Calcium (Ca2+) is a ubiquitous intracellular messenger, controlling a diverse range of cellular processes, such as gene transcription, muscle contraction and cell proliferation. The ability of a simple ion such as Ca2+ to play a pivotal role in cell biology results from the facility that cells have to shape Ca2+ signals in space, time and amplitude. To generate and interpret the variety of observed Ca2+ signals, different cell types employ components selected from a Ca2+ signalling ‘toolkit’, which comprises an array of homeostatic and sensory mechanisms. By mixing and matching components from the toolkit, cells can obtain Ca2+ signals that suit their physiology. Recent studies have demonstrated the importance of local Ca2+ signals in defining the specificity of the interaction of Ca2+ with its targets. Furthermore, local Ca2+ signals are the triggers and building blocks for larger global signals that propagate throughout cells.
Over the past decade, there has been an increasing awareness of local interactions between Ca2+ and its target molecules. The term ‘local’ has been variously used to describe microdomains of Ca2+ (dimensions ∼10-100 nm) at the mouth of channels or larger subcellular Ca2+ signals spanning several micrometers. All of these signals are local in the sense that they are spatially restricted rises in Ca2+ concentration within a cell. However, the sources and targets of the Ca2+ signals are usually very different.
Most cells utilise Ca2+ influx from the extracellular space and Ca2+ release from intracellular stores to generate intracellular signals. When activated, both Ca2+ entry and Ca2+ release channels can give rise to brief pulses of Ca2+ that form a small plume around the mouth of the channel before diffusing into the cytoplasm (reviewed by Neher, 1998). Such signals can remain localised and activate effectors within the immediate vicinity of the channels, they can recruit effectors (for example, see Maasch et al., 2000) or they can be summated to yield global increases that propagate throughout and between cells. Local Ca2+ signals therefore serve to activate specific targets before diffusion or regenerative mechanisms spread the Ca2+ across a cell.
A diverse array of local Ca2+ signals have been visualised in electrically non-excitable cells (Bootman, 1996). In addition, spatially restricted Ca2+ signals below the resolution of visible imaging techniques have been inferred from the activity of Ca2+-dependent processes. One type of local Ca2+ signal that appears to operate in many, if not all, electrically non-excitable cells is ‘Ca2+ puffs’ (Fig. 1). These elementary events have amplitudes typically ranging from ∼50-600 nM, a spatial spread of ∼6 μm and a total duration of ∼1 second. These Ca2+ puff dimensions indicate that there are <100 such sites in small somatic cells. Such events were first observed in Xenopus oocytes (for example, see Yao et al, 1995) but have subsequently been observed in HeLa cells (Bootman et al., 1997a), PC12 cells (Reber and Schindelholz, 1996; Koizumi et al., 1999) and endothelial cells (Hüser and Blatter, 1997). The non-stereotypic nature of Ca2+ puffs indicates that they probably arise from sites containing variable numbers of inositol 1,4,5-trisphosphate receptors (Ins(1,4,5)P3Rs; Sun et al., 1998; Thomas et al., 1998). The temporally and spatially coordinated recruitment of Ca2+ puffs is responsible for the generation of repetitive Ca2+ waves and oscillations observed during hormonal stimulation (Bootman et al., 1997b; Marchant et al., 1999; reviewed by Berridge, 1997). Although different non-excitable cell types give rise to distinct global Ca2+ spikes, similar Ca2+ puffs can be recorded from various cell lines expressing different combinations of the three Ins(1,4,5)P3R isoforms (Tovey et al., 2000). This suggests that cell-specific recruitment of a generic elementary signal underlies different global signals.
In HeLa cells (Thomas et al., 2000) and Xenopus oocytes (Marchant and Parker, 2001), Ca2+ puff sites that have a higher sensitivity to Ins(1,4,5)P3 consistently trigger Ca2+ waves (Fig. 1). What gives these pacemaking Ca2+ puff sites their enhanced sensitivity is unclear. In the case of somatic cells, the pacemaker sites tend to be distributed in a perinuclear region (Lipp et al., 1997), which raises the possibility that they can send Ca2+ signals specifically into the nucleus.
Apical Ca2+ spikes in pancreatic acinar cells
Another well-known local Ca2+ signal occurs in the apical region of secretory cells such as pancreatic acinar cells(reviewed by Petersen et al., 1999). In common with the pacemaker sites described above, the Ins(1,4,5)P3Rs that underlie the apical Ca2+ spikes are distinguished by a heightened sensitivity to Ins(1,4,5)P3 (Fogarty et al.,2000). Like Ca2+ puffs, such apical Ca2+ spikes probably arise from the coordinated release of Ca2+ from multiple Ca2+-release channels (Kidd et al., 1999). However, their slow rise time (>1 second; Kidd et al., 1999) and large spatial spread (∼10 μm; Kidd et al., 1999) distinguishes them from the Ca2+ puffs in HeLa cells and Xenopus oocytes. Recent evidence indicates that the apical spikes arise from a stimulus-dependent hierarchical activation of different types of Ca2+-release channel (Cancela et al., 1999;Cancela et al.,2000).
When cells are stimulated to low levels, the Ca2+ spikes stay restricted to the apical pole of the acinar cells, where they can activate ion channels and trigger limited secretion. Greater stimulation causes the Ca2+ spikes to trigger Ca2+ waves that propagate towards the basal pole. The restriction of the Ca2+ signal to the apical pole appears to be due in part to a ‘firewall’ of mitochondria that buffer Ca2+ as it diffuses from the apical pole and prevent the activation of ryanodine receptors (RyRs)in the basal pole (Tinel et al,. 2000; Straub et al., 2000).
Local signalling via Ca2+ entry
In addition to local Ca2+ signals arising through release from internal stores, Ca2+ entry can give rise to local signals with specific functions in electrically non-excitable cells. Cooper and colleagues have shown that both Ca2+-activated and Ca2+-inhibited isozymes of adenylyl cyclase are more sensitive to influx through store-operated Ca2+ channels (SOCC) than to Ca2+ release (Fagan et al., 2000a; Fagan et al., 2000b). Ca2+ influx through voltage-operated Ca2+ channels (VOCCs; Fagan et al., 2000b) or a novel arachidonic acid-activated Ca2+ channel (Shuttleworth and Thompson, 1999) does not modulate adenylyl cyclase activity to the same degree as SOCC entry. These data point to a close localisation of SOCC channels and adenylyl cyclases. One way in which this coupling might be achieved is through the co-localisation of SOCCs and target enzymes in specific cellular structures such as caveolae. These ‘flask-shaped’ invaginations of the plasma membrane have been proposed to contain Ca2+-entry channels and Ins(1,4,5)P3Rs (reviewed by Isshiki and Anderson, 1999). In addition to regulating adenylyl cyclase, localisation of Ca2+ entry to caveolae could underlie the specific requirement for Ca2+ entry for activation of endothelial nitric oxide synthase (Paltauf-Doburzynska et al., 1998; Lin et al., 2000).
Mitochondria sense local Ca2+ signals
Sequestration of Ca2+ signals by mitochondria (by both electrically excitable and non-excitable cells) serves at least two essential purposes: buffering of cytosolic Ca2+ loads and consequent activation of citric acid enzymes (for example, see Robb-Gaspers et al., 1998; reviewed by Duchen, 1999). Mitochondria sequester Ca2+ through a low-affinity, high-speed uniporter powered by the mitochondrial membrane potential. The in vitro affinity of the uniporter for Ca2+ is ∼10 μM, which is lower than the global Ca2+ concentrations observed in vivo during physiological responses. However, mitochondria have been shown to sense and modulate both Ca2+ entry and Ca2+ release (for example, see Hoth et al., 1997; Sheppard et al., 1997; Landolfi et al., 1998; Hajnoczky et al., 1999; Collins et al., 2000a). To explain the paradox that mitochondria accumulate Ca2+ during physiological responses despite the insensitivity of the Ca2+ uniporter, it has been suggested that mitochondria are located in close proximity to sites of either Ca2+ release or Ca2+ entry (for example, see Rizzuto et al., 1998; Csordás et al., 1999). In these locations, mitochondria can respond to Ca2+ changes within the microdomains around the mouth of Ca2+ channels.
It is therefore generally thought that microdomains of cytosolic Ca2+ may be the most efficient way of increasing mitochondrial matrix Ca2+ concentration. However, several studies have shown that modest cytosolic Ca2+ increases that do not form microdomains can also substantially increase mitochondrial Ca2+ levels (for example, see Collins et al., 2000b; Colegrove et al., 2000; Collins et al., 2001). Furthermore, mitochondrial Ca2+ uptake shows a steep dependence on the cytosolic Ca2+ concentration and might therefore facilitate its own sequestration (for example, see Colegrove et al., 2000; Collins et al., 2001; reviewed by Gunter et al., 1994).
Cardiac and skeletal muscle
In heart and skeletal muscle, release of Ca2+ from the sarcoplasmic reticulum (SR) by RyRs is the key event linking membrane depolarisation and mechanical activity during excitation-contraction coupling. RyRs occur in clusters that give rise to localised Ca2+-release events denoted ‘Ca2+ sparks’. These events are analogous to the Ca2+ puffs described above, although they are usually faster in onset and decline, and have a more restricted spread (∼1-3 μm). Spatio-temporal recruitment of Ca2+ sparks underlies the global Ca2+ signals that subsequently activate myocyte contraction (Fig. 2; reviewed by Cannell and Soeller, 1998; Niggli, 1999). The failure to recruit Ca2+ sparks appropriately can lead to defective excitation-contraction coupling in cardiac cells (Gomez et al., 1997).
In addition to forming the global Ca2+ transient underlying contraction, Ca2+ sparks can also cause depolarisation of cardiac cells and thereby enhance or corrupt the rhythm of the heart. An example of this is in primary pacemaker cells, which maintain the normal heartbeat. The electrical activity of cardiac pacemaking cells in the sino-atrial node depends on the interplay of several different sarcolemmal ion channels. In addition, Ca2+ release from the SR can also affect the frequency of generating action potentials (Rigg and Terrar, 1996). Hüser and colleagues observed that, following recovery from a previous action potential, there is a gradual increase in cytosolic Ca2+ concentration until the next depolarisation (Hüser et al., 2000). Imaging this slow Ca2+ rise revealed that it is due to the summation of infrequent subsarcolemmal Ca2+ sparks. Essentially, low-voltage-activated T-type Ca2+ channels provide a trigger Ca2+-influx current that evokes Ca2+ spark activity. In turn, the progressive increase in cytoplasmic Ca2+ caused by the Ca2+ sparks promotes electrogenic forward-mode Na+/Ca2+ exchange, and the resultant inward current drives the cell towards the threshold for depolarisation (Fig. 3).
A similar situation occurs in atrial myocytes stimulated with hormones that activate phospholipase C (PLC). Incubation of electrically-paced atrial myocytes with hormones such as endothelin 1 or phenylephrine causes the appearance of spontaneous subsarcolemmal Ca2+ sparks (Lipp et al., 1999a; Lipp et al., 1999b), which are probably due to the activation of Ins(1,4,5)P3Rs that co-localise with RyRs in these cells (Lipp et al., 2000). If the Ca2+ spark frequency is sufficient, a spontaneous action potential can be evoked.
These two examples of Ca2+ sparks driving myocyte depolarisation are entirely analogous. In both situations, a small trigger Ca2+ current (either from T-type Ca2+ channels or Ins(1,4,5)P3Rs) activates Ca2+ sparks that lead to enhancement of Na+/Ca2+ exchange (Fig. 3). Because of the strategic firing of subsarcolemmal Ca2+ spark sites, only a few such events may be necessary to drive a cell to the threshold for depolarisation (Hüser et al., 2000). The ability of a few Ca2+ sparks to enhance cardiac automaticity – that is, to increase the frequency of spontaneous action potentials – has potentially serious implications for the generation of cardiac arrhythmias and sudden heart failure.
The generation of spontaneous action potentials, and therefore a global Ca2+ response, by the Ca2+ sparks is analogous to the situation whereby Ca2+ puffs trigger regenerative Ca2+ waves in non-excitable cells (see above). Although the mechanisms involved are distinct, in both cases the infrequent firing of a limited number of elementary Ca2+-release sites leads to larger regenerative Ca2+ signals.
Smooth muscle – STOCs, STICs and STOICs
Another cell type in which local Ca2+ signals have a clear role is smooth muscle. Paradoxically, Ca2+ can activate smooth muscle contraction, but it can also inhibit contraction by causing hyperpolarisation of the sarcolemma (reviewed by Jaggar et al., 2000). The different effects of Ca2+ in smooth muscle are essentially determined by the spatial properties of the Ca2+ signal; global responses induce vasocontraction by activating Ca2+/calmodulin-dependent enzymes, whereas subsarcolemmal Ca2+ sparks promote vasorelaxation by activating Ca2+-dependent plasma membrane ion channels (Nelson et al., 1995).
It is well established that the Ca2+ sparks in smooth muscle can activate K+ and C- conductances, and thus give rise to brief currents known as STOCs (Spontaneous Transient Outward Currents; K+ current), STICs (Spontaneous Transient Inward Current; C- current) and STOICs (mixed K+ and C- currents). STOCs have been measured in a wide variety of smooth muscle cell types and serve to hyperpolarise the cell membrane by ∼20 mV. They primarily arise through activation of large-conductance Ca2+-activated K+ channels (BK channels). These Ca2+-activated channels have a low sensitivity to cytosolic Ca2+, requiring concentrations of >1 μM for significant activity (although their actual sensitivity also depends on membrane potential). It has been proposed that the BK channels are closely apposed to Ca2+ spark sites and sense rapid step-like Ca2+ changes during RyR activation (Fig. 4; ZhuGe et al., 2000). Walsh and co-workers have estimated that STOCs reflect the activation of a cluster of ∼15 BK channels on the sarcolemma, although the actual BK-channel:RyR stoichiometry may be variable (ZhuGe et al., 2000).
The global Ca2+ signals that activate smooth muscle cell contraction are largely due to activation of L-type VOCCs. Unlike cardiac muscle, in which Ca2+ influx also occurs through L-type VOCCs, there is little amplification of the entry signal by Ca2+-induced Ca2+ release (CICR) in most smooth muscle cell types (Jaggar et al., 2000), which means that few subplasmalemmal Ca2+ spark sites are recruited. Local and global Ca2+ signalling in smooth muscle cells is therefore segregated in terms of both source and function.
The dramatic effect of a few Ca2+ sparks on the automaticity of atrial cardiomyocytes (see above) is mimicked in smooth muscle cells, in which the membrane potential can be regulated by STOCs occurring at low frequencies (∼1 Hz; Nelson et al., 1995). Pharmacological inhibition of either BK channels or RyRs leads to cellular depolarisation and vasoconstriction. The modulation of vascular tone by vasoactive messengers might be largely due to changes in Ca2+ spark frequency (Jaggar et al., 2000). In some smooth muscle cell types, only one or a few ‘frequent-discharge sites’ are responsible for generating the Ca2+ sparks that activate STOCs (Gordienko et al., 2001). In addition to showing repetitive Ca2+ sparks, these sites also activate Ca2+ waves and may be analogous to the pacemaking Ca2+ puff sites in non-excitable cells (see above).
The BK channels are closely apposed to Ca2+ sparks sites to ensure that they can sense the narrow plume of Ca2+ around the mouth of the RyRs. In addition, the Ca2+-activated K+ channels express subunits that ‘tune’ their Ca2+ sensitivity and enable them to respond to Ca2+ sparks. BK channels comprise channel-forming α and auxilliary β subunits. Murine smooth muscle cells lacking the β subunit exhibit reduced sensitivity of BK channels such that they barely respond to the on-going Ca2+ sparks (Brenner et al., 2000; Pluger et al., 2000). In addition, the β-subunit-deficient animals display an elevated mean arterial blood pressure and cardiac hypertrophy.
Ca2+-activated C− channels are not as widely expressed in smooth muscle types, compared with BK channels. The STICs produced by the Ca2+-activated C− channels have an essentially opposite effect to that of STOCs in that they serve to depolarise smooth muscle cells (Fig. 4; Jaggar et al., 2000). The STIC channels are approximately one order of magnitude more sensitive to cytosolic Ca2+ than are STOC channels. This means that their activity more closely follows the decay of the Ca2+ spark than a STOC. In cells expressing both Ca2+-activated K+ and C− channels, mixed events denoted ‘STOICs’ have been recorded (ZhuGe et al., 1998). The ambient membrane potential reciprocally controls the flux through the Ca2+-activated K+ and C− channels. This suggests that such STOICs might function to stabilise membrane potential (ZhuGe et al., 1998).
Most smooth muscle cells express both RyRs and Ins(1,4,5)P3Rs. The relative contributions of these two channels to Ca2+ release is not entirely clear. In ureteric smooth muscle, the elementary events are ‘Ca2+ puffs’ derived solely from Ins(1,4,5)P3Rs (Boittin et al., 2000). In contrast, the Ins(1,4,5)P3Rs in pulmonary artery cells do not appear to form clusters that can produce significant elementary events (Janiak et al., 2001). Instead, Ins(1,4,5)P3Rs appear to reside within a cluster of RyRs, so that activation of Ins(1,4,5)P3Rs gives rise to an event whose characteristics are more similar to those of Ca2+ sparks (Boittin et al., 1998). A similar crosstalk between Ins(1,4,5)P3Rs and RyRs to give ‘mixed’ elementary events has also been described in neuronal (Koizumi et al., 1999) and cardiac (Lipp et al., 2000) cells.
The Ca2+ puffs observed during purinergic stimulation of colonic myocytes stimulate STOCs, which suggests that Ins(1,4,5)P3Rs and RyRs are functionally equivalent in generating local subplasmalemmal Ca2+ signals that regulate smooth muscle tone (Bayguinov et al., 2000). However, in contrast to purinergic stimulation of colonic myocytes, purinergic stimulation of cerebral artery smooth muscle cells causes vasoconstriction. In both colonic myocytes and cerebral artery smooth muscle cells, the purinergic agonists are coupled via P2Y receptors to PLC. The difference between the responses appears to be the relative actions of Ins(1,4,5)P3 and protein kinase C (PKC), the downstream effectors of PLC. In the colonic myocytes, Ins(1,4,5)P3 dominates and increases the frequency of Ca2+ puffs and associated STOCs, thus causing relaxation (Bayguinov et al., 2000). In cerebral artery smooth muscle cells, PKC-dependent desensitisation of RyRs causes a drop in the frequencies of Ca2+ sparks and STOCs, leading to contraction (Jaggar and Nelson, 2000). The effect of PLC on smooth muscle activity is therefore cell type dependent. Even more confusing are the observations that within the same smooth muscle cell type different PLC-linked agonists can cause contraction or relaxation because of their differential regulation of STOCs (Bayguinov et al., 2001).
Local Ca2+ signalling in dendritic spines
Subcellular Ca2+ signalling is the forte of neuronal cells. Because of their often intricate morphologies and the selective distribution of Ca2+ channels, neurons produce a dazzling array of local Ca2+ signals. Synaptic input can produce a hierarchy of Ca2+ signals. These can range from single spines or portions of the dendritic tree, up to responses that invade the soma (reviewed by Denk et al., 1996).
Ca2+ signals within spines are akin to elementary events in other cell types in that they are spatially constrained, short-lived responses that can trigger further propagation of a Ca2+ wave, depending on the underlying excitability of the cell (reviewed by Berridge, 1998). Spine morphology and number can be dramatically affected by Ca2+ signals that occur within their tiny volumes (reviewed by Segal, 2001).
Ca2+ signalling within spines underlies changes in synaptic plasticity such as long-term potentiation (LTP) and long-term depression (LTD). Paradoxically, Ca2+ is implicated in both stimulating and depressing the transmission of nervous signals. It seems that subtle changes in the amplitude, spatial or temporal arrangement of Ca2+ signals account for the different actions of Ca2+ in synaptic plasticity (reviewed by Yuste et al., 2000). In particular, the need for coincident signals can dictate the resulting change in synaptic coupling. LTD in cerebellar Purkinje fibres, for example, requires Ca2+ release from Ins(1,4,5)P3R within the spines in conjunction with a large Ca2+ signal emanating from the activation of VOCCs. Knockout mice that lack type 1 Ins(1,4,5)P3Rs in their Purkinje cells do not display LTD in response to a standard stimulation protocol (Inoue et al., 1998). Furthermore, Purkinje neurons lacking the motor protein myosin Va, which express Ins(1,4,5)P3Rs in dendritic shafts but not in spines, also fail to show LTD (Miyata et al., 2000). Although the Ins(1,4,5)P3Rs in the dendrites are <1 μm distant from those in the spines, they are not functionally equivalent. A complex interplay between local Ca2+ signals appears to underlie the polarity and spread of synaptic modification in CA1 hippocampal neurons (Nishiyama et al., 2000). A relay of Ca2+ between Ca2+-influx channels, RyRs and Ins(1,4,5)P3Rs within the same microdomain determines whether LTD or LTP is observed following synaptic stimulation (but see Fujii et al., 2000).
The production of Ins(1,4,5)P3 in spines is due to activation of metabotropic glutamate receptors, which are generally located on the periphery of these structures (reviewed by Blackstone and Sheng, 1999). The glutamate receptors are physically linked with Ins(1,4,5)P3Rs through members of the Homer protein family (Tu et al., 1998). By virtue of their coiled-coil domains, these proteins can multimerise and link other proteins bearing a proline-rich motif (found in Ins(1,4,5)P3Rs, RyRs and other signalling proteins). One member of the Homer family, Homer 1a, does not posses a coiled-coil domain, and can disrupt the coupling between glutamate receptors and Ins(1,4,5)P3Rs, which results in reduced Ca2+-release signals (Tu et al., 1998). Interestingly, Homer 1a is an immediate early gene that is rapidly transcribed after synaptic activity. The induction of this protein therefore provides an activity-dependent mechanism for regulating local Ca2+ release.
Local Ca2+ signals underlying repolarisation and hyperpolarisation of neurons
As in smooth muscle cells, neurons express Ca2+-activated K+ channels that repolarise or hyperpolarise the plasma membrane following an action potential. Events analogous to STOCs have been recorded in neuronal cell types. These spontaneous miniature outward currents (SMOCs) are probably also generated by Ca2+ sparks firing in close proximity to the plasma membrane (Berridge, 1998). Elementary Ca2+ signals analogous to Ca2+ sparks and Ca2+ puffs have also been observed in the neurites of PC12 cells and hippocampal neurons (Koizumi et al., 1999)
A significant function of the Ca2+-activated K+ channels in neuronal cells is to promote repolarisation of the cell after an action potential and to regulate the frequency of firing action potentials. Recent studies have illustrated complex relationships between VOCCs, Ca2+-release channels and Ca2+-activated K+ channels in different types of neurons (for a review on the subcellular distribution of VOCCs in neurons, see Catterall, 1998). In the soma of hippocampal pyramidal neurons, N-type VOCCs give a coincident activation of BK channels (Marrion and Tavalin, 1998). The communication between N-type VOCCs and BK channels is not inhibited by the rapid Ca2+ buffer BAPTA, which indicates that these two types of channel are probably within 30 nm of each other. In contrast, L-type VOCCs are associated with small-conductance Ca2+-activated K+ channels (SK channels) and are estimated to be ∼50-150 nm apart (Marrion and Tavalin, 1998). The close coupling of BK channels with N-type VOCCs allows them to be rapidly activated and cause cellular repolarisation. The spatial linkage between SK channels and L-type VOCCs is more intruiging. The SK channels are responsible for the ‘slow after hyperpolarisation’ (sAHP) – a period in which the cell remains hyperpolarised and action potential firing is inhibited. Although the close proximity of L-type VOCCs and SK channels allows communication between them, the AHP develops relatively slowly, and usually when the cell membrane has a negative potential (Berridge, 1998). In this case, the L-type Ca2+ channels must change their behaviour to be activated at negative membrane potentials, or the SK channels need to rely on local Ca2+ release from intracellular stores to generate the sAHP.
The involvement of intracellular Ca2+ release in the generation of AHPs seems clear (Berridge, 1998), but there can be an intricate interplay in the microdomains connecting the Ca2+ channels and K+ channels that modulate the membrane potential. In bullfrog sympathetic neurons, N-type VOCCs open during a single action potential and lead to the subsequent activation of BK channels that are responsible for repolarisation (Akita and Kuba, 2000). However, it appears that the Ca2+ signal from the VOCCs to the BK channels is relayed via RyRs, since inhibitors of CICR depress the rate of repolarisation and the extent of the subsequent AHP (Fig. 5). The situation is more complex when multiple action potentials are evoked, in that the AHP becomes less dependent on BK channels, and instead SK channels are recruited following RyR activation (Akita and Kuba, 2000). Akita and Kuba estimated that BK channels, RyRs and N-type VOCCs are within a ∼100 nm domain and that the SK channels are outside this region. The switch from BK to SK channels following multiple action potentials probably reflects the progressive diffusion of Ca2+ from the N-type VOCCs and RyRs beyond the 100 nm domain (for a detailed discussion, see Marchant and Parker, 2000). The SK channels can respond to a more diffuse Ca2+ signal, since they have a high sensitivity to Ca2+, compared with BK channels (Hirschberg et al., 1999).
The regulation of Ca2+-sensitive ion channels by local Ca2+-release signals in neurons and smooth muscle depends on the close apposition of Ca2+-release channels to their targets on the plasma membrane. In these cell types, projections of the ER/SR bring the Ca2+-release channels within 10-20 nm of the cell boundary (Berridge, 1998; Jaggar et al., 2000). These structures are akin to the diadic junctions of cardiac muscle, in which RyRs are brought to within ∼15 nm of sarcolemma (Cannell and Soeller, 1998). However, a significant difference between cardiac and neuronal/smooth-muscle cells is the degree of regenerativity of Ca2+ signals following activation of VOCCs. In cardiac myocytes, a large fraction of global Ca2+ signals is due to Ca2+ release (i.e. recruitment of Ca2+ sparks). By contrast, in neuronal and smooth muscle cells, the degree of regenerative Ca2+ release is variable, and when it does occur it can lag behind the Ca2+ influx by hundreds of milliseconds (Berridge, 1998; Jaggar et al., 2000). These data suggest that the coupling between L-type VOCCs and Ca2+-release channels in cardiac cells is much tighter than in neuronal and smooth muscle cells. Alternatively, loose coupling might introduce plasticity into the response. In both neuronal and smooth muscle cells, the Ca2+ stores can retain a ‘memory’ of previous stimulation, and Ca2+ release can be modulated by a variety of different factors (for example, see Andreasen and Lambert, 1995; Berridge, 1998; Jaggar et al., 2000).
The effect of T-type VOCCs on excitation-contraction coupling in guinea pig cardiac myocytes functionally resembles the loose coupling observed between VOCCs and Ca2+-release channels in neurons and smooth muscle. For the same Ca2+ current, T-type VOCCs activate much less CICR than does L-type VOCCs (Sipido et al., 1998), which suggests that they are not localised within the diadic junctions. However, blockade of the T-type VOCCs does cause a progressive decrease of the amplitude of action-potential-induced Ca2+ transients (Sipido et al., 1998). It is plausible that these channels are expressed outside the diadic junctions and are not needed for excitation-contraction coupling but function to refill depleted Ca2+ stores.
Local Ca2+ signalling and neurotransmitter release
Local Ca2+ signals control the release of neurotransmitter in the active zones of pre-synaptic nerve terminals and neuroendocrine cells (Neher, 1998). In rat cortical synaptosomes, release of small synaptic vesicles (SSVs) containing amino acids (glutamate and GABA) and large dense-core vesicles (LDCVs) containing neuropeptide (cholecystokinin) can be activated by N-, P- and Q-type VOCCs. However, these channels are not equivalent in their efficacy of releasing SSVs and LDCVs: P-type VOCCs preferentially release SSVs, whereas LDCVs are more sensitive to Ca2+ influx through Q-type channels (Leenders et al., 1999). Such selectivity between VOCCs and exocytotic vesicles is reminiscent of the linkage between VOCCs and BK/SK channels described earlier. The SSVs are released from the active zones of pre-synaptic nerve terminals, where they can be pre-docked and ready for fusion. In these locations, they are ideally placed to sense the high-amplitude transient Ca2+ microdomains upon activation of VOCCs (for example, see DiGregorio et al., 1999). LDCVs, in contrast, are more distant from the active zone, and their exocytosis requires prolonged depolarisation (Leenders et al., 1999). The significance of the relative distances of these vesicle types from the active zone is that, for the same Ca2+ signal, there will be a different secretory output. A single action potential, for example, will release SSVs but have little effect on LDCVs.
Ca2+ release from intracellular stores can also trigger exocytosis of neurotransmitters from pre-synaptic junctions (Emptage et al., 2001). Release of Ca2+ from stores in presynaptic boutons contributes to the enhancement of neurotransmitter release with repetitive stimulation (Emptage et al., 2001). In addition, spontaneous Ca2+-release events, akin to Ca2+ sparks and puffs, were recorded in synaptic boutons and appear to be responsible for unevoked synaptic signals. The continual release of neurotransmitter by such spontaneous events could cause long-term changes in strength of synaptic communication (Emptage et al., 2001).
Local Ca2+ signalling controls neuronal growth
Many neuronal cells display spontaneous Ca2+ signals necessary for growth, migration and differentiation (for example, see Komuro and Rakic, 1996; Owens and Kriegstein, 1998; Gomez and Spitzer, 2000; reviewed by Spitzer et al., 2000). In particular, it is well established that Ca2+ regulates the motility and directionality of neuronal growth cones. However, recent studies have shown that it is not always easy to predict the consequence of a rise in Ca2+ levels on growth cone behaviour. Large rises can cause growth cone collapse, whereas more modest signals can either slow or promote neurite outgrowth (Spitzer et al., 2000). Using localised release of caged Ca2+, Zheng showed that growth cones turn into or away from a focal Ca2+ source (Zheng, 2000). Local gradients of Ca2+ might also underlie the turning responses of growth cones to netrin 1, a molecule involved in axonal pathfinding in vivo (Hong et al., 2000). Interestingly, such local gradients must be set up by both Ca2+ influx and Ca2+ release, since abolition of turning responses occurs only if both of these Ca2+ sources are removed. Furthermore, whether a Ca2+ increase evoked attraction or repulsion depended on the basal level of Ca2+ (Zheng, 2000) and the activity of protein kinase A (Hong et al., 2000). The ability of such local Ca2+ signals to activate growth cone turning may be due to restricted rearrangement of the cytoskeleton on one side of a growth cone, perhaps by activation of the Ca2+-dependent phosphatase calcineurin (Lautermilch and Spitzer, 2000).
Just how the local Ca2+ signals are generated by growth cones is unclear. One suggestion is that they are due to activation of VOCCs at resting membrane potential by the action of arachidonic acid (Archer et al., 1999). Exogenous application of arachidonic acid can stimulate axon outgrowth that is blocked by L-type VOCC channel antagonists. Interestingly, in dorsal root ganglia, the Ca2+ signal resulting from arachidonic-acid-mediated VOCC activation cannot be visualised by imaging, but is blocked by BAPTA, which suggests that it is localised to a diameter of tens of nanometres (Archer et al., 1999).
The finger-like filopodia that project from growth cones can also relay local Ca2+ signals to regulate the direction of neurite outgrowth. Spitzer and colleagues have shown that the filopodia of Xenopus spinal neurons can discriminate between different types of substrate by generating local Ca2+ signals of varying frequency that are relayed back to the growth cone (Gomez et al., 2001). The neurons turn away from substrates that give higher frequencies of Ca2+ transients. The impression is that the filopodia act as scouts that sense the environment in front of the growth cone and direct the axon to preferential substrates.
Ca2+ effectors mediate local and global Ca2+ signals
The plasticity of Ca2+ as an intracellular messenger is almost matched by the versatility of some of its effectors. The ubiquitous Ca2+-binding protein calmodulin (CaM), for example, mediates many of the effects of Ca2+, and it can similarly work over different temporal and spatial scales (for example, see Török et al., 1998; reviewed by Chin and Means, 2000; Toutenhoofd and Strehler, 2000). Although a large portion of intracellular CaM is tethered, it can redistribute around cells in a Ca2+-dependent manner. In particular, CaM has been shown to move by facilitated diffusion into the nucleus following activation of post-synaptic Ca2+ channels (Deisseroth et al., 1998) or global Ca2+ waves (Craske et al., 1999). Furthermore, the extent of CaM translocation is determined by the nature of the cellular Ca2+ signal (Teruel et al., 2000). The example of nuclear CaM translocation in neurons demonstrates how local Ca2+ signals can have more distant longer-term consequences.
CaM has two pairs of ‘EF hand’ motifs that can bind one Ca2+ ion each. The binding of Ca2+ to the different EF hands may mediate distinct effects. In the case of L- and P/Q-type VOCCs, for example, CaM can mediate both facilitation and inhibition of the Ca2+ current by binding to a single site on the VOCC (Zuhlke et al., 1999; DeMaria et al., 2001). Although this implicates CaM as an essential component in short-term regulation of Ca2+ channels, it seems paradoxical that CaM can have opposing effects at a single binding site on its target. Interestingly, in the case of P/Q-type VOCCs, it appears that binding of Ca2+ to the N-terminal EF hand pair is responsible for inactivation, whereas the C-terminal EF hand pair underlies facilitation (DeMaria et al., 2001). The empirical observation that facilitation precedes inactivation (DeMaria et al., 2001) suggests that the two pairs of EF hands impart their effects in a temporally regulated manner, perhaps by causing different parts of CaM to interact with the Ca2+ channel. Since the different EF hands display distinct affinities for Ca2+, it is plausible they could sense different components of a Ca2+ signal (i.e. local versus global). Therefore, just as with Ca2+ itself, CaM can have distinct functions over distances of nanometres up to whole cells.
In addition to its Ca2+-regulated activities, CaM is known to bind to Ca2+ entry and Ca2+ release channels in its Ca2+-free form (apoCaM) (DeMaria et al., 2001; Rodney et al., 2001). Such binding of apoCaM may be a means of tethering a Ca2+ sensor for rapid detection of signalling events (DeMaria et al., 2001), or apoCaM may have regulatory effects distinct from those of Ca2+-bound CaM (for example, see Cardy and Taylor, 1998).
The many functions of local Ca2+ signals illustrate the versatility of Ca2+ as an intracellular messenger. No doubt, future studies will provide more examples of Ca2+ acting at the microscopic and nanoscopic levels within cells. The panoply of local Ca2+ signals and functions arises through the utilisation of a ‘Ca2+ signalling toolkit’ (Berridge et al., 1998; Berridge et al., 2000), whereby different cells express particular combinations of proteins that determine the nature and output of the Ca2+ signal. In addition, it is becoming apparent that many of the proteins involved in producing and sensing Ca2+ signals are organised into multiprotein complexes, which provide further means of shaping local events. These complexes include receptors that produce Ins(1,4,5)P3 (see Muallem and Wilkie, 1999), intracellular Ca2+-release channels (see Mackrill, 1999) and Ca2+-entry channels (see Blackstone and Sheng, 1999). Furthermore, many effectors of Ca2+ signals are scaffolded to their Ca2+ source (for example, see Isshiki and Anderson, 1999). Although local Ca2+ signals can have immediate effects within the vicinity of the channels from which they originate, the effects of such signals on cardiac pacemaking, growth cone directionality, neuronal membrane potential (see above) and gene transcription (reviewed by Bito et al., 1997) illustrate that brief local Ca2+ signals can have long-term consequences.