Correlating Ca2+ Responses and Secretion in Individual RBL-2H3 Mucosal Mast Cells*

The role of Ca2+ in stimulus-response coupling in nonexcitable cells is still not well understood. The Ca2+ responses of individual cells are extremely diverse, often displaying marked oscillations, and almost nothing is known about the specific features of these Ca2+signals that are important for the functional response of a cell. Using the RBL-2H3 mucosal mast cell as a model, we have studied the temporal relationship between changes in intracellular Ca2+ and serotonin secretion at the single-cell level using simultaneous indo-1 photometry and constant potential amperometry. Secretion in response to antigen never occurs until intracellular Ca2+ is elevated, nor is it seen during the first few oscillations in Ca2+. Exocytotic events tend to be clustered around the peaks of oscillations, but excellent secretion is also seen in cells with sustained elevations in Ca2+. Ca2+ release from stores in the absence of influx fails to elicit secretion. If refilling and continued release of Ca2+ from stores is prevented with thapsigargin, Ca2+ influx can still trigger secretion, suggesting that store-associated microdomains of Ca2+ are not required for exocytosis. Our findings demonstrate the importance of an amplitude-encoded Ca2+ signal and Ca2+influx for stimulus-secretion coupling in these nonexcitable cells.

Although it is generally agreed that an increase in Ca 2ϩ is both necessary and sufficient for the initiation of secretion in most excitable cells, the role of Ca 2ϩ in nonexcitable cells is less clear (1). The individual Ca 2ϩ responses of nonexcitable cells are often extremely heterogeneous, and many models have been proposed for the generation of these complex and often oscillatory patterns (2). Both amplitude-encoded and frequency-encoded Ca 2ϩ signals have been proposed (3,4), and the availability of both mechanisms would allow multiple signaling pathways to be activated by Ca 2ϩ in a single cell (5). Hepatocytes, with their repetitive oscillations of constant amplitude but variable frequency are prime candidates for frequencymodulated Ca 2ϩ signaling (6), and it has now been shown that mitochondrial NAD(P)H production in these cells is indeed regulated by the frequency of Ca 2ϩ oscillations (7,8). In contrast, it seems clear that the secretory response of single rat salivary acinar cells is tightly coupled to the amplitude of the Ca 2ϩ response (9), as is ciliary beating in tracheal epithelial cells (10). Furthermore, it has recently been shown that differential activation of transcription factors in B lymphocytes is achieved via non-oscillatory Ca 2ϩ signals of different amplitudes and durations (11).
In most cases, however, it has not been easy to determine the specific features of the Ca 2ϩ signal that are important for a physiological response. This is because sensitive methods for detecting function at the single-cell level and with high temporal resolution are not readily available. Furthermore, it now seems clear that additional signals, such as the activation of protein kinase C (12) or one or more G-proteins (13), are often required for the initiation of secretion in nonexcitable cells. We have therefore examined the Ca 2ϩ dependence of secretion at the single-cell level using the RBL-2H3 mucosal mast cell. The relatively slow time course of secretion (14) and the complex Ca 2ϩ signals seen in response to antigen (5,15,16) make this an excellent model system for studying the temporal relationship between Ca 2ϩ and exocytosis in nonexcitable cells.
EXPERIMENTAL PROCEDURES RBL-2H3 mucosal mast cells (17) seeded at low density on glass coverslips were sensitized with dinitrophenyl-specific immunoglobulin E and allowed to incorporate exogenous serotonin as described previously (14). Cells were incubated with 0.5 M indo-1/acetoxymethyl ester for 20 min at 37°C in modified Tyrode's solution (135 mM NaCl, 5 mM KCl, 5 mM CaCl 2 , 1 mM MgCl 2 , 5.6 mM glucose, 10 mM Na-HEPES, pH 7.4) containing 250 M sulfinpyrazone and 0.1% bovine serum albumin (16). Experiments were conducted in modified Tyrode's solution containing 250 M sulfinpyrazone and 0.05% gelatin. Cells were maintained at 35-37°C throughout the experiment and were stimulated by perfusing the sample chamber with 3 ml of warm solution containing 1 g/ml antigen (bovine ␥-globulin to which 10 -20 dinitrophenyl groups/ molecule had been coupled (18)). Fluorescence of the intracellular Ca 2ϩ indicator, indo-1, was excited using light from a Hg lamp filtered at 365 nm. Emitted light from a single cell was split using a 455 nm dichroic mirror and measured with a pair of photomultiplier detectors after filtering at 405 and 485 nm. Data were acquired, stored, and analyzed using Oscar or FeliX software (PTI Inc., South Brunswick, NJ) on a microcomputer equipped with an A/D interface board. The indo-1 ratio ((F 405 /F 485 ) Ϫ R min ) was calculated after correction for background fluorescence, and the intracellular Ca 2ϩ concentration was determined assuming a K d of 250 nM for indo-1 (19). Microelectrodes for amperometry were prepared by inserting an 8-m diameter carbon fiber into a glass capillary tube, which was then drawn in a patch pipette puller and sealed with cyanoacrylate glue. Once the glue had cured, the carbon fiber was trimmed back as close as possible (ϳ5 m) to the glass, and the pipette was back-filled with 3 M KCl. Unless otherwise stated, the electrode potential was set to ϩ550 mV, and the fiber tip was placed Ͻ4 m from the cell surface (20). Amperometric responses were filtered at 500 Hz, digitized, and stored on video tape. Amperometric signals were never detected in unstimulated cells or in cells that had not been loaded with exogenous serotonin (14), thus confirming that they are due to exocytotic release of serotonin. The total amount of serotonin detected was calculated from the integral of the current associated with all amperometric events (21) observed prior to and immediately after detergent lysis for a number of individual cells. From this we estimated that we can detect about 25% of the total cell serotonin (14), which is similar to other estimates for the efficiency of amperometric detection of exocytosis (21,22).

RESULTS AND DISCUSSION
The response of a typical mast cell when immunoglobulin E receptors on the cell surface are aggregated by antigen is shown in Fig. 1A (16). The Ca 2ϩ response was measured using * This work was supported by grants from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed.
the fluorescent indicator, indo-1 (19), and the release of serotonin from secretory vesicles in the same cell was monitored electrochemically (23) using constant potential amperometry (24). All amperometric events have the same asymmetric shape (Fig. 1B), which is characteristic of exocytotic release of single secretory vesicles (24). As has been shown in other cell types (21,25), some of the amperometric spikes are preceded by a small rising phase or "foot" (data not shown), which is thought to be due to the slow release of vesicle contents through a narrow fusion pore that forms before full vesicle fusion. Secretory events do not occur until intracellular Ca 2ϩ is elevated, and in cells where Ca 2ϩ is oscillating, amperometric events tend to be clustered around the peaks of the oscillations, and are infrequent or absent in the troughs between oscillations ( Fig. 1, A and B). In contrast to pituitary gonadotrophs, where exocytosis is largely confined to the rising phase of a Ca 2ϩ oscillation (26), secretory events in the mast cell do not appear to be specifically linked to either the rising or falling phase of the oscillation or to the peak itself (Fig. 1B). There is no obvious threshold concentration of Ca 2ϩ for the initiation of secretion, and secretion is not detected during every increase in Ca 2ϩ (Figs. 1 and 4). Histogram analysis of the Ca 2ϩ concentrations at which secretion is observed in oscillating cells (Fig.  2, narrow hatch marks) shows that the majority of amperomet-

FIG. 3. Secretion is only observed when intracellular Ca 2؉ is elevated and stops immediately when influx is prevented and
Ca 2؉ levels fall. A, antigen stimulation caused an increase in Ca 2ϩ and the initiation of secretion in this cell. Secretion was halted when the cell was depolarized in high K ϩ (all Na ϩ replaced with K ϩ ) but immediately resumed when Ca 2ϩ influx was restored by repolarizing the cell (Na ϩ ). B, addition of 100 nM thapsigargin (Tg) to the same cell caused a further increase in Ca 2ϩ and a burst of secretory events, which could be interrupted by transiently depolarizing the cell in high K ϩ . After a short delay, 1 M ionomycin (I) caused a large and sustained increase in intracellular Ca 2ϩ and additional secretion. Lysis of the cell with saponin (0.001% final) released all of the remaining intracellular serotonin (S). Amperometric data were digitized at 20 Hz. ric events occur at intracellular Ca 2ϩ concentrations above 300 nM, even though the time spent at such high Ca 2ϩ concentrations is relatively small (Fig. 2, wide hatch marks). These findings suggest that the Ca 2ϩ signal in mast cells is amplitude-encoded and that an increase in intracellular Ca 2ϩ is essential for the initiation of secretion in response to antigen.
Additional evidence comes from the experiment shown in Fig. 3. After antigen stimulation, intracellular Ca 2ϩ increases and secretory events are detected (Fig. 3A). When Ca 2ϩ influx is inhibited by depolarizing the cell in high (140 mM) K ϩ (27), intracellular Ca 2ϩ drops to resting levels, and secretion is immediately halted. Upon repolarization, intracellular Ca 2ϩ increases, and there is a burst of secretion. As intracellular Ca 2ϩ falls again, secretion is inhibited but can be restored (Fig.  3B) by stimulating the cell with the Ca 2ϩ -ATPase inhibitor thapsigargin (28), which depletes stores of Ca 2ϩ (29) and stimulates further Ca 2ϩ influx (30). Another cycle of K ϩ depolarization and repolarization leads to the same interruption of secretion during the decrease in intracellular Ca 2ϩ , and when intracellular Ca 2ϩ is increased further using the Ca 2ϩ ionophore ionomycin, there is another burst of secretion.
Secretion is rarely seen in cells with oscillations from near resting levels of Ca 2ϩ , despite peak Ca 2ϩ concentrations that match or exceed the levels seen in secreting cells (Fig. 4A). However, when a cell with oscillations from basal Ca 2ϩ levels and no secretion progresses to an oscillatory response superimposed on an elevated Ca 2ϩ level, secretion begins, even though there is only a small increase in the mean intracellular Ca 2ϩ concentration (Fig. 4B). Secretion then increases further during the subsequent high but non-oscillatory Ca 2ϩ response (Fig. 4B). Indeed, cells showing sustained, non-oscillatory Ca 2ϩ signals in response to antigen usually have excellent secretory responses (Fig. 4C). Secretion is also observed in response to thapsigargin, which elicits a sustained, non-oscillatory increase in intracellular Ca 2ϩ (Fig. 4D). Thus, Fig. 4 clearly demonstrates that Ca 2ϩ oscillations are neither necessary nor sufficient for exocytosis to occur. Again, this is consistent with an amplitude-encoded Ca 2ϩ signaling mechanism, in which a relatively prolonged increase in Ca 2ϩ is required for efficient secretion, and argues against a frequency-encoded mechanism (3,31).
One notable feature of the responses of all cells to antigen is the absence of secretory events during the initial part of the Ca 2ϩ response (Figs. 1 and 4). In contrast, when Ca 2ϩ influx is briefly interrupted, a burst of secretory events occurs without a delay once influx is restored and intracellular Ca 2ϩ increases again (Fig. 3). The relationship between the delay in the onset of the Ca 2ϩ response after the addition of antigen (Ca 2ϩ lag   FIG. 4  time) and the delay in the initiation of secretion (secretion lag time) is shown in Fig. 5. As expected, cells with long Ca 2ϩ lag times tend to have long secretion lag times, because secretion never occurs unless intracellular Ca 2ϩ is increased. There does not appear to be a fixed latent period between the onset of the Ca 2ϩ response and the initiation of secretion, but in general secretion occurs 30 -60 s after the increase in Ca 2ϩ (Fig. 5).

FIG. 4. Little or no secretion is detected in cells that oscillate from base-line levels of Ca 2؉ , but excellent secretion is seen during sustained non-oscillatory increases in intracellular
The initial Ca 2ϩ oscillations have been attributed to release of Ca 2ϩ from intracellular stores because they do not depend on Ca 2ϩ influx (16); influx is, however, necessary to replenish the stores, sustain the Ca 2ϩ response (16), and support secretion (27). Thus, a possible explanation for the lack of secretion during the first few oscillations is that secretion does not occur until stores have been depleted sufficiently to activate Ca 2ϩ influx. However, using the manganese influx technique to monitor the antigen-induced activation of the Ca 2ϩ influx pathway (32,33), we were unable to dissociate the initial increase in Ca 2ϩ from the activation of influx ( Fig. 6; n ϭ 5 cells). Thus the delay in secretion cannot be attributed to slow activation of Ca 2ϩ influx. It is possible that the first few Ca 2ϩ oscillations are required for the transport or docking of secretory vesicles (34). In excitable cells, both the recruitment of secretory vesicles and exocytosis itself are Ca 2ϩ -dependent (34). However, our finding that secretion is immediately halted when intracellular Ca 2ϩ is reduced to resting levels but resumes as soon as intracellular Ca 2ϩ increases again (Fig. 3) suggests that Ca 2ϩdependent vesicle docking may not be the rate-limiting step in the initiation of secretion from mucosal mast cells. Another  (32); this is shown as the normalized sum of the component wavelength responses, F tot ϭ F 405 ϩ (F 485 ϫ Ѩ) where Ѩ ϭ (increase in F 405 /decrease in F 485 ) during a substantial change in intracellular Ca 2ϩ (33). Cells were prepared as described in the methods, except that they were incubated with 0.5 M indo-1/acetoxymethyl ester for 1 h. MnCl 2 (100 M) was present throughout the trace.
FIG. 7. The Ca 2؉ signal for secretion does not originate from intracellular stores but requires Ca 2؉ influx. A, in the absence of extracellular calcium, no secretion was seen when intracellular stores released Ca 2ϩ in response to antigen (Ag), but when extracellular calcium was restored, Ca 2ϩ influx resulted in secretion (n ϭ 5 cells). B, when stores were depleted of Ca 2ϩ and prevented from refilling by exposure to 1 M thapsigargin (Tg) and antigen in the absence of extracellular calcium, secretion could still be elicited when extracellular calcium was restored (n ϭ 6 cells). Cells in both A and B were deprived of extracellular calcium by washing them in a Ca 2ϩ -free modified Tyrode's solution containing 0.1 mM EGTA approximately 3 min before the addition of antigen (1 g/ml). The transient spikes seen in the Ca 2ϩ traces when antigen was added are addition artifacts. The electrode potential was ϩ650 mV. Amperometric data were digitized at 50 Hz. possibility is that the initial Ca 2ϩ response is necessary for the production of other intracellular messengers (12,35). A detailed biophysical study of the Ca 2ϩ dependence of exocytosis in peritoneal mast cells dialyzed with a nonhydrolyzable analogue of guanosine triphosphate, GTP␥S, 1 also found little or no secretion during the initial increase in Ca 2ϩ (36). It was suggested that this might be due to the slow activation of protein kinase C (36), which is also required for secretion in mast cells (12), and this may also be an explanation for our results. Consistent with this, we found that addition of the protein kinase C activator, phorbol myristate acetate (50 nM), 5-8 min before antigen reduced the delay between the initial increase in Ca 2ϩ and the initiation of secretion from 34 Ϯ 17 s (mean Ϯ S.D., n ϭ 9 cells) to 16 Ϯ 8 s (n ϭ 10 cells; p Ͻ 0.025, Student's unpaired t test).
It is well established that Ca 2ϩ influx is required for antigeninduced secretion in RBL-2H3 mucosal mast cells (27,37), and Fig. 7A confirms that secretion does not occur until extracellular Ca 2ϩ is restored to an antigen-stimulated cell. However, the role played by Ca 2ϩ influx is not clear. Because the Ca 2ϩ response due to release from stores in the absence of influx is relatively transient and often consists of oscillations to baseline levels of Ca 2ϩ (Fig. 7A and Ref. 16), it may be that influx is simply required to sustain the Ca 2ϩ response for long enough for secretion to occur. Alternatively, it is possible that microdomains of high Ca 2ϩ close to the site of exocytosis, which are known to be important in excitable cells (38), might also occur in mast cells (39). Because the conductance of the calcium release-activated Ca 2ϩ channel found in mast cells and lymphocytes is much smaller (40) than that of the voltage-sensitive Ca 2ϩ channels found in excitable cells, it may not be possible to generate such steep gradients of Ca 2ϩ close to the plasma membrane in mast cells. In contrast, the Ca 2ϩ conductance of inositol trisphosphate receptor channels is high (41), so Ca 2ϩ microdomains in the vicinity of the stores may be more important for exocytosis in nonexcitable cells. If this is the case, then depleting stores of Ca 2ϩ and preventing them from refilling using the Ca 2ϩ -ATPase inhibitor thapsigargin should prevent secretion even when Ca 2ϩ influx is restored. However, as Fig.  7B shows, exocytotic events still occur under these conditions, provided that extracellular Ca 2ϩ is present. These results clearly demonstrate that Ca 2ϩ influx across the plasma membrane can directly trigger exocytosis and that refilling and continued release of Ca 2ϩ from stores and any store-associated microdomains of Ca 2ϩ are not essential for exocytosis in mucosal mast cells.
To our knowledge, this is the first study to examine in detail the temporal relationship between individual secretory events and Ca 2ϩ oscillations in nonexcitable cells, and it clearly demonstrates that the Ca 2ϩ signal in individual mucosal mast cells is amplitude-encoded. We have also confirmed at the single-cell level that there is an absolute requirement for elevated intra-cellular Ca 2ϩ in antigen-induced secretion from mucosal mast cells and that this secretion depends on Ca 2ϩ influx.