Microdomains of Intracellular Ca2+: Molecular Determinants and Functional Consequences

Rosario Rizzuto, Tullio Pozzan


Calcium ions are ubiquitous and versatile signaling molecules, capable of decoding a variety of extracellular stimuli (hormones, neurotransmitters, growth factors, etc.) into markedly different intracellular actions, ranging from contraction to secretion, from proliferation to cell death. The key to this pleiotropic role is the complex spatiotemporal organization of the [Ca2+] rise evoked by extracellular agonists, which allows selected effectors to be recruited and specific actions to be initiated. In this review, we discuss the structural and functional bases that generate the subcellular heterogeneity in cellular Ca2+ levels at rest and under stimulation. This complex choreography requires the concerted action of many different players; the central role is, of course, that of the calcium ion, with the main supporting characters being all the entities responsible for moving Ca2+ between different compartments, while the cellular architecture provides a determining framework within which all the players have their exits and their entrances. In particular, we concentrate on the molecular mechanisms that lead to the generation of cytoplasmic Ca2+ microdomains, focusing on their different subcellular location, mechanism of generation, and functional role.


A very steep gradient of Ca2+, over four orders of magnitude, exists across the plasma membrane of all eukaryotic cells: cytosolic free calcium is maintained roughly at 100 nM while the extracellular milieu generally has a [Ca2+] of over 1 mM. This gradient is maintained due to the action of plasma membrane Ca2+-ATPases (PMCA) and Na+/Ca2+ exchangers (NCX); whereas the former are ubiquitous, the latter are predominantly expressed in excitable tissues. Mammalian PMCAs are encoded by 4 genes, and ∼30 isoform variants can be generated via alternative RNA splicing of the primary gene transcripts; the expression of different PMCA isoforms and splice variants is regulated in a developmental as well as a tissue- and cell type-specific manner (45, 352, 353). Three distinct genes are known to exist for the NCXs, and several splice variants are known for NCX1 and NCX3 (277, 278, 297). A closely related family of exchangers, the Na+/Ca2+-K+ exchange (NCKX) gene family, has been recently discovered (138). This family of proteins, first discovered in the outer segments of vertebrate rod photoreceptors, now comprises four members (NCKX1 to 4); a partial sequence of a fifth human NCKX gene has also appeared in the databases. In situ NCKX1 and heterologously expressed NCKX2 operate at a 4Na+:1Ca2+ +1K+ stoichiometry.

The energy to extrude Ca2+ against its electrochemical gradient has different origins: ATP hydrolysis in the case of PMCAs, with one Ca2+ being extruded per ATP consumed, and the electrochemical Na+ gradient in the case of NC(K)Xs, with the extrusion of one Ca2+ at the expense of three Na+ (and 4 Na+ in the case of NCKX) (45, 73, 277, 278, 294, 297, 352, 353).

A general concept, obvious to specialists but often not appreciated enough by many cell biologists, is that the long-term steady-state free Ca2+ concentration in the cytoplasm ([Ca2+]c) depends exclusively on the equilibrium between the rates of the efflux mechanisms and the rate of inward Ca2+ “leak” across the plasma membrane. Indeed, any modulation of the activity of organelles that accumulate or release Ca2+, or even changes in the level of Ca2+ buffers within the cytoplasm, only affect [Ca2+]c in a transient manner and have no direct role in the long-term maintenance of this parameter. In other words, a cell could theoretically maintain its steady-state [Ca2+]c even without any Ca2+ storage organelles or cytoplasmic Ca2+ buffers. Thus inhibition of the sarco/endoplasmic reticulum Ca2+-ATPase (SERCA pumps; see below) causes the release of Ca2+ accumulated in the stores and a transient, reversible, increase in [Ca2+]c, but has no persistent effect on steady-state Ca2+, unless, as is the case in many cell types, it modifies the so-called capacitative Ca2+ entry (CCE; see below). Similarly, inhibition of mitochondrial Ca2+ uptake can only indirectly lead to long-term modifications of steady-state Ca2+ level in the cytosol, e.g., if this inhibition affects the cellular ATP levels or the plasma membrane potential. Along the same line of reasoning, the Ca2+ buffering capacity of the cell cytoplasm can vary by over an order of magnitude among different cells, yet this results in no obvious change of the resting [Ca2+]c (15, 20, 65, 95, 404). More than a mere biochemical curiosity, this concept is of primary importance: if it were not true, the classical methods used to monitor [Ca2+]c with fluorescent dyes would lead to gross alterations of the resting [Ca2+]c. Indeed, loading cells with Ca2+ indicators implies the addition of an exogenous Ca2+ buffer to the cytoplasm, with a variation of several tens to hundreds of micromolar (6); it is obvious that a different Ca2+ buffering capacity does influence the amplitude of the [Ca2+]c rises that are observed upon activation, but this can be taken into consideration and accounted for (181, 301).

With regard to the inward “Ca2+ leak,” its exact nature is still undetermined. Controlled, Ca2+ influx paths most likely exist to ensure this key aspect of cell physiology (2, 3, 406).

In addition to the largely mysterious leak channels mentioned above, cells express in their plasma membrane a variety of well-characterized Ca2+ channels. The gating properties of these channels can depend on membrane potential [in the so-called voltage-operated Ca2+ channels (VOC); Ref. 46], or ligand binding [in the case of receptor-operated Ca2+ channels (ROC); Ref. 213]. Furthermore, a heterogeneous group of channels [most of which belong to the so-called transient receptor potential family (TRPs)] are activated by a variety of factors, including mechanical stretch, osmolarity, temperature, second messengers, G proteins, protein-protein interactions, etc. (226, 227; see also Ref. 249). The molecular nature of an additional group of channels, in this case activated by arachidonic acid, is still undetermined (337). An exhaustive description of all these channels is well beyond the scope of the present text, and the interested reader is referred to recent reviews that provide detailed analyses of this topic (46, 226, 227). The different molecular nature of these channels, their subcellular distribution, and their regulation are the key components of the generation of subplasma membrane Ca2+ microdomains, the focus of this review, and this latter aspect of the problem is addressed in detail in the next section.

A similar conceptual framework, i.e., a Ca2+ pump and leak equilibrium, regulates the steady-state level of [Ca2+] within intracellular Ca2+ stores, i.e., the endo/sarcoplasmic reticulum, the mitochondria, and a heterogeneous group of vesicular organelles including the Golgi apparatus, endosomes/lysosomes, and secretory vesicles. Functionally speaking, the sidedness of the intracellular Ca2+ store membrane is identical to that of the plasma membrane, i.e., the active Ca2+ transport accumulates Ca2+ within the organelle lumen (that thus corresponds to the extracellular milieu) and the channels return it to the cytoplasm. Ca2+ uptake within the stores depends on mechanisms molecularly and functionally completely different from those of the plasma membrane (see below) and, as in the case of the plasma membrane, the leak mechanism is still highly mysterious, although evidence has been recently provided that the protein import complex of the ER, the so-called translocon (193), or, under some conditions, the inositol 1,4,5-trisphosphate (IP3) receptor themselves (253), may contribute to the Ca2+ leak from the ER. As to the regulated release of Ca2+ from intracellular stores, it also depends on a group of channels whose molecular properties have been the subject in the last years of intense investigation. Their distribution within cells and their physiological regulation will be discussed in some detail in the sections below.

The complexity of the Ca2+ handling machinery briefly summarized above is at the basis of the heterogeneity of Ca2+ distribution within cells, “the Ca2+ microdomains,” not only at rest, but also, and most important, during stimulation. The definition of Ca2+ microdomain is far from being univocous, and it has been used with different meanings (in particular as far as its spatial dimensions) by investigators in different fields. Thus neuroscientists, in general, use the term Ca2+ microdomains primarily referring to the very local elevations of Ca2+ in the vicinity (nm) of the mouth of presynaptic Ca2+ channels that play a pivotal role in regulating neurotransmitter release (9). Cell biologists, on the other hand, refer to Ca2+ microdomains in more general terms, i.e., any increase in cytoplasmic Ca2+ that remains localized in one part of the cell, from the very local hot spots close to Ca2+ channels, to larger subcellular Ca2+ gradients such as those formed between the complex arborization of astrocyte processes and the nuclear region (24, 268), in the apical pole of pancreatic acinar cells (8, 271, 370), during cytotoxic T lymphocyte attack of target cells (288) or oriented locomotion (119) (for recent reviews see, for example, Refs. 22, 32, 55, 291). In this contribution, the term Ca2+ microdomain is used in the most general way, i.e., all the increases in cellular Ca2+ that do not involve the generality of the cell cytoplasm, but remain localized to part of the cell.

In the 1970s and 1980s (and often even to date), the existence of Ca2+ microdomains was invoked to explain results otherwise uninterpretable, but without a solid, direct experimental support. The existence of Ca2+ microdomains is intrinsic to the physicochemical characteristics of the cell interior. In fact, the high Ca2+ buffering capacity of the cytoplasm and hence the very low Ca2+ diffusion rate, 10–50 μm2/s (4), the existence of highly organized subcellular structures endowed with the capacity to take up and release Ca2+, the clustering of Ca2+ channels (both of the plasma membrane and of organelles) in discrete membrane domains, all appear ideally suited to generate transient (or even prolonged) local elevations of Ca2+. However, discrete, transient, heterogeneities in Ca2+ levels within the cytoplasm of living cells were eventually experimentally measured only in the late 1980s when fluorescent Ca2+ indicators became widely available. One of the first direct measurements of a localized Ca2+ increase was reported in 1988: a subplasma membrane Ca2+ increase (in the cell body and growth cones) activated in sympathetic neurons by membrane potential depolarization or Ca2+ mobilization from intracellular stores (188, 189). Since then, localized increases of cytoplasmic Ca2+ have been repeatedly measured by direct imaging with fluorescent indicators or inferred from indirect approaches. The existence, amplitude, and functional role of Ca2+ microdomains have been the subject of intense investigation over the last decades, in particular, but not only, concerning their role in neurotransmitter release.

Overall, the review focuses on the [Ca2+] microheterogeneity detected in the cell domains participating in the generation and/or functional translation of cytosolic Ca2+ signals: the plasma membrane, sarco/endoplasmic reticulum, Golgi apparatus and secretory vesicles, and the mitochondria. We here discuss the heterogeneity in the molecular repertoire of Ca2+ handling of these cell domains, as well as the large differences in their resting and stimulated [Ca2+].

Section ii of this review is dedicated to a brief summary of the vast literature dealing with the generation of subplasma membrane microdomains, with particular emphasis on the microscopic local events occurring a few nanometers away from the mouth of plasma membrane Ca2+ channels in neuronal cells (see, for example, Ref. 398). The generation of subplasma membrane Ca2+ microdomains is, however, not a peculiarity of neuronal cells, and their existence and possible function have been described in many nonexcitable cells. In the latter cell types, such microdomains can again remain restricted to the few nanometers of cytoplasm around the inner mouth of the channels or they can be represented by more large heterogeneities (tens of μm2 from their site of generation). The second part of the first chapter is dedicated to discuss the functional role of the Ca2+ microdomains generated below the plasma membrane, from the activation of neurotransmitter release in synaptic terminals to the modulation of specific plasma membrane enzymes, e.g., Ca2+-sensitive adenylate cyclases, NO synthase, and, last but not least, Ca2+-dependent gene activation. This latter event, though occurring far away from the plasma membrane, yet in some cases depends on very local Ca2+ increases occurring at the mouth of plasma membrane Ca2+ channels.

Section iii is dedicated to reviewing the structural and functional bases that permit the generation of microdomains in cytoplasmic Ca2+ due to mobilization of Ca2+ from the sarco/endoplasmic reticulum. The nature and main structural features of the molecular machinery that allow the uptake, storage, and release of Ca2+ in the sarco/endoplasmic reticulum (e.g., Ca2+ pumps, Ca2+ release channels, and Ca2+ buffering proteins) and their heterogeneous distribution within the endomembrane system that forms the basis for the subcellular Ca2+ heterogeneities due to Ca2+ mobilization is discussed. The mechanism of generation of the microdomains named Ca2+ sparks and puffs and their importance as building blocks for the spreading of the Ca2+ signal to the whole cytoplasm (to ensure the physiological response of the cells) is dealt with in some detail, in particular in relation to their role in the physiopathology of the Ca2+ signal in cardiac myocytes.

Section iv is dedicated to mitochondria, organelles that are primarily the targets of cellular Ca2+ microdomains, rather than the generators of subcellular Ca2+ heterogeneity. The importance of cytoplasmic Ca2+ microdomains in shaping the capacity of mitochondria to accumulate Ca2+, the functional consequences of mitochondrial Ca2+ accumulation and their derangements under pathological conditions, particularly during programmed cell death, are reviewed in detail.

Finally, section v is dedicated to the mechanisms of Ca2+ handling by other cellular organelles (e.g., the Golgi apparatus, secretory vesicles, lysosomes) and to the potential role played by such noncanonical Ca2+ stores as further potential generators of Ca2+ microdomains during stimulation.


A. Generation of Subplasma Membrane Ca2+ Microdomains

Given the key importance of Ca2+ influx for the maintenance of cytoplasmic and organelle steady-state Ca2+ levels, as well as for the induction of Ca2+ increases during cell activation, it comes as no surprise that most cells express more than one type of Ca2+ channel. In addition, particularly in cells with a complex morphology, these channels can be differently distributed along the plasma membrane (see also below). Moreover, the repertoire of Ca2+ channels of any given cell can vary not only depending on its specific function and developmental stage, but also during its life span. For example, new channels can be inserted into the plasma membrane upon activation of specific receptors (see, for example, Refs. 340, 386); it has also been proposed that the mechanisms of Ca2+ influx that are activated by store depletion may depend exclusively on the insertion of channels into the plasma membrane (397). The Ca2+ selectivity of these channels is highly variable, ranging from very high (over 3 orders of magnitude in favor of Ca2+ compared with Na+, in the case of VOCs) to rather poor (in the case of some ROCs and some members of the TRP family). A common characteristic of the most Ca2+-selective channels is their capability of transporting efficiently monovalent cations when all divalent cations are removed from the medium, suggesting the existence in the Ca2+-selective channels of a common mechanism that ensures their ionic specificity.

Whatever the nature and gating properties of a channel, it is easily envisaged that upon its opening a high concentration of Ca2+ will be generated at its mouth, on the inner side of the plasma membrane (9). In terms of peak amplitude, it has been calculated that, at physiological extracellular Ca2+ concentration and at 0 mV of membrane potential, the Ca2+ concentration should be in the order of 100 μM within 100 nm of the mouth of an open Ca2+ channel (244). Even higher local concentrations will be transiently reached in voltage-clamped cells during the so-called tail current (i.e., during the repolarization phase). These values have been calculated for bovine adrenal medullary cells, a cell type where the nature of the channels, and the concentration and binding kinetics of the soluble Ca2+ buffers within the cytoplasm, are known in great detail (243); for most other cell types, these parameters are much less defined. In general terms, the peak amplitude and spatial diffusion of the Ca2+ microdomain formed at the mouth of a Ca2+ channel and its immediate neighborhood will depend on the conductance of the channel itself, its Ca2+ selectivity, the concentration of Ca2+ in the extracellular medium, the membrane potential, and the nature and amount of the intracellular Ca2+ buffers.

Over the last two decades, several experimental approaches have been adopted in the attempt to measure, directly or indirectly, the free Ca2+ concentration close to a Ca2+ channel’s mouth. Back in 1980, Llinas et al. (192) injected a low-affinity mutant of the Ca2+-sensitive photoprotein aequorin into squid giant axons and calculated that hot spots in the range of 200–300 μM are reached in the synapse during action potential discharge. Neher and co-workers (139), comparing the rate of exocytosis induced by uncaging Ca2+ from a high-affinity buffer and by physiological stimulation, calculated values in the order of 100–200 μM in the vicinity of the channels in synaptic terminals of retinal bipolar cells. Later studies in the large synapses of the calices of Held led to the conclusion that under physiological conditions fast vesicle release can be accounted for by increases in Ca2+ concentration to lower values, on the order of 20 μM (326).

The question of the [Ca2+] necessary to trigger fast vesicle exocytosis is not a mere problem of accuracy measurement but has major functional implications. Indeed, if a [Ca2+] of 10–20 μM is sufficient to trigger vesicle release, it can be concluded that channels other than those in close proximity to the vesicles are involved; if [Ca2+] in the 200–300 μM range is instead necessary to trigger vesicle fusion, the consequence is that only the channels closely apposed to the secretory site contribute to the neurotransmission event. A conservative conclusion on this point would be that, depending on the synapse type, either type of arrangements is possible.

Subplasma membrane [Ca2+] significantly higher than those measured in the bulk cytosol have been determined by several groups by localizing Ca2+-sensitive probes on the inner surface of the plasma membrane, both at rest and during opening of Ca2+ channels in cells other than neurons. Thus, by using a SNAP-25 aequorin chimera, Marsault et al. (202) concluded that on the inner surface of the plasma membrane of the smooth muscle cell line A7r5 at rest, the mean Ca2+ concentration is ∼1–2 μM (possibly reflecting the existence of hot spots in the vicinity of flickering channels, rather than an homogeneous high level underneath the plasma membrane), while during Ca2+ influx due to CCE activation, values as high as 50 μM can be reached (see Fig. 1). Similarly, Nakahashi et al. (241) concluded that an aequorin adenylate cyclase chimera bound to the inner surface of the plasma membrane is capable of detecting, upon activation of CCE, values in the order of 10–20 μM. Higher levels of Ca2+ in the subplasma membrane region (compared with the bulk cytoplasm) have been calculated also using green fluorescent protein (GFP)-based Ca2+ probes (cameleons): in particular, the same Ca2+ probe either selectively localized within caveolae, homogeneously distributed along the inner surface of the plasma membrane, or in the cytoplasm measured quite different values both at rest and upon activation of CCE (151). In particular, the [Ca2+] in the cytoplasmic rim around caveolae at rest was ∼320 nM, significantly higher than the mean value below the plasma membrane (260 nM) and ∼10-fold higher than that monitored in the bulk cytoplasm (22 nM). Upon activation of CCE, the [Ca2+] peak was much larger in the caveolae region (but the latter values were not quantified). More recently, using the same probe, Isshiki et al. (150) demonstrated that in resting conditions there is a rather inhomogeneous high Ca2+ level along the subplasma membrane cytoplasmic rim. This relatively elevated [Ca2+] could be abolished by removing Ca2+ from the medium and restored upon Ca2+ readdition (Fig. 2). Along the same line, also Nagai et al. (240), using new cameleons either bound to the inner leaflet of the plasma membrane or free in the cytoplasm, found a higher level of subplasma membrane [Ca2+] compared with that in the bulk cytoplasm, both at rest (∼80 vs. 30 nM, respectively) and upon stimulation (well above 1 μM below the plasma membrane compared with ∼700 nM in the cytoplasm). Less dramatic, but still measurable differences between the cytoplasmic rim at the inner surface of the plasma membrane and the bulk cytosol have been monitored with other probes, e.g., fluorescent Ca2+ dyes endowed with a lipid anchor or with other GFP probes bound to the inner surface of the plasma membrane. Thus, using the GFP-based Ca2+ indicator ratiometric pericam, it has been reported that the peak value of [Ca2+] in the bulk cytosol or on the inner leaflet of the plasma membrane in pancreatic β-cells depolarized by high K+ reached 1.38 and 1.82 μM, respectively (284).

FIG. 1.

[Ca2+] changes close to plasma membrane Ca2+ channels. Top: schematic view of the Ca2+ microdomains generated around an open channel. The pseudocolor scale (on the right) indicates the approximate values of [Ca2+], and the bar indicates the approximate distance. A and B: Ca2+ changes in the bulk cytosol and in the cytoplasmic rim underneath the plasma membrane as revealed by recombinant aequorin localized in the cytosol or bound to the inner surface of the plasma membrane. Bottom, A: A7r5 cells (a smooth muscle cell line) expressing cytosolic aequorin. B: same cells expressing a fusion protein between SNAP25 and aequorin that binds selectively to the inner leaflet of the plasma membrane. Please note that while Ca2+ mobilization from intracellular stores results in a very similar Ca2+ rise with either probe, the Ca2+ influx elicited by Ca2+ addition (to induce capacitative Ca2+ entry) causes a dramatically higher increase in [Ca2+] when measured with the aequorin bound to the plasma membrane. For other details, see Marsault et al. (202). [A and B from Marsault et al. (202).]

FIG. 2.

Changes of [Ca2+] on the inner side of plasma membrane rafts and their importance in the activation of specific cellular functions. Top: schematic view of the plasma membrane organization including membrane proteins and lipid rafts. Phospholipids forming rafts are depicted in greenish blue and cholesterol (enriched in rafts) as the red symbols. A: region of the cytoplasm close to the CCE channel. B: bulk cytosol. C: a membrane-anchored effector system, e.g., NO synthase. In this scheme the channel responsible for CCE is hypothesized to be located in the lipid raft. Bottom, A: endothelial cells expressing a raft targeted cameleon; measurement of [Ca2+] in the region depicted as A in the top panel. B: endothelial cells expressing a cytosolic cameleon; measurement of [Ca2+] in the region depicted as B in the top panel. Note the large change in fluorescence ratio (a measure of the change in [Ca2+]) upon changing the [Ca2+] of the medium only in the case of the cameleon bound to the inner side of the rafts. C: [Ca2+] increases elicited by activation of CCE in HeLA cells expressing an aequorin-NO synthase chimera; a cytosolic aequorin and the aequorin-NO synthase chimera, whose membrane-anchoring domain has been deleted. Note the much larger increase in [Ca2+] upon activation of CCE when using the aequorin-NO synthase construct. [A, B, and D from Isshiki et al. (150); C from Lin et al. (185).]

Although Ca2+ microdomains below the plasma membrane are primarily due to opening of plasma membrane Ca2+ channels, it should be mentioned that they can also depend on IP3-induced Ca2+ release from intracellular stores evoked in the immediate vicinity of the plasma membrane (see, for example, Ref. 26). Close contacts between IP3-sensitive Ca2+ stores and plasma membrane have indeed been found in several cell types (for a recent review, see Ref. 22). An emerging theme is the elucidation of the cytoskeletal scaffolds that allow the formation of signaling clusters on the inner side of the plasma membrane. In the Drosophila photoreceptor cells, INAD, a protein consisting of tandem arrays of five PDZ domains, was shown to assemble in the rhabdomere a complex including TRP and TRPL channels, protein kinase C (PKC), phospholipase C (PLC), rhodopsin, calmodulin, and myosin III, that is necessary for the G protein-mediated signaling cascade triggered by light exposure (182). Neurons express different isoforms of Homer, a scaffolding protein with an EV1 domain that binds to a consensus sequence present in various signaling proteins [e.g., mGluRs, IP3Rs, ryanodine receptors (RyRs), TRPs, and the Shank postsynaptic proteins, part of the NMDA-associated PSD-95 complex] and a coiled-coil dimerization domain (376). These intermolecular bridges may allow the formation of stable receptor complexes at the level of the synapse that favor the cross-talk of glutamate-dependent signaling pathways. This structural arrangement is not limited to excitable cells, although in other cell types the molecular information is still scant. It is likely, however, that similar, when not the same, scaffolding proteins play a role. Indeed, the Homer proteins are ubiquitously expressed, and in nonexcitable cells, Homer was shown to regulate the physical association between TRPC1 and the IP3R, that in turn controls the gating of these plasma membrane channels (400).

Polarized epithelial cells, such as those forming exocrine glands, provide another interesting example of membrane clustering of receptors and transducing proteins and allow some insight into the functional significance of these molecular arrangements. In these cells, there is no doubt that Ca2+ waves triggered by stimulation with hormones or neurotransmitters (cholecystokinin, bombesin, acetylcholine) always originate from the apical pole. However, the underlying mechanism of this polarization is still debated. According to one view, based on patch-clamp experiments with two separate stimulatory and recording pipettes, signaling complexes at the apical pole are more sensitive to agonist stimulation, implying a higher density and/or more efficient coupling in this portion of the cell (183). According to a radically different view, receptors are not clustered at the apical pole, but rather at the basal pole (318, 393). Nevertheless, the stimulation of receptors located in the basal membrane causes a cytosolic Ca2+ concentration ([Ca2+]c) rise that initiates in the apical pole (7, 370), because the latter region is endowed with a higher density of IP3 receptors (this conclusion is undisputed), and IP3 produced at the basal pole can rapidly diffuse throughout the cell. In support of this view is the observation that intracellular application of IP3 and/or cyclic ADP ribose (cADPR) (Fig. 3) through the patch pipette always causes Ca2+ spikes in the apical pole (370). Whatever the mechanism, the functional significance of this polarization is striking. Specific functions can be triggered at the apical pole (e.g., stimulation of exocytosis, but also activation of Ca2+-regulated Cl channels) and at the basal pole (e.g., regulation of nuclear transcription, but also activation of Ca2+ regulation of K+ channels). In general, it can be concluded that the spatial clustering of receptors and Ca2+ channels, and the ensuing spatial patterns of Ca2+ diffusion within the cell, as well as regulatory effects within membrane microdomains, contribute to increase the flexibility of Ca2+-mediated signals.

FIG. 3.

Localized Ca2+ changes in pancreatic acinar cells. Effects of cADPR (A), inositol 1,4,5-trisphosphate (IP3) (B), and cADPR + IP3 (C) on Ca2+-activated Cl current and localized increases of [Ca2+] at the apical pole in pancreatic acinar cells. [From Cancela et al. (43).]

B. Functional Role of Subplasma Membrane Microdomains

The generation of local hot spots at the mouth of Ca2+ channels (or in general just below the plasma membrane) critically controls a number of different cellular functions. By far the most intensely investigated is the secretion of neurotransmitters at the presynaptic membrane. This latter topic has been extensively and repeatedly reviewed in the last years. Accordingly, here we will limit ourselves to summarize the fundamental facts and concepts, referring the reader to other contributions for a more thorough analysis of this problem (23, 31, 61, 85, 162, 207, 208, 246, 247, 326, 345, 355). Below we will deal more extensively with other Ca2+-dependent events, similarly modulated by local [Ca2+] at the subplasma membrane level, that have received relatively less attention in the recent past.

As discussed above, although the Ca2+ levels reached at the mouth of the open channels and the spatial diffusion of Ca2+ within the cell may vary quite extensively depending on the channel type, synapse considered, and the buffering characteristics of the cytoplasm, there is a general consensus that under physiological conditions small synaptic vesicles require increases of [Ca2+]c at least 5–10 times higher than those reached in the bulk cytoplasm to efficiently and rapidly fuse with the plasma membrane. Synaptic vesicles are indeed known to fuse nonrandomly with the presynaptic membrane, but at specific sites, the so-called active zones. As discussed above, the latter are enriched with Ca2+ channels, and thus the vesicles containing the neurotransmitters are strategically located to take full advantage of the Ca2+ microdomains generated at, or close to, the mouth of the channels themselves. It should be stressed that the microdomains generated at the active zones are responsible for the fusion of small synaptic vesicles, while other slower exocytotic events, such as those occurring in endocrine tissues probably depend more on bulk, or less localized, cytosolic Ca2+ increases (52). The release of dense-core vesicles from synaptic terminals is also presumably dependent on bulk increases of [Ca2+]c (53).

Exocytosis is by no mean the only physiological event that depends on the amplitude of the hot spots occurring close to the mouth of plasma membrane Ca2+ channels. The modulation of the Ca2+ channels themselves by Ca2+ is another key event occurring at that site. Indeed, many Ca2+ channels are inactivated by the incoming cation, and most experimental evidence supports the notion that the site of this inactivation resides on the inner side of the membrane and thus depends on the amplitude of the Ca2+ level reached at the mouth of the channel itself (5, 92, 173, 344). Thus, not only the substitution of Ca2+ with Ba2+ prevents inactivation, but inclusion in the patch pipette of a fast Ca2+ buffer such as BAPTA or a slow one, such as EGTA, has dramatically different effects on the inactivation of the Ca2+ current, particularly in the case of some VOCs (184), i.e., BAPTA largely prevents inactivation, whereas EGTA does not. Ca2+ inhibition is not a unique characteristic of Ca2+ VOCs and other Ca2+-permeable channels (e.g., Ca2+ release-activated Ca2+ channel, CRAC) are potently inactivated by Ca2+ itself (191, 261, 407). Last, but not least, it should be mentioned that local [Ca2+] elevations may have the opposite effect in some channels, i.e., a facilitating role on Ca2+ channel gating (182).

A striking effect of the hot spots generated at the mouth of L-type Ca2+ channels, totally unrelated to secretion or Ca2+ channel modulation, is that reported in References 67, 71, and 163. These authors investigated the transmission of the Ca2+ signal evoked by opening of plasma membrane L-type Ca2+ channels to the nucleus using as models primary cultures of hippocampal or cortical neurons. This consists in the activation of CREB phosphorylation and downstream activation of gene transcription. The authors showed that in their model systems the phosphorylation of CREB is dependent on the nature of the Ca2+ channel (it depends critically on the activation of L-type channels and in particular on their COOH-terminal domain, while the role of other Ca2+ VOCs or NMDA receptors was negligible) and not on the amplitude of the bulk increases in [Ca2+]c. Most surprisingly for an event that occurs at a distance of several microns from the plasma membrane, CREB phosphorylation in the nucleus is critically dependent on the peak [Ca2+] reached close to the mouth of the channel itself. Indeed, the slow Ca2+ buffer EGTA was totally inefficient at blocking the activation of nuclear CREB phosphorylation, while the fast Ca2+ buffer BAPTA completely prevented it, despite the fact that the two Ca2+ chelators were equally effective at inhibiting the increases in bulk [Ca2+]c (67). Dolmetsch et al. (71) provided an elegant molecular explanation for this effect. In particular, they showed that the COOH terminus of L-type channels interacts with calmodulin in a Ca2+-dependent fashion that leads to the activation of the mitogen-activated protein kinase (MAPK) pathway; this in turn causes the phosphorylation of the CREB kinases Rsk1 and -2 and eventually to long-term CREB phosphorylation. All these events are thus critically dependent on the elevation of Ca2+ in the immediate vicinity of L-type channels and the isoleucine-glutamine (“IQ”) motif in the COOH terminus of the L-type Ca2+ VOCs (that binds Ca2+-calmodulin) is critical for conveying the Ca2+ signal to the nucleus from the mouth of the channels. Interestingly, while the phosphorylation of CREB depends on the microdomains of Ca2+ around the L-type Ca2+ channels, CREB-dependent gene transcription appears to depend also on the increases in nuclear [Ca2+], which occur after relatively intense stimuli. The dual regulation of signaling pathways by the [Ca2+] near the channels and in the nucleus may permit neurons to precisely tailor transcriptional activation to specific types of electrical or chemical stimuli and at the same time ensures that only robust stimuli that generate substantial nuclear [Ca2+] elevations are converted into long-term changes in gene expression (10, 135, 136; for a review, see Ref. 70).

Evidence not yet as clear linking local Ca2+ events to activation of gene transcription has been reported in other model systems and for other types of Ca2+ channels, namely, the P/Q type (358) and the NMDA receptors (364). In particular, using either HEK cells transfected with P/Q-type channels or cerebellar granule neurons, it was found that the synaptic protein syntaxin 1A is upregulated in response to P/Q-type channel activation. This signaling appears to require communication between the channel and internal Ca2+ stores because it is blocked by inhibitors of the SERCAs (155). Thus the P/Q channel regulation of syntaxin 1A may be a mechanism by which Ca2+ channels regulate the expression of proteins that participate in synaptic vesicle release. Along the same line, Greenberg’s group (64, 364) showed that stimulation of neurons in culture with ephrin B triggers clustering of NMDA and ephrin receptors and association of the receptor complex with the Src family of tyrosine kinases. In turn, the phosphorylation of the NMDA receptor increases the Ca2+ flux through the channels themselves and potentiates their activation of CREB-dependent gene expression. Unlike the situation of L-type Ca2+ channels, however, in the latter cases it is still unclear whether activation of gene expression depends on microdomains of Ca2+ around the channels or on bulk nuclear and cytoplasmic Ca2+ rises. Last, but not least, major functional differences in terms of conveying the Ca2+ signal from the channels to the nucleus have been demonstrated between the activation of synaptic versus nonsynaptic NMDA receptors (137). Activation of synaptic NMDA receptors resulted in a robust activation of CREB-dependent transcription, whereas activation of extrasynaptic NMDA receptors not only caused an inhibition of CREB activity, but also triggered cell death. The conclusion is that, given the different subunit composition of the NMDA receptors (NR2B extrasynaptic and NR2A synaptic), local events around the NR2A NMDA receptor subunit at synapses may underlie synapse-specific signaling to the nucleus.

Similarly striking appears the efficacy of Ca2+ entering through CCE channels in eliciting activation-inhibition of different isoforms of Ca2+-dependent adenylate cyclases (86). When recombinantly expressed, Ca2+-sensitive adenylate cyclases give rise to the anticipated changes in cAMP levels in response to CCE. However, CCE appeared far more efficacious at regulating their activity than other means of elevating bulk [Ca2+]c to similar levels, i.e., 1) release from intracellular stores mediated by IP3, 2) Ca2+ ionophores, or 3) arachidonic acid-mediated Ca2+ influx (81). It has thus been concluded that a very close apposition exists between Ca2+-sensitive adenylate cyclases and CCE channels, and more important, that the local microdomains of high [Ca2+] close to the channels are the key events in the control of cAMP synthesis by these enzymes (81, 225). Of interest, some of the most Ca2+-sensitive adenylate cyclases (AC5, AC6, and AC8) are concentrated in caveolae, from which the Ca2+-insensitive adenylate cyclase (AC7) is excluded (225, 342). We have mentioned above that the rises of Ca2+ underneath caveolae upon CCE activation are larger than those occurring in the cytosol or in other parts of the subplasma membrane cytoplasmic rim (150, 151).

Another physiologically important example is that provided by endothelial nitric oxide synthase (eNOS), another enzyme normally localized at the cytoplasmic surfaces of caveolae, via an acylation process (87, 88, 332). Mislocalization of eNOS outside the caveolar membrane markedly reduces the sensitivity of the enzyme to Ca2+ influx as a trigger for NO production. Furthermore, Isshiki et al. (150) showed that the increase in subplasma membrane Ca2+ elicited simply by shifting cells from Ca2+-free to Ca2+-containing medium results in robust activation of eNOS, under conditions where there is no appreciable increase in bulk [Ca2+]c, but significant increases in the subplasma membrane Ca2+ levels (150) (and see Fig. 2).

The role of subplasma membrane Ca2+ microdomains is also of key importance in modulating the Ca2+ uptake capacity of a mitochondrial subpopulation (304, 306); this issue will be addressed in detail below. Suffice it to say here that mitochondria appear to be both the target of these microdomains (thus allowing a very rapid and efficient uptake of Ca2+ by the organelles upon Ca2+ influx) as well as the key player for maintaining the Ca2+ flux through some channels, by buffering Ca2+ locally and thus preventing their Ca2+-dependent inactivation.

Finally, a particularly striking, and apparently counterintuitive, role of subplasma membrane Ca2+ microdomains is that occurring in smooth muscles. In these cells, a local mobilization of Ca2+ from strategically located ER cisternae can activate Ca2+-sensitive plasma membrane K+ channels leading to cellular hyperpolarization and muscle relaxation, an effect that is the opposite of that elicited by a global Ca2+ elevation, i.e., myofilament contraction (274).


The use of the term sarco/endoplasmic reticulum (SR/ER) was introduced (44) to suggest that the two tubular networks are the same cytological entity, with the SR being the muscle-specialized version of the ER of all other cell types (44). In reality, the similarities between ER and SR are not so obvious. Morphologically, the SR of striated muscles (cardiac and skeletal) comprises a network of tubules (longitudinal SR) and cisternae (TC, terminal cisternae) with a very specific anatomical distribution with respect to the plasma membrane invaginations, the transverse tubules (t tubules), and the myofilaments (98, 289, 346). On the contrary, since the classical studies of Palade in the 1950s (258–260), it was clear that the ER of all other cells does not have an obvious ordinate morphology, but fills up the cytoplasm with an intricate network of tubules and cisternae that are continuous with the nuclear membrane. Not only are the morphological features of the ER and SR dramatically different, but their protein composition is also quantitatively and qualitatively quite diverse. In the longitudinal SR, the SERCAs represent up to 90% of the total membrane proteins while the region of the SR juxtaposed to the t tubules, the junctional SR, is highly enriched in a Ca2+ release channel (also known as RyR) (44, 291). The RyRs are not present in any other location along the SR membrane. The SERCAs, on the contrary, represent up to 1% of the total membrane proteins of the ER and membrane regions so highly enriched in Ca2+ release channels such as the junctional SR have never been found in the ER. In addition, a large part of the ER is covered with ribosomes (the so-called rough ER) while no ribosomes are found on the surface of the SR. At the same time, however, the SR expresses several proteins typical of the ER, e.g., the immunoglobulin heavy-chain binding protein, BiP, calnexin, calreticulin and protein disulfide isomerase (PDI) (211). Finally, it must be stressed that a classical ER network does exist also in striated muscle, close to the nuclei, and endowed with all the typical morphological features of the ER of other cells, e.g., covered by ribosomes and continuous with the nuclear membrane. Thus a conservative answer to the question of whether the SR and the ER are the same organelle is that the SR is a differentiation of an ER-derived reticular network, that has progressively evolved into a unique highly specialized morphofunctional structure, characteristic of the striated muscle fibers.

Does the structural and functional heterogeneity of the SR, as far as Ca2+ handling is concerned, have an equivalent in the ER of other cell types? Tentatively one could try to answer the question using several criteria: 1) morphology, 2) membrane and luminal continuity, and 3) protein composition. These criteria are addressed separately below.

A. Organelle Morphology

Three morphologically distinct regions can be identified in the ER: the smooth, the rough ER, and the nuclear membrane. The first two regions differ by the presence of bound ribosomes (258–260), while the nuclear membrane is significantly different from the rough ER in as much as the ribosomes are bound only to the outer surface, it contains the nuclear pores, and the inner surface has a unique protein composition and function (121, 363, 382). To the best of our knowledge, however, to this morphologically distinct regions there is no correspondent qualitative functional differentiation in terms of Ca2+ handling: they all possess Ca2+ channels and pumps, and Ca2+ can be taken up and released from all three ER regions (212, 291). Quantitatively, though, the density and distribution of the proteins involved in Ca2+ handling are far from being homogeneous, an important aspect that is discussed in detail below.

B. Membrane and Luminal Continuity

There is strong, functional, and morphological evidence that both membrane and lumen are continuous throughout the ER network. Direct evidence for membrane and luminal continuity has been provided by experiments of fluorescence photobleaching in cells expressing an ER-located GFP. Prolonged light-induced photobleaching of GFP in a single small region of the ER results in diffuse reduction (up to complete disappearance) of fluorescence from the whole reticular network, including the nuclear envelope; on the other hand, short bleaching periods are followed by a rapid restoration of fluorescence due to the diffusion of GFP from neighboring ER tubules (77, 187, 230). Functional continuity within the ER has also been demonstrated in terms of Ca2+ handling properties. Petersen and co-workers (223) have demonstrated the concept of Ca2+ tunneling, in which ER in the apical region of acinar cells of the pancreas is rapidly refilled with Ca2+ originating from a pipette attached to the basolateral membrane (i.e., on the opposite site with respect to the apical pole), without measurable increases in bulk cytosolic Ca2+ concentration. More recently, direct monitoring of Ca2+ diffusion within the ER lumen has been carried out in pancreatic acinar cells by Park et al. (262), confirming the conclusion that, at least in pancreatic acinar cells, the ER lumen is continuous and Ca2+ is in rapid equilibrium within the network, from the cisternae located at the basolateral side up to the ER intermingled with the zymogen granules in the apical pole. A similar conclusion was reached by Verkhratsky (384) for different neurons. However, continuity in the ER lumen is not always the rule. In cultured hippocampal neurons, for example, vesicular ER compartments endowed with the capacity of accumulating and releasing Ca2+, but clearly luminally discontinuous from the rest of reticular ER, have been described in dendrites (13). Functional evidence of noncontinuity, at least in terms of equilibration of Ca2+ between different ER compartments, has been provided in several model systems. For example, in some smooth muscle cells it has been demonstrated that IP3 and caffeine (that activates RyRs) can release Ca2+ independently of each other, and the two Ca2+ pools apparently do not equilibrate (27, 122). Similar results have been obtained in chromaffin cells (48, 254; but see also Ref. 149), and the existence of subregions of the ER with different levels of Ca2+ in their lumen has been suggested not only by experiments using recombinant aequorin (230) but also by electron microscopic (EM) determination of Ca2+ content (275). However, it should be pointed out that it is still unclear whether these functional differences occur within the ER sensu strictu or in other tubovesicular compartments endowed with Ca2+ uptake and release mechanisms (see below).

C. Molecular Heterogeneity

By far the most convincing approach to demonstrate heterogeneity in the ER is to determine its protein composition and, using classical cell biological methods (subcellular fractionation and immunocytochemical localization), to establish whether indeed there are functionally distinct subdomains.

1. The SERCAs

A key component that could lead to heterogeneity in Ca2+ handling is obviously the mechanism of Ca2+ accumulation. The SERCA family includes the products of three genes, named SERCA1 (ATP2A1), SERCA2 (ATP2A2), and SERCA3 (ATP2A3), each giving rise to alternatively spliced mRNA and protein isoforms. Until recently, the situation appeared relatively simple. It has been known for some years that SERCA1 and SERCA2 have two 3′-end splice variants encoding isoforms differing in their COOH termini, mainly expressed in skeletal muscles of adult (SERCA1a) and neonatal fibers (SERCA1b), in cardiac muscle (SERCA2a), and in all other cell types (SERCA2b) (127, 194). A third SERCA2 variant has been recently identified and named SERCA2c (109). The third gene, SERCA3, was found in most nonmuscle cells (SERCA3a, according to the new nomenclature). More recently, the SERCA3 genes have been shown to possess a higher degree of complexity. Mice, rats, and humans express a variety of species-specific SERCA3 isoforms, in addition to the species-unspecific SERCA3a. Indeed, SERCA3b and -3c mRNAs and proteins were first described in mice and humans (287); these mRNAs originated from the partial or complete insertion of a new exon 21, respectively. Rats were then found to be devoid of SERCA3b and SERCA3c mRNAs, which are replaced by a so-called SERCA3b/c isoform (203). Next, it was found that the insertion of an additional exon 22 in human SERCA3b and SERCA3c mRNAs gave rise to the SERCA3d and SERCA3e isoforms (204). Finally, functional differences in the human SERCA3a, -3b, and -3c isoforms were also pointed out (69). An additional SERCA3f (h3f) mRNA has been very recently identified, derived from new alternative splicing of the human SERCA3 gene, and its distribution pattern in human cell lines of different origins and normal human tissues does not completely overlap with that of the other SERCAs (29).

Subcellular fractionation experiments of microsomal fractions have often revealed partial dissociation in the distribution of SERCAs and other ER marker proteins involved in Ca2+ handling (see, for example, Refs. 251, 316). They have been, to the best of our knowledge, mainly probed with nonisoform specific antibodies and accordingly an even further subcellular heterogeneity in SERCA distribution could have been underscored.

2. Ca2+ release channels

The second most important molecular component of the Ca2+ handling machinery of the ER is represented by the Ca2+ release channels. They belong to two families, the so-called IP3R and RyR. The heterogeneous distribution of these channels within intracellular membranes is potentially one of the key factors in determining a spatial heterogeneity in the Ca2+ signals and thus in the formation of Ca2+ microdomains; accordingly, this aspect will be dealt with here in some detail.

a) ryrs. RyRs are expressed in three different isoforms, the product of three independent genes (347, 357); a fourth, apparently fish-specific RyR (96, 234) will not be dealt with here. The gene products assemble in homotetramers, giving rise to a multi-subunit protein channel of very high molecular mass (>2,000 kDa). The isoforms show different degrees of tissue specificity: RyR1 is expressed essentially in skeletal muscle, although recent reports show its expression and activity in other cell types, such as cerebellum (168) and B lymphocytes (120, 329). RyR2 is expressed in heart (257) and, comparatively less abundantly, in brain (103). RyR3 is ubiquitously expressed at very low levels (including those tissues, such as cardiac and skeletal muscle, in which one of the other two isoforms is highly expressed) (349).

As to the subcellular distribution and the correlation with function, the simplest case is that of muscle cells, in which the abundant RyR repertoire represents the prevailing mechanism for rapidly generating the [Ca2+]c rise driving muscle contraction. In both skeletal and cardiac cells, RyRs are present in specialized portions of the SR, the TC facing the t tubule (97). In this domain, the bulky cytosolic protrusion of the RyR represents the “foot” (the electron-dense material interposed between the two membranes) and is part of the machinery that translates membrane depolarization into SR Ca2+ release. This process requires either the transfer of a conformational change from voltage-gated plasma membrane Ca2+ channels to RyR1 (in skeletal muscle) or the triggering of Ca2+-induced Ca2+ release (CICR) by Ca2+ entry in the case of RyR2 (in heart), as described below.

The brain represents a complex case, in which RyRs (mostly RyR2) are expressed at moderate levels in specific subsets of Ca2+ stores. Both glial cells and a variety of neuronal subpopulations (cerebral cortex, hippocampus, and cerebellum) have been shown to express RyRs (for a review, see Ref. 33), and both isoform specificity and expression levels appear to be developmentally regulated (233, 317). The wide expression of RyRs, and their fine regulation, appear to correlate with an important role of specific Ca2+ stores in modulating a variety of neuronal functions, ranging from morphogenesis (145) to synaptic plasticity (78), to cite but two interesting examples. In the first case, axonal guidance was correlated with RyR-mediated Ca2+ release; in the latter, in hippocampal dendritic spines, Ca2+ entry through NMDA receptors was shown to trigger a much larger CICR from RyR-dependent Ca2+ stores (in other reports, however, this effect was not observed, as store depletion did not reduce synaptically evoked Ca2+ signals; Ref. 164). The situation of RyR functions in glial cells is enigmatic. As mentioned above its expression is well documented (for a review, see Ref. 33), although most groups failed to reveal clear RyR-dependent Ca2+ release (see, for example, Ref. 267).

Finally, a challenging case is that of RyR3, given its very low levels of expression in essentially all cell types (117). What is its function, given that in most cases other Ca2+ release channels (RyR1 and -2 or IP3Rs) are much more abundant? An intriguing possibility was put forward in the most unexpected case, i.e., skeletal muscle, in which the amount of RyR3 is exceedingly small compared with the main isoform, RyR1 (348). In some nonmammalian vertebrates, such as frogs, RyR1 and RyR3 are expressed in similar amounts, and thus their localization and functional properties could be independently assessed. EM results showed that, although both isoforms are restricted to the TC, RyR3 are located in a more peripheral position (perijunctional feet) (84). As to function, CICR activity of RyR1 was shown be suppressed in situ, thus leaving RyR3 as the prevailing system for extending the Ca2+ signal by CICR in this cell model (236, 237). In these concepts (inhibition of RyR1 CICR in situ vs. high efficiency of RyR3) the apparent paradox of low level RyR3 expression in adult mammalian muscle could find an explanation. More in general, this weakly expressed RyR isoform could prove a relevant potentiating mechanism of Ca2+ signals. Coherent with this possibility, the RyR3 knockouts show an overt phenotype, which includes behavioral abnormalities (12, 105) but also reduced spark activity (59) and sensitivity to caffeine-induced muscle contracture (320).

Although by far the best-characterized mechanisms of activation of RyRs are on the one hand the protein-protein interaction between RyR1 and Ca2+ VOCs in skeletal muscle and on the other CICR in heart and most other cells (affecting primarily RyR2 and -3), over the last few years much interest has attracted the demonstration that cADPR can also modulate the opening probability of the RyRs (for a recent review see Ref. 176). Although it is undisputed that cADPR can cause the release of Ca2+ through the RyRs, it remains undetermined whether this molecule acts as a long-term modulator of Ca2+ release or whether its concentration can rapidly vary within cells upon receptor activation. It is worth mentioning here that another soluble messenger, NAADP, may also activate the RyRs, although the identification of the Ca2+ release channels sensitive to NAADP is still a matter of controversy, particularly in mammalian cells (see sect. v for a more detailed discussion of this point).

b) ip3rs. IP3Rs were first identified and cloned in the mouse in 1989 (104). Currently, three genes are known, with several alternatively spliced variants. The full-length sequences for three distinct IP3R genes from human and rodents are now available; the primary sequences of IP3Rs from other species (e.g., Xenopus, Caenorhabditis elegans, and Drosophila) have also been determined (for recent reviews, see Refs. 270, 368, 385). The type 1 isoform of the IP3R is spliced at three regions, with five distinct variants resulting from splicing at the S2 site. The type 2 isoform has also been proposed to undergo alternative splicing (for reviews, see Refs. 269, 270, 368, 385).

Whereas RyRs can form only homotetramers, IP3R channels can form homo- or heterotetramers. Characteristics such as structure-function relationship, activation by IP3, modulation by Ca2+, ATP, phosphorylation, or accessory proteins have all been thoroughly investigated, and several recent reviews have been published on this topic in the past few years (see, for example, Refs. 269, 270, 368, 385); here we limit ourselves to the discussion of IP3R distribution in the ER membrane as a potential source of heterogeneity in Ca2+ signaling. In cerebellar Purkinje neurons, the cell type with by far the highest expression of IP3Rs, IP3R1 is the most abundant (>90% of the total); these channels are found throughout the whole cell (cell body, dendrites up to the spines), apart from the axon, but within ER membranes the distribution is far from homogeneous (214, 319, 324, 365). The density on smooth ER cisternae is much higher than on rough membranes, and the density of the receptors on the outer surface of the nuclear membrane is comparatively rather low. No immunohistochemical evidence for the expression of IP3Rs on the inner surface of the nuclear membrane has ever been reported in these cells, nor significant labeling of the plasma membrane has been observed.

The situation in other cell types, that most often express more than one subtype of IP3Rs, is far more complex. In the central nervous system, for example, IP3R1 predominates in neurons [when present, IP3R3 is primarily at the level of the nuclear envelope (330)], and IP3R2 is exclusively expressed in astrocytes (333, 339). In the exocrine pancreas, a tissue where all three types of IP3Rs are expressed, the channels are concentrated in the apical pole, the trigger zone where all receptor-activated Ca2+ signals are initiated (235, 331, 370). This reflects a selective enrichment of the receptors in a subcompartment of the ER, given that the bulk of ER is localized in the basolateral region of the cell. A similar situation is true for other polarized epithelial cells, such as salivary glands (178, 179, 245) and parotid acinar cells. No such obvious polarization has been described for other cell types, whether they express primarily one or all subtypes of IP3Rs (for a review, see Ref. 385).

The distribution of IP3Rs on the nuclear membrane is still debated. Evidence has been provided that in some cells IP3R2 is particularly abundant on the nuclear envelope. The localization of IP3Rs on the nuclear membrane is not surprising, given that the nuclear envelope is continuous with the ER and the molecular composition of the outer surface of the nuclear envelope is known to be indistinguishable (as far as other protein markers is concerned) from the rest of the rough ER. Indeed, in isolated nuclei, IP3Rs have been clearly demonstrated on the outer surface also by electrophysiological, direct measurement of IP3-activated currents (197, 198). More debated is the expression of IP3Rs on the inner surface of the nuclear membrane, and conflicting results have been reported. Several functional data have been published in the last years supporting the possibility that IP3Rs are located also on the inner surface of the nuclear membrane (see, for example, Refs. 112, 113, 141, 273). Although these functional experiments are apparently convincing, to the best of our knowledge, however, no immunocytochemical evidence at the EM level for the presence of IP3Rs on the inner surface of the nuclear envelope (rather the opposite) has yet been obtained. It also needs to be stressed that the nuclear envelope often branches into the nucleus, and thus it may occur that the apparent IP3-induced release of Ca2+ from the inner nuclear envelope depends on these invaginations, rather than on IP3Rs located on the inner surface of the membrane. Indeed, Echevarria et al. (76) have recently provided convincing evidence for the existence of such a nucleoplasmic reticulum, continuous with the nuclear envelope and endowed with IP3Rs. More convincing in favor of the localization of the IP3R on the inner surface of the nuclear membrane appears the situation of starfish oocytes, but the situation in these cells may be unique (323).

In several cell types, the IP3Rs appear to be often concentrated close to the plasma membrane (e.g., in neurons, in some smooth muscle, or in atrial cardiomyocytes), and it has been also suggested that the IP3Rs may be endowed with the capacity to interact directly with some plasma membrane proteins, e.g., TRP channels (25, 400) or G protein (402) regulating their opening under a variety of conditions. Finally, the IP3Rs in the postsynaptic densities appear to be part of a complex scaffolding machinery composed of several proteins, including metabotropic receptors for glutamate, homer, etc. (377, 400; for recent reviews, see Refs. 270, 385).

Further heterogeneity can be provided by different functional modifications of the IP3Rs in selected compartments of the cells. Thus, in neurons, phosphorylated IP3Rs appear to be differently distributed between dendrites and cell bodies (279).

Still highly debated is the question of whether and to what extent IP3Rs, and in particular IP3R3, reach the plasma membrane. IP3R1 is endowed with strong ER retention signals (263), but whether this is the case or not for the other isoforms is not known. Good functional evidence for the existence of IP3-gated Ca2+ channels has been reported for the plasma membrane of olfactory neurons (41, 80, 190) and lymphocytes (167, 367).

To complicate matters even further, the distribution of the IP3Rs can vary during cell differentiation and upon stimulation. A major reorganization of the whole ER and of the IP3R distribution occurs during oocyte maturation (see, for example, Refs. 93, 354). Similarly, a spectacular redistribution of IP3Rs occurs in MDCK cells during the formation of the polarized epithelium (58), and a similar result had been observed in keratinocytes (101). In these cases, however, it is yet unclear whether the dramatic rearrangement involves complete or partial redistribution of the ER as a whole or whether it reflects a selective mobility of the IP3Rs themselves within the network.

Several proteins interact with IP3Rs. A few have a modulatory role, while others may be essential in determining its subcellular localization. Among the latter, a key role is probably played by actin, ankyrin, homer, and protein 4.1 (for recent reviews, see Refs. 270, 385). A unique regulation of IP3R1, dependent on pH, Ca2+ concentration, and redox state within the ER lumen is mediated by the resident ER protein of the thioredoxin family, ERp44 (142).

A last, but not secondary, question concerns the subtle microscopic heterogeneity of IP3R distribution, i.e., their aggregation in microscopic clusters. Indeed, ion channels and receptors in the cell membranes and internal membranes are often distributed in discrete clusters. With the use of mathematical modeling, it has been concluded that channel clustering can enhance the cell’s Ca2+ signaling capability (334, 335). Mapping of functional channels (in the ER of Xenopus oocytes) indicates indeed that IP3Rs tend to aggregate into microscopic (<1 μm) as well as macroscopic (∼10 μm) clusters. IP3R clustering may contribute to the spatiotemporal complexity of the [Ca2+]c signals (197, 198; see also below).

While the presence of IP3Rs in the SR of smooth muscle, from different tissues, and in cardiac myocytes, including the Purkinje fibers (125), is undisputed, the expression, localization, and functional role, if any, of IP3Rs in the SR of skeletal muscle is largely undetermined. The literature on this topic is highly contradictory: although initial studies in the 1980s provided strong evidence in favor of a role for IP3 in Ca2+ release from SR (383, 390), there is recent data that convincingly argue against this possibility (290). Unfortunately, this problem has been largely neglected in the last years.

3. Luminal Ca2+ binding proteins

Finally, heterogeneity in ER Ca2+ handling may depend on heterogeneous distribution of luminal Ca2+ binding proteins (215). Such heterogeneity is clear in the SR of striated muscle where the high-capacity, low-affinity Ca2+ binding protein calsequestrin (CS) is exclusively located in the TC (18, 42). Two isoforms of CS are known, one typical of skeletal and the other of cardiac muscle. This protein is expressed, in addition to striated muscle, in some smooth muscle cell types (250, 298, 388, 389) and in avian cerebellar Purkinje neurons (365, 387, 391). In these cells too, CS has an inhomogeneous distribution within the ER, presumably due to the unique sorting characteristics of this protein (107, 108). Other ubiquitous Ca2+ binding proteins (calreticulin, BiP, PDI, etc), on the other hand, appear to have a rather diffuse and homogeneous distribution (for a review, see Ref. 211), but this problem has not been analyzed in great detail.

D. Generation of Ca2+ Microdomains by RyRs

In cardiac and skeletal muscle, a highly structured morphological architecture allows the generation of Ca2+ microdomains at the surface of the SR; these microdomains are a key component of the trigger for firing the Ca2+ signals necessary for cell activation (97, 356). Indeed, in both cell types, the physiological, stimulatory signal leading to contraction is conveyed by an action potential: a plasma membrane depolarization travels via the opening of voltage-dependent Na+ channels and reaches the cell interior through invaginations (the t tubules) in which VOCs are located; this causes the influx of Ca2+ that is insufficient to induce the physiological response (the sliding of the acto-myosinic contractile apparatus), but represents the trigger for the release of Ca2+ (particularly in the heart) from the large intracellular Ca2+ reservoir (the SR), through the RyRs (302). In skeletal muscle, direct coupling between the two molecules (VOCs and RyRs) is believed to cause the opening of RyRs. In the latter tissue, thus the influx of Ca2+ through VOCs plays a facilitatory, but not necessary role. Indeed, the interaction between the two channels is thought to be the necessary event to cause activation of the RyR isoform of skeletal muscle (RyR1) (295), and then Ca2+ release through the RyR1 is amplified by CICR. In heart, no direct physical interaction occurs between the two types of channels, and thus a high [Ca2+] in the proximity of the RyR2 represents the essential activatory signal (238). Therefore, in heart Ca2+ influx through VOCs is necessary to trigger SR Ca2+ release through the RyR2 via the process of CICR.

In this context, the Ca2+ sensitivity of the RyRs, as well as the evaluation of the local Ca2+ concentration to which they are exposed, represents a key factor in determining the excitability of the muscle fiber and the efficiency and duration of contraction. For this reason, a great effort has been placed on identifying the “fundamental” Ca2+ signaling events and clarifying the mechanisms that allow their extension and lead to the contraction of the whole fiber. As will be discuss later, it is now clear that alterations in this ordered series of events leads to dysfunctions observed in a number of pathological conditions (ranging, in heart, from cardiac arrhythmias to heart failure). For reasons of brevity, we will focus on heart cells, in which both the regulatory mechanisms and the pathological consequences have been studied in great detail, to provide an outline of the molecular basis of elementary events, and the transition of these events into global rises in the cell. However, despite the known differences in the molecular identity of RyR isoforms and in the mechanism of channel activation, it should be remembered that the concept of spatially and temporally restricted “elementary” Ca2+ signals elicited by RyR opening (“sparks”) holds true also in skeletal muscle, and the regulatory mechanisms controlling their frequency and the extension into global signals are largely the same.

1. The Ca2+ sparks

Rapid Ca2+ imaging of both cardiac and skeletal muscle has revealed the occurrence of transient, local increases in Ca2+ concentration, denominated Ca2+ sparks (50) that were attributed to the opening of single RyR channels or, more likely, a cluster of RyRs (186, 375). Typically, these localized events are 2 μm wide and last for <0.05 s (10 ms to peak and 20 ms half-decay) (see, for example, Fig. 4). They represent elementary excitation events, are self-limited, and in a nonstimulated heart cell occur with a frequency of ∼100 Hz. Two aspects are relevant for the understanding of their basic properties.

FIG. 4.

Ca2+ sparks. A: typical Ca2+ sparks, taken by high-speed confocal microscopy, in cardiac myocytes loaded with fluo 3. On the left, control cells; on the right, cells from hypertrophic heart. Signal-averaged Ca2+ sparks are shown as line-scan images (top) and as surface plot (bottom). B: sulfo-rhodamine B was used to identify the extracellular space in the TT (red), whereas Ca2+ sparks were imaged simultaneously (green). These images demonstrate that Ca2+ release takes place where the TT and TC are located. Cells are from control (SD), salt-resistant animals (SR/Jr), and salt-sensitive, hypertensive animals (SS/Jr). [From Gomez et al. (123), copyright 1997 AAAS.]

The first regards the number of RyRs involved and their activation mechanism. Is a spark, by definition, the smallest [Ca2+]c rise detected by fluorescence microscopy with Ca2+ indicators, the opening of a single or multiple channels? This issue is still subject to debate, but the data currently available seem to argue against the possibility that a spark represents the activity of a single channel. Indeed, when taking into account the amplitude of the [Ca2+]c rise, the area involved and the buffering capacity of the cytosol, the ion flux generating a spark was estimated to be 2–4 pA (50). Electrophysiological estimates of the current through a single open RyR channel are ∼0.35 pA (and the expectation is that in physiological conditions in situ the current is even markedly smaller) (210). Thus it appears more likely that a spark is due to the synchronous opening of a cluster of RyRs. How does this occur? At least in part, the anatomical architecture provides the mechanism for the synchronous activation. Indeed, the crucial site for muscle excitability is the junction between the t tubules (containing the VOCs) and the TC of the SR (where clusters of RyRs are located). In this restricted cytosolic domain upon opening of the plasma membrane Ca2+ channels, the RyR clusters are exposed to a microdomain of high [Ca2+] that triggers RyR opening and releases SR Ca2+ (97, 356). In turn, the opening of the first RyRs of the cluster will further increase the local [Ca2+] and thus recruit a larger subset of RyRs, if not the whole cluster, thus generating a measurable Ca2+ spark.

The second essential property of the spark is its spatial limitation. In principle, this is a contradiction, given the inherent self-propagability of the Ca2+ signal (i.e., the fact that Ca2+ itself is the main activator of RyRs, and that the process of CICR is necessary for generating the Ca2+ spark, as described above). The simplest explanation for the spatial restriction rests most likely in the organization of the microdomain, which allows a closely packed group of RyRs to act synchronously (“sticky cluster”). Indeed, in normal conditions the [Ca2+]c rise in the microdomain elicited by the opening of one, or few, plasma membrane or RyR channels rapidly recruits most of the cluster, giving rise to the visible Ca2+ spark. In conditions of normal SR excitability, cytosolic Ca2+ buffers prevent the involvement of neighboring clusters (but see below the effect of SR Ca2+ overload). As to the activated cluster, local depletion of junctional SR leads to a rapid drop of Ca2+ flux through the tightly coupled group of RyRs, and thus promptly terminates the Ca2+ spark (343). This spatial limitation could be overcome either by the simultaneous opening of plasma membrane Ca2+ channels (such as the opening of VOCs triggered by the plasma membrane depolarization of the action potential) or by a higher open probability of RyRs, as we will discuss later.

Alternative possibilities for the spatial restriction of sparks cannot be excluded. These may include the heavily debated issue of Ca2+ inactivation, suggested by the evidence of the bell-shaped sensitivity of RyRs to Ca2+ (79), but questioned by the demonstration that Ca2+ inhibition occurs at ∼1 mM [Ca2+] for RyR1 (209) and 5–10 mM [Ca2+] for RyR2 and RyR3 (54, 60). Thus, if any physiological role has to be postulated for this process, it most likely refers to the termination phase of action potentials and the induction of a refractory phase. Another proposed mechanism is “Ca2+ adaptation” of RyRs, i.e., the slow (time constant ∼1 s) decay of channel opening (Po) observed in flash photolysis experiments (130), that could be relieved by a second Ca2+ pulse and thus appeared different from classical Ca2+ inactivation. Occurrence and significance of RyR adaptation is still debated (for a review, see Ref. 91). It should be noted that, at physiological Mg2+ concentration, the time constant is 10-fold lower (378), and thus the role of adaptation in rapid events, such as the control of RyR excitability in sparks, appears more plausible.

2. Generating a global Ca2+ signal

How, and when, do these elementary events turn into global signals, capable of causing the activation of the contractile apparatus, and thus the functional response of muscle cells? The simplest, and most physiological, mechanism is that of the action potential. In this case, the rapid propagation of a depolarization wave across the plasma membrane leads to the virtually synchronous opening of VOCs and thus the summation of Ca2+ sparks into a global rise (99). In this case, the expected increase in subplasmalemmal [Ca2+] (due to the concerted activity of multiple VOCs) leads to a dramatic enhancement of the spark rate (up to 106 higher than in resting conditions), which is further amplified by the greater efficiency of CICR. This leads to a large [Ca2+]c rise and the prompt activation of myocyte contraction.

In pathological conditions, such as cardiac arrhythmias, Ca2+ sparks can generate Ca2+ waves gradually diffusing through the cell. Specifically, conditions of “higher excitability” of SR Ca2+ release, such as that conferred by a higher luminal Ca2+ concentration (129, 341), may prevent the self-limitation of the elementary event and cause a feed-forward mechanism that causes the opening of neighboring RyR clusters, and thus the slow propagation of the Ca2+ rise to the whole cell (49). The asynchronicity of the Ca2+ rise (and consequently of the refractory period) may lead to the establishment of looping signals and thus generate alternative pace-making regions in the heart, leading to cardiac arrhythmias. Apart from the higher intrinsic excitability of RyRs, also the cross-talk of Ca2+ release channels in signaling microdomains has been shown to increase cardiac excitability (195). Type 2 IP3Rs, known to be expressed in heart (125), colocalize with RyRs in the junctional area, i.e., in the cell domain critical for triggering the [Ca2+]c rise. Thus stimulation by endothelin-1, angiotensin II, or α-adrenergic agonists (all known to be arrhythmogenic) lowers the threshold for SR Ca2+ release (195). This through the activation of depolarizing currents, such as ITI, the Ca2+-activated transient depolarizing current (159) causes delayed after-depolarizations (DADs) that may trigger premature action potentials and thus the generation of an arrhythmia.

3. The regulation of RyR activity in vivo

In this context, it is easy to understand why significant emphasis has thus been placed on identifying the regulatory mechanisms operating in the RyR microenvironment and controlling Ca2+ release through the channels (for a review, see Ref. 91). At least in vitro, it has been shown that a number of factors play an active role; they include Ca2+ itself, both on the cytosolic side (i.e., the positive feedback at the basis of CICR and the inhibitory effect described at millimolar Ca2+ concentrations) and on the luminal side (sensitivity of RyRs to cytosolic agonists increases at high levels of SR Ca2+ loading), Mg2+ and ATP (although it is unlikely that major fluctuations occur in their cytosolic levels), cADP ribose (that has been proposed to play an important regulatory role not only in nonexcitable cells, e.g., sea urchin eggs, as discussed elsewhere in this review, but also in heart), oxidation of critical residues, and phosphorylation/dephosphorylation events. In addition, several clear-cut examples of regulation by protein-protein interaction have been reported; these include calmodulin (with a reported inhibitory effect on RyR2 and a bell-shaped effect on RyR1, activatory and inhibitory at low and high Ca2+, respectively), FKBP12.6, and calsequestrin. The relative importance of these potential regulatory mechanisms in vivo is difficult to ascertain and remains inferred by either the interpretation of the in vitro studies or by analysis of informative clinical phenotypes.

On the first aspect, a large body of work correlates the state of filling of the SR store with both the excitability of RyRs (129) and the control of Ca2+-activated inward currents (56), such as ITI, two mechanisms that may concur in the generation of arrhythmias. SR Ca2+ overload may also increase the contractility of the myocardium, as directly demonstrated in a mouse model in which SR Ca2+ accumulation was increased by overexpressing a SERCA and ablating the SERCA inhibitor phospholamban (403).

Given that the relative importance of these potential regulatory mechanisms in vivo is difficult to ascertain, the information deduced from informative clinical phenotypes appears particularly useful. Hyperactivation of protein kinase A (PKA), reported to occur in the failing heart, hyperphosphorylates the RyR channels, and hence their binding to FKBP12.6, with consequent alteration of the open probability of the channel and defective channel function (205). Similarly, it was shown that PKC-α, by phosphorylating the inhibitor (I-1) of protein-phosphatase 1 (PP1), affects PLN phosphorylation and hence SR Ca2+ loading. As a consequence, PKC-α, by integrating different Ca2+-dependent signaling routes, guides the functional derangement of pathological cardiac hypertrophy (35). The conceptually simpler genetic diseases provide related examples. Recently, two genetic disorders have been ascribed to mutations in RyRs, a subset of arrhythmogenic right ventricular dysplasia cases (ARVD2), and stress-induced polymorphic tachycardia (VTSIP), in which, respectively, the prevailing or only clinical manifestation is a fatal arrhythmia (374). The diseases are caused by different missense mutations, located in the pore forming region (ARVD2) or in the regions binding the channel negative modulator FKBP12.6 (ARVD2 and VTSIP). Interestingly, in the latter case, the mutations of the two diseases differently affect FKBP12.6 binding, decreasing (ARVD2) or increasing (VTSIP) it (373). Thus it has been proposed that ARVD2 mutations, with different mechanisms, render RyR2 leaky, while the VTSIP mutations reduce the probability of channel opening. In the case of phospholamban, the clinical phenotype of human mutations adds complexity to the picture. In this case, null mutations (and the consequent hyperactivation of SERCA) lead to dilated cardiomyopathy (325), in complete contrast to the mouse knockout in which enhancement of cardiac activity was reported (196). A sensible explanation for the discrepancy is that while in mouse (which exhibits a high frequency of heart rate) phospholamban is mostly inhibited, in humans this is not the case, and chronic inotropic stimulation leads to development of heart hypertrophy (131).

E. Elementary and Global Signals in IP3-Dependent Systems

The occurrence of elementary signals, due to the opening of a spatially restricted group of channels (264) and denominated “Ca2+ puffs” (396), is also shared by the IP3-dependent cell systems. In Xenopus oocytes, imaging experiments at high spatial and temporal resolution demonstrated the appearance, at low agonist concentrations, of local increases that start abruptly, peak in ∼50 ms, and decay in 2–300 ms (265). These localized [Ca2+]c increases represent the opening of a tightly packed cluster of IP3Rs (336). As in the case of RyR-dependent systems, Ca2+ puffs denote the “Ca2+ excitability” of the cell, but the induction of a physiological response requires the coalescence of these elementary events into a larger rise, which may be limited to a portion of a polarized cell, or diffuse to the whole cell body in a truly global Ca2+ signal. Moreover, the “pacemaker” activity of Ca2+ puffs participates in controlling the frequency of repetitive Ca2+ spiking, the phenomenon known as “Ca2+ oscillations” that regulates downstream events such as gene transcription (72) or metabolism (133). The initiation and propagation of a global Ca2+ signal relies on the cumulative recruitment of Ca2+ puffs as the slow rise in [Ca2+] sensitizes neighboring IP3Rs, thus increasing the probability of puff occurrence and their merging into propagating Ca2+ waves (200). Interestingly, in a series of studies carried out in HeLa cells, it could be demonstrated that the “pacemaker” activity during the latency period (i.e., before the initiation of a global Ca2+ rise) originates from a limited number of puff sites, which repetitively fire during consecutive stimulations (369). Whether this represents a structural organization of IP3R clusters, or the convergence of modulatory inputs, remains to be ascertained and may vary in different cell types.

F. Turning Puffs Into Global Signals: Role of Other Second Messengers

Pancreatic acinar cells are an interesting example of how the site of [Ca2+]c increases determines different cellular effects. Agonist-evoked [Ca2+]c signals originate in the apical pole when Ca2+ puffs merge into a larger [Ca2+]c rise (370). These rises can either remain confined to the apical region or be extended to the whole cell (43, 179). In the former case, the effect is limited to the activation of the main physiological function of the cell (secretion of digestive enzymes), in the latter long-term effects are observed (ranging from activation of gene expression to cell death) (272, 372). In this complex system, in which different hormonal stimulations, e.g., cholecystokinin (CCK) and ACh, have distinct Ca2+ signaling and functional effects, interesting information has been acquired.

First of all, the diffusibility of local increases (puffs) by CICR differs in different portions of the cell. When a [Ca2+]c rise was induced by local uncaging of Ca2+, its capacity of triggering a Ca2+ wave differed in the various portions of this polarized cell (being higher in the apical region) and required the concerted action of RyRs and IP3Rs (8). Indeed, when either class of channel was inhibited, the diffusion of the Ca2+ wave was blocked. Thus the apical region, in which receptor clustering (165, 183) could account for the higher sensitivity, acts as the “pacemaker” of the agonist-evoked [Ca2+]c rise. As to the capacity of the [Ca2+]c wave to extend to the basal region, two aspects appear critical. The first is the intrinsic potency of the stimulus (256). Supramaximal stimulation with agonists always induces a Ca2+ wave that reaches the basal region. As to stimulation with lower, more physiological agonist concentrations, there is a general consensus that the [Ca2+]c rise generated by stimulation of different receptors occurs initially in the apical region, suggesting that the three types of channels are clustered in the same stores and activated by the appropriate receptor pathway (40). Local Ca2+ rises could be converted into global responses through the concerted action of different intracellular messengers (IP3, cADP ribose, NAADP), in an agonist-specific manner. ACh-evoked spikes could be efficiently transformed into global signals by further intracellular addition of NAADP and cADP ribose, CCK-evoked spikes by IP3 (43). According to Cancela et al. (43), low concentrations of any of these Ca2+ mobilizing stimulus elicited only Ca2+ elevations localized to the apical pole (see Fig. 3). At variance, however, some reports indicate that the RyR agonist cADP ribose and NAADP increased [Ca2+]c primarily in the basolateral portion of the cell (165, 180). This experimental discrepancy has not been solved yet.

Finally, ACh-triggered cADP ribose synthesis was eliminated in CD38 knock-out animals, and this resulted in different alteration of the Ca2+ signaling at low and high agonist concentrations, respectively (102). Thus conceptually different mechanisms (membrane receptor clustering, molecular diversity of intracellular stores, cross-regulation of second messenger biosynthesis) appear to control the generation and diffusion of intracellular Ca2+ signals in polarized cells. This adds a level of complexity to the system and a further mechanism for the integration of different receptor pathways into a common signaling route.

The extension from the apical to the basal region of the cell is then controlled by another checkpoint, i.e., the clustering of mitochondria, acting as fixed Ca2+ buffers, at the boundary between the two cell domains. As described in more detail in section iv, efficient uptake by tightly packed mitochondria represents a “firewall” transiently blocking the Ca2+ wave (372). To proceed, the Ca2+ wave must therefore overwhelm their uptake capacity and initiate a CICR-mediated wave downstream of the barrier. Overall, this biological system provides a clear example of how the concepts identified, and clarified, in simpler models (the occurrence of elementary IP3R-dependent events, the puffs, and the different modes for extending them to large cell domains, or the whole cell) have a precise physiological application in cell types in which different modes of Ca2+ signaling exert different functional effects. The elucidation of the regulatory mechanisms involved will provide new pharmacological clues for addressing pathological conditions (in the case of the pancreas, the medical emergency of acute pancreatitis) caused by deranged Ca2+ signals. On this topic, recent interesting work has provided insight into the pathogenesis of pancreatic damage induced by alcohol (62) or bile reflux (160, 392).


A. Stores, Sinks, or Inactive Elements in Ca2+ Homeostasis?

The capacity of mitochondria to rapidly move Ca2+ across their membranes was a relatively early notion in bioenergetics and cell biology. The principle of energy conservation in mitochondria (i.e., the translocation by protein complexes of H+ across an ion-impermeable inner membrane) generates a very large H+ electrochemical gradient that can be employed not only by the H+-ATPase for running the endoergonic reaction of ADP phosphorylation, but also to accumulate cations into the matrix. Indeed, research carried out in the 1960s by various groups demonstrated that energized mitochondria can rapidly take up Ca2+ from the medium and characterized the following fundamental principles (for reviews, see Refs. 75, 304): 1) an electrogenic pathway [i.e., most likely a channel, as was recently directly demonstrated (161)] that acts to equilibrate Ca2+ with its electrochemical gradient, and thus accumulates the cation into the matrix, and 2) two exchangers (with H+ and Na+, mostly expressed in nonexcitable and excitable cells, respectively), that utilize the electrochemical gradient of the monovalent cations to prevent the attainment of electrical equilibrium (that would imply, for a mitochondrial membrane potential, ΔΨm, of 180 mV and a cytosolic Ca2+ concentration of 0.1 μM, accumulation of Ca2+ into the matrix up to 0.1 M). It was thus logical to assume that mitochondria were loaded with Ca2+, possibly releasing it in a number of physiological and/or pathological conditions. In other words, mitochondria were very promising candidates for the role of rapidly mobilizable Ca2+ stores.

Experimental evidence obtained in the 1980s clearly disproved this possibility. In those years, while Ca2+ emerged as the ubiquitous, fundamental second messenger known to every biology student, it became immediately evident that mitochondria were not the active store in these signaling pathways. Indeed, the messenger shown to be produced upon stimulation of G protein-coupled or growth factor receptors, IP3, acts on ion channels located in the ER (351). Moreover, the latter organelle (and not mitochondria) was shown to contain the molecular elements of a Ca2+ store: a pump (to accumulate Ca2+ against electrochemical gradient), a channel (to rapidly release it), and buffering proteins (to increase the total amount of ion that can be stored; for a review, see Ref. 291). For this reason, the concept of mitochondria as cellular Ca2+ stores was largely dismissed, and most of the research in calcium signaling focused on the ER (and its specialized counterpart of muscle cells, the SR). As mentioned above, cellular and molecular work allowed to identify different classes (and subtypes) of Ca2+ channels and emphasis was mostly placed on the role of ER/SR (and its subcompartments) in spatio-temporally patterning the increases in Ca2+ concentration occurring upon cell stimulation.

At the same time, also the possibility that mitochondria receive Ca2+ after stimulation, thus acting as Ca2+ sinks, appeared dubious. Indeed, the availability of indicators that could be easily loaded into most cell types and calibrated into accurate [Ca2+] estimates (126) allowed us to verify that in living cells not only resting values (∼0.1 μM), but also those briefly reached after physiological stimulation (1–3 μM) are well below the affinity of mitochondrial Ca2+ transporters (i.e., those estimated in the earlier work with isolated organelles). Thus the general consensus was that mitochondria in most cases of cellular Ca2+ mobilization would receive little, if any, of it, while substantial organelle loading could be achieved only in the case of overt Ca2+ overload (i.e., in a number of pathological conditions, such as, for example, excitotoxic neuronal challenge). In contrast with this view, biochemical work demonstrated that important mitochondrial metabolic enzymes are regulated by Ca2+, thus suggesting fluctuations of matrix Ca2+ concentration and fine tuning of the enzymes in physiological conditions (206). However, given that no direct experimental evidence could support the notion that the Ca2+ concentration in the mitochondrial matrix ([Ca2+]m) rapidly changes upon cell stimulation, this observation did not modify the general perception of mitochondria as relatively inactive bystanders in the complex scene of cellular Ca2+ homeostasis.

This situation was completely reversed by the direct demonstration of mitochondrial Ca2+ uptake in intact cells, and by the appreciation of the importance in this process of subcellular heterogeneity in Ca2+ levels, the theme of the current review. This became possible when tools were developed that allowed the selective measurement of [Ca2+]m in living cells. This was first achieved by targeting to mitochondria a Ca2+-sensitive photoprotein, aequorin (311), and allowed us to demonstrate that a rapid [Ca2+]m peak, reaching values well above those of the bulk cytosol, parallels the [Ca2+] rise evoked in the cytoplasm by cell stimulation (305). Similar conclusions could be reached also with fluorescent indicators, such as the positively charged Ca2+ indicator rhod 2 (that accumulates within the organelle) (30, 74) and the more recently developed GFP-based fluorescent indicators (89, 90, 239). With the latter probes, endowed with a much stronger signal than the photoprotein, single-cell imaging of organelle Ca2+ can be carried out (90, 360). Thus it is possible to match the accurate estimates of [Ca2+]m values, obtained with the photoprotein, with detailed spatio-temporal analyses of [Ca2+]m transients (see Fig. 5). With these tools in hands, not only the notion was confirmed that mitochondria promptly respond to cytosolic [Ca2+] rises, but also that the [Ca2+]c oscillations, the typical response to agonists of many cell types, are paralleled by rapid spiking of [Ca2+]m, thus providing a frequency-mediated signal specifically decoded within the mitochondria, as clearly shown in hepatocytes (133), cardiomyocytes (313), and HeLa cells (90).

FIG. 5.

Microheterogeneity of mitochondrial Ca2+ response. A: schematic view of ratiometric pericam (239; and see Table 1). B: deconvoluted images of HeLa cells expressing the ratiometric pericam during a histamine challenge. On the left, the color scale represents the fluorescence excitation ratio, where blue represents low Ca2+ and red high Ca2+ levels. The pseudocolor image in the top left panel was taken 0.5 s after the addition of histamine, and the following images were taken 1 s apart. Note the different response of mitochondrial subpopulations at early times upon stimulus addition and the diffusion of Ca2+ at later times within the mitochondrial network. For experimental details, see Reference 360. [Modified from Szabadkai et al. (360).] C: kinetics of the nuclear and mitochondrial Ca2+ rises measured with a ratiometric pericams targeted to the mitochondrial matrix and to the nucleoplasm. In the top panel, the overall kinetics of the Ca2+ response of a single cell of the nucleus and of different mitochondrial regions; in the bottom panel, the same trace on an expanded time scale. Note the heterogeneities in the lag time between the rise of [Ca2+] in the nucleoplasm and in different mitochondrial populations within the same cell. [From Filippin et al. (90).]

A critical, and still somewhat controversial, issue was why mitochondria in situ behave so differently from what is expected based on the properties of their Ca2+ transporters established in vitro? In other words, why does [Ca2+]m rise, in a few seconds, to values above 10 μM (in some cell types up to 500 μM) in response to [Ca2+]c elevations that rarely exceed 2–3 μM, given that in the latter conditions mitochondrial Ca2+ accumulation should be very slow? A simple explanation would have been that the affinity of the transporters within cells is much higher than previously reported. However, direct perfusion of permeabilized cells with buffered [Ca2+] similar to those measured in the cytoplasm of stimulated cells induced a relatively inefficient Ca2+ loading of mitochondria (thus confirming the notion of a low-affinity uptake system) (305). Conversely, discharge of Ca2+ from the ER triggered by the direct perfusion of IP3 (thus causing the opening of the physiological Ca2+ release pathway also in permeabilized cells) induced mitochondrial Ca2+ uptake almost as efficiently as in intact, stimulated cells (305). This observation led to the proposition that mitochondria, upon cell stimulation, are exposed to [Ca2+] of the microenvironment of the open IP3-gated channel, that are much higher than those measured in the bulk cytoplasm. In other words, “privileged,” local signaling between the Ca2+ store (the ER) and mitochondria appeared to be the key to the participation of this organelle in intracellular Ca2+ homeostasis. In this scheme, mitochondria could be proposed to act as detectors of the microheterogeneity of cellular Ca2+ signaling, participating in decoding it into defined cell action, as discussed below. The close proximity of mitochondria and the main intracellular store of agonist-releasable Ca2+, the ER, had been observed by electron microscopy in fixed samples of several cell types, also taking advantage of quick-freezing techniques, that avoid artifacts due to tissue fixation procedures (276). In our experiments, we could confirm this proximity in intact, living cells, in which the two organelles were labeled with specifically targeted GFP mutants (GFP for the ER, erGFP, and the blue variant for mitochondria, mtBFP), and resolved with a high-resolution imaging system (optimized for fast acquisition and equipped with software for deconvolution and three-dimensional rendering of the image). Figure 6 shows EM images of ER-mitochondria close apposition and three-dimensional reconstruction of such images. In keeping with the idea of “local” Ca2+ cross-talk between the mitochondria and the ER, we recently demonstrated, through fast single-cell imaging of mitochondrial [Ca2+]m with targeted Ca2+-sensitive GFPs (pericams and cameleons), that [Ca2+]m increases originate from a discrete number of sites and rapidly diffuse through the mitochondrial network. Indeed, fragmentation of the network induced by overexpressing components of the organelle fission machinery, such as the dynamin-related protein 1 (drp-1), greatly reduces the global amplitude of the agonist-induced [Ca2+]m elevation (360).

FIG. 6.

Mitochondria/ER interaction. Electron micrographs (A and B) and three-dimensional reconstruction (C and D) of the ER/mitochondria interaction in an hepatocyte. Please note the very close apposition of the ER (blue in C and D) and the outer mitochondrial membrane (pink). In yellow is the inner mitochondrial membrane, and in green is the three-dimensional reconstruction of the inner cristae. (Figure kindly provided by Dr. Carmen Mannella, Albany, NY.)

A first obvious question is how general is the concept of fast mitochondrial responses to physiological Ca2+ signals, allowed by the “sensing” of microdomains close to the mouth of Ca2+ channels, the concept originally put forward and experimentally evaluated in HeLa cells and hepatocytes? The occurrence of rapid [Ca2+]m responses has been conclusively shown in a wide variety of cell systems in which Ca2+ is released from the ER through IP3Rs (e.g., L929 fibroblasts, 143B osteosarcoma and primary hepatocytes, to name but a few) (30, 133, 303). However, this is not an exclusive property of this signaling mechanism, as the same pattern of mitochondrial Ca2+ signaling (a rapid peak and a slower return to basal values) has been demonstrated in muscle cells (both cardiac and skeletal), in which Ca2+ release from the SR occurs through the ryanodine receptors (36). Particularly striking is the case of heart cells, in which, using Ca2+ probes as diverse as fluorescent dyes and targeted photoproteins, a beat-to-beat relay of [Ca2+]c spiking to [Ca2+]m transients could be demonstrated; every [Ca2+]c pulse induced a [Ca2+]m response, which returned to basal values also at high spiking frequencies (313, 361). Recently Rudolf et al. (321) demonstrated that also in skeletal muscle in a live mouse mitochondria are capable of taking up Ca2+ during a single muscle twitch (321). As to the functional consequences, the pacing of cytosolic, and hence, mitochondrial oscillations (that is under hormonal control) controls in turn the organelle metabolism, thus tuning ATP production to the increased needs of a rapidly contracting muscle, as discussed below.

What about neurons and other cell types in which influx through plasma membrane channels represents the prevailing mechanism for generating a [Ca2+]c rise? Also in these cells, a rapid [Ca2+]m rise parallels the localized or global Ca2+ signal, and a number of experimental observations indicate that exposure to local domains of high [Ca2+] generated in proximity of the opening channels accounts for the efficiency of the mitochondrial response. Specifically, X-ray microanalysis of sympathetic neurons demonstrated substantial Ca2+ uptake into mitochondria during depolarization-induced Ca2+ signals and a marked heterogeneity between individual organelles, with subplasmalemmal mitochondria undergoing the largest increases (285). In hippocampal neurons, opening of NMDA-gated channels induced in a subset of mitochondria a Ca2+ uptake large enough to completely bleach its photoprotein probe content, which could be replenished by diffusion from neighboring mitochondria in a 30-min recovery period (14). In adrenal chromaffin cells, it was demonstrated that mitochondria clustered at neurotransmitter release sites are exposed to high [Ca2+] upon opening of both plasma membrane and ER channels, thus implying the existence of highly structured signaling domains, and undergo the largest [Ca2+]m responses so far documented (up to 0.5 mM). In turn, mitochondrial Ca2+ uptake modulates the availability of Ca2+ for exocytosis, and indeed, uncouplers greatly enhance catecholamine secretion (229). Interestingly, rapid responses to subplasma membrane [Ca2+] microdomains have been documented also in nonexcitable cells; specifically, mitochondria were shown to sense Ca2+ gradients near CRAC channels. This process, by clearing Ca2+ in the proximity of the channel, has been shown to modulate the Ca2+-dependent inactivation of the channels and thus determine the net influx of Ca2+ and the ensuing biological response (118, 146).

Two issues remain to be assessed: 1) what is the magnitude of the Ca2+ rise to which mitochondria are exposed in the microdomain of the above-mentioned examples, and 2) is a rapid, pulsatile increase dependent on microdomains always necessary, or conversely in some cases mitochondria can sense the bulk cytosolic increases, and the consequent slow [Ca2+]m increases account for the biological response? On the first issue, work was carried out by various groups using different Ca2+ probes and cell models and a coherent set of estimates was obtained. In permeabilized RBL-2H3 and H9c2 cells, maximal rates of mitochondrial Ca2+ uptake were achieved by exposing the organelle to ∼16 and >50 μM, respectively (63, 361); in the latter cell type, a rate of uptake comparable to that observed in intact cells upon maximal stimulation of RyRs was achieved by perfusing 30 μM Ca2+. Finally, in chromaffin cells, rapid Ca2+ uptake into mitochondria (in turn needed for efficient vesicle release) requires exposure to a local [Ca2+] >5 μM, a value that is reached upon opening of P/Q-type, but not L-type voltage-gated channels (228). Two concepts thus emerge: 1) the Ca2+ concentration in the microdomain sensed by mitochondria is at least one order of magnitude higher than in the bulk cytosol, and 2) the sensitivity to microdomains (rapidly declining with the distance from the channel) enhances the selectivity of different signaling routes, as it allows mitochondria to discriminate, based on their localization, the opening of channels giving rise to comparable bulk responses.

As to the second issue (i.e., the responses to the small [Ca2+] increases), it should be remembered that mitochondrial Ca2+ uptake occurs also at submicromolar Ca2+ levels. Even in resting conditions, a small amount of Ca2+ is accumulated in the organelle, which is released by dissipation of the mitochondrial membrane potential and causes a small, but detectable, [Ca2+] rise in the cytoplasm and a decrease in the matrix Ca2+ concentration. Thus, in principle, a sustained elevation of [Ca2+]c should always cause a [Ca2+]m rise. This concept (and its physiological relevance) was conclusively demonstrated in steroid-producing cells (adrenal glomerulosa and luteal primary cultures). In both glomerulosa and luteal cells, [Ca2+]m increases were detected by perfusing permeabilized cells with [Ca2+] <0.2 μM and by triggering capacitative Ca2+ influx in intact cells (a procedure that generated a [Ca2+]c rise <0.3 μM). Under those conditions, an increase of mitochondrial NADH production, a Ca2+-dependent matrix reaction, was also detected (350, 359). This provides a clear example of the biological significance of mitochondrial Ca2+ responses to “bulk” cytosolic rises: Ca2+-dependent mitochondrial NADH production provides, through the activity of transhydrogenases, the NADPH necessary for the side chain cleavage of cholesterol (the first, rate-limiting step of steroid hormone biosynthesis). Mitochondrial responses to “bulk” [Ca2+]c rises are not limited however to steroid-producing cells. An additional example is provided by ECV304 endothelial cells, in which stimulation with ATP causes a prolonged [Ca2+]c elevation (a plateau of 0.7 μM that is maintained throughout agonist application), that is paralleled by a sustained [Ca2+]m rise of comparable amplitude (172). As to the exact correlation between [Ca2+]c and [Ca2+]m levels, and the establishment of a “threshold” for mitochondrial Ca2+ responsiveness, information can be obtained from the work by Bootman and co-workers in HeLa cells (57) who exposed mitochondria to [Ca2+] increases of different amplitude and source. Mitochondrial Ca2+ uptake was slow at 2–300 nM and steeply increased >400 nM, and IP3-dependent Ca2+ release from the ER was more efficient than any other source, favoring the concept of privileged Ca2+ communication between the two organelles.

B. Mitochondrial Heterogeneity in Ca2+ Handling

If microdomains are critical determinants of mitochondrial Ca2+ responsiveness, it can be argued that only some mitochondria (or domains of a continuous network) are appropriately placed to “sense” the high [Ca2+]c domains and thus undergo large-amplitude [Ca2+]m increases. Thus one would expect that significant heterogeneity in Ca2+ uptake should be detected within the mitochondrial population of the cell. This seems the case in most cell types, although there are clear differences in the size of the rapidly responding mitochondrial pool. In HeLa and chromaffin cells, this information was achieved taking advantage of the intrinsic properties of the photoprotein aequorin, that is “consumed” during the experiment, since it can emit only one photon in the Ca2+-dependent reaction (229, 303). Thus highly responding mitochondria rapidly exhaust their pool of active photoprotein and in consecutive stimulations the Ca2+ rise essentially reflects only the activity of “low-responder” organelles. Interestingly, a similar estimate was made in the two cell models: ∼30% in HeLa cells and 20–40% in chromaffin cells. Conversely, in other cell types, most mitochondria rapidly respond to agonist-induced Ca2+ signals. This was conclusively shown in skeletal myotubes, both with the aequorin approach (36) and by measuring [Ca2+]m with rhod 2 in permeabilized cells, comparing RyR-mediated responses with maximal uptake due to perfusion of high [Ca2+] (361). Similar data were obtained in RBL-2H3 cells, in which saturation of mitochondrial Ca2+ uptake was observed during IP3-induced Ca2+ release from the ER (63). A conclusive explanation for these differences is difficult, given that only recently tools have become available for directly imaging the [Ca2+]m hot spots with good spatial and temporal resolution. It is however reasonable to assume that the density of ER/SR-mitochondria juxtaposition, and the clustering of Ca2+ release channels (as well as of the yet unidentified mitochondrial Ca2+ transporters) will prove to play a major role.

A second, more macroscopic mechanism for mitochondrial heterogeneity, however, was clearly shown to be operative in those cells in which specialized functions pose strict requirements to cell morphology, and intracellular distribution of organelles and signaling mechanisms. Thus mitochondrial distribution to portions of the cell with radically different functions (and channel repertoires) may cause per se an intrinsic major heterogeneity in organelle response. Neurons are an obvious example of complex cell architecture and highly localized signals. Not surprisingly, numerous reports have appeared on the selective involvement of resident mitochondria in spatially restricted Ca2+ signals. While we refer to specific reviews for a detailed coverage of this broad topic (100), we simply point out the functional significance of such an organization with an example regarding one of the most important and intensely studied processes in neurobiology: synaptic plasticity. In two different experimental systems (crayfish and Xenopus neuromuscular junctions), Ca2+ efflux from mitochondria localized in the synapse appears to contribute in part to posttetanic potentiation (PTP) of neurotransmitter release (366, 395). In the latter case, PTP was observed also if tetanic stimulation was applied in Ca2+-free media, thus suggesting that mitochondria are, at least partially, loaded with Ca2+ also in resting situation, and thus act as a bona fide pool of releasable Ca2+. Apart from neurons, polarized epithelial cells provide another clear example of highly organized mitochondrial distribution and Ca2+ responsiveness, with direct implication for the physiological regulation of cellular function. Specifically, in pancreatic acinar cells, in which the Ca2+ response to a low-dose agonist stimulation is restricted to the apical pole (where it causes granule secretion) by the action of the largest group of mitochondria, clustered between the apical and basolateral portions of the cell and acting as a barrier to the spread of the Ca2+ signal (372). Only stimuli that can overcome this mitochondrial “firewall” induce a global Ca2+ response, and then the functional effects radically change: the nucleus is reached by the [Ca2+]c rise, with consequent modification of the gene expression profile and thus long-term alterations of cellular activity. A conceptually related mechanism has been shown to occur in parotid acinar cells. In these cells, however, fluid secretion requires the activation of both Ca2+-dependent Cl channels of the apical membrane and Ca2+-dependent K+ channels of the basolateral membrane (to maintain hyperpolarization and hence the driving force for Cl movement). Thus, not surprisingly, no restriction of the [Ca2+]c signal to the apical pole is observed, nor existence of a perigranular mitochondrial belt. Conversely, mitochondria are clustered around the nucleus, where they delay the propagation of the Ca2+ wave as well as probably provide a source for ATP needed in various transport or enzymatic functions occurring in this domain (39).

C. Functional Role of Mitochondrial Ca2+ Uptake

What is the function of mitochondrial Ca2+ uptake, and how does it correlate with the microheterogeneity of Ca2+ signaling? Given the focus of this review, we do not discuss in detail the physiological consequences of mitochondrial Ca2+ uptake. A brief overview, however, gives some insight into the potential targets of the different types of organellar Ca2+ signals, described above.

1. Ca2+ effects within mitochondria

Three key metabolic enzymes (the pyruvate, α-ketoglutarate, and isocitrate dehydrogenases) are activated by Ca2+, by different mechanisms: in the case of pyruvate dehydrogenase through a Ca2+-dependent dephosphorylation step, in the latter two cases through the direct binding of Ca2+ to the enzyme complex (134, 206). Recently, also some metabolite transporters were shown to be regulated by Ca2+ and also participate in the enhancement of aerobic metabolism upon cell stimulation (171). One issue remains to be clarified: why do [Ca2+]m rises reach values up to 500 μM, given the matrix dehydrogenases are activated by [Ca2+] in the low micromolar range (a similar K0.5 of 0.5–2 μM was reported for the three enzymes). Several explanations could be given. The first is that most mitochondrial Ca2+ concentration ([Ca2+]m) estimates were obtained in prolonged, supramaximal stimulations with Ca2+-mobilizing agonists, and it is likely that these conditions do not fully mimic the response to physiological challenges, that give rise to smaller and more transient [Ca2+]c increases. The second is that a large [Ca2+]m response could be [together with other mechanisms, such as the enzymatic delay in the rephosphorylation of the pyruvate dehydrogenase (PDH) complex] a route to extend the metabolic activation well beyond the duration of the [Ca2+]c rise (133, 157, 312).

Mitochondrial Ca2+ overload and/or the combined action of apoptotic agents or pathophysiological conditions (e.g., oxidative stress, Refs. 153, 154) may induce a totally different effect (i.e., a profound alteration of organelle structure and function, Refs. 281, 362) leading to cell death by necrosis and apoptosis. The reduction of ER Ca2+ loading, such as that caused by Bcl-2, reduces mitochondrial Ca2+ uptake and thus the efficacy of apoptotic stimuli (16, 280, 281). Conversely, both Bax and other proapoptotic proteins (albeit with different mechanisms) enhance mitochondrial Ca2+ loading, and thus the sensitivity of cells to apoptosis (252, 253; for recent reviews, see Refs. 255, 308, 328).

2. Effect of mitochondrial Ca2+ homeostasis on cellular Ca2+ signals

Mitochondria, distinctly from cytosolic proteins, are highly sophisticated, “tunable” buffers that vary their activity in different phases and functional states of the cell; indeed, their number, shape, and distribution (147, 158) and most likely their responsiveness to Ca2+ (232, 282) are controlled by converging signaling pathways. This Ca2+ buffering activity influences cytosolic Ca2+ signals in two conceptually different ways, i.e., 1) by acting as high-capacity sinks placed on the way of a propagating Ca2+ wave and 2) by clearing Ca2+ in restricted microdomains (such as the microenvironment of a Ca2+ channel). In the first case, spatial clusters of mitochondria have been demonstrated to isolate functionally distinct domains of polarized cells, namely, a mitochondrial “firewall” was shown to prevent the spread of Ca2+ signals from the apical (secretory) region of pancreatic acinar cell from the basolateral region, containing the nucleus (372). Similarly, neuronal mitochondria have been shown to buffer [Ca2+] increases in defined cellular regions, e.g., the presynaptic motoneuron ending (66). As to the second case, a thoroughly investigated example is the regulation of Ca2+ release through IP3Rs. In Xenopus oocytes, the energization state (and thus the capacity to accumulate Ca2+) was shown to modify the propagating Ca2+ waves induced by IP3 (156). In permeabilized blowfly salivary glands, it was observed that perfusion of IP3 induced ER [Ca2+] oscillations, the frequency of which increased with the dose of IP3. Such an effect was observed only upon energization of mitochondria, implying a primary role of these organelles in regulating the Ca2+ microdomain in the proximity of IP3Rs and thus the oscillatory pace of stimulated cells (405). In mammals, this effect has been seen in many cell systems, including hepatocytes, HeLa cells, astrocytes, and BHK cells. As to the cellular consequence, very different effects were observed given the bell-shaped sensitivity of IP3Rs to Ca2+ concentration on the cytosolic side. In astrocytes and hepatocytes, cytosolic excitability appeared enhanced when mitochondrial Ca2+ uptake was inhibited, indicating that mitochondrial clearance of the Ca2+ microdomain reduced the positive Ca2+ feedback on the IP3R and/or buffered substantial Ca2+ loads (30, 132, 338). Conversely, in BHK cells, inhibition of mitochondrial Ca2+ uptake resulted in reduction of ER Ca2+ release (169), thus indicating that mitochondria prevent the Ca2+-dependent inactivation of the channel.

Recently, a provocative new hypothesis, also related to mitochondrial regulation of Ca2+ microdomains, was put forward on the basis of studies carried out on the postsynaptic ending of the neuromuscular junction. Earlier investigations showed that slower release of Ca2+ from these mitochondria (through the Na+/Ca2+ exchanger) can maintain relatively high cytosolic Ca2+ for a period following intense stimulation, thus participating in the phenomenon of posttetanic potentiation (66, 366). In a later paper, the contribution of mitochondrial Ca2+ release to posttetanic potentiation of neuromuscular junctions was observed also in Ca2+-free medium (395). These results suggest that Ca2+ release from mitochondria through the Na+/Ca2+ exchanger, triggered by plasma membrane depolarization, may induce during the action potential a primary release of Ca2+ from mitochondria, rather than causing the reextrusion of Ca2+ loaded during the action potential. In this case, the mitochondria would act as a bona fide Ca2+ store that is loaded with Ca2+ at rest and releases it upon cell stimulation. Obviously, the general relevance of this observation must be confirmed, given that in the vast majority of cells with a variety of experimental approaches, i.e., measurement of free Ca2+ with aequorin, GFP-based probes or dyes (74, 90, 303), and of total Ca2+ by X-ray microanalysis (285) or electron energy loss imaging (276), mitochondria were demonstrated to contain at rest very little calcium and maintain a matrix [Ca2+] similar to that of the bulk cytosol.


Although by far the best-characterized intracellular compartments endowed with the capacity of accumulating and releasing Ca2+ are the ER/SR and mitochondria, other vesicular compartments of the cells are known to accumulate Ca2+ (the Golgi apparatus, secretory granules, lysosomes, and endocytic vesicles) and to be capable of releasing it under some conditions. To our knowledge nothing is yet known about the Ca2+ handling capacity of peroxisomes, and thus these organelles will not be further discussed.

Unlike the ER and mitochondria that are usually rather homogeneously distributed within the cytoplasm, most of these organelles are often highly localized within cells, and accordingly, their capacity to take up or release Ca2+ in selected regions of the cells may be ideally suited to generate functionally relevant local microdomains of Ca2+. Several lines of evidence support the notion that the Ca2+ uptake and release mechanisms of these organelles may be molecularly distinct from that of the ER/SR and therefore at least potentially the differential activation of the classical Ca2+ stores or of these other stores may be a tool to selectively and locally activate specific cell functions.

The mechanisms of Ca2+ homeostasis by these organelles, however, are far less well understood than that of the ER/SR and of mitochondria, and accordingly, at the moment most conclusions must be based on indirect evidence or mere speculation.

A. Endosomes

Although intensely studied from many points of view, the literature concerning Ca2+ handling by endocytic vesicles is very scarce. A careful analysis of their Ca2+ handling properties has been carried out by Petersen’s group (114) a few years ago, and the results of this study appear clear cut and thus far undisputed. In particular, the authors showed that their Ca2+ content depends on the inclusion, during vesicle formation, of extracellular medium in their lumen, while upon endocytosis Ca2+ is rapidly lost into the cytoplasm, concomitant to (or because of) endosome acidification. The author also concluded that in cells with active endocytic activity, the Ca2+ taken up with endocytosis may account for a substantial part of the Ca2+ influx rate under resting conditions; Pryor et al. (296) concluded that intraorganellar Ca2+ plays a key role in late endosome-lysosome heterotypic fusion.

B. Golgi Apparatus

The Golgi is well known to be highly heterogeneous in terms of morphology, protein composition, and functions (170, 224). Similarly heterogeneous appears its Ca2+ handling mechanisms. Indeed, Ca2+ uptake in the cis- and intermediate Golgi is mediated by the SERCA pumps of the ER/SR while in the trans-Golgi the existence of intracellular Ca2+ pumps other than the classical SERCAs is now well documented. Specifically, much information has recently been obtained on the mammalian homolog of PMR1, the Mn2+/Ca2+-dependent ATPase of Saccharomyces cerevisiae (322). Its mammalian homolog was identified (128) and investigated, providing new interesting insight into the signaling properties of the Golgi apparatus (19, 380) and its dysfunction in a human genetic disease (for a detailed coverage of this topic, see Ref. 379). In brief, this pump comprises a group of related P-type ion-motive ATPases (two independent genes, ATPC1 and ATPC2, have been identified, and ATPC1 was shown to encode four different splice variants). Evolutionary analysis of the PMR-1-related ATPases (expressed in yeast, worms, insects, and mammals and closely resembling some bacterial ATPases) suggests that they represent the most ancient class of Ca2+-ATPases. As to their function, these ATPases markedly differ from the SERCAs in two major aspects: 1) they transport equally well Ca2+ and Mn2+, and 2) they are located in the Golgi stacks (where they coexist with SERCAs), the trans-Golgi network, and the secretory vesicles. Based on the latter property, the mammalian pump is now referred to as SPCA (secretory pathway Ca2+-ATPase). As to the roles in cell physiology, two different functions can be envisaged. The first is to transport Mn2+ across endomembranes, thus accounting for the high [Mn2+] of the Golgi and post-Golgi compartments. This allows one to introduce into the lumen Mn2+ to be utilized by resident enzymes (such as lactate synthase, that requires Mn2+ for activity). Indeed, in lactation, SPCA was shown to be overexpressed. In addition, it may allow one to prevent cellular Mn2+ overload, which has been implicated in the pathogenesis of neurodegenerative disorders, such as Parkinson’s disease (143). The second function, shared with the SERCA in the Golgi apparatus and entirely carried out by SPCA in the later compartments, is that of accumulating Ca2+ into the organelle (see Fig. 7). In this case, much remains to be understood on the specific relevance of this Ca2+ pump, and in general of the post-ER Ca2+ stores. Two nonmutually exclusive roles can be envisioned. The first is that these stores cooperate with the larger ER compartment in setting up the fundamental properties of cellular Ca2+ signaling. In support of this notion, transfected SPCA was shown to sustain baseline [Ca2+]c oscillations in normal conditions and after inhibition of SERCAs by thapsigargin (216). The data indicate that after ER Ca2+ discharge, Ca2+ influx sustains Golgi loading until this store is fully loaded. At this point, no further uptake can occur, and [Ca2]c slowly increases. This sensitizes the resident IP3Rs and causes the release of Ca2+ from the organelle. A second Ca2+ spike is thus produced, that ends with the depletion also of the Golgi store. The reloading of the organelle, and the reinitiation of the cycle, could support repetitive [Ca2+]c oscillations. In this view, the Golgi apparatus may play the role of the poorly IP3-sensitive Ca2+ store, which was proposed to cooperate with the highly IP3-sensitive ER store in triggering the repetitive baseline oscillations that follow the initial large transient [Ca2+]c rise caused by agonist stimulation (the two-pool model put forward by Berridge, Ref. 21). As to later compartments, no significant agonist-dependent Ca2+ release was detected from the portions of the Golgi containing only SPCA (381). Conversely, RyR-dependent Ca2+ release from SPCA-containing secretory granules was observed in pancreatic β-cells, and it was proposed that local CICR from docked vesicles could participate in triggering exocytosis (219). The second role for the Ca2+ pumping activity of SPCA is that of maintaining a steadily high [Ca2+] within the lumen, which is needed for the constitutive activity of a number of resident enzymes (such as those involved in the posttranslational processing and sorting of secreted proteins). This aspect may bridge SPCA function to a human disease. Indeed, the autosomal dominant skin disorder known as Hailey-Hailey disease (characterized by erosions and lesions at sites of trauma) is due to mutations of the gene encoding SPCA1 (148). It may appear surprising that a ubiquitous Ca2+ pump of the Golgi and post-Golgi membranes causes selectively a skin disease, but the peculiar properties of this tissue may provide a reasonable explanation. Skin integrity (and resistance to mechanical stress) requires the tight association of the multilayered epithelium, which is in turn dependent on adhesion structures (the desmosomes). Maturation and membrane exposure of scaffolding proteins, as well as their assembly into the functional units, depends on luminal Ca2+-dependent reactions, and this could account both for the high expression level of SPCA in these cells and for their high vulnerability to genetic lesions of this Ca2+ pump (309).

FIG. 7.

[Ca2+] changes within the ER and Golgi lumens. A and B: immunocytochemical localization of recombinant aequorins targeted to the ER and Golgi lumens in HeLa cells. Please note the typical perinuclear localization of the signal and the absence of visible staining of other structures in B. C and D: kinetic changes of the [Ca2+] concentration measured with the ER and Golgi located aequorins upon addition of ATP (Pozzan and Rizzuto, unpublished data). The ER and Golgi Ca2+ were initially depleted by treatment with ionomycin in EGTA-containing medium. Where indicated, CaCl2 (1 mM) was added to refill the organelles with Ca2+, and finally, ATP was added to cause IP3 generation and Ca2+ release. For experimental details, see Pinton et al. (283).

C. Secretory Granules

Although often lumped together under this name, these organelles are clearly quite different in different cells and thus, not surprisingly, the mechanisms through which they take up Ca2+ from the cytosol are heterogeneous. To date, most direct information is available for insulin granules, and detailed indirect information is also available for other secretory compartments such as zymogen and chromaffin granules and synaptic vesicles. As to insulin granules, after the initial demonstration that they marginally contribute to Ca2+ homeostasis in permeabilized cells (292, 293), detailed studies have been carried out with selectively localized aequorin leading to the conclusion that they accumulate Ca2+ by means of a vanadate-sensitive Ca2+-ATPase, possibly SPCA itself (219). According to these studies, although the lumen of the granules is notoriously acidic, the H+ gradient plays no role in Ca2+ accumulation, in as much as Ca2+ uptake by the granules is unaffected by drugs that collapse such gradients as the H+ pump inhibitor bafilomycin or the H+/Na+ ionophore monensin (219). As to zymogen granules, the mechanism for their Ca2+ accumulation has not been investigated in great detail, but it has been clearly demonstrated that it does not depend on a thapsigargin-sensitive SERCA (115), and a similar conclusion has been reached for the histamine-containing granules of mast cells (40, 248). As to chromaffin granules and synaptic vesicles, the general consensus, based on rather old data, is that they load via a pH-driven H+ (Na+)/Ca2+ antiport (40, 94, 166).

The [Ca2+] in the lumen of the granules has been measured both with selectively targeted Ca2+ indicators (insulin and mast cell granules) and by indirect means (chromaffin granules). The [Ca2+] within insulin granules has been estimated to be ∼100–200 μM (219), while that of mast cell and chromaffin cell granules amounts to ∼10–40 μM (40, 248). It needs to be stressed that the difference between the total Ca2+ content (tens of millimolar) and the free [Ca2+] in the vesicle lumen indicates the existence of a very strong Ca2+ buffer within these organelles.

D. Lysosomes

As to lysosomes, it should be stressed that in addition to Ca2+ storing organelles, they can behave also as classical, Ca2+-sensitive, secretory vesicles. This is true for bona fide lysosomes (see, for example, Ref. 314) or for lysosome-related organelles such as azurophil granules of neutrophils, cytotoxic T-lymphocyte granules, and mast cell granules. The mechanism for their loading with Ca2+ is presently undefined, but clearly it does not depend on SERCAs.

A Ca2+ store of undefined cytological nature has been functionally identified a few years ago in several cell types. These stores do not use SERCA for taking up Ca2+ and load only when the cytosolic Ca2+ level is massively increased and for prolonged periods of time (286).

E. IP3R and RyR Expression

While it is undisputed that Golgi, lysosomes, and secretory vesicles contain in their lumen high amounts of Ca2+, much less clear, and debated, is not only whether indeed they can release their Ca2+ into the cytoplasm under physiological conditions, but also what type of Ca2+ channel is present in their membrane. We consider here the recent literature in the field pointing out the important discrepancies still existing.

First of all, what about the expression of the classical intracellular channels (IP3Rs and RyRs) in these other stores? The situation appears relatively clear for the Golgi complex. The cis and presumably the intermediate Golgi express not only, as mentioned above, the SERCAs, but also IP3Rs (216, 283) (see Fig. 7), while it is yet unclear whether functional RyRs are present on these membranes. Accordingly, these Golgi compartments behave largely as extension of the ER and local release of Ca2+ during activation may be of importance for some specific functions localized therein (see below). The trans-Golgi, on the other hand, apparently only expresses the SPCA pump, and thus far no evidence has yet been obtained indicating that physiological agonists can mobilize Ca2+ from its lumen (217, 380).

Far more complex is the situation of the secretory compartment. Petersen and co-workers (115) provided initial functional evidence that zymogen granules from the exocrine pancreas are indeed sensitive to IP3, but other studies (401) have argued against this conclusion. Along the same line, Nguyen et al. (248) have suggested that mast cell, histamine-containing, granules do express IP3Rs, but not SERCAs (248). Both the experiments of Petersen’s group (115) and those of Nguyen et al. (248) have been carried out with highly sophisticated image analysis of fluorescence images carried out in purified isolated single granules, and thus it is difficult to argue against the conclusion that IP3 can indeed release Ca2+ from these organelles; yet, at least in the pancreatic acinar cells, there is no doubt that the vast majority of the Ca2+ released upon receptor activation comes from ER cisternae intermingled with the zymogen granules (262). However, the possibility that Ca2+ released from zymogen granules plays a key role under pathological conditions (e.g., in pancreatitis) has been recently suggested (299). Finally, thus far no conclusive immunocytochemical evidence for the presence of IP3Rs on secretory vesicles of any type has been provided. The original report (28) showing expression of IP3Rs in insulin-containing granules has been dismissed later as an artifact of antibody cross-reactivity (300). The existence of functional IP3Rs in insulin granules has been more recently clearly excluded also by direct measurement of its luminal Ca2+ content that is unaffected by receptor-induced IP3 generation (219). More convincing, yet not conclusive, is the localization of the IP3Rs on the chromaffin granule membranes (371). The modulation by chromogranin of the IP3Rs opening probability cannot by itself be considered an argument in favor of IP3R localization in the granules given that the latter proteins are present, though at lower concentration, also in the ER.

It should be also stressed, in addition, that at least the IP3R1 isoform contains strong ER retention signals in its transmembrane domains (263), and thus it is difficult to envisage how it could proceed much further in the secretory pathway than the more proximal cisternae of the Golgi where the ER retrieval mechanisms are located. However, as shown above, evidence for the presence of IP3Rs on the plasma membrane has been provided (in some cell types), and accordingly IP3Rs, at least in transit along the secretory pathway, must thus exist.

Better documented is the presence of RyRs, in particular in insulin-secreting granules. By means of both immunocytochemical localization at the EM level, functional studies with selectively targeted Ca2+ indicators and subfractionation experiments, there is solid evidence supporting the idea that Ca2+ can be released from the insulin granules through the RyRs under physiological conditions (218, 219). Insulin-secreting granules appear unique also from other points of view as Ca2+ stores: not only, as mentioned above, do they express RyRs (type 1 and 2), but also SPCA (while most other granule apparently have other type of Ca2+ accumulation mechanisms), and are also sensitive to NAADP (see below).

Finally, as far as lysosomes (or in general acidic lysosome-like organelles) are concerned, there is no positive evidence for the expression of IP3Rs or RyRs, at least in mammalian cells, while in some lower eukaryotes (e.g., Plasmodia) indirect evidence has been provided indicating that a Ca2+-containing, acidic compartment, is sensitive to IP3 (266).

With the exception of insulin granules, the calculated values for the free Ca2+ concentration (10–40 μM) within the lumen of the secretory vesicles is about one order of magnitude lower than in the ER. This suggests that the contribution of the granules to IP3 (or RyR)-induced bulk cytoplasmic Ca2+ rises may be, anyway, relatively small. These Ca2+ release channels are in fact nonselective cation channels, and thus they transport relatively little Ca2+ at these low luminal Ca2+ concentrations.

F. Channels Other Than the IP3Rs and the RyRs and Their Subcellular Localization

IP3Rs and RyRs have been cloned, purified, reconstituted in lipid bilayers, recombinantly expressed, and extensively characterized for their subcellular distribution and functional characteristics. The situation with other intracellular Ca2+ release channels is far less defined and still rather debated. Over the last few years, much interest has been devoted to the channels regulated by NAADP. Up to now, however, these channels not only remain elusive from the molecular point of view, but their subcellular distribution is highly debated. In sea urchin egg homogenates, very clear evidence has been provided that the vesicular compartments that release Ca2+ in response to NAADP are distinct from those sensitive to cADPR or to IP3. This conclusion has been reinforced by experiments carried out in intact eggs. Most important, the classical SERCA inhibitor thapsigargin empties the stores sensitive to IP3 and cADPR, leaving unaffected the response to NAADP. This observation, coupled to the fact that inhibitors of H+-ATPases such as bafilomycin block the NAADP effect led to the, so far undisputed, conclusion that in the sea urchin egg the NAADP-sensitive compartment is distinct from the ER and its lumen is acidic (also defined as “lysosome-like”) (1, 51, 106, 110, 174177).

The situation is far less clear in mammalian cells, where clear-cut evidence by Gerasimenko and colleagues (112, 113) demonstrates that NAADP not only releases Ca2+ from the same compartment as cADPR and IP3R do, but also that NAADP acts on the RyRs expressed on the ER and nuclear envelope. Direct demonstration that NAADP can activate single RyRs incorporated in lipid bilayer has been provided recently by Hohenegger et al. (144). Mitchell et al. (218), on the other hand, provided very good evidence that NAADP releases Ca2+ from insulin granules in B cells of the pancreas, whose lumen is indeed acidic, but clearly not a “lysosome-like” organelle. It remains to be established whether this sensitivity to NAADP depends, as suggested by Gerasimenko and co-workers (112, 113), on the sensitivity of the RyRs to this second messenger, or whether this is due to the yet unknown NAADP receptor.

To complicate matters even further, Galione and co-workers (394) very recently showed [using the very same cells as Gerasimenko and co-workers (112, 113) and Mitchell et al. (218)] that NAADP acts on Ca2+ channels distinct from RyRs and IP3Rs, and they suggested that the stores should be identified with lysosome-like organelles. It should be stressed that acidic “lysosome-like” Ca2+ store is an operational definition (based essentially on pharmacological sensitivity to bafilomycin and other drugs targeted to acidic compartments), and thus it is not entirely clear to which cytological entity the reported sensitivity to NAADP should be attributed. In mammalian cells we noticed that the distribution of the NAADP-sensitive pool described by Yamasaki et al. (394) tends to be very similar to that of secretory granules, e.g., in the apical pole in pancreatic acinar cells or diffuse in pancreatic β-cell. It is interesting to note another major experimental discrepancy, i.e., that bafilomycin, an inhibitor of H+-ATPases, blocked NAADP-induced Ca2+ release from the lysosome-like stores according to Reference 394, but it had no effect on Ca2+ accumulation/release by insulin granules according to Reference 218. On the contrary, Ca2+ uptake by insulin granules is blocked by vanadate (219), an inhibitor of P-type ATPases, and is abolished also by silencing of the gene encoding SPCA (220).

A clarification of these clearly contradictory findings is urgently needed. For the time being, we believe that it is fair to conclude that the existence of NAADP-sensitive channels pharmacologically distinguishable (and presumably located in different compartments) from the IP3Rs and RyRs is firmly established in lower eukaryotes, but still highly debatable in mammalian cells.

NAADP-sensitive channels are not the only noncanonical intracellular Ca2+ channels for which experimental evidence has been provided. Considerable interest in fact has been devoted to the understanding of the nature and role of Ca2+ channels regulated by sphingonoids. In particular, evidence has been provided for the existence of intracellular Ca2+ channels regulated by sphingosine 1-phosphate (S1P) and sphingosinphosphoryl choline (SPCI). As to the latter, a putative intracellular receptor/channel, named Scamper, has been cloned, recombinantly expressed in oocytes and characterized electrophysiologically (199). Most recently, however, these data have been reevaluated and, based on overexpression data, presumed membrane topology and functional evidence, it has been concluded that Scamper is not an intracellular Ca2+ channel (327). Last, but not least, Cavalli et al. (47) recently concluded that Scamper may be located on the t tubules of cardiac cells, i.e., in a specialized region of the plasma membrane and not on intracellular stores. Due to these inconsistencies, it seems fair to conclude that for the moment there is no convincing evidence for a SPCI-gated intracellular Ca2+ channel.

Less confused, but still far from clarified, is the situation for S1P, which can act both as a primary messenger acting on a family of G protein-coupled receptors and intracellularly as a second messenger. A first report was published in 1990 (116), and scattered data on this topic have appeared in the most recent literature. Quite recently, however, Nahorski’s group (399) rejuvenated the field by providing new evidence in support of the notion that some agonists (e.g., LPA in particular) may mediate their Ca2+-mobilizing action exclusively through S1P production; in their hands, S1P not only releases Ca2+ from stores independently of RyRs and IP3Rs, but also with a subcellular localization and kinetics quite distinguishable from that mediated by IP3. It must be also mentioned here that S1P has been recently also proposed to play a role as a soluble messenger activating CCE (152).

Finally, there is experimental evidence that is most often neglected when considering the role of acidic compartments (secretory granules, lysosomes, or lysosome-like organelles) in physiological Ca2+ mobilization: this release, whether due to IP3Rs, RyRs, or any other channel, must occur in the absence of extracellular Ca2+ and should be insensitive to Ca2+ ionophores such as ionomycin or A23187. This is because these ionophores, which are obligate 2H+/Ca2+ exchangers, do not release Ca2+ from acidic compartments (82, 83). To our knowledge, with the exception of the insulin granules (219), this simple test has either not been carried out or has led to complete inhibition of Ca2+ release (thus excluding the role of acidic compartments from physiological Ca2+ mobilization).

The existence of such alternative Ca2+ storage compartments and of their channels could be of major relevance for the generation of localized Ca2+ microdomains. For example, it seems possible that, in the case of both insulin-containing secretory vesicles and sea urchin lysosomes, the formation of spatially constrained domains of high Ca2+ concentration in the immediate vicinity of an individual organelle may play a key role either for the fusion of the secretory vesicle with the plasma membrane or for heterotypic fusion between lysosomes and endosomes. Furthermore, the possibility of selective activation of Ca2+ release pathway from organelles with a specific cellular location, e.g., the Golgi and secretory granules in highly polarized cells, may be of central role in local activation of other signaling pathway distinct from those activated by the more classical IP3R- or RyR-dependent release.


In the past, the existence and functional significance of Ca2+ microdomains (i.e., localized regions with a [Ca2+] significantly different from that of the bulk cytoplasm) were sometimes seen as working hypotheses, evoked to explain otherwise mysterious experimental results. This is no longer the case, as Ca2+ microdomains are experimentally verified facts with known biological consequences, and interwoven with older notions of Ca2+ homeostasis (an electron micrograph summarizing the general conclusions about the microheterogeneity of intracellular Ca2+ is presented in Fig. 8). Much remains to be understood, however. In future work, the availability of Ca2+ probes with high spatial selectivity will be fundamental. Table 1 shows a list of available recombinant probes that were highly valuable in deciphering Ca2+ homeostasis in intracellular organelles. It is easy to predict that this list will rapidly grow with new probes for monitoring the microenvironment of Ca2+ channels and/or transducers. Exciting roads lie ahead, not only regarding the mechanisms behind this heterogeneity, but, most importantly, regarding the identification of the molecular actors responsible for these localized events, as well as their functional consequences. In this contribution, we attempted to present an updated analysis of what is currently known on this topic, while pointing out the situations where, in our opinion, confusion or contradictions still reign. Our own experience in the field may have caused a subconscious bias in the selection of references, but we strove to present a comprehensive view. Mistakes and omissions regarding important contributions are practically unavoidable when dealing with such a large body of literature, and for this we must accept responsibility and present our apologies.

FIG. 8.

Heterogeneity of the [Ca2+] within a typical mammalian cell. Electron micrograph of an epithelial cell. For each compartment, an approximate [Ca2+] is given, as it results from different experimental approaches. (Electron micrograph kindly provided by Dr. Carlo Tacchetti, Genova, Italy.)

View this table:

Recombinant genetically encoded Ca2+ indicators


The original work by the authors has been supported over the years by grants from the Italian Ministry of University (Cofin and Firb), the Italian Association for Cancer Research (AIRC), Italian Telethon, European Union, Human Science Frontier Program, the Italian Consiglio Nazionale delle Ricerca, and the Harvard Armenise Foundation.


We are indebted to P. Magalhães for critically reading the manuscript and for many suggestions.