Regulation of Inositol 1,4,5-Trisphosphate Receptors

Cellular responses as diverse as fertilization, cell growth and development, secretion, muscle contraction, sensory perception and neuromodulation depend critically on elevations in intracellular Ca²⁺ concentration ([Ca²⁺]_i) (Berridge, 1993). The introduction of sophisticated imaging techniques has revealed astonishing complexities in the temporal and spatial characteristics of these Ca²⁺ elevations in single cells (see review by Cheek and Barry, 1993). Repetitive, transient elevations in intracellular [Ca²⁺] (Ca²⁺ spikes or oscillations) have been recorded in a number of cell types (see review by Thorn, 1993); these can be arranged spatially as repetitive Ca²⁺ waves originating from a specific point within the cytosol (see review by Gillot and Whitaker, 1993), or even as regenerative Ca²⁺ spirals (Lechleiter et al. 1991).

Whilst the organization of these signals has been extensively characterized, a more difficult problem has been to establish their physiological significance. A recurring criticism is that the techniques employed to measure Ca²⁺ signals, such as the incorporation of Fura-2, a mobile Ca²⁺ buffer and fluorescent Ca²⁺-indicator dye, into the cytosol may themselves contribute to the complex spatial organization. A few examples have been reported, however, in which oscillatory Ca²⁺ signals correlate with periodic cellular responses. In monolayers of ciliated airway epithelia, Ca²⁺ waves that propagate across, and between, cells correlate with waves of increased beating-frequency of cilia (Sanderson et al. 1990), a response which may be important for the removal of foreign objects from the airways. In rat gonadotrophs, Ca²⁺ oscillations have been shown to correlate with periodic elevations in membrane capacitance, reflecting exocytotic vesicle fusion (Tse et al. 1993).

Rises in intracellular Ca²⁺ concentration are generated by a combination of the release of stored Ca²⁺ from organelles and the stimulation of Ca²⁺ influx across the plasma membrane, although the relative contributions from these two sources of Ca²⁺ varies between tissues. In skeletal and cardiac muscle, for example, Ca²⁺ is stored in the sarcoplasmic reticulum (SR) from which it is released through ryanodine-sensitive Ca²⁺ channels (ryanodine receptors, RyRs) following depolarization of T-tubules: however, whilst skeletal muscle RyRs are activated directly by voltage-sensing dihydropyridine receptors in the T-tubular membrane, release of Ca²⁺ through cardiac muscle RyRs is triggered by the Ca²⁺ that enters the cells in response to depolarization (reviewed by Sorrentino and Volpe, 1993).

In many cell types, intracellular Ca²⁺ stores are mobilized by the soluble second messenger inositol 1,4,5-trisphosphate [Ins(1,4,5)P₃]. Ins(1,4,5)P₃ is produced at the plasma membrane following receptor activation and mobilizes Ca²⁺ through specific receptors that are structurally similar to the RyRs of skeletal and cardiac muscle (Berridge and Irvine, 1984, 1989); the depletion of the Ca²⁺ stores by Ins(1,4,5)P₃ is thought to provide a signal for Ca²⁺ influx across the plasma membrane (Putney, 1986). Initially, Ins(1,4,5)P₃ was proposed to provide only the initial trigger for the onset of Ca²⁺ responses (Oron et al. 1985; Parker and Miledi, 1986; Berridge and Galione, 1988), but growing evidence now supports a role for Ins(1,4,5)P₃ in the temporal and spatial organization of Ca²⁺ signals (Harootunian et al. 1991; DeLisle and Welsh, 1992; Miyazaki et al. 1992) and possibly even a direct role in the stimulation of Ca²⁺ influx across the plasma membrane (Irvine, 1991). Indeed, with the identification of multiple Ins(1,4,5)P₃ receptor subtypes, and the growing number of cytosolic factors known to regulate these receptors allosterically, it is likely that the control of Ins(1,4,5)P₃-stimulated Ca²⁺ mobilization is important in contributing to the organization and diversity of Ca²⁺ signals.

Ins(1,4,5)P₃ levels within cells peak during the first few seconds of agonist stimulation, after which they are rapidly reduced either to basal levels or, in the continued presence of agonist, to a slightly elevated plateau. These rapid, transient changes in Ins(1,4,5)P₃ concentration result from the competing metabolic pathways associated with Ins(1,4,5)P₃ generation and Ins(1,4,5)P₃ metabolism.

Ins(1,4,5)P₃ is generated in cells by the receptor-mediated activation of certain phospholipase C (PLC) isoforms (Fig. 1). Two classes of cell-surface receptor are involved. Receptors coupled to GTP-binding (G) proteins of the G_q family undergo a conformational change following agonist binding which allows them to interact with, and activate, the G-protein. An active subunit of the G-protein stimulates PLC-β. Receptors with intrinsic tyrosine kinase activity dimerize in response to agonist stimulation: the dimers phosphorylate each other on tyrosine residues within the cytosolic domains of the receptors, and these phosphotyrosine residues provide a site to which PLC-γbinds and becomes activated. Both PLC-β and PLC-γhydrolyse the membrane-associated lipid, phosphatidylinositol 4,5-bisphosphate, to produce 1,2-diacylglycerol and Ins(1,4,5)P₃. 1,2-Diacylglycerol remains in the plasma membrane where it activates proteins kinase C (Nishizuka, 1988), whereas Ins(1,4,5)P₃ diffuses into the cytosol and binds to specific receptors through which it mobilizes intracellular Ca²⁺ stores (Taylor and Richardson, 1991).

Ins(1,4,5)P₃ is inactivated by two metabolic pathways. It is either phosphorylated by a 3-kinase to yield inositol 1,3,4,5-tetrakisphosphate [Ins(1,3,4,5)P₄] (Batty et al. 1985; Irvine et al. 1986) or dephosphorylated by a 5-phosphatase to give inositol 1,4-bisphosphate [Ins(1,4)P₂] (Downes et al. 1982; Berridge et al. 1983). Ins(1,3,4,5)P₄ may itself regulate intracellular Ca²⁺ concentration by promoting Ca²⁺ sequestration (Hill et al. 1988), by synergizing with Ins(1,4,5)P₃ to mobilize intracellular Ca²⁺ stores (Cullen et al. 1990; Gawler et al. 1990) or by regulating agonist-stimulated Ca²⁺ entry across the plasma membrane (Irvine and Moor, 1986). The presence of specific Ins(1,3,4,5)P₄-binding sites in a number of tissues further supports a role for Ins(1,3,4,5)P₄ in the phosphoinositide signalling pathway (Bradford and Irvine, 1987; Theibert et al. 1987; Enyedi and Williams, 1988; Donié and Reiser, 1989). Ins(1,4)P₂, in contrast, probably plays no part in signal transduction and is rapidly metabolized to inositol (Storey et al. 1984).

Evidence from a variety of sources has indicated that Ins(1,4,5)P₃ binds to a specific Ins(1,4,5)P₃ receptor/Ca²⁺ channel complex located on intracellular Ca²⁺ stores where it directly stimulates release of Ca²⁺ into the cytosol. Comparison of the abilities of various inositol phosphates to release Ca²⁺ from permeabilized cells demonstrated a structure–activity relationship indicative of the binding of inositol phosphates to a specific receptor (Burgess et al. 1984). Ins(2,4,5)P₃ and Ins(4,5)P₂ were both able to mobilize stored Ca²⁺, although less potently than Ins(1,4,5)P₃, whereas Ins(1,4)P₂ was without effect. Thus, the ability of the inositol phosphates to mobilize intracellular Ca²⁺ depends on the presence of phosphate groups at both the 4 and 5 positions of the inositol ring and is enhanced by a phosphate at the 1 position.

High-affinity, saturable binding sites specific for Ins(1,4,5)P₃ have been identified in a number of tissues (Baukal et al. 1985; Spät et al. 1986; Worley et al. 1987; Guillemette et al. 1987), and binding of Ins(1,4,5)P₃ to these sites has been shown to correlate with Ins(1,4,5)P₃-stimulated Ca²⁺ mobilization (Spät et al. 1986; Guillemette et al. 1987). Ins(1,4,5)P₃ receptors have since been purified from a variety of tissues and species (Supattapone et al. 1988b; Maeda et al. 1991; Mourey et al. 1990; Chadwick et al. 1990; Parys et al. 1992). When reconstituted into lipid vesicles, the purified receptor mediates Ca²⁺ flux in response to Ins(1,4,5)P₃, providing firm evidence that both the Ins(1,4,5)P₃-binding site and the Ca²⁺ channel reside in the same protein complex (Ferris et al. 1989).

The structure and membrane topology of Ins(1,4,5)P₃ receptors is shown in Fig. 2. Each receptor consists of four similar subunits which are non-covalently associated (Mignery et al. 1990; Maeda et al. 1991). From electron microscopy of purified receptors, the four subunits have been shown to adopt a four-leaf-clover-like structure (Chadwick et al. 1990), at the centre of which is thought to be the Ca²⁺ channel itself. This bears a marked resemblance to the structure of ryanodine receptors, the Ca²⁺ release channels of skeletal and cardiac muscle (Lai et al. 1988), and indeed these proteins show considerable homology in their primary sequences (Mignery et al. 1989), higher-order structures and in allosteric regulation (see below). The two receptors may be derived from a common ancestral gene. Our idea of how these intracellular Ca²⁺ channels operate has been greatly influenced by the parallels that appear to exist between the two proteins.

Immunolocalization studies have revealed that both the amino and carboxy termini of the Ins(1,4,5)P₃ receptor subunits are cytosolic (Mignery et al. 1989; Nakade et al. 1991). Deletion of amino acid residues from the N-terminal region of the receptor either abolishes (Mignery et al. 1990) or greatly reduces (Miyawaki et al. 1991) binding of Ins(1,4,5)P₃, suggesting that the Ins(1,4,5)P₃-binding site is located within the first 788 residues of the N terminus of each subunit. The Ca²⁺ channel is thought to reside close to the C terminus of the receptor and indeed monoclonal antibodies directed against the C terminus have been shown to inhibit Ca²⁺ release (Nakade et al. 1991). Hydrophobicity analysis has identified between six and eight membrane-spanning domains close to the C terminus (Mignery et al. 1990; Yoshikawa et al. 1992), which are important for anchoring the protein into membranes (Mignery and Südhof, 1990). The four putative transmembrane regions closest to the C terminus, which are flanked by a large number of net negative charges, show the highest degree of homology with ryanodine receptors, and they are therefore presumed to be important in the gating of Ca²⁺ (Mignery et al. 1990). The C terminus is also thought to be important in the formation of tetrameric complexes, since deletion of this portion of the receptor results in the formation of monomeric Ins(1,4,5)P₃-binding proteins (Mignery and Südhof, 1990; Miyawaki et al. 1991).

Located between the N-terminal Ins(1,4,5)P₃-binding domain and the C-terminal Ca²⁺ channel is a stretch of more than 1500 amino acid residues which is the target of a number of allosteric regulators of the Ins(1,4,5)P₃ receptor (discussed below). When Ins(1,4,5)P₃ binds to the N terminus, a large conformational change ensues (Mignery and Südhof, 1990) which is thought to link Ins(1,4,5)P₃ binding, via this long coupling domain, to opening of the Ca²⁺ channel. This may be an oversimplification, however, since we do not yet know how the Ins(1,4,5)P₃ receptor folds within the membrane of intracellular stores, and it is possible that regions further from the N terminus also influence Ins(1,4,5)P₃ binding (see Furuichi et al. 1989).

Ins(1,4,5)P₃ receptors are highly concentrated in the Purkinje cells of the cerebellum (the physiological significance of such high receptor density is not understood), with lower levels being found in other regions of the brain (Worley et al. 1987, 1989) and in a variety of peripheral tissues (Chadwick et al. 1990; Guillemette et al. 1990; Sharp et al. 1992). At a subcellular level, immunohistochemical studies in Purkinje cells, Xenopus oocytes and pancreatic epithelial cells have localized Ins(1,4,5)P₃ receptors to the rough and smooth endoplasmic reticula and the subplasmalemmal cisternae, with conflicting reports as to the presence of Ins(1,4,5)P₃ receptors in the nucleus and an apparent absence of receptors in the mitochondria and Golgi cisternae (Ross et al. 1989; Maeda et al. 1990; Parys et al. 1992; Sharp et al. 1992). More precise localization has been achieved using immunogold labelling, resulting in the localization of receptors to specific subcompartments of the endoplasmic reticulum (Satoh et al. 1990). Consistent with the Ins(1,4,5)P₃-sensitive Ca²⁺ stores residing in specialized regions of the endoplasmic reticulum, vasopressin, an agonist linked to phosphoinositide metabolism, reduces the Ca²⁺ content of the rough endoplasmic reticulum in rat liver (Bond et al. 1987).

In some cells, Ins(1,4,5)P₃ receptors have been found to be associated with the plasma membrane (Kuno and Gardner, 1987; Khan et al. 1992), prompting speculation that Ins(1,4,5)P₃ may play a direct role in regulating hormone-dependent Ca²⁺ influx. These plasma-membrane-associated Ins(1,4,5) P₃ receptors appear to show less specificity for Ins(1,4,5)P₃ over other inositol phosphates compared with Ins(1,4,5)P₃ receptors localized in the endoplasmic reticulum (Kalinoski et al. 1992). Attention has therefore been drawn towards the Ins(1,4,5)P₃ metabolite, Ins(1,3,4,5)P₄, as a candidate for a physiological regulator of Ca²⁺ influx (Irvine, 1991). Interestingly, Ins(1,4,5)P₃ and Ins(1,3,4,5)P₄ appear to act synergistically to stimulate Ca²⁺ influx into sea urchin eggs (Irvine and Moor, 1986) and lacrimal cells (Morris et al. 1987), and in endothelial cells a Ca²⁺ channel within the plasma membrane appears to be activated by Ins(1,3,4,5)P₄, but is insensitive to Ins(1,4,5)P₃ (Lückhoff and Clapham, 1992).

The mechanism(s) underlying receptor-mediated Ca²⁺ influx remain elusive, but it is clear that the products of phosphoinositide metabolism are likely to feature prominently in at least some cell types.

Two distinct genes have been identified in rodent cerebellum that encode full-length sequences for Ins(1,4,5)P₃ receptors (Furuichi et al. 1989; Mignery et al. 1990; Südhof et al. 1991). The type 1 and type 2 receptors (2749 and 2701 amino acids in length respectively) are 69% identical in their amino acid sequences and are structurally very similar, both to each other and to the ryanodine receptor Ca²⁺ channel of skeletal muscle (Takeshima et al. 1989). The type 2 receptor is much less abundant than the type 1 receptor (Südhof et al. 1991). Significantly, the ligand-binding site and most of the transmembrane domains comprising the Ca²⁺ channel appear to be the most conserved regions between the receptor subtypes, whereas the coupling domain shows the greatest diversity (Südhof et al. 1991). The major functional differences between the two receptors could therefore be associated with their respective patterns of allosteric regulation. Partial sequences have now been obtained indicating at least one further subtype (Ross et al. 1992), which suggests that a large family of Ins(1,4,5)P₃ receptors may exist.

Full-length sequences encoding Ins(1,4,5)P₃ receptors from Xenopus oocytes (Kume et al. 1993) and Drosophila melanogaster (Yoshikawa et al. 1992) have now been determined. The Xenopus receptor appears to be a species variant of the type 1 receptor, displaying 90% sequence homology with the mouse cerebellar type 1 receptor, whereas the Drosophila receptor differs from both the type 1 and type 2 receptors to a similar extent (Yoshikawa et al. 1992). Once again, the receptor from Drosophila shows considerable sequence homology with the mouse type 1 and Xenopus receptors in both the ligand-binding domain and Ca²⁺ channel region, but displays considerable diversity in the coupling domain. Most significantly, the Drosophila receptor lacks consensus sequences for phosphorylation in the coupling domain, indicating that it is not a substrate for regulation by cyclic-AMP-dependent protein kinase (PKA).

Ins(1,4,5)P₃ receptor diversity can also be generated by differential splicing of receptor transcripts. Two major splice sequences, SI and SII, have been identified within the type 1 receptor. SI encodes a 15 amino-acid sequence located within the N-terminal Ins(1,4,5)P₃-binding domain (Mignery et al. 1990; Nakagawa et al. 1991) and may therefore influence the binding characteristics of Ins(1,4,5)P₃. SII is larger, encoding 40 amino acids within the regulatory domain of the receptor (Danoff et al. 1991), and is made up of three smaller splice variants (A, B and C: 23, 1 and 16 residues respectively) which can each be independently removed to generate a series of different Ins(1,4,5)P₃ receptor transcripts (Nakagawa et al. 1991). The SII splice sequence is located between two PKA phosphorylation consensus sequences (Danoff et al. 1991), and it is therefore not surprising that receptors that differ in the expression of this splice sequence also appear to differ in their patterns of phosphorylation by PKA (see below; Danoff et al. 1991).

At present we have little idea as to the physiological significance of these multiple receptor transcripts, although the conservation of alternative splice events between species (Danoff et al. 1991) suggests an important role in the regulation of Ins(1,4,5)P₃ receptor activity. Recently, different Ins(1,4,5)P₃ receptor splice variants have been shown to be expressed in both a developmentally specific and tissue-specific manner (Danoff et al. 1991; Nakagawa et al. 1991). The formation of synapses between Purkinje cells and granule cells during cerebellar development, for example, coincides with peak expression of receptors containing the SI splice sequence, whereas other forms of the receptor show a progressively increasing level of expression throughout this period of development (Nakagawa et al. 1991). In mouse, Ins(1,4,5)P₃ receptors containing the SII splice sequence are exclusively localized in neuronal tissue, whereas the shorter forms of the receptor lacking this SII domain are found largely in the periphery (Danoff et al. 1991). The predominance of SII-containing receptors in neuronal tissue is, however, not common to all species, since Ins(1,4,5)P₃ receptors cloned from Drosophila head lack the SII splice sequence (Kume et al. 1993).

It is clear that a number of subtypes and isoforms of the Ins(1,4,5)P₃ receptor exist and that these can be differentially expressed both in different tissues and at different stages of development. Furthermore, cerebellar Purkinje cells are now known to express at least two, and possibly three, distinct Ins(1,4,5) P₃ receptors (Ross et al. 1992), implying that, even within a single cell, the responsiveness to Ins(1,4,5)P₃ may be regulated by a heterogeneous population of receptors. Whether heterogeneity can exist between the subunits in a single Ins(1,4,5)P₃ receptor remains to be determined, but this would provide the basis for further diversity reminiscent of many cell surface ligand-gated ion channels (Unwin, 1989).

Ins(1,4,5)P₃ receptors are the targets of a number of allosteric regulators, including protein kinases, adenine nucleotides, pH and divalent cations, all of which may play a part in determining the responsiveness of cells to Ins(1,4,5)P₃. Further complexity is now apparent with the discovery that different Ins(1,4,5)P₃ receptor subtypes and splice variants show different patterns of allosteric regulation.

Cross-talk between the cyclic AMP and phosphoinositide signalling pathways is well documented (Berridge, 1975; Jenkinson and Koller, 1977; Rasmussen and Barrett, 1984). In hepatocytes, for example, the frequency of Ca²⁺ spikes triggered by hormones linked to Ins(1,4,5) P₃ production is increased by activation of receptors coupled to adenylate cyclase. Cyclic AMP is now known to phosphorylate Ins(1,4,5)P₃ receptors via the activation of PKA (Walaas et al. 1986). Two putative consensus sequences for phosphorylation by PKA (X-Arg-Arg-X-Ser-X) have been identified in the coupling domain of both the type 1 (Furuichi et al. 1989; Mignery et al. 1990) and type 2 (Südhof et al. 1991) Ins(1,4,5)P₃ receptors, fuelling speculation that this region of the receptor is the major target for allosteric regulators. A third putative consensus sequence located in the Ca²⁺ channel domain has also been reported (Mignery et al. 1990). Although these consensus sequences are often conserved between tissues, phosphorylation of Ins(1,4,5)P₃ receptors by PKA appears to have different effects in different cell types. Cerebellar microsomes, for example, are 10-fold less sensitive to Ins(1,4,5)P₃ following cyclic-AMP-dependent phosphorylation (Supattapone et al. 1988a), whereas Ca²⁺ release from permeabilized hepatocytes is potentiated by pretreatment of the cells with the catalytic subunit of PKA (Burgess et al. 1991). In platelets, PKA also phosphorylates Ins(1,4,5)P₃ receptors but there are conflicting reports as to the effects of PKA on Ca²⁺ release (Enouf et al. 1987; Quinton and Dean, 1992). PKA markedly stimulates Ca²⁺ uptake into the platelet dense tubular system, whereas the reported effects of PKA on Ins(1,4,5)P₃-stimulated Ca²⁺ release are modest by comparison (Enouf et al. 1987; Quinton and Dean, 1992) and may be confused with regulatory effects of luminal Ca²⁺ on the Ins(1,4,5)P₃ receptors (see below).

The tissue-specific effects of PKA on Ins(1,4,5)P₃-stimulated Ca²⁺ fluxes may arise from differences in the phosphorylation patterns of distinct Ins(1,4,5)P₃ receptor subtypes and splice variants (see above). For example, while the long, neuronal form of the type 1 Ins(1,4,5) P₃ receptor is phosphorylated on both PKA consensus sequences in the coupling domain (Ser-1589 and Ser-1756), the shorter splice variant, localized in peripheral tissues, is phosphorylated at lower substrate concentrations and almost exclusively on Ser-1589 (Danoff et al. 1991). Differences in patterns of phosphorylation also exist between species: the recently cloned Drosophila Ins(1,4,5)P₃ receptor, for example, appears not to contain potential PKA phosphorylation sites (Yoshikawa et al. 1992).

In addition to PKA, Ins(1,4,5)P₃ receptors are now known to be targets for phosphorylation by protein kinase C (PKC) and Ca²⁺ calmodulin-dependent protein kinase II (CaM kinase II) (Yamamoto et al. 1989; Ferris et al. 1991). The three enzymes phosphorylate the receptor at different sites (Ferris et al. 1991), suggesting that each may have a specific and independent role in controlling Ins(1,4,5)P₃ receptor activity. In each case, phosphorylation is reported to be stoichiometric (Supattapone et al. 1988a; Ferris et al. 1991), with each subunit of a receptor being phosphorylated at a single site by each kinase. However, in the case of phosphorylation by PKA, this may be an oversimplification since Ins(1,4,5)P₃ receptors are known to show distinctive patterns of phosphorylation depending on the expression of various alternatively spliced sequences (Danoff et al. 1991; see above).

At present, the functional significance of these phosphorylation events for Ins(1,4,5)P₃ receptor activity is poorly understood. PKC-mediated phosphorylation of Ins(1,4,5)P₃ receptors is reported to accelerate Ca²⁺ release (Matter et al. 1993), whereas the effects of phosphorylation by PKA may be either stimulatory or inhibitory (as discussed above). It is clear that phosphorylation of Ins(1,4,5)P₃ receptors provides a means by which different limbs of the intracellular signalling pathways may overlap to regulate tightly the release of intracellular Ca²⁺ stores. The recent finding that purified Ins(1,4,5)P₃ receptors themselves have intrinsic protein kinase activity, such that they are able both to autophosphorylate and to phosphorylate exogenous substrates (Ferris et al. 1992), further suggests that phosphorylation of the receptor is an important mechanism for regulating the actions of Ins(1,4,5)P₃.

ATP regulates the activity of Ins(1,4,5)P₃ receptors in a biphasic manner. Small elevations in ATP concentration (in the range 10–500 μmol l^-1) increase the binding of Ins(1,4,5)P₃ to its receptor (Spät et al. 1992) and potentiate Ins(1,4,5)P₃-stimulated Ca²⁺ fluxes (Smith et al. 1985; Suematsu et al. 1985; Ferris et al. 1990; Iino, 1991), whereas at higher concentrations (>500 μmol l^-1) ATP inhibits both Ins(1,4,5)P₃ binding (Guillemette et al. 1987; Willcocks et al. 1987; Nunn and Taylor, 1990) and Ca²⁺ release (Ferris et al. 1990; Iino, 1991).

The stimulatory effects of ATP are mimicked by other adenine nucleotides (Ferris et al. 1990; Iino, 1991) but not by GTP (Ferris et al. 1990), suggesting that the adenine base is important in mediating these effects. Electrophysiological studies of purified Ins(1,4,5)P₃ receptors reconstituted into planar lipid bilayers have demonstrated that ATP acts directly on the Ins(1,4,5)P₃ receptor to promote Ca²⁺ release either by increasing the open probability of Ca²⁺ channels (Ehrlich and Watras, 1988) or by promoting the formation of a larger-conductance Ca²⁺ channel (Maeda et al. 1991). Two putative adenine-nucleotide-binding sites (Gly-X-Gly-X-X-Gly) have been identified in the coupling domain of the cerebellar Ins(1,4,5)P₃ receptor (Furuichi et al. 1989; Mignery et al. 1990), and a third site is present in non-neuronal tissues following the deletion of the SII splice sequence. Binding of ATP to these sites is thought to mediate the stimulatory effects of ATP on the Ins(1,4,5)P₃ receptor (Ferris et al. 1990; Maeda et al. 1991).

The inhibitory effect of ATP arises from competition between ATP and Ins(1,4,5)P₃ for the Ins(1,4,5)P₃-binding site (Willcocks et al. 1987; Nunn and Taylor, 1990; Iino, 1991). ATP-mediated inhibition of [³H]Ins(1,4,5)P₃ binding to purified receptors is mimicked by GTP and pyrophosphate (Maeda et al. 1991), suggesting that it is the binding of the pyrophosphate group of ATP to the Ins(1,4,5)P₃ receptor that is important in mediating the inhibitory effects of ATP.

Although ATP levels in the cytosol may be important in setting the sensitivity of Ins(1,4,5)P₃ receptors, there is still doubt as to whether Ins(1,4,5)P₃ receptors respond to changes in ATP concentration following agonist stimulation. From nuclear magnetic resonance studies in smooth muscle bundles, the intracellular ATP concentration has been shown to remain fairly constant, in the range 1–2mmol l^-1, even following maximal agonist stimulation (Kushmeric et al. 1986). It is still conceivable, however, that local changes in ATP concentration, possibly in the vicinity of the Ca²⁺-ATPase pumps, which may be maximally activated following cell stimulation, may be important in controlling the activity of Ins(1,4,5)P₃ receptors in intact cells.

Binding of Ins(1,4,5)P₃ to its receptor and Ins(1,4,5)P₃-stimulated Ca²⁺ release are both favoured by alkaline pH, optimally in the range 7.5–9 (Brass and Joseph, 1985; Worley et al. 1987; Joseph et al. 1989). These effects are mediated by an increase in the affinity of Ins(1,4,5)P₃ receptors for their ligand rather than by an increase in the number of available Ins(1,4,5)P₃-binding sites (Joseph et al. 1989).

The steep dependence of Ins(1,4,5)P₃ binding on pH suggests that small increases in intracellular pH may have a stimulatory effect on Ca²⁺ release in an intact cell. Indeed, growth factors and hormones that stimulate phosphoinositide turnover are also known to cause alkalization of the cytosol (by up to 0.3pHunits) via the activation of protein kinase C (PKC) (Horne et al. 1981). Once activated, PKC is able to stimulate the membrane-bound Na⁺/H⁺ antiporter, which extrudes H⁺ from the cytosol. This raises the attractive possibility of crosstalk between the two limbs of the phosphoinositide signalling pathway.

In a number of cell types, sulphydryl reagents such as thimerosal and oxidized glutathione trigger Ca²⁺ oscillations (Swann, 1991; Bootman et al. 1992). In hamster eggs, these oscillations can be inhibited by monoclonal antibodies directed against the C terminus of the Ins(1,4,5)P₃ receptor, suggesting a central role for Ins(1,4,5)P₃ receptors in the responses (Miyazaki et al. 1992). It has been proposed that the stimulation of Ca²⁺ oscillations by oxidized glutathione and thimerosal is dependent on their ability to increase the affinity of Ins(1,4,5)P₃ receptors for Ins(1,4,5)P₃ (Hilly et al. 1993) and subsequently to sensitize the release of Ins(1,4,5)P₃-sensitive Ca²⁺ stores (Missiaen et al. 1991), although several groups have reported inhibitory effects of thimerosal on Ins(1,4,5)P₃ receptors (Adunyah and Dean, 1986; Guillemette and Segui, 1988). These findings are reconciled by the observation that, at low concentrations, thimerosal sensitizes Ins(1,4,5)P₃-stimulated Ca²⁺ release from cerebellar microsomes, but at higher concentrations it is inhibitory (Sayers et al. 1993).

In resting cells, the concentration of oxidized glutathione is more than 20-fold lower than the concentration of the reduced form, and is out of the range at which it is believed to regulate the activity of Ins(1,4,5)P₃ receptors. We do not yet know whether physiological changes in redox state within cells can influence the activity of Ins(1,4,5)P₃ receptors.

The regulatory effects of cytosolic Ca²⁺ on the Ins(1,4,5)P₃ receptor are confusing. Ca²⁺ inhibits [³H]Ins(1,4,5)P₃ binding to cerebellar membranes (Worley et al. 1987), stimulates binding in liver (Pietri et al. 1990), but has no effect on the binding of Ins(1,4,5)P₃ to its receptor in vas deferens (Mourey et al. 1990). Similarly, Ca²⁺ reportedly inhibits Ca²⁺ release in some instances (Joseph et al. 1989; Willems et al. 1990; Zhao and Muallem, 1990) but not in others (Brass and Joseph, 1985). While some tissue specificity may exist, other factors have certainly contributed to these inconsistent observations. The inhibitory effect of Ca²⁺ on cerebellar Ins(1,4,5)P₃ receptors has been ascribed to an accessory protein, termed calmedin, which is enriched in cerebellum but exists at much lower concentrations in the periphery (Danoff et al. 1988). Once purified, the cerebellar Ins(1,4,5)P₃ receptor is insensitive to Ca²⁺, but sensitivity is restored by addition of solubilized cerebellar membranes enriched in calmedin activity (Supattapone et al. 1988b). However, recent evidence suggests that rather than being a Ca²⁺-binding accessory protein, calmedin may be a Ca²⁺-sensitive isoform of PLC (Mignery et al. 1992). Apparent Ca²⁺-dependent inhibition of Ins(1,4,5)P₃ binding might then simply reflect Ca²⁺-stimulated production of endogenous Ins(1,4,5)P₃, which would compete with radiolabelled Ins(1,4,5)P₃ for the receptor. Another complexity arises from the observation that, at high concentrations, a number of frequently used Ca²⁺ buffers, including the fluorescent Ca²⁺ indicator dye Fura-2, have been shown to act as competitive antagonists at the Ins(1,4,5)P₃ receptor (Richardson and Taylor, 1993). As free Ca²⁺ concentrations are adjusted in experimental protocols, so too are the concentrations of free Ca²⁺ chelators.

Finally, regulation of Ins(1,4,5)P₃ receptors by cytosolic Ca²⁺ is now known to be highly concentration-dependent. Small elevations in cytosolic [Ca²⁺] (<300nmol l^-1) potentiate both Ca²⁺ release and channel opening, whereas at higher concentrations Ca²⁺ is inhibitory (Iino, 1990; Bezprozvanny et al. 1991; Finch et al. 1991). These effects of Ca²⁺ operate on a rapid time-scale, suggesting that they may play an important role in the regulation of Ins(1,4,5)P₃-stimulated Ca²⁺ release in intact cells (Iino and Endo, 1992). In liver, at least three distinct affinity and conductance states of the Ins(1,4,5)P₃ receptor have been identified (Fig. 3). When cytosolic Ca²⁺ is elevated (<700nmol l^-1), receptors are converted from a state with low affinity for Ins(1,4,5)P₃ to a state with high affinity (Pietri et al. 1990), both states being able to conduct Ca²⁺ in response to Ins(1,4,5)P₃ (Marshall and Taylor, 1993). As Ca²⁺ is further elevated (>1 μmoll^-1), receptors maintain their high-affinity conformation but are unable to conduct Ca²⁺ when bound to Ins(1,4,5)P₃.

Considering the oscillatory changes in cytosolic Ca²⁺ during cell activation, Ca²⁺ itself has the potential to regulate it own release from Ins(1,4,5)P₃-sensitive Ca²⁺ stores. Such regulatory control is, therefore, a key component in several recent models used to describe the complex organization of Ca²⁺ signals (discussed below).

Ca²⁺ within the lumen of the sarcoplasmic reticulum has been shown to enhance opening of ryanodine receptors in skeletal muscle (Nelson and Nelson, 1990), prompting speculation of a similar role for luminal Ca²⁺ in regulating Ins(1,4,5)P₃ receptors (Irvine, 1991). Evidence for such a role, however, has been difficult to obtain for a number of reasons. Unlike the situation in the cytosol, where Ca²⁺ buffers can be used to titrate the cytosolic [Ca²⁺], the [Ca²⁺] within the lumen of stores cannot be reliably measured. Experimental procedures have therefore relied on the manipulation of luminal [Ca²⁺], either by overloading the Ca²⁺ stores (Missiaen et al. 1991) or by progressively discharging full stores using Ca²⁺ ionophores or Ca²⁺ pump inhibitors (Nunn and Taylor, 1990; Marshall and Taylor, 1993). To compound these problems, the effects of luminal Ca²⁺ on Ins(1,4,5)P₃ receptor activity are modest compared with those of cytosolic Ca²⁺. Depleting Ca²⁺ stores reduces their affinity for Ins(1,4,5)P₃ only twofold (Oldershaw and Taylor, 1993), with a similar effect on the EC₅₀ for Ca²⁺ release (Nunn and Taylor, 1990). The failure of some groups to detect effects of luminal Ca²⁺ on Ins(1,4,5)P₃ receptors (Shuttleworth, 1992; Combettes et al. 1993) may be due to the extensive store depletion that is required before these effects become evident (Marshall and Taylor, 1993).

Although the effects of luminal Ca²⁺ on the Ins(1,4,5)P₃ receptor are modest, there is some evidence to suggest that changes in luminal [Ca²⁺] also modulate the regulatory effects of cytosolic Ca²⁺ (Missiaen et al. 1992): thus, a three-way balance may exist in which Ins(1,4,5)P₃, cytosolic Ca²⁺ and luminal Ca²⁺ interdependently regulate the opening of Ins(1,4,5)P₃ receptors.

Mg²⁺ inhibits both Ins(1,4,5)P₃ binding (Varney et al. 1990; White et al. 1991) and Ins(1,4,5)P₃-stimulated Ca²⁺ mobilization (Volpe et al. 1990) at concentrations slightly lower than those found in intact cells (Brooks and Bachelard, 1989). At present, we have little information concerning Mg²⁺ concentration fluctuations during agonist stimulation, but it is possible that, when cytosolic [Ca²⁺] is increased, Mg²⁺ may be displaced from Ca²⁺/Mg²⁺-binding proteins, such as parvalbumin, thereby transiently elevating intracellular Mg²⁺ levels. Interpreting how Mg²⁺ is affecting Ins(1,4,5)P₃ receptors may be more difficult: apart from its putative role as an allosteric regulator, Mg²⁺ also stimulates metabolism of Ins(1,4,5)P₃via activation of the 5-phosphatase and is essential for Ins(1,4,5)P₃ receptor autophosphorylation (Ferris et al. 1992).

When Ins(1,4,5)P₃ binds to its receptor, there is a conformational change in the receptor complex (Mignery and Südhof, 1990) and an increase in the frequency of Ca²⁺ channel opening (Ehrlich and Watras, 1988), resulting in the release of Ca²⁺ from intracellular stores (Streb et al. 1983). Rapid kinetic measurements of Ca²⁺ release in permeabilized (Champeil et al. 1989; Meyer et al. 1990) and intact (Ogden et al. 1990) cells have revealed a delay of less than 100ms between the addition of Ins(1,4,5)P₃ and the onset of Ca²⁺ mobilization, indicating that the latency of receptor-mediated [Ca²⁺] changes is primarily determined by the rate of Ins(1,4,5)P₃ formation.

Electrophysiological studies suggest that four subconductance states exist for single Ins(1,4,5)P₃ receptors (Watras et al. 1991); these may arise from different numbers of Ins(1,4,5)P₃ molecules binding to the tetrameric receptor complex. With the possibility that heterogeneous Ins(1,4,5)P₃ receptor subunits may exist in a single receptor, there is certainly scope for a large family of Ins(1,4,5)P₃ receptors that differ in their Ca²⁺ conductances. Some groups have reported cooperative Ins(1,4,5)P₃-stimulated Ca²⁺ release from intracellular stores (Champeil et al. 1989; Meyer et al. 1990). However, such cooperativity is not evident under conditions in which the cytosolic [Ca²⁺] is clamped (Finch et al. 1991) and is, therefore, more likely to be a consequence of the stimulatory effects of cytosolic Ca²⁺ on Ins(1,4,5)P₃-stimulated Ca²⁺ release (see above). Ca²⁺ release in intact cells may, nevertheless, show steep dependence on Ins(1,4,5)P₃ concentrations as a result of this regulatory effect of cytosolic Ca²⁺.

A further complexity in the mechanism of Ins(1,4,5)P₃-stimulated Ca²⁺ release is the phenomenon of ‘quantal’ Ca²⁺ mobilization, whereby low concentrations of Ins(1,4,5)P₃ are able to mobilize only a fraction of the Ins(1,4,5)P₃-sensitive Ca ²⁺ stores (Muallem et al. 1989; Taylor and Potter, 1990), despite the inability of Ins(1,4,5)P₃ receptors to desensitize intrinsically (Oldershaw et al. 1992). Although a number of explanations for this phenomenon have been proposed, neither the mechanism nor the physiological significance of quantal Ca²⁺ mobilization is understood.

The Ca²⁺ signals evoked by receptor activation are complex, occurring in many cells as repetitive Ca²⁺ spikes (see Berridge, 1990), which may be arranged spatially as regenerative waves (Jacob, 1990; Rooney et al. 1990) or even as spirals of Ca²⁺ release (Lechleiter et al. 1991). Models which describe the generation and propagation of Ca²⁺ signals are plentiful (reviewed by Berridge, 1990) and differ in the extent to which Ins(1,4,5)P₃ contributes. Ins(1,4,5)P₃ plays an important role in triggering regenerative Ca²⁺ signals in many cells (Oron et al. 1985; Parker and Miledi, 1986; Wakui et al. 1989), and more recently it has been shown to be essential for the propagation of Ca²⁺ waves in Xenopus oocytes (DeLisle and Welsh, 1992) and fertilized hamster eggs (Miyazaki et al. 1992). Consistent with these observations, recent models to describe how Ca²⁺ waves propagate across cells invoke the sequential release of Ins(1,4,5)P₃-sensitive Ca²⁺ stores: it is necessary, therefore, to determine how these stores are regulated so as to release their Ca²⁺ in a sequential manner.

If diffusion of Ins(1,4,5)P₃per se were to account for the sequential discharge of Ca²⁺ stores, one would expect Ca²⁺ waves to propagate across cells at a non-uniform rate, a feature not easily reconciled with experimental observations (Rooney et al. 1990).

Another possibility is that release of Ins(1,4,5)P₃-sensitive Ca²⁺ stores is regulated by a signal transmitted concomitantly with the Ca²⁺ wave; the complex control of Ins(1,4,5)P₃ receptor sensitivity by cytosolic Ca²⁺ (discussed above) suggests that Ca²⁺ itself is a likely candidate. The Ca²⁺ wave might then propagate across the cell as follows (Fig. 4). Prior to receptor activation, resting cytosolic [Ca²⁺] is low (approximately 100nmol l^-1), so that Ins(1,4,5)P₃ receptors are in a conformation with low affinity for Ins(1,4,5)P₃ (see Fig. 3). Activation of cell surface receptors triggers production of Ins(1,4,5)P₃ at the plasma membrane, which stimulates neighbouring Ins(1,4,5)P₃-sensitive stores to release their Ca²⁺ into the cytosol. The Ca²⁺ released from these stores then sensitizes Ins(1,4,5)P₃ receptors on stores deeper inside the cell, allowing them to respond to the lower levels of Ins(1,4,5)P₃ further from the plasma membrane. The Ca²⁺ wave would then propagate by the concerted actions of Ins(1,4,5)P₃ and Ca²⁺ on Ins(1,4,5)P₃ receptors. Once a store releases its Ca²⁺, Ins(1,4,5)P₃ receptors on that store may become inactive as a result of the inhibitory effects of the high local cytosolic [Ca²⁺] and the reduced luminal [Ca²⁺] (discussed above). This would render Ins(1,4,5)P₃-sensitive Ca²⁺ stores refractory to Ins(1,4,5)P₃, allowing them to refill in preparation for a subsequent Ca²⁺ wave. Support for this model comes from the study of agonist-evoked oscillations in pancreatic acinar cells, where the frequency of oscillations is increased in response to small elevations in cytosolic Ca²⁺ (<200nmol l^-1), but reduced when cytosolic Ca²⁺ is raised above 500nmol l^-1 (Zhang et al. 1992).

Recently, single spontaneous Ca²⁺ spikes have been recorded in populations of permeabilized hepatocytes (Missiaen et al. 1991, 1992); they have been attributed to the synchronized discharge of Ins(1,4,5)P₃-sensitive Ca²⁺ stores. Overloading of these stores with Ca²⁺ appeared to sensitize Ins(1,4,5)P₃-stimulated Ca²⁺ release, allowing the stores to respond to basal levels of Ins(1,4,5)P₃. These responses were preceded by a pacemaker rise in cytosolic Ca²⁺ concentration, which may reflect the involvement of a positive feedback effect of cytosolic [Ca²⁺] on Ins(1,4,5)P₃-stimulated Ca²⁺ release during the rising phase of the spikes. Interestingly, these spontaneous Ca²⁺ spikes in permeabilized cell suspensions are critically dependent on ATP concentration, pH, cytosolic [Ca²⁺] and cell density, and are sensitized by sulphydryl reagents (Missiaen et al. 1992). An attractive possibility is that they represent a crude demonstration in vitro of the Ca²⁺ spikes seen in intact cells: here, the responses are evident in a system in which the cytosol from a population of cells has been effectively pooled. If this proves to be the case, it is clear that a number of cytosolic regulators of Ins(1,4,5)P₃ receptor activity are critically important in setting the conditions for the generation of Ca²⁺ signals.

Work from the authors’ laboratory is supported by the Wellcome Trust and an SERC–CASE award with SmithKline Beecham to I.C.B.M. C.W.T. is a Lister Institute Research Fellow.