Mechanisms of Intracellular Calcium Release During Hormone and Neurotransmitter Action Investigated with Flash Photolysis

There are many tissues in which hormone or neurotransmitter action on certain G-protein-coupled surface receptors results in the release of Ca²⁺ from intracellular stores by InsP₃. The rise in free [Ca²⁺] activates physiological processes, such as ion channel opening, contraction or secretion, or metabolic processes, such as glycogenolysis. This process has been best described in non-excitable tissues, such as liver or exocrine cells. In general, it has a characteristic time course, showing delays that can be many seconds after receptor activation and periodic fluctuations of the intracellular Ca²⁺ concentration when hormonal stimulation is maintained. The complex kinetics suggests the presence of both regenerative and inhibitory interactions in the second-messenger reactions that regulate the rise and fall of free Ca²⁺ concentration. Kinetic methods, including flash photolysis and fast mixing, have been used to identify important interactions in the sequence of reactions. These will be reviewed here.

The existence of a similar system in neuronal tissue is supported by the presence of G-protein-coupled ‘metabotropic’ receptors, the presence of high densities of InsP₃ receptors and the demonstration of polyphosphoinositide metabolism and Ca²⁺ mobilisation in tissue derived from central nervous system (CNS). However, kinetic experiments on uniform neuronal cell populations or single cells are more difficult. As a result, the physiological role, means of activation and properties of InsP₃-evoked Ca²⁺ signalling in neurones are unclear. A comparison will be made here between data from neuronal and non-neuronal tissue.

Flash photolysis of photolabile ‘caged’ precursors is a technique now quite widely used in cell physiology to make rapid, precise changes of substrate or ligand concentration at sites within the cytosol or at cell surface receptors. The information obtained from experiments of this kind can be simply qualitative, for instance gaining evidence of the participation of a ligand such as calcium ions in a cellular process, or can be more quantitative, providing data on mechanism such as equilibrium constants and reaction rates. In the latter case, the time course of the re-equilibration following a concentration jump can give the rates of biochemical steps – ligand binding, conformation changes or ion translocation rates – that constitute the physiological response and can show the site of regenerative or inhibitory steps in the overall process. The technique is most usefully applied at sites within unstirred regions of the extracellular or intracellular space only slowly accessible by diffusion, where perfusion methods would be too slow to study kinetics.

Diffusional delays distort both the time course and steady-state concentration of physiological ligands applied by conventional perfusion methods, because of metabolism, uptake and binding. With flash photolysis, a physiologically inert, photolabile ‘caged’ derivative is allowed to reach diffusional equilibrium before photolysis by a brief light pulse releases a known concentration of free ligand at the site of action. Diffusional delays and metabolic breakdown are minimised and the rate of application and concentration of the ligand are determined by the speed and efficiency of photolysis.

The problem can be illustrated by comparing the action of intracellular InsP₃ when perfused from a whole-cell patch pipette with that seen with flash photolysis of caged InsP₃. Diffusional equilibration of small ions from the pipette is known to take tens of seconds (Fenwick et al. 1982) and for larger molecules may require many minutes (Pusch and Neher, 1988). The time course of loading the Ca²⁺ dye Fura-2 into a parotid gland acinar cell by diffusion from a patch pipette is shown in Fig. 1A by the rise of the fluorescence traces at 350 and 390nm excitation. Even in these small cells, several minutes were required to reach a steady level of fluorescence, indicating equilibration of dye concentration between pipette and cytosol, after establishing access to the cytosol (at the beginning of the trace). Applying 10 μmol l^-1 InsP₃ simultaneously with Fura-2 produced an erratic and small increase of free [Ca²⁺], as monitored by the ratio of Fura-2 fluorescence signals and shown in the top trace. Loading 1 μmol l^-1 ‘caged’ InsP₃ with the Fura-2 for 3min (Fig. 1B) followed by release of 0.5 μmol l^-1 InsP₃ in the cytosol by a 1ms flash, given at the time indicated by the arrow, produced an immediate fast rise of free [Ca²⁺], followed by a slower decline to baseline levels. The time course of this response is more readily interpreted in terms of the kinetics of InsP₃-evoked Ca²⁺ release from stores and subsequent Ca²⁺ metabolism than that of Fig. 1A. In this system, InsP₃ released by photolysis produced Ca²⁺ release at concentrations as low as 200nmol l^-1, compared with concentrations of 10 μmol l^-1 required with perfusion from the pipette (Evans and Marty, 1986; Capiod et al. 1987; Wakui et al. 1989). This suggests that metabolism of InsP₃ reduces considerably the concentration reaching the stores from the patch pipette. Similar considerations apply also to the second messengers inositol tetrakisphosphate (InsP₄), cyclic AMP and cyclic GMP.

Although the principle of flash photolysis is straightforward, there are a number of technical problems which will be reviewed briefly with reference to caged InsP₃.

Given that the photolytic yield and rate of release of the ligand are good enough to be useful, it is essential to know the extent to which the caged compound itself and by-products formed during photolytic decay interfere with the physiological system under investigation.

Caged compounds developed as physiological probes since 1978 have 2-nitrobenzyl or 2-nitrophenyl moieties (or close derivatives) as photolabile protecting groups attached covalently to the ligand (see Kaplan et al. 1978; McCray and Trentham, 1989). The reduction in activity of the caged ligand is presumably due to lowered binding affinity, produced either by changing the charge near an important binding group or by some steric effect impeding access to the binding site. Because of the remaining structural similarity, the caged ligand may act as a weak agonist or antagonist at the receptors for the free ligand.

The photochemical properties of caged compounds, particularly the mechanism of photolysis as exemplified by caged ATP, have been reviewed in detail recently by McCray and Trentham (1989). The photolysis reaction proceeds essentially in two steps, absorption of a photon of sufficient energy to generate an activated intermediate (or intermediates), followed by one or more slower dark reactions in which the intermediates decay to release products. On absorbing a photon, the activated intermediate is formed very quickly, on a nanosecond time scale. However, the decay to release products during the dark part of the reaction is relatively slow, in some cases in microseconds, but with some reactions very much slower, for example in 5–10ms with caged InsP₃. Whether these compounds are useful depends on the rates of the physiological process being investigated. The efficiency and rate of photolysis depend on the nature of the bond cleaved and the substituent groups on adjacent atoms. Changing these substituents can be used to eliminate residual biological activity and to modify the rate and efficiency of photolysis. Descriptions of the effects of chemical substitutions can be found in reviews by Wootton and Trentham (1989), McCray and Trentham (1989), Nerbonne (1986) and Lester and Nerbonne (1982). The importance of the experimental conditions in photolysis is discussed by McCray and Trentham (1989).

The synthesis and purification of 2-nitrobenzyl derivatives, the effects of substituents on their properties and problems such as the presence of stereoisomers with differing biological activity are discussed in a methodological review by Walker (1991). In particular, he gives recipes for the synthesis and purification of several compounds, including caged InsP₃. The approach used to ‘cage’ ATP, with a 2-nitrophenylethyl ester to protect the terminal phosphate (Walker et al. 1988), was applied by Walker et al. (1989) to generate the three photolabile isomeric 2-nitrophenylethyl phosphate esters of InsP₃, substituted on the 1, 4 or 5 phosphates. The isomers were separated by anion-exchange HPLC, characterised with nuclear magnetic resonance and their photochemical and physiological properties evaluated. The rate of the dark reaction was determined from the decay of aci-nitro and the formation of nitrosoacetophenone. The scheme of the photolysis reaction is shown in Fig. 2. At physiological ionic strength and pH (0.2mol l^-1 and pH7.1), the P4 and P5 isomers photolysed with rates of 225 s^-1 and 280 s^-1, respectively (half-times close to 3ms), and with quantum yields Q_p of 0.65, similar to the value of 0.63 determined for caged ATP in the same conditions. The interactions of the isomers with the InsP₃-evoked Ca²⁺ release process in smooth muscle and with enzymes involved with InsP₃ metabolism were tested. At 10 μmol l^-1, the P1 isomer produced maximal tension in permeabilised smooth muscle, indicating Ca²⁺ release from stores. The P4 and P5 caged isomers were inactive at 50 μmol l^-1 in releasing Ca²⁺ and were ineffective in blocking Ca²⁺ release evoked by InsP₃. Similar results have been obtained for Ca²⁺ release from stores in isolated guinea pig hepatocytes, in which the caged InsP₃ was applied from whole-cell patch-clamp pipettes. The potency of InsP₃ released photochemically with low levels of photolysis, so that a high caged InsP₃ concentration remained, is similar to that obtained with high levels of photolysis and a low cage concentration. The potency in single hepatocytes with whole-cell perfusion and flash photolysis is similar to that obtained by applying InsP₃ to permeabilised hepatocyte suspensions (Burgess et al. 1984). A similar concentration range, 0.2–10 μmoll^-1, is required for activation of Ca²⁺ release in both cases, suggesting that inhibition by the unphotolysed cage was small.

The three isomers were tested by Walker et al. (1989) as substrates and inhibitors of 5-phosphatase and 3-kinase. The P1 isomer was hydrolysed to caged Ins(1,4)P₂ by phosphatase and the P4 and P5 isomers were neither substrates nor inhibitors of this enzyme. The P5 isomer inhibited 3-kinase, which phosphorylates InsP₃ to InsP₄, by about twofold at 10 μmol l^-1, and the P4 isomer was inactive on this enzyme. Neither was phosphorylated to InsP₄. Thus, the commercially available mixture of P4 and P5 isomers may inhibit 3-kinase. Further, the P1 caged Ins(1,4)P₂ has been used as a control for the effects of photolysis, release of the inactive bisphosphate in smooth muscle and liver showing no effect on the hormonal response of the system (Walker et al. 1987; Ogden et al. 1990).

The aromatic nitroso ketones released as by-products are reactive particularly with –SH groups of proteins and are potentially toxic. It is necessary to make control experiments in which the by-products of photolysis are released into the system without the ligand of primary interest. In some instances, custom-made analogues such as caged InsP₂ may be available, but compounds such as caged ATP or caged phosphate may be suitable. As a guide, it has been found that 50 μmol l^-1 or less of nitrosoacetophenone released into cells appears to have no effect, but higher concentrations approaching 1 mmol l^-1 are toxic. The toxicity can be alleviated by including thiols – mercaptoethanol, dithiothreitol (DTT) or glutathione – in the solution to react with nitrosoketones released.

The optimal wavelengths for photolysis of nitrophenyl derivatives are in the near ultraviolet, 300–360nm, longer than those that damage proteins or nucleic acids and in a region where cells are relatively transparent. Light flashes in the near ultraviolet with sufficient intensity for flash photolysis can be produced by lasers, flashlamps or shuttered arc lamps. The energy density required is about 50–500mJcm^-2 at 300–360nm. The pulse should be short enough, 1ms or less, not to limit the rate of photolysis. Details of light sources and their utilisation have been reviewed by McCray and Trentham (1989).

Xenon arc flashlamps produce pulses of 0.5–1ms duration and energies up to 150mJ at 300–350nm (Schott UG11 or Hoya U 350 filter). Details of the design and performance of a lamp available commercially are given by Rapp and Guth (1988). The lamp arc can be focused directly to an image of about 4mm in diameter at a focal length of 3cm.

The quantum yield, Q_p, of a caged molecule is the proportion of molecules which, having absorbed a photon of light, then break down to release products. To determine Q_p requires measurement of the light absorbed as well as the extent of photolysis, and is useful for photochemical comparison. For routine use of flash photolysis, it is important to know simply the percentage conversion of cage to product under experimental conditions. How easy this is to determine depends on the complexity of the experimental arrangement. Two approaches have been used. In the most straightforward case, where the light is focused to a large spot from above, a 5–10 μl drop of solution can be exposed to a flash and the changes estimated chemically, for instance by HPLC for caged phosphates (Ogden et al. 1990; Walker et al. 1988). If photolysis is carried out in combination with microspectrofluorimetry, protons released stoichiometrically during photolysis can be measured directly with a pH-sensitive dye (Walker et al. 1988; Khodakhah and Ogden, 1993). As a guide to the extent of photolysis, caged ATP and caged InsP₃ (the 2-nitrophenylethyl esters) are photolysed by about 50% with a 500 m Jcm^-2 pulse of 300–360 nm light from a xenon arc flashlamp applied to a 10 μl drop.

Experiments in which free [Ca²⁺] is changed by flash photolysis are designed either to study the effects of [Ca²⁺] on a physiological variable, such as contractile tension, membrane conductance or secretion, or to study the mobilisation and sequestration of free Ca²⁺ by the cell. In some instances, the Ca²⁺-sensitivity of the tension or conductance may be known and provides a measure of the free [Ca²⁺]. However, it is often necessary to make independent measurements of free [Ca²⁺] with an optical technique, such as Ca²⁺-binding dyes, and the use of both simultaneously may give greatly enhanced information about the system. The light pulse can produce direct optical interference in two ways, from saturation of the photomultipliers (or other photosensitive element) during the flash and by a slowly decaying visible phosphorescence induced by the flash in glass components, such as coverslips and objective lenses. The direct effect of the flash can be removed by a fast shutter or by electronically gating the dynodes of the photomultiplier during the flash. The phosphorescence is less easily dealt with and requires a near-ultraviolet filter between the preparation and the microscope optics and the use of quartz components for the chamber (Ogden et al. 1991).

A second form of interference arises from continuous photolysis of caged compounds by excitation light during [Ca²⁺] measurement with fluorescent dyes. The most widely used Ca²⁺ probes are the fluorescent dyes Indo-1 and Fura-2, which are excited at near-ultraviolet wavelengths and can be used without appreciable photolysis only with weak excitation intensity. Indo-1 and Fura-2 have the advantage of spectral shifts on binding Ca²⁺, permitting [Ca²⁺] measurements independently of dye concentration and geometric factors by the use of the ratio of fluorescence at two wavelengths. However, a weak excitation intensity means, in practice, that time resolution is limited by the need to integrate the emitted light over 100ms or more to achieve an acceptable signal/noise ratio (as in Fig. 1), and the good time resolution possible with flash photolysis is lost. There are three commonly used dyes excited at longer wavelengths, Fluo-3 or Calcium Green at fluoroscein (480nm) and Rhod-2 at rhodamine (520nm) wavelengths, which can be used at high excitation intensity.

Time-resolved measurements of free [Ca²⁺] require a detection system that is linear over the whole [Ca²⁺] range of even local transients, which may reach 100 μmoll^-1 adjacent to Ca²⁺ channels for short times (e.g. Chad and Eckert, 1984). Distortions arise when the dye is locally approaching saturation and does not contribute proportionately to the spatial average. The fluorescent dyes commonly used (Fura-2, Indo-1, Fluo-3) have equilibrium dissociation constants near the resting [Ca²⁺] and therefore give a good signal at low [Ca²⁺] levels. However, their response characteristics approach saturation at relatively low [Ca²⁺] and, as a result, transient detection and time resolution is poor. A second problem is that they are extensively bound to Ca²⁺ even at resting levels, buffering Ca²⁺ to a greater extent than the native buffers. As a result, they distort spatial aspects of Ca²⁺ signalling that arise from Ca²⁺ binding to fixed cytosolic buffers by acting as a mobile Ca²⁺ carrier and speeding Ca²⁺ diffusion. A fluorescent dye for which the problems of distortion and buffering are minimal is MagFura-2 (Furaptra) which, when excited at 420nm, is quenched by Ca²⁺ to almost zero fluorescence, like Fura-2, and so is easily calibrated, but has the advantage of a relatively low Ca²⁺ affinity, around 50 μmol l^-1 (see Konishi et al. 1991).

Calcium-mobilising hormones bind to cell surface receptors and set in motion a sequence of reactions comprising activation of a pertussis-insensitive G-protein, phospholipase C, and generation of InsP₃, which releases Ca²⁺ from a component of the endoplasmic reticulum. Hepatocytes are a good ‘model’ system for single-cell studies, partly because they have provided much information about the biochemistry of second-messenger action and also because Ca²⁺ release from stores is the only means of Ca²⁺ signalling. In hepatocytes, the Ca²⁺ has a primary function of activating phosphorylase kinase via calmodulin, producing glycogenolysis and glucose release. A secondary action (in guinea pig and rabbit) is to activate K⁺-permeable and Cl^--permeable ion channels in the surface membrane, resulting in rapid KCl loss from the cell (Burgess et al. 1981). The potassium ion conductance is an apamin-sensitive Ca²⁺-activated channel that can serve as a monitor for intracellular free Ca²⁺ concentration in the range 0.3–2.0 μmol l^-1 (Capiod and Ogden, 1989b). The whole-cell patch-clamp technique permits both the perfusion of the cell with caged InsP₃ and the measurement of the K⁺ conductance to monitor Ca²⁺ release. Single-cell microelectrode or patch-clamp recordings have shown that, when noradrenaline acting through a-adrenoceptors is applied rapidly at high concentration to single hepatocytes, the K⁺ conductance increase follows with a variable delay of minimum duration 2s and up to 90s or more (Field and Jenkinson, 1987; Capiod and Ogden, 1989a; Ogden et al. 1990). Flash photolysis of caged InsP₃ has been used in a qualitative fashion to indicate the point in the sequence of reactions where the delay may occur. A number of studies have compared the time courses of responses to Ca²⁺-mobilising hormones and to InsP₃ released by photolysis from caged InsP₃. Fig. 3 shows, for comparison, in the same hepatocyte the Ca²⁺-activated K⁺ conductance following with 8s delay after application of noradrenaline (upper trace) and flash photolytic release of 0.5 μmol l^-1 InsP₃, with a delay of 0.3s, in the lower trace. A pulse of InsP₃ at moderate concentration, 0.5–1 μmoll^-1, initiates a rise of [Ca²⁺]in 100–200ms in most tissues studied, compared with delays in the onset of the effect of Ca²⁺-mobilising hormones of 1s or longer (smooth muscle: Walker et al. 1987; hepatocytes: Capiod et al. 1988; Xenopus oocytes: Miledi and Parker, 1989; vascular endothelia: Carter and Ogden, 1992). The delays in Ca²⁺ mobilisation by hormones are therefore at a step prior to, or during, the generation of InsP₃. The delays with hormone vary from one cell type to another, and in some cases have been shown to be shorter at high than at low hormone concentration, even though the ensuing response has a rise time and amplitude that are much the same. It should also be mentioned that responses of this kind have, in some cases, been shown to propagate across and between cells, and, in the case of primary cultures of hepatocytes, that the response originates at the same point each time in a particular cell (Rooney et al. 1990). In exocrine acinar cells, there is evidence that the site of InsP₃ production is at the opposite pole of the cell from the region of Ca²⁺ release (Kasai and Augustine, 1990; see Marty, 1991). Thus, the spatial arrangement of the different elements of the signalling pathway and diffusion of intermediates may contribute to the delay.

It is possible that the delay is due to the time taken to accumulate sufficient local concentration of a mediator, InsP₃ or Ca²⁺ or both, to set off a regenerative reaction sequence (Marty et al. 1989; Parker and Miledi, 1989; Crossley et al. 1991). This idea could be tested by release of InsP₃ by photolysis during the delay in hormone action to see whether the response can be triggered. Attempts to do this have shown no ‘triggering’ action of InsP₃ in hepatocytes (T. Capiod and D. Ogden, unpublished observations) but it must be remembered that InsP₃ is released by this means throughout the cell and, if a spatial element is important, experiments with photolysis restricted to small areas of the cell would be needed.

The GDP/GTP exchange reaction and dissociation of subunits of the G-proteins coupling receptors to phospholipase C (PLC) are also thought to be slow (rate 0.03min^-1, Pang and Sternweiss, 1990) but catalysed by the activation of hormone receptors. Caged GTP-γ-S is potentially a useful agent to activate G-protein independently of hormonal stimulation, as has been done with G-protein interactions with Ca²⁺ channels (Dolphin et al. 1988). Experiments with a caged 2-nitrophenylethyl ester of GTP-γ-S show first a delay of several seconds in activation of Ca²⁺ release and, second, that this delay is not reduced by application of noradrenaline, which would accelerate GTP/GDP exchange. On the basis of this observation, the delay would appear to be after G-protein activation and before InsP₃ action on stores, i.e. during production of Ins P₃ by PLC.

Flash photolysis of caged InsP₃ introduced into single hepatocytes or other cells during whole-cell patch-clamp can give information on the concentration-dependence, kinetics and regulatory interactions during InsP₃ action that would otherwise only be obtained with fast mixing experiments in permeabilised cells or subcellular fractions. Microinjection of InsP₃ itself or caged InsP₃ followed by photolysis does not give information on the concentration applied because of the difficulty of estimating the quantities injected and the cell volume.

The simplest hypothesis of how the Ca²⁺ release might occur is gating by cytosolic InsP₃ of an ion channel in the store-cytoplasm membrane. In this case, the rate of Ca²⁺ flux into the cytosol will be proportional to the open probability of the channels. The rate of change of Ca ²⁺ concentration in the cytosol, d[Ca²⁺]/dt, is proportional to the net flux of Ca²⁺ and provides the best measure of open probability of the InsP₃ channels within each cell. However, it may be expected to vary from cell to cell, not least because of geometric factors such as the ratio of InsP₃ receptor density to cytosolic volume.

Experiments with caged InsP₃ and the Ca²⁺ dye Fluo-3 in permeabilised smooth muscle (Somlyo et al. 1992), in hepatocytes (Ogden et al. 1990, 1991) and in vascular endothelial cells (Carter and Ogden, 1992; see Fig. 4) have shown a delay in the activation of Ca²⁺ efflux of several hundred milliseconds at low InsP₃ concentration (0.1–0.5 μmol l^-1), which becomes much shorter, less than 20ms, at high concentration (10–100 μmol l^-1). The short delay at high concentration is not much longer than the time required for photolysis and suggests direct activation of the Ca²⁺ efflux channel by InsP₃. The concentration range of InsP₃ activation of receptors is given by the range over which the delays were observed to become shorter, from about 0.2 to 5 μmol l^-1. These data from photolytic release of InsP₃ in single hepatocytes and other tissues agree with data from rapid mixing experiments with permeabilised hepatocytes (Champeil et al. 1989). In permeabilised rat basophilic leukaemia cells, a similar reduction of the delay at high concentration was found with rapid mixing experiments, but in this case over a lower concentration range and with a longer delay (65ms) at high concentration (Meyer et al. 1990). In all tissues tested with both photolytic and fast mixing experiments, the speed of activation at high concentration is consistent with a directly gated channel, being much faster, for instance, than rates of activation of G-protein-coupled channels in the surface membrane (Hartzell et al. 1991).

The delay at low InsP₃ concentrations is not predicted for simple models of ligand gating, and two general explanations have been proposed. The first is a high (n=4) cooperativity in InsP₃ binding seen in permeabilised rat basophilic leukaemia cells (Meyer et al. 1990). In permeabilised smooth muscle, the delay is consistent with n=1 or n=2 (Somlyo et al. 1992). The second explanation is a cooperative or co-agonist effect between InsP₃ and Ca²⁺ at low free Ca²⁺ concentrations. This idea is supported by observations of the Ca²⁺-dependence of InsP₃ action in smooth muscle (Iino, 1990; Iino and Endo, 1992), brain microsomes (Finch et al. 1991) and reconstituted InsP₃ receptor (Bezprozvanny et al. 1991). Although a co-agonist action of Ca²⁺ and InsP₃ can explain the delay quite well, direct experimental tests are difficult. The experiments of Iino and Endo (1992) show that, at strongly buffered [Ca²⁺] of 300nmol l^-1, the delay to photoreleased InsP₃ is removed and the sigmoid initial rise of [Ca²⁺] is absent and is replaced by a single exponential rise, consistent with loss of the cooperative action of Ca²⁺.

To summarise, the presence of a delay suggests cooperativity in InsP₃ action, as multiple InsP₃ binding and/or a co-agonist action of Ca²⁺. The very short delays at high InsP₃ concentration are consistent with a direct ligand-gated mode of channel activation.

The concentration range of InsP₃ action, from longest to shortest delay, is 0.1–5 μmol l^-1 in the peripheral tissues studied.

To study channel activation as a function of InsP₃ concentration or on changing other variables, the maximal rates of rise of free [Ca²⁺] should be measured in conditions where other Ca²⁺ fluxes into or out from the cytosol are small. In whole-cell recording from guinea pig liver cells, the Ca²⁺-activated K⁺ conductance of the surface membrane provides a good measure of the submembrane free [Ca²⁺], with a steep Ca²⁺-dependence in the range 0.3–2 μmol l^-1 Ca²⁺. The rate of rise of the conductance increased with InsP₃ concentration between 0.5 and 10 μmol l^-1 when compared within the same cell (Ogden et al. 1990). These results have been confirmed with the Ca²⁺ indicator dye Fluo-3 (Ogden et al. 1991). In endothelial cells, an increase in the rate of rise of the Fluo-3 signal was found in the range 0.2–5μmol l^-1 InsP₃ (Carter and Ogden, 1992). A dependence of the rate of change of the Fluo-3 fluorescence on InsP₃ concentration over a similar range has been found in permeabilised smooth muscle (Somlyo et al. 1992).

It is of interest to know the slope of the relationship between d[Ca²⁺]/dt and log [InsP₃]. The Hill coefficient, n, in principle, gives a lower-limit estimate of the number of InsP₃ molecules binding to produce activation. In single intact cells, data with Fluo-3 have proved quantitatively unreliable because of dye binding to proteins and because the high affinity for Ca²⁺ will underestimate the rate of change of free [Ca²⁺], especially if there is spatial localisation of the [Ca²⁺] rise to regions near the release sites. Data on the efflux/InsP₃ concentration relationship have come mainly from fast mixing experiments. A high potency of InsP₃, active at 4nmol l^-1, and a large Hill slope of 3–4 at low concentrations, suggesting high cooperativity, were found in permeabilised rat basophilic leukaemia cells (Meyer et al. 1988). In permeabilised liver cells (Champeil et al. 1989), the maximum rates of rise of [Ca²⁺] measured with Quin-2 increased with concentration in the range 0.1–10 μmol l^-1 with a Hill coefficient of 1.6, indicating a small degree of cooperativity. The cooperativity seen in hepatocyte suspensions was reduced at high (10 μmol l^-1) [Ca²⁺] (Combettes et al. 1993). In an InsP₃-sensitive microsomal preparation from the CNS, the concentration range measured by ⁴⁵Ca²⁺ efflux was 30nmol l^-1 to 10 μmol l^-1 with a Hill coefficient of 1.0, under conditions of constant buffered free Ca²⁺ concentration (Finch et al. 1991). Similar results were found with single-channel recording from InsP₃ receptors isolated from cerebellum and incorporated into bilayers (Bezprozvanny et al. 1991). The maximum InsP₃-evoked open probability, of up to 15% at 2 μmol l^-1, occurred at a free [Ca²⁺] of 0.3 μmoll^-1 and declined at higher [Ca²⁺]. In smooth muscle, a Hill coefficient of 1–2 was reported by Somlyo et al. (1992) and of 2 by Iino and Endo (1992), both studies performed using flash photolysis of caged InsP₃ and with Fluo-3 as the Ca²⁺ indicator in permeabilised strips.

The strong dependence of InsP₃ potency on the free [Ca²⁺] discussed above suggests that the discrepancies between different studies, particularly in cell suspensions, may result from local Ca²⁺ accumulation. The data of Iino and Endo (1992), which gave n=2, were carried out with strongly buffered Ca²⁺ in order to avoid the influence of variable

[Ca²⁺] on the InsP₃ cooperativity estimated, and were measured at the foot of the InsP₃ concentration range. It would appear that the value of the Hill coefficient is between 1 and 2, suggesting that at least two InsP₃ molecules need to bind to produce channel opening, but some degree of uncertainty remains about the value and significance of this number. An analogy may be made with studies of the nicotinic receptor, where good measurements of the Hill coefficient were difficult to make, even in this more accessible preparation, and in fact the firmest evidence for two bindings in activation was from the biochemical demonstration of two binding sites in each receptor oligomer.

There are a number of reports indicating that regenerative phenomena occur during InsP₃ action. Parker and Ivorra (1990a, 1993) found that increasing InsP₃ concentration by incremental photolysis of caged InsP₃ showed a ‘threshold’ concentration for activation of Ca²⁺-dependent Cl^- conductance and for the peak [Ca ²⁺] estimated by dye in Xenopus oocytes. In these experiments, [Ca²⁺] was monitored with Rhod-2 fluorescence confocally within the oocyte from a spot of a few micrometres in diameter, to avoid spatial averaging of the signal from a large number of Ca²⁺ storage elements that may have had different sensitivities. InsP₃ concentrations were generated by increasing the duration or transmission of a steady ultraviolet irradiation of caged InsP₃ microinjected into the oocyte, but the absolute concentrations were not known. Above threshold, the amplitude of the Ca²⁺ fluorescence was constant with increasing InsP₃ concentration, indicating a constant ‘quantal’ Ca²⁺ release. The rate of rise of the Ca²⁺ fluorescence signal was greater as InsP₃ concentration was increased, suggesting that the rate of Ca²⁺ efflux from stores, and therefore receptor activation, increased with concentration even though the peak did not. The data therefore suggest that some process quickly terminated Ca²⁺ efflux so that the final concentration of Ca²⁺ did not increase much at high concentration (see discussion below). The advantage of Parker and Ivorra’s approach is the confocal localisation of the observations to a small region of the oocyte’s Ca²⁺ store, but observations of conserved ‘quantal’ Ca²⁺ release have been reported in other cells in response to thio-InsP₃ or hormonal stimulation (Muallem et al. 1989; Taylor and Potter, 1990). The reasons for the threshold and ‘quantal’ behaviour are not known, but they may be linked with Ca²⁺ regulation of InsP₃ sensitivity.

The complex kinetic behaviour of the response of many cell types to Ca²⁺-mobilising hormones indicates the presence in the second-messenger system of steps that are regenerative or highly cooperative, and also of inhibitory mechanisms that restore the system. Some mechanisms of this kind have been shown to occur in the InsP₃-evoked Ca²⁺ release step. First, it is well documented that, immediately following InsP₃-evoked Ca²⁺ release, the release mechanism becomes refractory to further InsP₃ action and that recovery of sensitivity has a time course of tens of seconds. Fig. 5 shows data from experiments in which consecutive twin pulses of InsP₃ were released photolytically in a hepatocyte. The sensitivity of the Ca²⁺ release mechanism to InsP₃, measured by the rate of rise of the Ca²⁺-activated conductance, is suppressed to zero immediately after the first pulse and recovers with a half-time of 10–20s as the pulse interval is increased. It occurs in Limulus photoreceptors (Payne et al. 1988, 1990), guinea pig hepatocytes (Ogden et al. 1990), Xenopus oocytes (Parker and Ivorra, 1990b), brain synaptosomes (Finch et al. 1991), rat hepatocytes (Combettes et al. 1992, 1993) and in single-channel recordings from cerebellar InsP₃ receptors in bilayers (Bezprozvanny et al. 1991). It has been shown to be due to raised cytosolic [Ca²⁺] and is attributed to an action directly on the channel or with a closely associated protein. This mechanism terminates InsP₃-evoked Ca²⁺ release in guinea pig hepatocytes (and most probably in other tissues). It has a rapid onset (Finch et al. 1991; Iino and Endo, 1992; Levitan et al. 1993) and may be important in the declining phase of hormonally induced Ca²⁺ spikes and in conserving stored Ca²⁺. Estimates of the Ca²⁺ concentration needed to produce half-inhibition in the steady state are 0.5–1 μmol l^-1 in brain microsomes (Finch et al. 1991) and 3–15 μmol l^-1 in permeabilised hepatocytes (Combettes et al. 1993). The onset of inhibition has been reported to have a time constant of 580ms (brain microsomes, Finch et al. 1991). An immediate inhibition was reported by flash-released Ca²⁺ in smooth muscle (Iino and Endo, 1992) and complete inhibition within 200ms in Limulus photoreceptors (Levitan et al. 1993). The time course of recovery from inhibition is 10–50s. It is likely that the inhibition characterised by Ca²⁺ release measurements in cells has a counterpart in a Ca²⁺-induced increase of InsP₃ binding affinity, coupled with a Ca²⁺-induced desensitisation of the InsP₃ receptor, described in hepatocyte microsomes by Pietri et al. (1990).

An inhibitory effect of the decline in Ca²⁺ concentration within the store has been proposed as a regulatory mechanism (Irvine, 1990) but experimental tests do not support an important role for this mechanism (Combettes et al. 1992; Shuttleworth, 1992).

A positive interaction has been demonstrated between InsP₃ and cytosolic Ca²⁺ in preparations from CNS and smooth muscle. Finch et al. (1991) described a strong co-agonist action of Ca²⁺ and InsP₃ with an apparent equilibrium constant for Ca²⁺ of 0.66 μmol l^-1 and a Hill coefficient for InsP₃ action of 1 at constant [Ca²⁺].

Bezprovzanny et al. (1991) described an increase in open probability of cerebellar InsP₃ channels at Ca²⁺ concentrations up to 0.5 μmol l^-1. In smooth muscle, Iino (1990) first described an increased rate of InsP₃-induced Ca²⁺ release at a free [Ca²⁺] up to 0.3 μmoll^-1 and showed subsequently that, if [Ca²⁺] was suddenly increased to about 0.3 μmoll^-1, the rate of InsP₃-induced efflux was immediately increased (Iino and Endo, 1992; see also Finch et al. 1991; Combette et al. 1993). Thus, InsP₃-evoked Ca²⁺ release has the elements – positive feedback by Ca²⁺ to increase Ca²⁺ efflux – that could produce a regenerative component to the response. There is, however, no direct evidence that this occurs in intact cells, particularly that Ca²⁺ applied in the presence of InsP₃ can initiate Ca²⁺ efflux from stores. This is a point that needs to be tested.

The calcium ions which accumulate in the cytosol adjacent to sites of Ca²⁺ efflux would initially increase InsP₃-evoked release and, at later times, inhibit release by the mechanism described above, which comes into play at higher Ca²⁺ concentrations. The result would be a pulse of Ca²⁺ released into the cytosol. Given that Ca²⁺ is heavily buffered by immobile binding sites to an extent of 50–100 bound/free (see, for example, Neher and Augustine, 1992), Ca²⁺ diffusion will be considerably slowed from that in free solution by about the same proportion. A pulse of Ca²⁺ released at the store will diffuse slowly, on a time scale of 50–100ms over 1 μm. Therefore, conditions exist in which the Ca²⁺ concentration adjacent to the receptor could rise to high levels as Ca²⁺ permeates through the channel. Conditions like this are thought to exist in the cytoplasm adjacent to Ca²⁺ channels in the surface membrane. Calculations based on the dimensions of a channel with adjacent buffering of Ca²⁺ indicate that a moderate Ca²⁺ flux, corresponding to a channel current of about 0.5pA, will quickly, in 1–2ms, produce a large Ca²⁺ concentration of 10 μmol l^-1 or higher within 0.1 μm of the channel (see, for example, Chad and Eckert, 1984), sufficient in this case for inhibition of the InsP₃-evoked efflux (50% at 3–15 μmol l^-1, Combettes et al. 1993). The brief, local high pulse of Ca²⁺ concentration will diffuse into the adjacent cytosol, becoming diluted over distances of about 1 μm in 100ms. On reaching adjacent receptor channels, the Ca²⁺ will contribute to their activation by the InsP₃ present. It is easy to envisage this cycle occurring at each receptor/channel and producing a spatially propagated wave, possibly locally restricted if channel open time were relatively brief and if the time course were dominated by the diffusion of Ca²⁺. Control of the cycle could be, in some circumstances, by local Ca²⁺ buffering and by the channel density.

Hormonal stimulation of non-excitable cells often results in repetitive fluctuations of [Ca²⁺], which may be periodic spikes every 20s to 4–5min, as shown for a single guinea pig hepatocyte with 10 μmol l^-1 noradrenaline in Fig. 6B, or disorganised fluctuations about a plateau level, as shown for a different hepatocyte in Fig. 6A (for reviews, see Berridge, 1990; Jacob, 1990; Petersen and Wakui, 1990; Cobbold et al. 1991; Marty, 1991). Models for systems with periodic fluctuations in the concentration of reactants require one or more autocatalytic or cooperative steps, to provide the rise in concentration, and an inhibitory step, or exhaustion of a reactant, to account for the fall. Two approaches can be made experimentally with flash photolysis, one to isolate regenerative and inhibitory steps under controlled, non-fluctuating conditions, the other to probe the sensitivity to intermediates such as InsP₃ during different parts of the cycle. The fluctuations of [Ca²⁺] seen in the continuous presence of InsP₃ or its stable thio derivative in several cell types (Evans and Marty, 1986; Capiod et al. 1987; Wakui et al. 1989; Ogden et al. 1990; Payne and Potter, 1991) indicate that regenerative and inhibitory elements are present even in the simplest case of InsP₃ interaction with the Ca²⁺ store. Taking the first approach, the cooperative and inhibitory interactions of Ca²⁺ with InsP₃, described above for measurements under stable conditions, could form the basis of the simple type of InsP₃-evoked Ca²⁺ fluctuations with periods of 10–30s. However, it is clear that there are three or more different patterns of Ca²⁺ fluctuations or spiking that occur with hormonal stimulation and may involve additional steps, such as interactions between the receptor/G-protein/PLC and the diacylglycerol/protein kinase C/Ca²⁺ pathways (Cobbold et al. 1991; Llano and Marty, 1987; Sanchez-Bueno et al. 1990). In slowly spiking systems, the interval between spikes, of 1min or more, is too long to be accounted for by the Ca²⁺ inhibition of InsP₃ action discussed above, which has a half-time for recovery of 10–20s. Taking the second approach, photolytic release of InsP₃ during the interspike interval in slowly spiking REF52 fibroblastic cells produced a sharp rise in [Ca²⁺], confirming in this case that the release process was not inhibited and that cytosolic InsP₃ was below the concentration required to initiate Ca²⁺ release (Harootunian et al. 1988). Although this would appear to be a good experimental approach, systematic studies with caged InsP₃ or with caged Ca²⁺ chelators to test the sensitivity to InsP₃ and [Ca²⁺] during different parts of the cycle have yet to be made.

Glial cells in culture, both astrocytes and Schwann cells, respond to neurotransmitters with Ca²⁺ mobilisation from intracellular stores. The response may be evoked with excitatory amino acids, monoamines, peptides, ATP and cholinergic ligands. In more physiological conditions, propagated Ca²⁺ signalling has been observed among glia in organotypic cultures of hippocampal slices following stimulation of input pathways (Dani et al. 1992). Experiments with caged InsP₃ in cultured cerebellar astrocytes show Ca²⁺ mobilisation with characteristics similar to those seen in nonexcitable cells; the concentration range is 0.2–5 μmol l^-1 and responses are transient, lasting 5–20s, and show delays of up to 200ms at low concentrations (Khodakhah and Ogden, 1993). The role of Ca²⁺ mobilisation in glial physiology has not been demonstrated but, by analogy with other cells, it may play a part in the regulation of ion fluxes and volume, via Ca²⁺-activated channels, and in the regulation of metabolic processes such as glycogenolysis. It is difficult to record from glia in situ, and the significance of data in culture is unclear since it is likely that glial characteristics, such as the sensitivity to neurotransmitters and the role of second-messenger systems, change in culture.

In neurones there is evidence of Ca²⁺ mobilisation by neurotransmitters that is independent of extracellular Ca²⁺ (dorsal root ganglion cells, Murphy and Miller, 1988; cerebellar Purkinje cells, Llano et al. 1991). The kinetics of InsP₃-evoked Ca²⁺ release has been studied with experiments in Purkinje neurones in thin cerebellar slices. A large, rapid Ca²⁺ mobilisation from stores was found with photolytic release of InsP₃. The properties of Ca²⁺ release in Purkinje cells differ from those in peripheral tissues, described above. The InsP₃ concentrations needed are in the region of 10 μmol l^-1 or higher, 20 times higher than those needed in astrocytes and other nonexcitable cells, with maximal efflux still not attained at 80 μmoll^-1 InsP₃ (Khodakhah and Ogden, 1993). The overall characteristics are similar, but with shorter delays of up to 70ms at a low concentration, 10 μmol l^-1, of InsP₃, this being reduced to less than 20ms at concentrations above 33 μmol l^-1. The fluorescence increase was very large compared with the Ca²⁺-induced fluorescence produced in the same cell either by depolarisation in voltage-clamp or with action potential discharge. The rate of rise of the Ca²⁺ concentration is much greater, up to 1mmol l^-1 s^-1, more than 30 times the maximum seen in hepatocytes, and the peak levels of Ca²⁺ concentration are also much higher. The decline of the Ca²⁺ transient is quite fast in most cells, the response to a pulse of InsP₃ lasting a few seconds as in other tissues. Examples of the Ca²⁺ transients evoked by three concentrations of InsP₃ in a Purkinje cell are shown in Fig. 7. It was essential to monitor Ca²⁺ concentration with the low-affinity dye Furaptra (Magfura-2) in whole-cell patch-clamp. The basis for the difference in properties between liver and Purkinje cell InsP₃ receptors may be in the isoform of receptor present, suggesting that the type 1 isoform present in Purkinje cells may have a lower InsP₃ affinity. The density of receptors per unit cytosolic volume, which is very high in Purkinje cells, would result in a high d[Ca²⁺]/dt compared with peripheral tissues.

The physiological role of ‘metabotropic’ Ca²⁺ release from stores in neurones is difficult to demonstrate, but Ca²⁺ release by InsP₃ will result in short-term changes in excitability due to activation of Ca²⁺-gated K⁺ channels (Khodakhah and Ogden, 1993), and possibly in slower changes, such as long-term depression in cerebellar Purkinje cells. The reason for the lower apparent affinity of InsP₃ receptors and the much faster and larger Ca²⁺ mobilisation by InsP₃ in Purkinje cell may be related to a synaptic rather than hormonal function of the phosphoinositide system.

The rise of cytosolic [Ca²⁺] during hormonal Ca²⁺ mobilisation has been shown, in many cell types, to have an initial phase independent of the presence of external Ca²⁺. This phase is therefore thought to be due entirely to release from Ca²⁺ stores. A later phase of the response, in which the [Ca²⁺] often shows periodic fluctuations dependent on the presence of external Ca²⁺, is thought to require Ca²⁺ influx (see reviews by Jacob, 1990; Marty, 1991). Current hypotheses concerning the regulation of this influx invoke InsP₃, InsP₄ or a combination of the two, possibly acting via Ca²⁺ stores either by means of a ligand-regulated connection of the store to the external solution or by depletion of the store itself, which then activates refilling from the external solution. Evidence has been obtained for InsP₃-induced Ca²⁺ influx, by both 1,4,5-and 2,4,5-isomers alone (e.g. Kuno and Gardner, 1987; Parker and Miledi, 1987; Matthews et al. 1989), for influx requiring both InsP₃ and InsP₄ (e.g. Morris et al. 1987) and for influx requiring InsP₄ alone (Luckhoff and Clapham, 1992). The problems of access and metabolism encountered with InsP₃ applied from the patch pipette also occur with InsP₄, but a ‘caged’ InsP₄ is not available to obtain evidence of its role or of the time course and concentrations required.

Ca²⁺ influx current via store depletion (Hoth and Penner, 1992) has been demonstrated in several tissues and appears to require stringent buffering of intracellular [Ca²⁺] and drastic depletion of stores compared with that expected with physiological stimulation. Photolysis of caged InsP₃ has been used to show a Ca²⁺ influx in T-cells with milder conditions (McDonald et al. 1993). The time course of Ca²⁺ influx suggested that it may be via the stores, but similar experiments with caged InsP₃ in other systems have not been reported. The question of whether Ca²⁺ influx is via the stores or directly through the surface membrane in physiological conditions has not been satisfactorily resolved.

The role of cyclic AMP as a second messenger in cytosolic Ca²⁺ regulation has been investigated with flash photolysis in two systems: enhanced Ca²⁺ influx through non-inactivating high-threshold Ca²⁺ channels in cardiac muscle; and Ca²⁺ release from stores in hepatocytes. Two derivatives of cyclic AMP have been used as caged precursors, both as esters of the phosphate: the 4,5-dimethoxy-2-nitrobenzyl ester and the 2-nitrophenylethyl ester. The photochemistry of the caged cyclic nucleotides is discussed by Wootton and Trentham (1989). The 2-nitrophenylethyl esters photolyse slowly, with rates of about 5 s^-1 at pH7 and 100mmol l^-1 salt, and fairly efficiently, with a Q_p about

0.4. The 4,5-dimethoxy-2-nitrobenzyl esters photolyse more rapidly but with lower efficiencies. Caged cyclic nucleotides are lipophilic and may partition within cell membranes, where photolysis rates and efficiencies may differ from those in the cytosol. Photolytic release of cyclic AMP increased the amplitude and the rate of decay of Ca²⁺ current in heart, probably through phosphorylation via protein kinase A (Nerbonne et al. 1984; Charnet et al. 1991). It may be noted that cyclic AMP released in the cytosol acts with little delay, unlike a-adrenoceptor stimulation by noradrenaline, which augments Ca²⁺ current after a delay of several seconds.

The Ca²⁺-mobilising actions via adenylyl cyclase of β-receptor stimulation in hepatocytes are less well documented than those via InsP₃. This pathway can produce Ca²⁺ mobilisation alone in a smaller proportion of cells, often after very long delays of 1 min or more (Capiod et al. 1991). Cyclic AMP released from either cage mobilises Ca²⁺ with an average delay of 3s. The cyclic AMP mechanism is insensitive to heparin, providing evidence that InsP₃ is not involved by an indirect route (Noel and Capiod, 1991).

The authors’ work was supported by the MRC, INSERM, The British Heart Foundation and Glaxo PLC.