II, ATP, ADP, and bile acid in this cell type (42, 65, 243, 244, 250, 309). Such oscillations are normally dependent on cyclic release from and uptake into intracellular calcium stores (usually ER).

Different agonists produced specific calcium signaling patterns in hepatocytes. For example, the duration of calcium spikes was shortest for phenylephrine, whereas ADP-induced spikes were more prolonged. The duration of spikes was even longer when stimulated by vasopressin, angiotensin II, or endothelin-1 (57, 287). The rate of calcium rise and peak calcium level were similar for transients induced by different agonists, whereas the rate of decline showed considerable variability and was agonist specific. Therefore, it is the differences in the rate of decline which underlies the agonist-dependent differences in transient duration (57, 287). ATP triggered complex Ca²⁺ responses that could be comprised of short (~9 s in duration) spikes and much longer (~50 s) transients. It is interesting to note that increasing the concentration of some agonists (e.g., phenylephrine) resulted in an increase of frequency of oscillations while the shape and duration of transients remained unchanged. This led to the suggestion that the intensity of the stimulus could be encoded by frequency modulation of calcium signals rather than by amplitude modulation (57, 240, 243, 287, 309).

Oscillatory calcium behavior was also described in gastric parietal cells stimulated by histamine, carbachol (156), or gastrin (63). Spontaneous calcium oscillations, calcium oscillations triggered by mechanical stimulation, and oscillations induced by ATP were found in experiments on mammary epithelial cells (87). Electrophysiological and optical measurements identified calcium oscillations in lacrimal acinar cells stimulated by muscarinic agonists (21, 167). In parotid acinar cells, stimulation of cell surface muscarinic and -adrenergic receptors induces intracellular calcium oscillations (101).

It was in parotid acinar cells that a very interesting and unusual type of oscillation was discovered. These oscillations are triggered by thapsigargin, an inhibitor of the sarcoplasmic/endoplasmic reticulum calcium pumps (SERCA), which normally replenish intracellular calcium stores. The inhibition of store uptake initiates oscillations that are based on variations of calcium exchange between the cell and the external media. This exchange is mediated by calcium influx channels and by the plasma membrane pumps (83, 84).

Calcium oscillations in exocrine pancreas were described in both duct cells (266) and acinar cells (226, 319). Pancreatic acinar cells display complex patterns of oscillatory behavior, which are specific for particular hormones and neurotransmitters. The circulating hormone, cholecystokinin (CCK), produces one type of pattern. At submaximal doses of this hormone, the oscillations are usually composed of individual calcium elevations from a baseline that is similar to the resting calcium level (226, 272, 321). Bombesin produces a similar pattern of calcium oscillations (172). However, the neurotransmitter acetylcholine (ACh), in the same cell type, produces faster calcium oscillations that are superimposed on a plateau that is considerably elevated compared with the resting calcium level (321). At low doses, ACh and CCK can generate oscillating calcium signals that are confined to a small region of the cell (132, 210, 290).

The role of individual second messengers in generating oscillations was examined in a number of studies. It was found that calcium oscillations can occur at constant concentration of the calcium-releasing messenger IP₃ (306). The bell-shaped sensitivity to Ca²⁺ of the IP₃R suggests that Ca²⁺ itself, by action on the IP₃R, could provide the negative feedback necessary to generate oscillations (20). Some recent studies on nonepithelial cells indicate that IP₃ oscillations after agonist stimulation can occur and underlie oscillatory calcium responses (114, 192, 193, 208).

An interesting hypothesis on the mechanisms determining calcium signaling patterns was put forward in studies from Muallem's laboratory (159, 311, 324). The hypothesis suggests that regulators of G protein signaling (RGS) proteins specifically interact with different G_q-coupled receptors for calcium-releasing secretagogues. The effect of RGS proteins was to inhibit phospholipase C (PLC)- activity and consequently to suppress IP₃ generation and downstream calcium signaling. Thus it is proposed that RGS proteins may be responsible for the control and formation of oscillations in pancreatic acinar cells via the interaction with the G protein-receptor complex. It is suggested that cyclic binding of Ca-CaM to RGS proteins is essential for generating oscillations. Although this hypothesis provides a conceptually novel mechanism for control of oscillations, important questions concerning the dynamics of IP₃ production and Ca-CaM binding to RGS proteins remain to be tested by direct experiments.

An interesting aspect of these studies is that the effect of the RGS proteins was found to be determined by the nature of the receptors and was very different for different secretagogues (311, 324). For example, both responses to carbachol and to CCK were inhibited by RGS4, but responses to carbachol were 33 times more sensitive to RGS4 than responses to CCK (311); the difference was even more dramatic (~1,000-fold) for RGS1. It was suggested that the interaction of amino terminus of RGS proteins with receptors determined the specificity of the RGS protein action (311, 324). The specificity of this interaction may determine the individuality of agonist-induced oscillation patterns.

In addition to IP₃, there is strong evidence that "novel" calcium-releasing messengers (cADP-ribose and NAADP) may have a role in determining patterns and receptor specificity of oscillations in pancreatic acinar cells (40, 55, 289).

During the oscillatory phase, the rise of each individual calcium transient inside the cell is heterogeneous in both space and time. There are some interesting general mechanisms that underlie the formation and propagation of these polarized calcium signals. The way in which these mechanisms are activated determines the characteristics of the transients and thus forms the basis for agonist-specific oscillation patterns. The properties of these mechanisms are discussed in the following sections.

Soon after the discovery of calcium oscillations, imaging techniques became widely available for studies of cellular calcium distribution. Results from the last 10 years of videoimaging experiments have demonstrated that a number of secretory epithelial cell types, hepatocytes, pancreatic acinar cells, lacrimal acinar cells, submandibular gland acinar cells, and submandibular gland duct cells, exhibit a common property. In all these cell types, calcium transients are initiated in the apical region. This initiation is usually followed by the formation of calcium waves, which propagate into the basal part of the cells (131, 148, 195, 290, 293). However, some types of smaller calcium signals can have an entire life cycle (initiation, local spreading, and decline) that is confined to only a small, subcellular compartment, the apical region of the cells (132, 290, 314) (example shown in Fig. 1A).

View larger version (51K): [in this window] [in a new window]   Fig. 1. Examples of the subcellular dynamics of secretagogue-induced calcium responses in secretory epithelia. All images show pseudocolor representation of fluo 3 fluorescence obtained using confocal mocroscopy. A: an isolated pancreatic acinar cell, shown in transmitted image (top left image), produces a calcium spike that is localized wholly in the apical (secretory granule) region when stimulated with low doses of ACh (10 nM). Scale bar = 10 µm. B: the same cell stimulated later with a higher concentration of ACh (50 nM) undergoes a global calcium transient that spreads as a wave in an apical-to-basal direction. C: in blowfly salivary gland epithelial cells, stimulation with 3 nM 5-hydroxytryptamine triggers calcium waves that travel in the opposite direction to those in pancreatic acinar cells. The calcium response begins in the basal area of the cell (cell delineated by white box) and spreads as a wave toward the apical region. Scale bar = 20 µm. [C from Zimmermann (326), copyright Harcourt Publishers.]

This form of localized signaling is probably very energy efficient. Calcium signals simply occur where they are needed. The apical membrane of exocrine epithelial cells is the place where secretion, both fluid and exocytotic secretion, takes place. The primary purpose of the calcium signals is to initiate the secretion process, and therefore, it is a most economical strategy to produce the signals in the region where the cell's business is taking place.

Interestingly, in pancreatic acinar cells, short applications of ACh at very high concentration produce calcium responses localized to the apical (secretory granule) region. ACh is a neurotransmitter and is expected to have a short lifetime after its release from the synaptic terminals of peripheral neurons, due to activity of ACh esterase present in pancreas. It is therefore conceivable that this localized apical calcium response is the main calcium signaling mode of this important neurotransmitter in exocrine pancreas (90).

There is, however, a necessity to inform the rest of the cellular organelles about the apical secretion so that all the functions of the cell can be coordinated. For example, to replenish the store of proteins lost as a result of secretion, mRNA transcription and translation should be stimulated. Propagating calcium waves could play the role of long-range, coordinating messenger by delivering the signals to the nucleus, rough ER, mitochondria, and other organelles outside apical region. It would seem most reasonable, evolutionarily, for initiation of these waves to use the same mechanism that is involved in generation of localized apical calcium responses. Therefore, the waves should start in the apical part of the cells (example in Fig. 1B).

It is important to note that to organize localized calcium signals, small cells need to have a high calcium buffering capability. Without buffering, calcium will rapidly spread through the cytoplasm by diffusion, limiting the possibility of localized signaling. The numerical measure of buffering is the calcium binding capacity (for definition, see Ref. 199). Pancreatic acinar cells are the only exocrine secretory cell type in which calcium binding capacity was estimated (182). The buffering was found to be very high, with a binding capacity of ~2,000. This means that out of 2,000 calcium ions added to the cytoplasm of the cell, only 1 will remain free and the rest will be bound by buffer (182). The high binding capacity is probably important for localized apical calcium signaling and must be overwhelmed to allow propagation of calcium waves.

The nuclei of these cells appear to be particularly well protected against calcium rises induced by short applications of ACh. Large calcium gradients of many hundreds of nanomolar per micrometer can form along a line connecting the secretory region and the nucleus. Despite their size, these gradients eventually dissipate without spreading into the nucleus (90). It is not yet clear whether this nuclear protection is mediated by the nuclear envelope acting as a physical barrier to diffusion or by elevated nuclear calcium buffering capacity.

The discovery of apical-to-basal calcium waves and their correlation with agonist-stimulated ion currents in pancreatic acinar cells led Kasai and Augustine (131) to propose the "push-pull" model for unidirectional Cl secretion (131). This model is based on early activation of apically located Ca²⁺-dependent Cl channels leading to secretion of Cl into the lumen, the "push" phase. Cl uptake through the basal membrane channels only begins later when the calcium wave reaches that area of the cell; this is the "pull" phase. The sequential activation of the two phases would be necessary because the depolarization of the cell membrane potential caused by the push phase is prerequisite to the pulling of Cl into the cell. This was the first description of how a subcellular calcium gradient triggered by agonists could be vital in the control of cell function. The model obviously relies on the existence of both basal and apical Ca²⁺-dependent Cl channels and has been thrown into some doubt by work from Greger's laboratory in which the basal channels could not be found (265). Recent experiments using local uncaging of caged Ca²⁺ to activate the Cl current have also demonstrated the absence of Cl channels in the basal and lateral regions of the plasma membrane while confirming the presence of the channels on the apical membrane (217). Furthermore, in the fluid-secreting blowfly salivary gland, agonist-induced Ca²⁺ waves initiate in the basal region and spread to the apical part of the cell (326). This leaves the situation somewhat unresolved. Although Cl secretion appears to be activated by the Ca²⁺ transient and will be dependent on cell membrane potential as Kasai and Augustine showed, the exact mechanism of transcellular Cl movements remains elusive.

One of the earliest descriptions of polarized calcium signals came from a study on hepatocytes in which Rooney et al. (244) reported that intracellular calcium oscillations originate repeatedly in the same region of the cell, adjacent to the plasma membrane. The initiation site remained the same during sequential stimulation with different agonists (244, 287). Polarized calcium signaling in hepatocytes was later confirmed in experiments on intact, perfused liver. In the periportal hepatocytes, the initiation of calcium signals occurred repeatedly in particular intracellular regions, the sinusoidal domains (240). As noted above, the calcium waves in hepatocytes propagate in an apical-to-basal direction (195). Importantly, IP₃Rs are found in apical region of hepatocytes (but also in the nucleus) (195). Initiation is therefore suggested to be IP₃ dependent. The propagation speed and the amplitude of the propagating wave are, however, reduced by substances that inhibit RyRs (dantrolene, ryanodine). This suggests the involvement of RyRs in propagating the calcium signal into the basal part of the cell (195).

The geography of calcium signal initiation was advanced in work by Kasai et al. (132). It was found that calcium signals are initiated in a small subregion of the apical part of pancreatic acinar cells termed the "trigger zone" (132). After initiation in the trigger zone, the signals spread over the granular area. Injection of IP₃ from the patch pipette resulted in calcium signals originating in the trigger zone, indicating the presence of high-affinity IP₃Rs in this part of the cell. These IP₃Rs are essential in the initiation of local and global calcium signals in pancreatic acinar cells (132). A more recent study, which probed the cytosol of pancreatic acinar cells with local uncaging of IP₃, revealed that apical uncaging systematically produced calcium signals, whereas uncaging in basal regions produced calcium responses in only 50% of the experiments. The amount of uncaged IP₃ necessary to generate responses in basal part of the cells was also larger than in the apical region (81). These data suggest that high-affinity IP₃Rs concentrate in the apical part of pancreatic acinar cells, whereas the basal part contains IP₃Rs of lower affinity that are distributed nonuniformly (present only at a number of discrete sites) (81). The existence of IP₃-induced release in the basal region suggests that these receptors could contribute to propagation of calcium waves.

The distribution of release channels underlying polarized calcium signaling was directly elucidated using immunostaining. IP₃Rs (type 3) were found at the extreme apex of pancreatic acinar cells (197).

The suitability of different types of IP₃Rs for initiating calcium signals recently became a topic of lively discussion. Hagar et al. (107) considered that the absence of inhibition of IP₃R3 by high levels of calcium makes it a particularly suitable channel for initiation sites. Later work, however, indicated that IP₃R3 has a bell-shaped dependency on calcium (274). In the presence of IP₃, it is activated by low concentrations of calcium and inhibited by higher concentrations (274). In other words, it behaves similarly to other types of IP₃Rs. This does not preclude, however, that IP₃R3 could play an important role in initiating calcium signals. Indeed, studies using electrophysiological analysis of recombinant IP₃Rs in oocytes have suggested that, although many of the general properties of the different subtypes are similar, there are sufficient differences to suggest specific roles for certain isoforms (163, 164). The relatively high sensitivity to IP₃ of subtype 3 implicates this isoform in initiation of calcium signals while the highly Ca²⁺-sensitive IP₃R1 is ideally suited to amplification and propagation of the signal (163).

In another study on pancreatic acinar cells, IP₃R1, IP₃R2, and IP₃R3 were resolved in the apical part of the cells close to luminal and lateral membranes. In this work, the location of initiation sites was compared with the distribution of immunofluorescent staining obtained with different types of antibodies for IP₃Rs. In some pancreatic acinar cells, calcium initiation sites correlated with the position of IP₃R2 (148), although there is a substantial overlap in distribution of all the subtypes in the apical region (148, 197).

The position and the density of receptors can apparently be influenced by secretagogues. Treatment with secretagogues causes redistribution and downregulation of IP₃Rs in the pancreas. The downregulation utilizes the ubiquitin/proteasome pathway but is not limited by this pathway (308). Calcium release by IP₃Rs is also subject to regulation by other intracellular pathways. IP₃R3 is rapidly phosphorylated by cAMP-dependent kinase (PKA) leading to inhibition. This phosphorylation occurs after stimulation with physiological concentrations of CCK but not with ACh. PKA and IP₃R3 are colocalized in the apical region of pancreatic acinar cells. Treatment with dibutyryl-cAMP or forskolin (activators of PKA) inhibits local apical calcium transients induced by uncaging of small doses of caged IP₃. Brought together, these data suggest that PKA-mediated phosphorylation of IP₃R3 receptors may be important in shaping the pattern of hormone-induced calcium signaling in these cells (94).

The notion that IP₃Rs are the only type of receptors responsible for the activation of calcium signals in the apical region of pancreatic acinar cells was shaken by experiments with recently discovered calcium-releasing substances. cADP-ribose, a potential activator of RyRs, induces repetitive calcium transients localized to the secretory granule region of pancreatic acinar cells. This suggests that both RyRs and IP₃Rs could be involved in generation of local calcium signals in the apical region of pancreatic acinar cells (289). The calcium-releasing channels involved in activation of calcium signals could be agonist specific. The cADP-ribose antagonist 8-NH₂-cADP-ribose inhibits local calcium responses induced by cADP-ribose but not responses triggered by IP₃. These results were not unexpected but, more surprisingly, 8-NH₂-cADP-ribose also blocks apical calcium oscillations triggered by small, physiological doses of CCK and bombesin (29, 38, 39). In contrast, ACh-induced calcium signals were unaffected by 8-NH₂-cADPr (29). It was mentioned previously that agonists produce different patterns of calcium signals, and these results suggest that it may be the recruitment of different calcium release channels following agonist-stimulated messenger production that underlies this specificity.

Another recently discovered (89, 146) calcium-releasing compound, NAADP, also induces calcium responses in pancreatic acinar cells. Calcium transients can be local (confined to the apical region) or global. The shape of NAADP-triggered responses is similar to those triggered by CCK. An interesting property of NAADP is its autoinhibition; short exposure to a high concentration of NAADP inhibits further NAADP responses for a long period of time. Remarkably, high autoinhibitory concentrations of NAADP also block CCK-induced calcium signaling. These data invite the suggestion that NAADP could serve as a second messenger linking CCK receptors with the calcium signaling cascade (36).

At the moment, it is not clear whether different agonists produce signals that have absolutely identical initiation sites. One study found no difference in initiation sites of signals induced by CCK and by ACh (37), whereas another work reported that the ACh analog carbachol and CCK initiate signals at different sites (260). However, both studies agree that for all agonists the signals originate in some part of the apical region.

In pancreatic acinar cells, then, IP₃, cADP-ribose, and NAADP can all release calcium and all seem to be involved in the initiation of agonist-evoked signals. This has led to the idea of cooperativity between the release mechanisms being necessary for initiation of the calcium signals in the apical region of the cell (37). This postulate obviously requires that calcium stores available to each messenger should be present in this highly sensitive region. Experiments performed on cultured pancreatic acinar cells permeabilized by streptolysin-O (SLO) indicate the existence of three types of calcium stores. The first type possesses receptors to cADP-ribose and IP₃, the second type has only IP₃Rs, and the third type contains Ca²⁺ that cannot be mobilized by IP₃ or cADP-ribose but can be released with ionomycin (97). This finding is in contradiction with a previous study (also performed on SLO-permeabilized pancreatic acinar cells) which reports uniformity of IP₃ sensitivity in different parts of the cell (301). Recent experiments with photobleaching of indicators trapped in the lumen of ER and uncaging of caged calcium loaded into the ER lumen showed considerable connectivity of this major calcium store in pancreatic acinar cells (218). The intracellular stores of hepatocytes are also luminally connected, allowing full access to all the available stored calcium after opening of any release channel (109). Luminal connectivity of internal calcium stores means that the localization of signal initiation is very dependent on the positioning of release channels on the store membrane rather than on the positioning of the stores themselves.

Recent work from Muallem's group (260) reported that apical initiation of calcium signals may be aided by preferential localization of ACh and CCK receptors to apical and lateral membranes. This potentially interesting new finding is, however, in contradiction with classical work from the Jamieson laboratory (247) that describes the presence of CCK receptors on the basal membrane and explicitly states the absence of CCK receptors on the apical membrane (247). The results of the Muallem group are also in apparent contradiction to biophysical studies in which stimulation of apical calcium release was induced by basal application of ACh (290). The clarification of agonist receptor localization is important for our understanding of the process of calcium signal initiation in pancreatic acinar cells.

Duct cells of the submandibular gland (SMG) present an interesting example of polarized calcium signals which appears to rely on unconnected intracellular calcium stores. As with pancreatic acinar cells, calcium waves initiate in the apical region of SMG duct cells and propagate toward the basal part of the cells (314). The calcium elevation in the basal region is small and not statistically distinguishable from noise, which indicates that calcium waves are confined to the apical and middle regions of the cells while the basal part remains inert with respect to calcium signaling. Immunofluorescence imaging localized IP₃R types 2 and 3 to the apical region of SMG duct cells, whereas IP₃R1 was not detected. The amount of IP₃R2 in the duct cells was higher than in any other tissue examined by the authors. Immunoelectron microscopy identified the IP₃R2 on small apical vesicles of some SMG duct cells (314). This study is unique since it actually ascribes IP₃Rs to particular separate organelles, namely, the tiny vesicles concentrated in subplasmalemmal region in the apical part of the cell. The implication of these findings is that these discrete stores provide the calcium for signals that follow IP₃-mediated agonist stimulation.

There was considerable heterogeneity of the amount of IP₃R2 between different cells. The calcium signals in the submandibular gland duct cells were also nonuniform from cell to cell. Some showed rapid calcium mobilization, whereas others responded slowly. These findings led to the concept of "pioneer cells" that possess IP₃R2 in considerable numbers. As a result of their higher release capability, these cells are the first to initiate calcium responses and therefore coordinate calcium signaling throughout the duct by transmitting their signals to neighboring cells (314) (see sect. V). Somewhat different results regarding IP₃Rs in this cell type were reported by Lee et al. (148). In SMG ducts, they resolved IP₃R1 and IP₃R2 close to luminal and lateral membranes. Furthermore, one IP₃R3 antibody used in this study detected receptors around the nucleus of SMG duct cells, whereas another type did not stain the duct at all (148). Both papers report the presence of more than one type of IP₃R in the apical region of SMG duct cells.

RyRs confined to the luminal pole and lateral membrane were reported in SMG duct cells (148). Both caffeine and cADP-ribose release calcium from internal stores in these cells (148). The RyR stimulators, caffeine and cADP-ribose, both release calcium in SMG acinar and duct cells, and this release was not inhibited by the IP₃R antagonist heparin (148).

Acinar cells of the SMG give another example of apical-to-basal pattern of propagation of calcium waves triggered by secretagogues. In these cells, IP₃R1, IP₃R2, and IP₃R3 were detected close to luminal and lateral membranes (148).

Lacrimal acinar cells also conform to the general rule of apical initiation of calcium signals. ACh and ATP induce calcium signals that are initiated in the secretory (apical) pole of the lacrimal acinar cells and rapidly spread toward the basal pole (293).

An interesting exception from the rule of apical-to-basal propagation of calcium waves is found in acid-secreting parietal cells of the stomach. The polarity of calcium signaling seems to be reversed in these cells. Histamine stimulation triggers calcium signals that are initiated in the basal part of the cells (although closer to the central part of the cell rather to the extreme basal membrane, judging by the traces on Fig. 3 from this paper) (9).

The mechanism underlying the propagation of hormone-induced calcium signals into the basal part of polarized epithelial cells was touched upon above, and the suggestion was made that propagation could involve both IP₃Rs and RyRs. There are some arguments that the propagation of the calcium signals is mainly dependent on IP₃Rs. Of particular interest in this respect is a study in which different types of calcium responses (local and global calcium spikes), which can normally be triggered by ACh, were mimicked by uncaging of IP₃. Rapid and transient uncaging of IP₃ produces local calcium signals similar to that triggered by ACh. To reproduce the shape of global calcium transients, gradual slow IP₃ uncaging was necessary, indicating that global calcium transients may depend on continuous IP₃ production (125).

One study compared the importance of RyRs for calcium signaling in pancreatic and SMG acinar cells (148). This paper described that RyRs were found in SMG acinar and duct cells but not in pancreatic acinar cells (148). Caffeine, which increases the Ca²⁺ sensitivity of RyRs, elevated calcium in SMG acinar and duct cells but not in pancreatic acinar cells. cADP-ribose was very effective in mobilizing calcium from the stores in SMG acinar and duct cells. Although cADP-ribose can activate calcium release from the stores of pancreatic acinar cells, it does so less effectively. Furthermore, the cADP-ribose-induced calcium response was blocked by the IP₃R antagonist heparin (148). These data argue that, in pancreatic acinar cells, RyRs are not important for calcium signaling, so propagation should be IP₃ mediated. Intriguingly, though, infusion of IP₃ through the patch pipette mainly produces localized nonpropagating calcium signals (227). Furthermore, there are data supporting the central role of RyRs in propagating the calcium signals into the basal part of pancreatic acinar cells. The existence of RyRs in pancreatic acinar cells and their localization to the basal part of the cells was reported by Nathanson's group (150). This was confirmed recently in what seems to be a very detailed study from Pandol's group (80). RT-PCR identified RyR isoforms 1, 2, and 3 in both rat pancreas and dispersed pancreatic acini. Isoforms 1 and 2 were found using single-cell RT-PCR on cells that were individually selected under a microscope. The subcellular distribution of these isoforms of RyRs in pancreatic acinar cells was studied using Western blot analysis. RyR1 and RyR2 were present in all fractions of the ER (both rough and smooth). Immunofluorescence labeling using general antibodies against RyRs stained most of the cytoplasm but was excluded from the nucleus. Palmitoyl-CoA, an activator of RyRs, released calcium from the stores of permeabilized pancreatic acinar cells. Notably, the release was induced in both the apical and basal regions of the cells. Upon palmitoyl-CoA activation, basal release occurred simultaneously or immediately before the release in apical region. Alone, neither ryanodine nor caffeine had much effect. However, both substances potentiated the releasing action of palmitoyl-CoA (80).

Strong evidence in favor of RyR-dependent propagation of calcium signals into basal part of pancreatic acinar cells was recently presented by Straub and colleagues from Yule's group (268). In their experiments, strong uncaging of IP₃ results in global calcium signals. The application of ryanodine to inhibit RyRs limits these uncaging-induced transients to the region slightly larger than the trigger zone. In other words, when RyRs were inhibited, the IP₃-induced apical responses were not converted into global responses. The conclusion from this work was that IP₃ triggers responses in the apical part of pancreatic acinar cells, but the propagation of the calcium wave depends on RyRs (268). The role of cADP-ribose receptors in converting local apical signals into global calcium waves was demonstrated in studies by Cancela et al. (37). These views of calcium wave propagation in pancreatic acinar were recently given further weight by work which used region-specific uncaging of IP₃ and cADP-ribose to investigate the role of IP₃Rs and RyRs (149). Leite et al. (149) concluded from their studies that release from IP₃Rs in the apical region coordinates with cADP-ribose-sensitized release from basal RyRs to generate global ACh-induced calcium waves. NAADP is another candidate for the role of "globalization" messenger. With the use of whole cell electrophysiological recordings, it was found that globalization of local ACh-induced calcium signals in pancreatic acinar cells can be achieved by infusion of NAADP (40).

In pancreatic acinar cells, nocodazole (a substance that disrupts microtubular networks) inhibits IP₃-induced localized apical calcium spikes. Interestingly, treatment with nocodazole also caused a redistribution of the ER, resulting in a considerable decrease of ER density in the apical region. Other substances that modify the status of the microtubular network, colchicine and taxol, also inhibited local calcium spikes. These data argue that microtubule-dependent positioning of the ER in the apical region of pancreatic acinar cells is essential for maintaining localized calcium signals (82).

Pancreatic acinar cells also provide another example of polarized distribution of cellular organelles. Most of the active mitochondria in these cells are clustered around secretory granule region. This strategic positioning of mitochondria allows them to play a barrier function in limiting propagation of calcium signals. Indeed, inhibition of mitochondrial metabolism (and therefore mitochondrial calcium uptake) with antimycin and oligomycin or with carbonyl cyanide m-chlorophenylhydrasone (CCCP) converts IP₃-induced, local calcium oscillations into global calcium transients (291). These results were challenged by studies of Gonzales et al. (98), who found that mitochondrial calcium accumulation in pancreatic acinar cells occurs with a time delay of ~11 s after the rise of cytosolic calcium. In this paper, the mitochondrial clustering in the perigranular region was not found. On the basis of their apparent slow Ca²⁺ uptake and uniform distribution, the authors concluded that mitochondria do not participate in limiting calcium signals to the apical part of the cell.

However, the study by Straub et al. (268), mentioned previously, confirmed the importance of mitochondria to localization of apical calcium signaling in pancreatic acinar cells. Using IP₃ uncaging, they also found that local apical calcium signals are converted into global elevations by inhibition of mitochondria. The authors also confirmed that mitochondria were concentrated mainly in the perigranular region. These findings further demonstrated the barrier role of mitochondria in limiting the spread of calcium signals from the apical region. In the same paper, staining of the RyRs by BODIPY-ryanodine showed that they were also concentrated in the perigranular region, although there was some diffuse staining throughout the rest of the cytoplasm. The experiments of this group indicate that mitochondria can play an important role in modulating the nearby RyRs. Inhibition of this mitochondrial modulation makes propagation of the wave through the area enriched with RyRs more likely (268).

Recent results from our group demonstrated the existence in pancreatic acinar cells of three distinct groups of mitochondria: perigranular (described previously), perinuclear, and subplasmalemmal. Fast calcium uncaging experiments showed that each group is capable of rapid calcium uptake. The individual positioning of each group of mitochondria allows them to influence distinct calcium signals (216).

In hepatocytes, elements of the ER and mitochondria are positioned closely to one another so that IP₃-induced calcium release has considerable effects on adjacent mitochondria. IP₃-induced loss of calcium from the ER was reflected in a rise of intramitochondrial calcium and increase of mitochondrial metabolism (108). Conversely, mitochondria can modulate IP₃-induced calcium release. Inhibition of mitochondria results in an enhancement of calcium loss from ER upon IP₃ stimulation. The authors of this study interpret this effect as an indication of mitochondrial suppression of positive calcium feedback on IP₃Rs. So, under normal conditions, by taking up some of the calcium that comes out of the ER, the mitochondria make it less likely that calcium-induced calcium release will occur. The ability of mitochondria to effectively take calcium released from ER is based on the neighborliness of the two organelles. Such structural juxtaposition of ER and mitochondria has been demonstrated by Rizzuto et al. (239) using organelle-targeted GFP in HeLa cells. Interestingly, nonuniform distribution of mitochondria produces heterogeneity in IP₃ sensitivity of different ER regions. The regions with a high density of mitochondria are less sensitive to IP₃ than the regions with lower density (108).

The geography of cellular organelles in pancreatic acinar cells presents an interesting puzzle in terms of effective calcium signaling. The problem is that the apical region of these cells is packed with secretory granules. That leaves very little space for ER, which is the major calcium-releasing store. Despite there being only a very small volume of ER in the apical region, it must be able to generate repetitive local calcium spikes, which are quite resistant to removal of external calcium, and provide effective initiation signals for propagating calcium waves. How does the cell manage this when it would seem that there is so little calcium stored in ER located in the secretory granule region? It seems that the cell has resolved this problem by connecting the small calcium stores in the apical part of the cell with a massive calcium store, namely, the basally located ER. Direct conformation of ER connectivity was achieved by monitoring FRAP. Mag-fluo 4 equilibrated quickly in the ER lumen of pancreatic acinar cells after local photobleaching. This dye is a particularly suitable indicator for such experiments, since it is much more fluorescent in ER than in the cytosol. Similar fast equilibration (confirming luminal connectivity of ER) was found for Mag-fura 2 (218). Uncaging of calcium was used to directly visualize calcium movement in ER. Cells were loaded with the caged calcium compound NP-EGTA by incubation with the AM form of this probe. After that, the cytosolic NP-EGTA was removed via a patch pipette (the pipette also contained high concentration of calcium buffer, to prevent any calcium changes in the cytosol). Localized uncaging of NP-EGTA trapped in the ER lumen allowed monitoring of calcium movements inside this organelle. In experiments of Park et al. (218), fast intra-ER calcium movement was found, confirming that calcium can indeed "tunnel" from one side of the cell to the other (218).

Fast calcium movement through the ER lumen suggests that after calcium release, the apical calcium stores can be effectively refilled by calcium moving from the basal side through the lumen of ER (183, 218). This tunneling of calcium from the basal to the apical part of the cell means that the apical stores effectively become undepletable. ER continuity is not peculiar to pancreatic acinar cells. In hepatocytes, the connectivity of the stores was assessed by using Mn²⁺ quench of fura 2 in the cellular organelles (109). This study indicated that the stores in intact hepatocytes were luminally connected, and this connectivity was lost after disruption of the cytoskeleton (using cytochalasin B). With the use of FRAP of GFP fusion proteins targeted to the aqueous phase of ER lumen (61, 273), ER continuity was also demonstrated in nonepithelial cells.

Measurements of calcium binding capacity indicated relatively weak calcium buffering in the ER compared with the cytosol. This should make the ER route very convenient for fast calcium translocation (182). Another advantage of this transport route is that the cell could move calcium through the ER tunnel without risk of triggering calcium-induced calcium release. This should make the "through ER" reloading route safer than via the cytosol. For small and intermediate doses of agonists, one can envisage bidirectional calcium fluxes in the pancreatic acinar cell: apical-to-basal flux in the cytoplasm and simultaneous basal-to-apical movement through the ER lumen.

In contrast, the major IP₃-dependent store in SMG duct cells seems to be disconnected from the basal ER. As was mentioned above, Mikoshiba's group (314) described small vesicles containing IP₃Rs located in the apical part of these cells.

Calcium pumps are essential in shaping calcium signals in exocrine secretory cells. They maintain the resting calcium level, provide intracellular calcium stores with releasable calcium, and play an essential role in recovery of cytosolic calcium after calcium release from the store. Calcium pumps are divided into two classes: plasma membrane calcium pumps (PMCA) and SERCA (43, 160).

Recently, immunofluorescent staining revealed a highly polarized distribution of different types of SERCA pumps in pancreatic acinar cells. Pumps of SERCA2a type were found almost exclusively at the luminal pole of these cells, while SERCA2b was found preferentially in the basal region (147). There is considerable similarity in the distribution of IP₃Rs and SERCA2a calcium pumps. Close colocalization between calcium pumps and calcium release channels leads to suggestion of short-range interactions between the two mechanisms. Thapsigargin, the specific inhibitor of SERCA pumps, depletes the calcium store of pancreatic acinar cell (183). It was also shown to induce a cytoplasmic calcium rise as calcium leaks out of the store. Interestingly, thapsigargin induces a diffuse, spatially homogeneous calcium rise (293). Recently (and unexpectedly), it was found that thapsigargin not only produces uniform calcium release itself, but also abolishes the apical-to-basal polarity of calcium release triggered by carbachol. In the presence of thapsigargin (but evidently before the calcium store is completely discharged), carbachol induces a uniform calcium rise (147). This further suggests that there is a close functional relationship between pumps and release channels. The mechanism of this interaction is not yet clear. One possible effect of the calcium pump activity could be formation of high-affinity calcium buffering in close vicinity to the calcium-releasing channels. As a transporter, the calcium pump is relatively slow, with a maximum rate of a few tens of ions per second. However, because of its sluggishness, it must be expressed in relatively high concentrations and its calcium binding sites form a buffer of high affinity. Such buffering should be an effective mechanism for both localizing calcium responses to a particular region of the cell and also for introducing a delay in the propagation of calcium waves. Total calcium buffering in pancreatic acinar cells is very considerable (182), but the nature and the distribution of buffers are a complete enigma in this and in many other cell types. It seems feasible that the calcium-binding sites of the pumps could form a substantial portion of this buffering.

SERCA2b has also been localized in cells of the SMG. It was found predominantly in the apical pole of duct cells and, in acinar cells, it is close to luminal and lateral regions of plasma membrane (147). Interestingly, in these cell types, the apical-to-basal polarity of SERCA2b distribution is exactly opposite to pancreatic acinar cells (see above). SERCA3 was found in the basal pole of both types of submandibular gland cells but was absent in pancreatic acinar cells (147).

PMCA are responsible for setting the resting level of free calcium in the cytosol. This, of course, has important consequences for the total calcium content of the cell and for the level of loading of calcium stores. Recent experiments showed that PMCA pumps also have an important role in the regulation of the frequency of calcium oscillations. In isolated hepatocytes, inhibitors of PMCA pumps, glucagon-(1929) and carboxyeosin, increased the frequency of oscillations triggered by calcium-mobilizing agonists (102). The distribution of PMCA in hepatocytes is nonuniform, with the pumps found predominantly in the blood sinusoidal regions of the cells (134).

With the use of pancreatic acinar cells, direct measurements of calcium extrusion fluxes at the cellular level were made by placing a single isolated cell into a very small droplet (~10¹⁰ liter) that contains fluorescent calcium indicator. This "droplet technique" allows simultaneous measurements of intracellular calcium concentration and calcium extrusion (using different calcium probes in the cell and in the extracellular droplet solution) (282, 283). The technique was used to characterize calcium dependency of calcium extrusion (33) and for measurements of calcium buffering of the cytoplasm (182). The extrusion of calcium appears to be very active. When maximally stimulated, calcium pumps of plasma membrane can decrease total cellular calcium content by more than 400 µM/min (33). It was possible to resolve oscillations of calcium extrusion rate that occurred synchronously with spikes of intracellular calcium (284). It was estimated that during global calcium transients triggered by secretagogues, >20% of the calcium released from the store is extruded from the cell by the PMCA pumps (284).

The localization of calcium extrusion at the single-cell level was studied after the development of the "indicator-dextran" technique in our laboratory (16, 281). This technique utilizes calcium indicators linked to heavy dextrans (mol wt from 70,000 to 500,000) in combination with confocal microscopy. Under usual experimental conditions, when calcium is extruded from the cell, it would diffuse away quickly and be diluted in the vast volume of extracellular solution, making it impossible to measure. The idea behind the dextran-indicator technique is to slow down diffusion of extruded calcium so that the fluorescence changes in the extracellular solution can be resolved close to the site of extrusion. The labeled high molecular weight dextrans in the extracellular solution are both calcium indicators and immobile (in comparison with small ions like calcium) calcium buffers. This means that they simultaneously slow down calcium diffusion and measure the calcium level. Free calcium concentration is kept low (0.2-0.4 µM) so that the indicator dextran is able to bind calcium. Combined imaging of intracellular calcium (e.g., using fura-red) and of calcium extrusion with calcium green-1 dextran is possible (15). The sensitivity of the method was high enough to allow detection of calcium released by exocytosis of a single secretory granule (14). The technique was used to demonstrate that extrusion occurs preferentially from the apical regions of the pancreatic acinar cells (15, 17), suggesting that PMCA are mainly localized in this part of the cell. This was very soon confirmed by immunostaining, which directly demonstrated preferential localization of plasma membrane calcium pumps in the luminal and lateral domains of the plasma membrane of pancreatic acinar cells (147). Much less pronounced punctate staining with antibodies against PMCA could be found on the basal membrane of the pancreatic acinar cells (147). PMCA was also preferentially localized to the luminal regions of submandibular acinar cells and submandibular duct cells (147). The preferential distribution of PMCA pumps in the apical region seems logical since it suggests that the region of the cell that specializes in calcium release also has greater ability to restore the resting calcium level.

	IV. TRANSCELLULAR CALCIUM TRANSPORT

Top Previous Next References

The nonuniform distribution of calcium channels and calcium extrusion mechanisms can result in considerable epithelial transcellular calcium transport. This calcium transport is clearly an important aspect of body calcium homeostasis, but it is not always immediately clear whether it is essential for epithelial function or simply a consequence of the use of calcium as an intracellular messenger. One clear example of the functional importance of transport of calcium into the exocrine fluid is found in mammary epithelia, where the process serves a clear nutritional purpose. Calcium concentration in rabbit milk can approach 100 mM (258). Amazingly, this is approximately a million times higher than the level of free calcium inside the cell and 50 times higher than total calcium in the blood. Human milk also contains a large amount of calcium, with the level around 12 mM (103). Calcium secretion occurs synchronously with the major milk proteins casein and lactose, indicating that the transport is transcellular rather than paracellular and that it occurs via Golgi-derived secretory vesicles (201).

Calcium released from stimulated pancreas also correlates with secreted protein, indicating that the mechanism of calcium release is mainly via exocytosis (165, 286). The total calcium concentration of pancreatic juice is in the range of a few millimolar (165).

In saliva, an approximate calcium concentration of 1 mM was reported (173). The results of measurements of calcium release from SMG could also be interpreted as an indication of release by transcellular transport. ACh and epinephrine cause ⁴⁵Ca²⁺ release from prelabeled gland, indicating that these agonists release calcium that is "bound intracellularly" (205).

Recent, elegant experiments have directly demonstrated the transepithelial movement of calcium following stimulation of intact g