Signal transduction mechanisms involved in hormonal Ca2+ fluxes.

This article reviews literature up to mid-1988 covering recent developments pertaining to agonist-induced Ca2+ signaling in various cell types. A large amount of experimental evidence supports a mechanism involving specific guanine nucleotide-binding proteins (G-proteins) as transducing factors between occupancy of a wide variety of receptors by many different agonists and activation of polyphosphoinositide specific phospholipase C enzymes. Although many different G-proteins and phospholipase C enzymes have been purified and cloned, successful reconstitution of the components has not been achieved. Hence, many questions concerning the specificity of coupling between particular receptors to a particular G-protein and phospholipase C subtype remain unresolved. Phospholipase C subtypes isolated from the membrane and soluble fractions of the cell are directly activated by Ca2+ and, preferentially, hydrolyse phosphatidylinositol 4,5-bisphosphate (PIP2) and phosphatidylinositol 4-phosphate (PIP). The role of the G-protein is to stimulate inositol lipid breakdown at free Ca2+ concentrations (0.1-0.2 microM) typical of unstimulated cells. Overwhelming evidence supports the concept that Ins 1,4,5-P3, the product of PIP2 hydrolysis, is responsible for the initial agonist-induced Ca2+ transient by mobilization of Ca2+ from a specialized intracellular store. An Ins 1,4,5-P3 receptor has been purified that may correspond to the postulated Ins 1,4,5-P3 gated Ca2+ channel. Despite a growing understanding of the complexities of the metabolism of Ins 1,4,5-P3 and a successful purification of many enzymes involved, including the ATP-dependent 3-kinase that converts Ins 1,4,5-P3 to Ins 1,3,4,5-P4, the role of Ins 1,3,4,5-P4 as a putative second messenger remains enigmatic. Multiple forms of protein kinase C have been described and the role is well established for a 1,2-diacylglycerol, the second product of PIP2 hydrolysis, as its physiological activator. Although protein kinase C has been shown to phosphorylate and modulate the activity of several proteins involved in the Ca2+ signaling pathway and Ca2+ transport, the physiological significance of the protein kinase C in agonist-stimulated cell function requires further elucidation. The extension of measurements of hormone-induced Ca2+ changes to single cells has shown that the occurrence of Ca2+ oscillations is a common phenomena. Elucidation of the biochemical mechanisms causing this oscillatory response and its physiological significance represents an important challenge for future studies.


Introduction
The cytosolic free Ca2" concentration functions as an important intracellular signaling mechanism whereby hormones and growth factors regulate many different cellular processes such as secretion, metabolism, neurotransmitter release, cell growth, and differentiation. Signal transduction by the ligandactivated receptor is mediated by specific guanine nucleotide binding proteins (G-proteins), which activate phospholipase C-mediated hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2) in the plasma membrane. The products of this reaction are 1,2-diacylglycerol, which is retained in the plasma membrane, and D-myo-inositol-1,4,5-trisphosphate gene products involved in cell growth (5,6). These latter effects may be mediated by protein kinase C isozymes that are different from those responsible for the short-term effects (7)(8)(9).
Despite a general acceptance of the overall sequence of events responsible for linking receptor activation to increases of cytosolic free Ca2" (10)(11)(12)(13), many details of the mechanisms underlying each step remain unresolved. Recent advances suggest a high level of cellular diversification and the presence of an intricate network of positive and negative controls providing continuous regulation of the response elicited by receptor activation.
This review will describe recent findings concerning various types of novel G-proteins and phospholipases found in different cells, and the mechanisms involved in regulation of Ca2' flux across the plasma and intracellular membranes. These mechanisms, together with a variety of feedback controls, could account for diverse cellular Ca2+ responses seen in different tissues, as well as the marked heterogeneity of response seen in single cells within a population, including an oscillatory behavior of the Ca2' signal.

Generation of Intracellular Second Messengers
The feature that appears to be common to different cell types for agonist-induced Ca21 mobilization involves an interaction in the plasma membrane between three different types of proteins, namely, receptors, G-proteins, and phospholipase C. Most cells contain many different types of receptors in the plasma membrane which, upon stimulation by suitable agonists, cause an increase of cytosolic free Ca2' (10,(14)(15)(16). The number of receptors and their affinity determines the relative sensitivity of the cell to each agonist and the extent of phospholipase C activation. Since a particular agonist is often capable of interacting with several different subtypes of receptors, the specificity of signal transduction is likely to be determined by the nature and properties of the G-protein that interacts with a particular receptor subtype and effector system (17).

The Role of G-Proteins in Activating Phospholipases
The structure and molecular properties of G-proteins and their role in signal transduction have been reviewed recently (18,19). Those involved in receptor coupling in the plasma membrane act as both on/off switches and a mechanism for signal channelling and amplification. Generally they are present in greater abundance than the receptor proteins. G-proteins typically have a heterotrimeric apy structure. The asubunit is the site for GTP-binding, and differences in their primary structure (20)(21)(22)(23) provides the basis for their specific roles as transducing agents. A general model for G-protein action has been proposed, which is based on numerous studies concerned with the mechanism of activation of adenylate cyclase and photoreceptor function (18,24). Binding of agonist to the receptor causes the exchange of guanosine diphosphate (GDP) for guanosine triphosphate (GTP) on the G-protein a-subunit. This is accompanied by a dissociation of the a-subunit from the heterotrimer and a decreased binding affinity of the agonist to the receptor. The activated (GTP-bound) a-subunit then interacts with and modifies the activity of the effector protein. The G-protein a-subunits have an intrinsic GTPase activity, yielding inactive a-GDP complexes, which reassociate with the available pool of Py-dimers to reform the heterotrimeric G-protein, which can then undergo further cycles of activationdeactivation with signal amplification. Different forms of Pand -y-subunits also exist, but the primary role of the more lipophilic Py complexes, or myristoylated a-subunits, (25) might be to anchor the G-protein in the membrane, possibly adjacent to a particular receptor subtype.
Several properties of the G-proteins provide useful tools for investigating their functional role in cellular processes. Nonhydrolyzable GTP analogues such as GTPyS cause a persistent activation of G-proteins, but GDP,S, which mimics GDP, is inhibitory. Another property of the G-protein a-subunits is their susceptibility to ADP-ribosylation by bacterial toxins. Cholera toxin catalyses ADP-ribosylation at an arginine site of several G-proteins, and pertussis toxin causes ADP-ribosylation at a cysteine site, located near the carboxyl terminus, and uncouples the G-protein from the receptor. The a-subunits of different G-proteins may be ADP-ribosylated by cholera or pertussis toxins (18). Notably, the G-protein responsible for stimulating adenylate cyclase (Gs) is a substrate for cholera toxin, while the G-protein associated with inhibition of adenylate cyclase (Gi), and a G-protein prevalent in brain (Go) are substrates for pertussis toxin.
The original evidence leading to the conclusion that G-proteins are involved in receptor coupling of Ca2+mobilizing agonists to PIP2 breakdown was reviewed recently (12,19,26,27). The basic observations are that G-protein activating agents such as GTP, GTPyS or AlF3 (which mimics GTP binding) when added to permeabilized cells or plasma membrane preparations cause an activation of phospholipase C and a decrease of agonist binding affinity. Further evidence is provided by a) the enhancement of phospholipase C activation by suboptimal concentrations of GTPyS in the presence of agonist, b) an agonist-induced stimulation of GTPase activity, c) the inhibition of phospholipase C activation by GDP,S, d) an inhibition of agonist-induced activation of phospholipase C by toxins (some cell types only), and e) the retention of receptor-G-protein complexes (28)(29)(30) or G-proteinphospholipase C complexes (31) during purification. Current evidence suggests that a variety of different types of G-proteins are responsible for regulating receptor coupling to phospholipase C in different cell types. This was first suggested by the fact that agonist-mediated stimulation of inositol phosphate formation and Ca2' mobilization could be inhibited by pertussis toxin pretreatment in some cell types, but not others [for references and details of cell types see (26)]. Ca2' mobilization induced by activation of different receptors can also be selectively inhibited by pertussis toxin pretreatment in the same cell type. In hepatocytes, epidermal growth factor (EGF) responses are pertussis-toxin sensitive, whereas those to vasopressin or angiotensin II are not (32). In platelets, pertussis toxin inhibits thrombin but not U46619 (a thromboxane A2 analogue) responses (33). Further, heterogeneity in the type of G-protein able to couple to phospholipase C is suggested by the ability of cholera toxin to inhibit the effects of Ca2'-mobilizing agonists in a variety of cell types (27,(34)(35)(36)(37). This effect of cholera toxin was independent of ADP-ribosylation of G. and changes of cyclic AMP.
A number of novel G-proteins that may be involved in coupling to phospholipase C have been identified in various cells. In neutrophils and human leukemia HL-60 cells (30,(38)(39)(40), a pertussis toxin-sensitive 40 kDa a-subunit has been purified and shown to be immunologically distinct from the a-subunits of Go and Gi. The copurification of this 40 kDa polypeptide with the chemoattractant receptor suggests that it is responsible for activation of phospholipase C (30). This same 40 kDa a-subunit from HL60 cells is ADP-ribosylated by cholera toxin in the presence of the Ca2'-mobilizing chemotatic peptide fMet-Leu-Phe, as well as by pertussis toxin (34,35).
A pertussis toxin-sensitive 40 kDa a-subunit has been isolated from brain (41) and shown to be structurally and immunologically similar to Gia40 from neutrophils and HL-60 cells (21,22). Reconstitution studies performed by adding Gi or Go prepared from brain to plasma membranes from HL-60 cells previously treated with pertussis toxin showed that they were equally effective in restoring fMet-Leu-Phe plus GTP-stimulated formation of inositol phosphates (42). Addition of Go or Gi also restored bradykinindependent activation of GTPase and inhibition of adenylate cyclase in membranes from NG108-15 cells treated with pertussis toxin (43). Furthermore, Go or Gi stimulated the activity of partially purified phospholipase C from platelet membranes (44). These results suggest an apparent promiscuity of G-protein coupling to phospholipase C, but since the preparations of Go or Gi used in these experiments probably contained Ga40, it seems more likely that this G-protein, which also stimulates phospholipase C activity (44), provides specificity for G-protein-phospholipase C coupling.
A different pertussis toxin target protein with an apparent molecular mass of 43 kDa has been partially purified from human erythrocytes (45). This protein has the properties expected for a G-protein a-subunit and is also present in membranes from brain, GH4C1 pituitary cells, leukocytes, and liver (45). A 54 kDa protein that was ADP-ribosylated by pertussis toxin and that copurified with an inositol lipid specific phospholipase C has been identified in the cytosolic fraction of thymocytes (31), while a pertussis toxin insensitive GTP-binding protein was found associated with purified vasopressin receptors in rat liver (28,29). Some indirect evidence suggests the possibility that an unidentified pertussis toxin sensitive Gprotein may be involved in inhibiting phospholipase C (32,46,47). Several members of a group of G-proteins with Mr values betwen 20,000 and 25,000, including the well-characterized p21 ras oncogene product (48), have been purified but their possible function in relation to inositol lipid metabolism has not been elucidated (see 49).
In summary, growing evidence suggests that several different G-proteins are responsible for activating phospholipase C. A number of novel G-protein asubunits have been identified by protein purification, and multiple forms of Gi-and Go-like a-subunits with high-sequence homology have been identified by cDNA cloning (20,21) and by reaction with different antibodies (22,50,51). These studies also indicate that the Gia41 that couples with receptors to inhibit adenylate cyclase activity is unlikely to activate phospholipase C. A new G-protein a-subunit has been identified from molecular cloning of cDNA, which is Gi-like but has a lower sequence homology than other known G-protein a-subunits. This subunit is particularly interesting because it lacks the cysteine site at the carboxyl terminus for ADP-ribosylation by pertussis toxin (52,53). If this G-protein is found to be widely expressed, it may account for pertussis toxin insensitivity of receptor coupling to phospholipase C observed in many tissues.

Phospholipases as G-Protein Effector Enzymes
Several inositol lipid-specific phospholipase C enzymes have been found in both the soluble and plasma membrane fractions of the cell (53). Membrane-bound phospholipase C has been purified after detergent solubilization from brain, with a molecular mass of 150 to 154 kDa (55,56); 61 kDa and 62 kDa enzymes have been purified from platelets (57) and uterus (58), respectively. Molecular cloning and determination of the complete amino acid sequence has been achieved for a number of phosphatidylinositolspecific phospholipase C enzymes from brain (59)(60)(61)(62) and uterus (63), which show a remarkable lack of sequence homology. Generally, these enzymes will hydrolyse phosphatidylinositol (PI), phosphatidylinositol 4-phosphate (PIP), and PIP2 at high Ca2" concentrations, but they are more active toward PIP and PIP2 at CA2+ concentrations below 10 gM. Forms of phospholipase C, apparently specific for PI and not for polyphosphoinositides, have also been recognized (64,65). As shown in intact membrane preparations from a number of cell types, the major effect of activated G-protein on PIP2-specific phospholipase C activity is to decrease the Ca21 concentration required for enzyme activity to the physiological level of about 0.1 iM (66)(67)(68)(69). In platelets, where most of the phospholipase C activity is recoverable from the soluble cell fraction, the partially purified enzyme is stimulated by GTPyS (70)(71)(72). In contrast, soluble phospholipase C from liver is not stimulated by GTPyS (69). At present, it remains an open question whether or not G-protein a-subunits in some tissues may dissociate from the membrane after receptor activation and, subsequently, interact with intracellular effector enzymes (73).
Some tissues also contain a phospholipase C specific for phosphatidylcholine, which appears to be coupled to receptors by an unidentified G-protein (74,75). Other types of phospholipase C specific for a PI-glycan structure involved in membrane-anchored proteins and insulin action have been described [see (76) for review]. Since diacylglycerol is produced by the action of all the different types of phospholipase C, it is evident that protein kinase C can be activated in the absence of Ins 1,4,5-P3-induced Ca21 mobilization.
G-proteins also control the hydrolysis of other lipids by coupling to phospholipase D (77,78) and phospholipase A2 (79)(80)(81), which produce phosphatidic acid and arachidonic acid, respectively. Of the range of products from phospholipid hydrolysis, only Ins 1,4,5-P3 is firmly established as a physiological messenger for Ca2' release, although future studies may clarify the roles postulated for arachidonic acid (82) and phosphatidic acid (77) as Ca2+-releasing agents.
However, further metabolism of arachidonic acid to eicosenoids such as thromboxane A2 (33) and the C6sulfidopeptide leukotrienes (83) causes Ca21 mobilization by their receptor coupling to phospholipase C. It thus appears that cells contain a heterogeneity of phospholipase effector enzymes as well as G-proteins, which further suggests the possibility that chemical signaling from different receptors is uniquely channeled to specific effector enzymes.

Regulation of Ins 1,4,5-P3 Production
In the intact cell Ins 1,4,5-P3 is formed together with small amounts of the 1,2 cyclic isomer only by phospholipase C-mediated hydrolysis of PIP2 (see 84 for review). The amount of PIP2 that is available for hydrolysis is only sufficient to maintain Ins 1,4,5-P3 production for a few minutes. However, this hormone-sensitive PIP2 pool is rapidly replenished via PIP from part of the much large PI pool by the action of PI and PIP kinases (11,12). These enzymes exist in multiple forms, and there is evidence that some types are activated by tyrosine kinases (85) or by protein kinase C (86). The large PI pool and the high activity of the enzymes synthesizing PIP2 mitigates against the regulation of Ins 1,4,5-P3 production by a limitation of substrate availability.
The production of Ins 1,4,5-P3 is subject to several kinds of feedback regulation. A rapid inhibition of agonist-induced Ins 1,4,5-P3 production and Ca2" signaling is observed in many cell types by phorbol ester-induced activation of protein kinase C (3,10,12,79). This interaction can either be at the receptor or a post-receptor site. Some receptors, notably a1-adrenergic and EGF receptors, are phosphorylated by protein kinase C with decreased agonist binding and prevention of agonist-mediated signal transduction to the G-proteins. (87). Phosphorylation of the a1-adrenergic receptor by agonists has been demonstrated in DDT1 MF-2 smooth muscle cells (88). The purified a1-adrenergic receptor is also phosphorylated by protein kinase C (89). Receptor phosphorylation, however, does not seem to be a general mechanism for phorbol ester (PMA)-induced inhibition of agonist effects. Thus, an additional post-receptor site for protein kinase C-mediated interaction is suggested from studies with plasma membrane preparations. These showed no effect of PMA pretreatment on agonist binding affinity nor the ability of GTP analogues to suppress high-affinity agonist binding (90), but they did show an inhibitory effect on GTPyS-stimulated hydrolysis of PIP2 (90,91). Since the production of Ins 1,4,5-P3 as well as Ca21 mobilization is inhibited by the activation of protein kinase C, the site for the post-receptor interaction is thought to be at the G-protein-phospholipase C complex. Although several G-protein a-subunits (notably Gia41) have been shown to be phosphorylated by protein kinase C (92,93), clear evidence for the phosphorylation of an a-subunit functionally coupled to phospholipase C is presently lacking. Purified pholpholipase C from uterine smooth muscle can be phosphorylated by protein kinase C, but the ability to hydrolyze PIP2 was unaffected (58). Resolution of the different mechanisms responsible for regulating Ins 1,4,5-P3 production by activating protein kinase C will undoubtedly be aided by isolating and reconstituting functional receptor-G-protein phospholipase C complexes and using selective protein kinase C inhibitors in intact cells.

Intracellular Ca2+ Mobilization
The Hormone-Sensitive Ca2`Pool Recent evaluations of the tissue distribution of total Ca2+ between the cytosol and the different organelles have led to the conclusion that the bulk of the sequestered Ca2+ is in the endoplasmic reticulum (105,106). It is also generally accepted that this pool represents the source of the Ca2+ released by Ca2+-mobilizing agonists. However, the endoplasmic reticular structures are themselves heterogeneous, as are the pools of sequestered Ca2". This is revealed by the fact that, in numerous studies after adding agonists to intact cells or Ins 1,4,5-P3 to permeabilized cells and isolated microsomal preparations, only a fraction of the sequestered Ca2" is released (10). Recent studies have, in fact, suggested that a morphologically discrete organelle termed the calciosome, which generally copurifies with endoplasmic reticulum membranes, may represent the source of hormone releasable Ca2" in nonmuscle cells (107). In some cells, such as smooth muscle (108), and Limulus photoreceptors (109), the sites for Ca2" release may be adjacent to the plasma membrane. In Xenopus oocytes, both shallow and deep injections of Ins 1,4,5-P3 caused a local release of Ca2 , suggesting that Ins 1,4,5-P3sensitive Ca2`pools are distributed throughout the cell (110,111). For most cell types, however, the distribution and morphology of Ins 1,4,5-P3-sensitive Ca2" pools have not been established. The mitochondria are not involved in Ins 1,4,5-P3-mediated Ca21 release, but respond to the increased cytosolic free Ca2' by a net uptake of Ca2`and an increase of the matrix-free Ca2+ concentration with consequent activation of mitochondrial dehydrogenases (e.g., a-ketoglutarate dehydrogenase) and respiration (112,113). Sequestration of Ca2" by the endoplasmic reticulum occurs by an ATP-driven Ca2" pump. A high-affinity Ca2+-ATPase with a molecular mass of 116 kDa and a Km for Ca21 of 0.1 to 0.2 jM has been purified from liver (114) and other tissues (106). This enzyme belongs to the class of E1P.E2P-ATPases and is biochemically, kinetically, and immunologically similar to the enzyme in skeletal muscle sarcoplasmic reticulum (115). In some cell types it is also calmodulindependent (106). As with sarcoplasmic reticulum, the Ca21 sequestered by the endoplasmic reticulum is in equilibrium between free and bound forms. It is of considerable interest that calsequestrin (the major Ca2+-binding protein of sarcoplasmic reticulum with a Kd for Ca21 of 100 gM) or a similar protein has recently been detected in the endoplasmic reticulum of nonmuscle cells (115), including the calciosome (107). The total amount of Ca2+-binding protein in the vesicular compartments presumably provides a limit to the amount of Ca21 that can be accumulated. The free Ca21 concentration in the endoplasmic reticulum has not been measured directly, but is probably two to three orders of magnitude higher than that in the cytosol, in order to provide a suitable electrochemical gradient for rapid Ca2' efflux via opening of the Ins 1,4,5-P3-sensitive Ca21 channel. The rate of Ca21 uptake into liver microsomes is inhibited by the intravesicular Ca21 concentration with an apparent Ki of 250 to 300 jM (116), presumably by inhibiting the dissociation of Ca2+ from the phosphoenzyme. This effect probably accounts for the stimulation of MgATP-dependent Ca2+ accumulation into liver microsomes by glucose 6-phosphate, which on hydrolysis with liberation of inorganic phosphate in the lumen, augments the intravesicular Ca2"-buffering capacity (117).
The endoplasmic reticulum Ca2" pump of pancreatic acinar cells has been shown to be activated after hormonal stimulation with a two fold increase of Vmax for Ca2" uptake and a decrease of the Kd for Ca2" from 0.26 to 0.09 tM (118). Evidence obtained with saponized platelets suggests that the activation of protein kinase C may account for a similar stimulation of Ca2" sequestration after thrombin addition (119). If the Ins 1,4,5-P3-sensitive Ca21 pool is distributed throughout the cell in discrete vesicular compartments, it is expected that with submaximum agonist stimulation, there would be a heterogeneous population of Ca2"-depleted and nondepleted vesicles, depending on their spatial location and possible gradients of Ins 1,4,5-P3 concentration within the cell.
Therefore, an activation of the Ca2' pump by either a Ca2+/calmodulin mechanism or by a protein kinase Cmediated phosphorylation may augment the Ca2" content of both Ins 1,4,5-P3-sensitive and Ca2"-insensitive pools. Since the activation of the Ca21 pump by the protein kinase C mechanism could persist after the initial Ca21 transient, the Ca2" pools may be refilled from the extracellular Ca2' pool without a simultaneous increase of the cytosolic free Ca2 , as occurs when an agonist is displaced from its receptor (120-124).

Ins 1,4,5-P3-Induced Ca2" Release
The ability of Ins 1,4,5-P3 to release Ca21 from intracellular, nonmitochondrial, vesicular stores appears to be a ubiquitous property of cells. Its role in activating many hormonal responses and in stimulussecretion coupling is firmly established (10). It is also involved in excitation-contraction coupling in smooth muscle, but whether Ins 1,4,5-P3 is similarly involved in skeletal muscle is doubtful (125). The amount of Ca2' released by maximal concentrations of Ins 1,4,5-P3 is variable with different cell types, being 20 to 50% of the total cellular Ca2' content (10,126). However, the increase of the cytosolic free Ca21 is limited to about 1 gM because of the presence of a variety of Ca2+-binding proteins in the cytosol (106). The Ca21buffering capacity in the cytosol of neutrophils, for instance, has been estimated to be 0.76 mM, with an average dissociation constant of 0.55 jM (127). The peak increase of the cytosolic free Ca21 occurs within a few seconds of its initiation.
Except for one report (123), the formation of Ins 1,4,5-P3 has been shown to occur prior to or simultaneously with the increase of cytosolic free Ca2' after addition of agonists to intact cells (10,12). It is important to note, however, that the density of certain receptor types in a given cell may be so high that agonist stimulation causes an accumulation of Ins 1,4,5-P3 far in excess of the amount required for a maximal release of Ca2". Thus, with vasopressin stimulation of of hepatocytes, the peak increase of Ins 1,4,5-P3 accumulation occurs after the peak of the cytosolic free Ca2" (12), whereas in other cells [e.g., A431 carcinoma (129)], the accumulation of Ins 1,4,5-P3 is more transitory and appears to correlate with a transient rise and fall of the cytosolic free Ca2". Studies using a stopped-flow measurement of the cytosolic free Ca2" with the fluorescent Ca2" indicator Fura-2 after agonist addition in parotid acinar cells (130) (144), while a monoclonal antibody raised against platelet microsomal proteins inhibited thrombin (plus GTP) and Ins 1,4,5-P3-induced Ca2' release in mixed dense tubular system vesicles and plasma membranes from platelets (145). As studied in permeabilized cells, Ins 1,4,5-P3-mediated Ca2" release does not become desensitized (126,146). In preparations of microsomes from some tissues, notably liver (141), addition of Ins 1,4,5-P3 produces a negligible release of Ca2', although a large fraction of the sequestered Ca2" can be released by Ins 1,4,5-P3 in microsomal fractions from other tissues (146)(147)(148) and saponin-permeabilized hepatocytes (149). Dawson (150) first showed that the Ins 1,4,5-P3-stimulated Ca2" release by liver microsomes was greatly enhanced by an addition of polyethylene glycol and low concentrations of GTP. It subsequently became evident that GTP itself promoted Ca2' release with variable facilitation by polyethylene glycol and by a separate mechanism from Ins 1,4,5-P3-induced Ca2" release (137,(151)(152)(153)(154). GTP has been reported to promote fusion between microsomal vesicles, which correlated with the rate of GTP-mediated Ca2" release (151). On the other hand, Nicchitta et al. (153) showed that GTP induced a nonspecific permeability of the liver microsomal membrane. The GTP effect on Ca21 release in liver microsomes is associated with a polyethylene glycol-promoted increase of GTPase activity (155), suggesting the involvement of a GTP-binding protein. In contrast to the effects of GTP-yS on Gproteins involved in receptor signaling, GTP-yS inhibits the GTPase activity and GTP-mediated Ca2" release (155).
Experiments with permeabilized neuronal and smooth muscle cell lines showed that Ins 1,4,5-P3 released Ca2" from the same pool as GTP, while GTP also released Ca2" from an additional nonmitochondrial pool (156). In further studies, Gill and his colleagues suggested that uptake of Ca21 into the Ins 1,4,5-P3-sensitive Ca2' pool is controlled by a GTPregulated Ca21-translocating mechanism (157). In permeabilized hepatocytes, under certain conditions GTP enhanced the Ins 1,4,5-P3-induced Ca2" release (158). At present, the physiological significance of the GTP effect is not clear. Although it is unlikely that GTP activates a specific Ca2" channel, it is possible that through the mediation of an intracellular GTPbinding protein, the GTP may unmask Ins 1,4,5-P3binding sites, increase the size of the Ins 1,4,5-P3sensitive Ca2' pool, or be involved in the translocation of the calciosomes closer to the plasma membrane where Ins 1,4,5-P3 is generated.
In a number of studies using membrane fractions derived from liver, adrenal cortex, macrophages, neutrophils, anterior pituitary, and brain (159,160) the density and affinity of Ins 1,4,5-P3 binding sites have been characterized. Several reports indicate that plasma membrane-enriched fractions of liver (159) and platelets (161) contain a higher density of Ins 1,4,5-P3-binding sites than do fractions derived from the endoplasmic reticulum. These findings are consistent with the proposal that Ins 1,4,5-P3-binding and Ca2+ release occur in specialized organelles (calciosomes) that are not physically or functionally part of either the plasma or endoplasmic reticular membranes. The specificity of different inositol phosphates to displace radiolabeled Ins 1,4,5-P3 from these binding sites is similar to their relative potency in causing Ca2`release, namely Ins 1,4,5-P3 > Ins 2,4,5-P3> Ins 4,5-P2 > Ins 1,4-P2, Ins P5, Ins P6. Other inositol phosphates such as Ins 1,3,4-P3 and Ins 1,3,4,5-P4 were 1 to 5% as effective as Ins 1,4,5-P3 in causing a displacement of bound Ins 1,4,5-P3. The number of binding sites for Ins 1,4,5-P3 is higher in the cerebellum, compared with other parts of the brain (162), and is about 100 times greater than that observed in peripheral tissues. The half-maximum binding (40-80 nM) in cerebellar membranes (160) is also higher than values of 1 to 10 nM reported in other tissues (159). Thus, there is an apparent, large discrepancy between these low values for half maximum Ins 1,4,5-P3 binding and the half maximum concentration of Ins 1,4,5-P3 (0.1 to 2 FM) required to release Ca2+ from permeabilized cells or isolated microsomes (10). Several factors are known to affect the Ins 1,4,5-P3 binding affinity that may account for this apparent discrepancy. Notably, Ins 1,4,5-P3 binding is lower at physiological pH than at the more alkaline pH values typically used for binding studies (160) and is also inhibited by ATP (159). Furthermore, the transient nature of the Ca2' release elicited by Ins 1,4,5-P3 in permeabilized cells and membrane preparations reflects its further metabolism to relatively inactive products, so that the Ins 1,4,5-P3 concentrations present in the medium might be overestimated.
The Ins 1,4,5-P3 receptor has recently been purified from rat cerebellum using a heparin affinity column (163). This solubilized membrane protein was shown to have a molecular mass of 260 kDa by SDS-PAGE and to bind Ins 1,4,5-P3 with a Kd of 100 nM at pH 8.3. In contrast to studies with cerebellar membrane fractions, where Ca2+ was shown to inhibit Ins 1,4,5-P3 binding with a half-maximal effect at 300 nM (160), the binding of Ins 1,4,5-P3 to the purified receptor was unaffected by Ca21 (163). The Ca2+ sensitivity is apparently conferred by a separate protein present in solubilized brain membranes having an estimated Mr of 300,000 by gel filtration (164). The effect of Ca21 in cerebellar membranes is to decrease the affinity of Ins 1,4,5-P3 binding to the receptor (165). An increase of extravesicular Ca2' in the range from 1 to 10 jM has been shown to inhibit Ins 1,4,5-P3-induced Ca2+ release in permeabilized neuronal cells and microsomes (137,166). In studies with cerebellar microsomes, the concentration of Ca2+ required for inhibition was lower at suboptimal Ins 1,4,5-P3 concentrations (165). Feedback inhibition by Ca2+ to decrease the affinity of Ins 1,4,5-P3 binding to the receptor might provide a mechanism for generating oscillations in cytosolic Ca2" concentration without changes in Ins 1,4,5-P3 concentration. (See later section.) Ca2" Efflux from the Cell An early event following stimulation of cells by Ca2"-mobilizing agonists is an increased efflux of Ca2", with a consequent fall in the total cellular Ca2" content (1,2,16). Evidence with hepatocytes has indicated that the onset of net Ca2" efflux is delayed by 5 to 10 sec after agonist addition, indicating that there is an initial inhibition of the plasma membrane Ca2+ efflux pump (167). The mechanism of this effect has not been ascertained, but it may relate to the altered inositol lipid environment of the Ca2+-ATPase (168) or to a direct inhibitory effect of Ins 1,4,5-P3 (169). Subsequently, however, this inhibition is overcome so that Ca2+ efflux is stimulated above the resting rate. This can be attributed partly to the Ins 1,4,5-P3induced increase of cytosolic free Ca2 , which stimulates the calmodulin-dependent plasma membrane Ca2+-ATPase and the electrogenic Na+/Ca2+ exchanger. The latter system is present in the plasma membrane of many cells (170). The removal of extracellular Na+ has been shown to inhibit agoniststimulated Ca2`efflux in some cells such as arterial smooth muscle (171), but not in others, such as platelets (172). However, in most cells, the efflux of Ca2' by the Ca2+-ATPase probably predominates. A net loss of Ca21 from the cell occurs over the first few minutes until there is a balance between the rate of Ca2' efflux and the agonist-stimulated rate of Ca2+ influx (122). The loss of Ca21 from the Ins 1,4,5-P3-sensitive Ca21 pool makes the cell unresponsive to further Ca21 mobilization by additional agonist stimulation until the pool is refilled from the extracellular medium.
The Ca2+ efflux rate is not solely dependent on the cytosolic free Ca21 concentration since the rate of exchange of Ca2' across the plasma membrane is increased to 2-to 4-fold during the sustained phase of the Ca2+ transient when the cytosolic free Ca2" is only slightly elevated above control levels (1). Studies of a variety of cells have indicated that activation of protein kinase C by phorbol esters stimulates Ca2' efflux (172)(173)(174). The mechanism of this effect has been investigated by studies with the erythrocyte Ca2+-ATPase, which showed that calmodulin increased the Vmax 14-fold with a 4-fold decrease of the Km to 0.3 gM, while the addition of activated protein kinase C independently increased the Vmax by 5to 7-fold without changing the Km (175).
A Ca2+-independent activation of the Ca2+-ATPase by protein kinase C may thus account for the higher Ca2+ efflux rate in cells after prolonged agonist stimulation compared with control cells, despite the small elevation of cytosolic free Ca2+. This dual regulation of the Ca2+-ATPase by Ca2+/calmodulin and by protein kinase C, together with different contributions of the Na+/Ca2" exchanger to Ca2" efflux, probably accounts for the many detailed differences in the relative duration and shape of the Ca2`transient observed upon addition of different agonists to various cell types.

Extracellular Ca2+ Mobilization
A sustained hormonal response requires an influx of extracellular Ca2`to maintain an elevated cytosolic free Ca2+ concentration. Evidence for an agonist dependent stimulation of Ca2" influx is based on the dependence of the sustained phase upon extracellular Ca2' and an increase of unidirectional 45Ca2`influx (1,2). The temporal relationship between intracellular Ca2+ release and stimulated influx appears cell specific, since rapid kinetic measurements show that the onset of stimulated Ca2`influx occurs at least as rapidly as intracellular Ca2" release in parotid cells (130) and platelets (131). In A10 smooth muscle (132) and GH3 pituitary (176)  There is a 10,000-fold concentration gradient of Ca2+ across the plasma membrane, which requires a strict regulation of its permeability to Ca2 . Several types of voltage-sensitive Ca2`channels that open with membrane depolarization have been well characterized (177). In some secretory cells-notably GH3 pituitary (176) and adrenal glomerulosa cells (121)-Ca2+ entry by voltage-sensitive Ca2+ channels may contribute towards the sustained phase of the agonist-induced Ca2+ transient. However, in other cells such as hepatocytes (178), the response to Ca2+ mobilizing agonists may be associated with a hyperpolarization because of a secondary Ca2+ activation of K+ and Clchannels. Alternatively, agonistinduced Ca2" entry is considered to be regulated by a process usually referred to as a receptor-operated Ca2`channel. This entry of Ca21 is only inhibited by agents that block voltage-sensitive Ca2`channels when they are added at very high concentrations (120). In fact, several mechanisms may mediate activation of Ca2`channels by receptor occupation. One mechanism is by direct ligand activation, which appears to occur with the ATP-activated high conductance cation channel in smooth muscle (179). Another mechanism is by receptor-activated G-proteins, which have been demonstrated to regulate several types of K+ and Ca2`channels in a variety of cells (reviewed in 180). Evidence consistent with a role for a G-protein in activating receptor operated Ca2+ channels is suggested from the observations that pretreatment with pertussis toxin inhibited Ca2+ influx, which was induced by vasopressin in hepatocytes (181) and by angiotensin II in adrenal glomerulosa cells (182), but did not affect intracellular Ca2+ mobilization. A nonselective cation channel activated by a rise of cytosolic free Ca2+ has been reported (183,184), but in most cells the onset of Ca21 influx does not lag behind intracellular Ca2`mobilization. Stimulated Ca2+ influx persists after the initial Ca2" transient, and it is dependent on receptor occupancy, rather than on the cytosolic free Ca2' concentration (120, 121,123,185).
An attractive possibility that has generated considerable attention is that Ins 1,4,5-P3, itself, or one of its metabolic products may be responsible for agoniststimulated Ca2" influx. Ins 1,4,5-P3-induced Ca2" release has been demonstrated in Na+-loaded plasma membrane vesicles from platelets (161). Mitogenic stimulation of T lymphocytes has been shown to be associated with activation of a low conductance voltage-independent Ca2" channel in the cell-attached patch, which could be directly activated by Ins 1,4,5-P3 after excision of the patch (186).
In summary, it must be concluded that there is insufficient evidence to deduce the mechanism for receptor-dependent Ca2+ entry into cells, although it is likely that different mechanisms may predominate in different cell types. Whether Ins 1,3,4,5-P4 or another inositol phosphate are involved remains an open question. Specific Ins 1,3,4,5-P4 binding sites have been demonstrated in membranes of several tissues that are distinct from Ins 1,4,5-P3 binding sites (193,194), but their function has not yet been ascertained.

Oscillations in Cytosolic Free Ca2+
In contrast to the many studies with bulk cell suspensions where agonist-induced Ca2+ transients are typically biphasic, the development of methods for measuring changes of Ca2' in single cells has revealed that the response may be oscillatory. Measurements with the bioluminescent protein, aequorin, in single hepatocytes stimulated with agonist showed a series of periodic increases in cytosolic free Ca2' to above 600 nM, each returning to resting levels within 7 sec (195,196). The frequency of these oscillations was a function of phenylephrine concentration, with a period varying from 20 to 240 sec. Low concentrations of vasopressin and angiotensin II, which produce nonsaturating amounts of Ins 1,4,5-P3, caused similar oscillations, suggesting that the occurrence of Ca2+ oscillations is favored when tissue Ins 1,4,5-P3 concentrations are low. In contrast, Fura-2 loaded hepatocytes rarely exhibit Ca2' oscillations after addition of agonists (132). However, in other cell types, oscillations of the cytosolic free Ca2' in a proportion of cells have been observed using Fura-2 as the Ca 2+ indicator. For example, BC3H-1 cells stimulated with phenylephrine exhibited a burst of Ca2`spikes that ceased after several minutes (197), low concentrations of angiotensin II-elicited Ca2' oscillations in adrenal glomerulosa cells (133) and macrophages induced to spread onto a suitable substratum (frustrated phagocytosis) showed Ca2' oscillations (198). In parotid acinar cells a damped series of oscillations could be evoked by carbachol, but in a concentration-independent manner (199); rat peritoneal mast cells stimulated with antigen responded variably with one, two, or more spikes of increased Ca2' (200). Similarly, the addition of carbamylcholine to insulinoma HIT (T-15) cells loaded with Fura-2 produced a series of characteristic but damped Ca2`oscillations in each responding cell (201). In general, measurements of Ca2`with Fura-2 show a more transient, irregular pattern of Ca2' fluctuations than those obtained with aequorin and with a smaller proportion of cells generating oscillations.
The difference between the data obtained with aequorin and Fura-2 even in the same cell type, suggest that they may arise from the different properties of the Ca2' indicators. Since Fura-2 is a Ca2+ chelator, it is possible that it diminishes the peak rise of the Ca2' transients, which may be part of a negative feedback interaction responsible for generating the oscillations. On the other hand, the nonlinear relationship betwen aequorin bioluminescence and Ca2' concentration may amplify spatially localized oscillations of Ca2'. Thus, a series of transient, localized increases in Ca2' might be detected by aequorin as large, transient spikes, decaying rapidly as Ca2+ diffuses into the remainder of the cytosol, whereas Fura-2 might not detect such a response because of its lower sensitivity to local changes of the Ca2`concentration. However, sustained Ca21 oscillations have been observed in oocytes after fertilization using either Ca2+-selective microelectrodes (202) or aequorin to measure the Ca2c hanges (203).
Indirect evidence for oscillations of the cytosolic free Ca21 in response to Ca21-mobilizing hormones has been obtained from electrophysiological measurements of the oscillatory behavior of Ca2"-activated K+ and Cl-channels. In HeLa cells, histamine stimulated a pattern of Ca2+-dependent channel activation, characterized by a series of bursts, which were separated by silent periods. These correlated with membrane potential oscillations and occurred in the absence of extracellular Ca21 (204). Similarly, a1-adrenergic stimulation of voltage-clamped guinea pig hepatocytes induced oscillatory membrane conductance increases to K+ and Cl-, which were attributed to Ca2+-activated K+ and Cl-channels (205). A close correlation between oscillatory Ca2"-activated K+ channel activity and intracellular Ca2' has been reported in hamster oocytes (202), with similar responses being observed after injection of GTPyS (206).
Mechanisms for the generation of Ca21 oscillations appear to fall into two categories: oscillations that occur secondary to spontaneous action potentials, as in some secretory cells (207), and oscillations induced by Ca21-mobilizing hormones. The latter can be initiated in the absence of extracellular Ca2 , indicating that intracellular Ca2' release is the primary source of Ca2+, although extracellular Ca2' is required for their maintenance (197,198,204,205). A minimum oscillating system requires a feedback loop; for sustained oscillations, some delay step is required to generate the periodicity (208). Theoretically, oscillations in the Ins 1,4,5-P3 concentration could be produced by a negative feedback effect on the receptor or the Gprotein coupled to activation of phospholipase C by protein kinase C. The delay step could be caused by a slow rate of activation of the G-protein or of protein kinase C by diacylglycerol. However, such a mechanism involving a periodically activated protein kinase C is difficult to reconcile with the decrease in the frequency of the Ca21 oscillations observed by addition of exogenous diacylglycerol or phorbol ester to hepatocytes (209). Furthermore, in guinea pig hepatocytes, oscillations in Ca2+-dependent ion conductances could be induced by an injection of Ins 1,4,5-P3, suggesting that changes in the Ins 1,4,5-P3 concentration are not required (205). A possible mechanism for negative feedback is suggested by the inhibitory effect of Ca2+ on Ins 1,4,5-P3 binding to its receptor, as described earlier. Also, the time required for reaccumulation of Ca21 into the Ins 1,4,5-P3-sensitive pool could provide the delay to generate a characteristic periodicity.
In summary, there is good evidence that some oscillatory Ca2+ response occurs in stimulated cells. Oscillations are only seen in single cells, since individual cells in suspensions respond asynchronously. Possible advantages of a periodic frequency-modulated signal, compared to a graded amplitude-modulated signal, include protection from desensitization or preventing mitochondrial Ca21 overload (196). An alternative possiblility is suggested by an analogy with neural transmission where a self-propagating action potential allows transmission of information over distances. There is considerable evidence that cells are in communication with each other through gap junctions (210), which is also suggested by the synchronization of oscillations in the intact perfused liver (211). If oscillations at low frequency induced in one cell facilitate responses in adjacent cells, the overall tissue response would be increased. Should such a mechanism exist, the regulation of gap junction permeability by a variety of agents (210) would enable fine modulation of the sensitivity of the tissue response.

Summary
The involvement of inositol lipid metabolism in agonist-mediated Ca2" signaling by Ins 1,4,5-P3 has become firmly established. Recent advances have led to a better understanding of the proteins associated with signal transduction in the plasma membrane. A number of specific receptors, G-proteins, phospholipases, and inositol lipid kinases have now been purified and characterized. An Ins 1,4,5-P3 receptor has also been purified, which is presumably involved in mediating the Ca2" efflux from intracellular stores. The morphological site of the hormonesensitive Ca2" pool has been tentatively identified as discrete, specialized intracellular structures (calciosomes); however, further studies are required to demonstrate that these contain Ins 1,4,5-P3-gated Ca2" channels. Receptor occupancy by Ca2"-mobilizing agonists also stimulates Ca2" entry into the cell, but the mechanism for the activation of voltage-insensitive Ca2" channels and the possible involvement of Ins 1,4,5-P3, Ins 1,3,4,5-P4 and/or G-proteins in this process has not been established.
The Ca2" signaling pathway is subject to multisite feedback regulation by Ca2", itself, and by a diacylglycerol-mediated activation of protein kinase C.
Potential sites for Ca2" interaction are displacement of Ins 1,4,5-P3 from its receptor by a Ca2"-dependent mechanism, promotion of Ins 1,3,4,5-P4 formation by the Ca2+/calmodulin-regulated Ins 1,4,5-P3 3-kinase and efflux of Ca2" from the cell, or sequestration into intracellular Ca2" stores by Ca2+/calmodulin-regulated Ca2+-ATPases. Protein kinase C activation potentially affects the rate of generation of Ins 1,4,5-P3 by negative feedback to the receptor-G-proteinphospholipase C transduction system and possibly also the rate of Ins 1,4,5-P3 removal by activation of an inositol polyphosphate 5-phosphomonoesterase and/or Ins 1,4,5-P3 3-kinase. It may also attentuate the Ca2" transient directly by increasing the activity of Ca2+-ATPases associated with the plasma membrane and the endoplasmic reticulum. Cell-to-cell heterogeneity in the relative control strengths of these different mechanisms may explain the differences in the Ca2" signal in different tissues and even in different cells within a population. The ability of Ca2" and protein kinase C to provide negative feedback at various points in the signal transduction pathway suggests that a complex mechanism involving multiple feedback loops is likely to regulate the generation of Ca2" oscillations seen in some cells. The precise interactions between these feedback mechanisms and the crosstalk with other intracellular signaling pathways enables each cell to respond in an appropriate and unique manner.