Rat mast cells degranulate in response to microinjection of guanine nucleotide

The mechanism of regulated exocytosis has been investigated in many cell types by manipulating the composition of the cytosol following permeabilisation of the plasma membrane. Mast cells in suspension may be permeabilised by a variety of agents, including bacterial cytolysins such as streptolysin O (Howell and Gomperts, 1987). Streptolysin O generates lesions that allow the rapid leakage of the soluble contents of the cytosol, including lactate dehydrogenase (Howell and Gomperts, 1987) and actin (Koffer and Gomperts, 1989). At the same time, this treatment provides access to the cytoplasm, so that if the cells are permeabilised in the presence of guanine nucleotide analogues such as guanosine 5′-[γ-thio]triphosphate (GTPγS), and if Ca²⁺ is buffered in the micromolar range, then exocytotic secretion occurs (Howell et al. 1987; Gomperts et al. 1987).

The use of patch-clamp techniques has also led to many new insights. By the use of the patch-pipette in the wholecell mode, the cytoplasm is brought into contact with an effectively infinite volume of fluid within the pipette, causing rapid dilution of the cytosol and permitting control over its subsequent composition. The patch-pipette can be used to measure the capacitance of the cell membrane and thus to monitor exocytosis in real time. In this way it has been shown that non-hydrolysable analogues of GTP can serve as intracellular effectors while, under these conditions, Ca²⁺ acts as a modulator, serving only to accelerate the process (Fernandez et al. 1984; Fernandez et al. 1987; Lindau and Nüsse, 1987; Neher, 1988; Penner, 1988).

These approaches have led to the identification of at least two sites of action of guanine nucleotide, termed Gp and GE (Gomperts et al. 1986; Fernandez et al. 1987; Penner, 1988; Cockcroft et al. 1987; Gomperts and Tatham, 1988). However, the manner in which these influence or control the activation of intact, as opposed to permeabilised, cells can only be inferred. In this paper we describe the use of the microinjection technique to introduce nucleotides into intact rat mast cells. The injection procedure involves the introduction of a small volume of fluid directly into the cytoplasm of a cell by means of a fine glass needle. As the needle enters it forms a seal with the plasma membrane and after injection, as it is withdrawn from the cell, the membrane reseals without apparent loss of cell contents. The process takes about a second and, unlike the methods described above, inflicts minimal damage.

Exocytosis in mast cells is accompanied by a substantial morphological change as the numerous intracellular granules fuse with the plasma membrane and release their contents. We have used this ‘degranulation’ event, which is readily detected in the light microscope, to assess individual cell responses following injection. We define here the conditions for the microinjection of soluble substances into individual rat mast cells and describe the effects of injecting nucleoside triphosphates in terms of exocytotic degranulation. We find that the pattern of responses to different nucleotides and nucleotide analogues is similar but not identical to that observed in permeabilised cells. However, the microinjected cells exhibit a Ca²⁺ requirement for activation and so do not resemble patch-clamped mast cells in the whole-cell configuration (Neher, 1988), but are more similar to streptolysin O-permeabilised cells, which require both Ca²⁺ and guanine nucleotide for full activation (Howell et al. 1987). Activation by nucleotide injection is inhibited when Mg²⁺ is included in the injected solution.

Mast cells were obtained from adult male Sprague Dawley rats and purified by sedimentation through Percoll as described elsewhere (Tatham and Gomperts, 1990). The purified cells were resuspended at ∼5 × 10⁵ ml⁻¹ in a buffered salt solution containing 137 mM NaCl, 2.7 DIM KOI, 1.8 mM CaCl₂, 1 mM MgCl₂, 5.6 ma glucose, 1 mgml⁻¹ bovine serum albumin and 20 mM NaHepes, pH 7.2; 200 pl samples of this suspension were then placed in 35 mm diameter plastic Petri dishes (Falcon). These had previously been thermally impressed with a lettered grid (BB-Form, Mecanex, Vuarpillière, Switzerland) to enable the subsequent location of individual cells. These were allowed to adhere to the covered dishes at room temperature for 20–40 min before the addition of a further 2 ml of medium. After 1–2 h, adherence was usually complete. Before injections were commenced, the buffer was aspirated from the Petri dish and replaced with fresh medium. For most experiments this consisted of the standard buffer solution described above. Ca²⁺-free medium was prepared by omitting CaCl₂ from the standard buffer solution and including 1 mM NaEGTA, while low Na medium was obtained by replacing NaCl in the standard buffer with potassium gluconate (145 HIM). Transfer to a different medium was always preceded by two washes.

Cell microinjection was carried out using glass needles prepared from thin-walled filament capillaries (1 mm diameter, Kwik-fil, Clarke Electromedical Instruments, Pangbourne, UK) drawn to a fine tip with a vertical needle puller (Kopf 720, Tujunga, CA USA) and then back-filled with injection solution (1–2 μl), using a fine plastic tube connected to a micrometer syringe. Filled needles were mounted in the holder of a Leitz micromanipulator (Leitz Wetzlar, FRG) and a hand-held syringe provided continuous gentle positive pressure, which was maintained as the needle was moved from cell to cell. Each injection took approximately Is.

In order to evaluate the effect of injecting the nucleotide solutions each was introduced into a group of at least 60 cells, usually occupying one grid square in the Petri dish. Before treatment, a video image of the field was printed. After injection (10–15 min later) the cells within the field that had not been injected were marked on the picture and after a further 10–20 min the responses of the remainder were evaluated. Mast cell degranulation is a dynamic transformation that may be observed in the light microscope as a characteristic morphological change (Bloom and Chakravarty, 1970; Rolich et al. 1971). Under phase-contrast illumination, it is manifest as a sequence of events in which the cells successively lose their sharp circular contours and take on a diffuse, grey appearance. This is frequently accompanied by the emergence of free granule matrices undergoing Brownian motion. Fig. 1 shows a phase-contrast image of mast cells some of which

were previously microinjected. The five cells indicated by arrows were injected with an aqueous. GTPγS solution (1 mM in the needle). Typical degranulation morphology is apparent and these cells are readily distinguishable from control cells (asterisks) injected with a similar volume of water. These are indistinguishable from the remaining uninjected cells. In this paper responses are reported as the percentage of injected cells that degranulate. Only full degranulations in which the refractive changes described above were observed were scored as positive responses.

Medium was removed by aspiration from Petri dishes, each of which contained approximately 10⁵ adherent mast cells, and was replaced by buffered saline lacking divalent cations and containing 500 μM ATP, 1 mM EDTA and a range of concentrations of ethidium bromide. Since high concentrations of the dye caused intact mast cells to degranulate, neomycin (2.5 mM), which inhibits phospholipase C (Cockcroft and Gomperts, 1985; Cockcroft et al. 1987), was also added to this medium. Alternatively, the cells were pretreated in glucose-free medium with metabolic inhibitors (5 μM antimycin A and 6 mM deoxyglucose) to prevent activation. After incubation of the dishes for 40 min at 37 °C in a humidified atmosphere, the cells were rinsed twice with buffered saline lacking CaCl₂ but containing MgCl₂ (2mM) and either NaEGTA (3mM) or EGTA/CaEGTA (3mM, pCa>8) to reseal the membranes and trap the dye. Under these conditions, cell fluorescence could be measured reproducibly over a period of at least 45 min.

The fluorescence of ethidium-stained cells was measured with a modified Leitz Fluovert microscope fitted with a fixed stage and an MPV photometer, the output of which was fed to a pen recorder. Fura2 fluorescence measurements in individual cells were performed using an apparatus previously described (Swann and Whitaker, 1986). Estimations of [Ca²⁺]_i were calculated from the ratio of the fluorescence signals obtained by exciting alternately at 350 and 380 nm (Grynkiewicz et al. 1985).

Intracellular ATP was measured by an adaptation of a method described by Johansen (1987). Briefly, samples of cell suspension were added to equal volumes of perchloric acid (final concentration 0.33 M) and kept on ice for at least 15 min. The precipitated protein was then removed by centrifugation (1800g). The supernatants were neutralised by adding Tris/KOH and were kept on ice for a further 15 min before another spin to remove the precipitated potassium perchlorate. The final supernatants were mixed with luciferin/luciferase reagent solution (ATP Bioluminescence CLS, Boehringer-Mannheim) and the constant bioluminescence light signal was measured using an LKB Luminometer. Estimates of the concentration of cytosolic ATP were made assuming a mean cytosol volume of 540 fl (see Results).

GTPγS was supplied by Boehringer-Mannheim as a solution of its tetralithium salt (100 mM). This stock solution was diluted appropriately in water before injection. No significant difference was observed between the responses of cells injected with 1 mw GTPγS obtained by diluting the stock solution in either water, potassium glutamate (150mM, pH 7.2) or KOI (150mM, pH 6.8). All the remaining nucleotides were obtained as their sodium salts from Boehringer-Mannheim, except for GTP and guanosine 5′-[βγ-imido]triphosphate (GppNHp), which were supplied as lithium salts, and xanthosine triphosphate (XTP), which was provided as a sodium salt by Sigma Chemical Co.; 100mM stock solutions of each of these were obtained by dissolving them in water with sufficient Tris to take the pH to 7. GTP was made up in a similar way to 50 mM.

Ethidium bromide, compound 48/80, potassium gluconate, deoxyglucose and antimycin A were obtained from Sigma. The pentapotassium salt of fura2 was supplied by Molecular Probes Inc.

Since mast cells are terminally differentiated and are therefore (predominantly) diploid, the nuclear DNA can be used, in an inverse fashion, as a probe for measuring the concentration of ethidium as its fluorescent DNA adduct. Calibration of cell fluorescence in terms of cytosol ethidium concentration can be achieved by equilibrating permeabilised cells with known concentrations of ethidium bromide. At increasing levels of intracellular dye, binding to DNA increases until saturation is achieved, and so by comparing either the saturating concentrations or the concentrations for 50 % saturation for injected and permeabilised cells, it is possible to evaluate the dilution factor.

To make accurate measurements of fluorescence from individual cells, it is necessary to remove any external dye. Unfortunately, changing the medium to one that is dye-free causes the nuclei of permeabilised cells to destain. This problem can be overcome, however, by adopting a method of permeabilisation that is reversible, so that the dye is trapped in the cells. For mast cells this can be achieved by exposing them to ATP⁴⁻ and after dyeequilibration removing the free ATP by adding MgC1₂ (Gomperts, 1983; Tatham and Gomperts, 1990). The details of the procedure for plated cells are given in Materials and methods.

Fig. 2 shows the relationship between ethidium concentration and cell fluorescence for both injected and ATP-permeabilised cells. The pattern of binding, saturation and apparent quenching observed in the permeabilised cells is repeated at much higher concentrations (in the needle) for the dye-injected cells. The shift between the two curves indicates the dilution of the injected dye as it entered the cytosol. The concentration of dye giving 50 % saturation of the fluorescence signal when admitted to the cells through permeability lesions was 20 μM. For injected dye the needle concentration giving 50% saturation was 2.5 mw and the ratio of these is 1:125. Taking into account the fact that variations in the quantity injected will have most effect at the half-saturation points and allowing for other inaccuracies (see Discussion), we estimate that the dilution is between 60- and 160-fold. Henceforth, as a rule of thumb we assume a factor of 100.

The mean cell diameter of rat mast cells has been estimated to be 13.5 μm (Kruger et al. 1974), and the fluid space within the cytoplasm of these cells occupies 42 % of the cell volume (Helander and Bloom, 1974), so we deduce a mean cytosol volume of 540 fl. On this basis the volume injected is approximately 5 fl.

The effects of injecting various nucleoside triphosphates and nucleotide analogues into mast cells are summarised in Fig. 3. GTP (50 mM, needle concentration) and XTP (100 mM) failed to bring about degranulation of a significant proportion of the cells, while CTP and UTP at 100 mM activated only a very limited number. Inosine triphosphate (TTP) and GppNHp were considerably more effective, while GTPγS was the most potent. Fig. 4 illustrates the concentration dependence of degranulation in response to injections of the poorly hydrolysable guanine nucleotide analogues, GTPγS and GppNHp. At concentrations above 100 μM, corresponding to an intracellular concentration of ∼1 μM, GTPγS caused the majority of cells to degranulate within 1–5 min. The EC₅₀ is 10 μM, equivalent to ∼100 nm inside the cell. GppNHp was less effective and somewhat slower, degranulating no more than about 20 % of the cells with an EC50 of 3 mM (∼30 μM, intracellular). Since GTP, GTPγS and GppNHp were provided as lithium salts, we tested the effects of injecting LiCl on its own and together with GTPγS; 400 μM LiCl did not cause degranulation when introduced into cells nor did it cause inhibition when provided at this concentration together with GTPγS (100 μM).

When GTPγS was injected into cells maintained in Ca²⁺-free medium (lmM EGTA, 1 mM MgCl₂; pCa>8), the response was greatly reduced at all concentrations (Fig. 5). This dependence on external Ca²⁺ is unlikely to be due to depletion of the ion from the cells, since they were not transferred to the Ca²⁺-free medium until just before they were injected. Under these conditions, plated cells kept at room temperature remained responsive to the mast cell agonist, compound 48/80 (10 μg ml⁻¹) for over 2 h (2=95% degranulation). Although the time between penetration and withdrawal of the injection needle is brief and occurs without apparent loss of cell contents, the seal between the needle and plasma membrane is likely to possess an impedance similar to that observed when cells are impaled by conventional glass microelectrodes, i.e. in the megohm range. In the presence of a steep Ca²⁺ concentration gradient across the plasma membrane, it is likely that each injection event will be accompanied by a brief inward flux of Ca²⁺. This will then contribute a transitory signal component that, together with the injected guanine nucleotide, becomes sufficient momentarily to trigger exocytosis. Restoring Ca²⁺ (1.8 mM) to dishes of cells previously injected with GTPγS (1 mM) in the absence of Ca²⁺ (<10⁻⁸M) had virtually no effect on the injected cells, increasing the proportion of degranulated cells from 4±1% to 6±2% (S.E., n=3). This suggests that an increase in cytoplasmic Ca²⁺ during the injection is necessary for secretion in these experiments.

In an attempt to detect an influx of Ca²⁺, cells were loaded with the fluorescent Ca²⁺-indicator fura2. Since mast cells incubated with the acetoxymethyl esters of Ca²⁺-indicator dyes accumulate substantial amounts of dye in their granules (Bibb et al. 1986; Aimers and Neher, 1985), it is more satisfactory to load the free acid form of the dye directly by microinjection. During this process considerable quantities of dye escape from the needle prior to and after injection, preventing the monitoring of [Ca²⁺]_i during the operation. However, after allowing the external dye to disperse or after washing, it is possible to make a subsequent estimation of [Ca²⁺]_i. Thus, cells in the standard buffer were injected with fura2 (1–3 mM as its potassium salt in water) and 5 min or more later, when the background was negligible, values of [Ca²⁺]_i were calculated. This treatment gave rise to two groups of cells in which resting [Ca²⁺]_i was 143±39nM (S.D., 9 cells) or 474±33nM (S.D., 4 cells).

Since our data implicate Ca²⁺ in the activation of mast cells by nucleotide injection, we have tested the effects of including the ion, or the chelator EGTA in the injection solution. However, in order to have a significant effect on the ultimate [Ca²⁺]_i by this means, it is necessary to overcome the substantial Ca²⁺-buffering capacity of cytoplasm. Injections of CaCl₂ at needle concentrations between 10 μM and 1 mM failed to induce degranulation but did lead to cell injury. In response to the higher concentrations of CaCl₂ the cells immediately formed blebs. These tended to be reabsorbed during the ensuing minutes but the cells remained swollen and had a damaged appearance. When 500 mM NaEGTA (which had no effect when provided alone) was injected into cells together with GTPγS (1 mM) the proportion of degranulating cells was still very high (90±2 %, data not shown). It is likely that even when EGTA is introduced into cells at a final level of 5 HIM, a Ca²⁺-transient caused by leakage from the outside during impalement is not eliminated. (This leakage will commence as soon as the needle enters the cell and before the chelator has been delivered.)

We also tested the effect of including MgCl₂ in the needle solutions of both GTPγS and ITP. These experiments are illustrated in Fig. 6, where it can be seen that this ion has a substantial inhibitory effect on the actions of the two nucleotides, when provided at 100 mM. We would expect this to increase the intracellular concentration of Mg²⁺ by ∼1 mM.

Under the conditions of a leaky seal, other ions (Na⁺, K⁺ and Cl^-) may also move into or out of the cell down their respective concentration gradients. To determine if such ion movements are necessary for the activation of cells by guanine nucleotide, injections were performed in medium containing potassium gluconate instead of NaCl, in order to reduce or eliminate the normal concentration gradients of the principal monovalent ions. Na⁺ was not completely absent from this solution, since it was buffered by NaHepes. The EC₅₀ of GTPγS and the maximum extent of degranulation for cells maintained in this medium are not significantly different from the values obtained from cells in the standard buffer (data not shown), indicating that leakage of Na⁺, K⁺ or Cl⁻ is not essential for secretion in our experiments.

Work with permeabilised mast cells has shown that there is no absolute requirement for ATP when exocytosis is induced by an effector combination consisting of GTPγS and Ca²⁺, although ATP does control the effective affinities for these two effectors (Howell et al. 1987). In order to gain some insight into the role of ATP in intact cells, we have examined the effect of reducing intracellular ATP levels by metabolic inhibition prior to microinjection of GTPγS. In cells suspended in the standard buffer solution (including glucose) the estimated intracellular concentration of ATP is 3.4±0.3mM (S.E.M., n=5 experiments). Fig. 7A shows the time-course of the fall in [ATP], following metabolic inhibition at room temperature. Antimycin A (5μM) and deoxyglucose (6mM) were added at zero time to cells in medium lacking glucose. Also shown is the concomitant decline of responsiveness of cells to compound 48/80 (10μg ml⁻¹). At 45 min, when the cells had become refractory to the agonist, [ATP], was 60 ±18 μM (S.E.M., n=3); (note that the loss of ATP and the loss of responsiveness to agonist stimulation is much more rapid when metabolic inhibition is carried out at 37°C; P. E. R. Tatham, unpublished results; Howell et al. 1987). Fig. 7B shows the effect of injected GTPγS on mast cells that had been subjected to antimycin A (5 μM) and 2-deoxyglucose(6 mM) for at least 45 min in standard medium lacking glucose. Although the ECgo of GTPγS for the inhibited cells is ∼10 times that of untreated cells, the nucleotide is fully effective at the higher concentrations. Thus, under conditions where compound 48/80 is unable to bring about degranulation, injection of GTPγS can still activate the cells.

We have derived a relationship between the cytosolic level of an injected solute and its concentration in the injection needle by comparing the fluorescence of cells injected with ethidium bromide with that of cells permeabilised in the presence of the dye and resealed after an equilibration period. The emission from the DNA-bound dye depends on the cytoplasmic dye level in each case. Factors that contribute to the uncertainty in these measurements include variations in the sizes of individual mast cells and in the amount iryected into each cell. Indeed, at intermediate and low dye concentrations, the observed fluorescence was variable from cell to cell. Moreover, it is possible that although the ATP-treated cells were equilibrated with the dye for 40 min before resealing, it may not have reached complete equilibrium across the membrane pores. This would lead to an overestimate of the intracellular ethidium level (i.e the left-hand curve of Fig. 2 should lie further to the left) and hence an underestimation of the dilution. Conversely, ATP⁻⁴-treated mast cells can leak out proteins such as actin (Koffer and Gomperts, 1989) and this may allow the dye to permeate the cytoplasm more extensively than it can in the injected cells. In this case, even when the levels of cytoplasmic dye are equivalent, the nuclear fluorescence would be brighter in the permeabilised cells (effectively shifting the left-hand curve of Fig. 2 to the right) leading to an overestimation of the dilution. In view of these uncertainties we assume an approximate dilution factor of 100.

The mobility and ultimate distribution of substances injected into cells is determined by the submicroscopic structure of the cell interior. Not only are the physical characteristics of cytoplasm far from isotropic but local structural elements may also undergo temporal changes, and these properties will affect the way in which injected solutes are accommodated within the cell. For example, substances of high molecular mass may be restricted, at least initially, to the site of injection and even low molecular mass solutes may not be able to permeate the entire space (Luby-Phelps et al. 1988). The notion of (final) cytosolic concentration is therefore a simplification and from the outset its estimation must be viewed as approximate.

Of the nucleotides tested, only ITP, GTPγS and GppNHp gave rise to substantial degranulation when injected into cells bathed in medium containing Ca²⁺ (Fig. 3). Table 1 compares these responses with the ability of the same nucleotides to support secretion in permeabilised preparations of mast cells and neutrophils. In making these comparisons, it is important to remember that MgATP (present in intact cells at millimolar levels) has a number of modulatory effects upon secretion in the permeabilised preparations. In permeabilised neutrophils and HL60 cells, the presence of MgATP augments the extent of lysosomal enzyme secretion. The rather low levels of release (indicated in Table 1) from permeabilised neutrophils evoked by GTP, ITP, XTP, CTP and UTP reflect the omission of MgATP in these particular experiments (Barrowman et al. 1987). Thus, the nucleotides that activate mast cells upon injection, appear to be a subset of those that support secretion in the permeabilised cells. The exceptions are XTP and GTP, which failed to activate the cells when injected. XTP does support secretion in permeabilised mast cells, neutrophils and HL60 cells, and GTP is effective in permeabilised mast cells and neutrophils (Table 1). GTP is of course already present in intact cells and the injection will have merely increased its intracellular level by approximately 1 mM. XTP, like ITP, is a potential substrate for intracellular nucleotidases, and this may account for its failure to activate upon injection. This may also explain why ITP is the most potent nucleoside triphosphate when provided to permeabilised mast cells under conditions where its supply is effectively infinite (Howell et al. 1987), and yet is only partially effective when it is injected.

The most effective activator of mast cells by injection is GTPγS. This poorly hydrolysable analogue of GTP is well established as an activator of G proteins in both cell-free and permeabilised cell systems, and it has been used extensively to investigate the involvement of G proteins in intracellular processes. When it is injected into intact cells, surface receptor events are bypassed and cell responses are activated directly, presumably through interactions with G proteins such as GE, a hypothetical mediator of exocytosis for which there is evidence in a wide range of cell types including mast cells (Gomperts, 1990a,b). Direct activation can also be achieved using permeabilised cells, but it must be recognised that these are modified or depleted systems. Cells with small permeability lesions exchange ions and small molecules such as nucleotides with the surroundings, while substantially permeabilised cells also leak out soluble proteins. Leakage of cytosol proteins is not necessarily inhibitory as it has been found that mast cells permeabilised lightly (by ATP⁴⁻) require a concentration of GTPγS to elicit secretion that is several hundred times that required by streptolysin O-permeabilised cells at the same [Ca^z+] (Koffer and Gomperts, 1989). Thus care must be exercised when interpreting the results of experiments on permeabilised cells that are likely to have lost modulatory factors. During the microinjection of intact cells, the positive pressure in the needle, the brevity of the injection and the rapid resealing of the membrane, allow substances to be introduced into the cytosol without significant leakage. Thus there is an important complementarity between the different methods of gaining access to the cytosol. Streptolysin O-permeabilisation and the application of patch-pipettes produce substantially depleted cells; ATP⁴⁻ and agents such as o-toxin (see Ahnert-Hilger and Gratzl, 1988) cause less damage and microinjected cells sustain the least damage.

In permeabilised cells GTPγS is always more potent than GppNHp (Howell et al. 1987), and the microinjected cells adhere to this pattern. The concentration dependence of the response to GTPγS (Fig. 4) indicates an EC₅₀ in the region of 100 nM (final). This is similar to the level required for streptolysin O-permeabilised cells buffered at pCa5 but lower than that required (10μM) by ATP⁴⁻-permeabilised cells also at pCa5 (Koffer and Gomperts, 1989). To compare our data with these results, it is necessary to have some idea of [Ca²⁺], in the injected cells. Since the effect of injected GTPγS is almost completely dependent on the presence of Ca²⁺ in the external medium and since late provision of Ca²⁺ does not induce exocytosis in pre-injected cells, a transitory Ca²⁺ increase must accompany each injection event. The extent of the influx may be considerable. (Indeed, fura2 fluorescence revealed that some of the cells retained high levels of Ca²⁺.) Inclusion of EGTA in the GTPγS solution to a final concentration of ∼5 mM did not inhibit the degranulation response. This level should be adequate to overcome normal intracellular Ca²⁺-buffering but is not likely to be sufficient to withstand a substantial inward leak of Ca²⁺ down a concentration gradient from millimolar extracellular levels. Moreover, the leakage of Ca²⁺ into the cell will commence as soon as the needle penetrates the membrane and will continue during the injection, so that [Ca²⁺]_i may increase before the EGTA has reached its final concentration. In contrast with this, the permeabilised cell data were obtained under Ca²⁺ clamp conditions (pCa5) and this makes direct comparison of GTPγS requirements with those of injected cells somewhat difficult. For further clarification it will be necessary to define the Ca²⁺ requirement in terms of both concentration and timing.

Mast cells can be activated by treatment with ionophores such as A23187 and ionomycin, which allow the entry of external Ca²⁺, and can also release the ion from intracellular storage sites (Foreman et al. 1973; Bennett et al. 1980). In a similar vein, permeabilisation of mast cells by ATP⁴⁻ in the presence of Ca²⁺ gives rise to secretion (Cockcroft and Gomperts, 1980). In consequence, it might be expected that microinjection of CaCl₂ would also trigger secretion. In agreement with an earlier report (Tasaka et al. 1970), we find that injection of CaC1₂ solution into cells in Ca²⁺-containing medium does not lead to degranulation, but can cause cell damage. However, degranulation has been reported to occur when CaC1₂ is introduced by iontophoresis into mast cells attached to mesentery (Kanno et al. 1973). Since these are presumably gentler conditions, it may be that pressure injection of high concentrations of Ca²⁺ in the presence of extracellular Ca²⁺ has a deleterious effect on the exocytotic apparatus, perhaps through damage to structural elements such as the filamentous network within the cytoplasm.

Although we have been unable to trigger cells by injection of Ca²⁺, it is clear that under our experimental conditions Ca²⁺ is required for degranulation induced by injection of GTPγS. In intact rat mast cells extracellular Ca²⁺ is required for activation through the IgE receptor pathway (Foreman and Mongar, 1972), but not through non-immunological polycationic stimuli such as compound 48/80 (Atkinson et al. 1979). On the other hand, when GTPγS is introduced into cells through a patch-pipette, degranulation can occur even when [Ca²⁺]_i is clamped at pCa8 (Neher, 1988), although the response is slow and occurs after a time-lag of minutes. Since cells both in the whole-cell configuration of the patch-pipette and in streptolysin O-permeabilised preparations are substantially depleted by leakage, it is important when making comparisons to give consideration to the differences in these systems. In both cases the cells lose soluble factors from their cytosols and these may possess modulatory activities. In metabolically inhibited mast cells permeabilised by ATP⁴⁻ 10 μM GTPγS elicits 50% secretion at pCa5. When the cells are more substantially permeabilised by streptolysin O at the same [Ca²⁺], only 100 nM GTPγS is required to produce the same level of secretion. This may be explained by the loss from the streptolysin O-penneabilised cells of factors that can suppress the sensitivity to Ca²⁺ and GTP analogues (Koffer and Gomperts, 1989). Since ATP⁴⁻ permeabilisation causes less membrane damage than permeabilisation by streptolysin O, it might appear that ATP⁴⁻-treated cells should be more comparable to microinjected cells. However, as stated above, a direct comparison here is not possible, since [Ca²⁺] was held at pCa5 in the permeabilised cells but could vary in the injected cells.

In permeabilised cells it is also necessary to provide a suitable ionic environment. In patch-clamp experiments the intracellular solution conventionally contains glutamate as the principal anion (Neher, 1988). In permeabilised mast cells glutamate stands out from other anions (chloride, acetate, succinate etc.) in that it can support a measure of Ca²⁺-independent secretion (Churcher and Gomperts, 1990).

In assessing the dependence of GTPγS-induced secretion on intracellular ATP levels, we are hampered by the fact that, using metabolic inhibition, we are not able to eliminate ATP completely from intact cells. We are therefore limited to concluding that the ATP-dependent processes that mediate compound 48/80-induced secretion are not part of the pathway that is activated by GTPγS and Ca²⁺.

It is difficult to explain the inhibitory effect of Mg^24- in our experiments, but we note that it also inhibits secretion from electrically permeabilised, bovine adrenal chromaffin cells; both the extent of release and the apparent affinity for Ca²⁺ are reduced (Knight and Baker, 1982). Mg²⁺ also suppresses Ca²⁺-induced secretion from rabbit neutrophils when admitted to the cells by ionophore (Di Virgilio and Gomperts, 1983). Mg²⁺ has been shown to be necessary for the dissociation of activated, GTP-bound Gs into its α and βγ subunits in rat liver, membranes (Iyengar and Bimbaumer, 1982; Iyengar et al. 1988). Binding sites for Mg²⁺ have also been detected on G-protein-receptor complexes in HL60 membrane preparations (Gierschik et al. 1989), and these are associated with the modulation of the number and affinity of formylmethionyl peptide receptors. However, it is not clear how such sites could affect the activation by injection of guanine nucleotide when [Mg²⁺]_i is elevated from approximately 1 mM to 2mM.

This work was supported by grants from the Wellcome Trust, the SmithKline (1982) Foundation and the Central Research Fund of the University of London. Support was also provided under NATO grant RG.0908/87 and in this regard we are indebted to Dr Dafna Bar-Sagi for her advice and assistance in setting up the microinjection facility, to Dr Michael Whitaker for his assistance in the measurements of intracellular [Ca²⁺], and to Professor David Saggerson and Dr Jenny Fordham for their cooperation in ATP measurement.