ATP-Independent Luminal Oscillations and Release of Ca²⁺ and H⁺ from Mast Cell Secretory Granules: Implications for Signal Transduction

Abstract

InsP₃ is an important link in the intracellular information network. Previous observations show that activation of InsP₃-receptor channels on the granular membrane can turn secretory granules into Ca²⁺ oscillators that deliver periodic trains of Ca²⁺ release to the cytosol (T. Nguyen, W. C. Chin, and P. Verdugo, 1998, Nature, 395:908–912; I. Quesada, W. C. Chin, J. Steed, P. Campos-Bedolla, and P. Verdugo, 2001, Biophys. J. 80:2133–2139). Here we show that InsP₃ can also turn mast cell granules into proton oscillators. InsP₃-induced intralumenal [H⁺] oscillations are ATP-independent, result from H⁺/K⁺ exchange in the heparin matrix, and produce perigranular pH oscillations with the same frequency. These perigranular pH oscillations are in-phase with intralumenal [H⁺] but out-of-phase with the corresponding perigranular [Ca²⁺] oscillations. The low pH of the secretory compartment has critical implications in a broad range of intracellular processes. However, the association of proton release with InsP₃-induced Ca²⁺ signals, their similar periodic nature, and the sensitivity of important exocytic proteins to the joint action of Ca²⁺ and pH strongly suggests that granules might encode a combined Ca²⁺/H⁺ intracellular signal. A H⁺/Ca²⁺ signal could significantly increase the specificity of the information sent by the granule by transmitting two frequency encoded messages targeted exclusively to proteins like calmodulin, annexins, or syncollin that are crucial for exocytosis and require specific combinations of [Ca²⁺] “and” pH for their action.

INTRODUCTION

The dynamics of lumenal pH in the secretory pathway is critical for the proper function of a broad range of cell processes including protein sorting, enzyme activation, and biogenic amines loading (Bell-Parikh et al., 2001). In secretory granules, changes in the preexocytotic lumenal pH are thought to be necessary for the final steps of exocytosis (Williams and Webb, 2000; Barg et al., 2001; Han et al., 1999). The regulation of intralumenal pH (pH_G) has been thought as a simple H⁺ influx/efflux equilibrium with pumps as H⁺ source and “leakage” to the cytosol as a H⁺ sink (Mitchell et al., 2001; Demaurex, 2002). However, the characteristic polyelectrolyte matrices present inside virtually all granules offer a novel alternative as intralumenal H⁺ sink/donors. The strong polyanionic properties of these polymer networks can function as efficient ion exchange resins controlling the bound/free turnover of intralumenal cations (Uvnas and Aborg, 1977, 1989; Verdugo, 1994; Nguyen et al., 1998; Quesada et al., 2001; Nanavati and Fernandez, 1993; Marszalek et al., 1997; Chin et al., 2002). Thus, the level of free ionized Ca²⁺, K⁺, and H⁺ inside the granule results not only from the influx/efflux equilibrium of these ions in/from the granule but also from their complex exchange with the secretory matrix. However, most of the work on the control of intralumenal cations, particularly Ca²⁺, remains focused almost exclusively on the action of pumps and export channels (Mitchell et al., 2001; Demaurex, 2002). The key role of the secretory matrix as cation sink/donor in the granule has been highlighted by recent observations that reveal that the granule can function as an intracellular Ca²⁺ oscillator, and that InsP₃-induced intralumenal Ca²⁺ oscillations—and corresponding oscillatory release of Ca²⁺ to the cytosol—results from Ca²⁺/K⁺ exchange in the matrix (Nguyen et al., 1998; Quesada et al., 2001). According to this new model, the frequency-encoded Ca²⁺ signaling system of the granule results from the interplay between the Ca²⁺/K⁺ ion-exchange properties of the secretory matrix and two Ca²⁺-sensitive channels located at close proximity on the membrane of secretory vesicles: an ASK_Ca channel that mediates K⁺ entry into the vesicular lumen, and an InsP₃-R channel that releases Ca²⁺ to the cytosol (see Fig. 1). Stimulation of the cell induces production of InsP₃ leading to InsP₃ binding to the InsP₃-R channel, release of Ca²⁺ from the granule, and decrease of intralumenal [Ca²⁺]_IL. Cytosolic Ca²⁺ ([Ca²⁺]_C) increases around the granule, turning open nearby ASK_Ca channels and closing InsP₃-R channels. K⁺ imported into the vesicular lumen exchange for Ca²⁺ in the polyanionic matrix and together with the closure of the InsP₃-R channel results in an increase of [Ca²⁺]_IL in the granule's lumen. As diffusion and cytosolic Ca²⁺-buffering restore the [Ca²⁺]_C to lower levels, the InsP₃-R channel opens again, starting a new cycle that recurs for as long as the InsP₃-R remains activated (Nguyen et al., 1998; Quesada et al., 2001). In goblet cells, these [Ca²⁺]_IL oscillations are accompanied by corresponding pH_G oscillations that can increase the gain of the Ca²⁺/K⁺ exchange leading to increased Ca²⁺ unbinding and a rise in the flux of diffusion-driven Ca²⁺ release (Chin et al., 2002).

FIGURE 1.

Dynamics of H⁺ and Ca²⁺ in secretory granules. This model involves two pools of intralumenal cations: a pool of free ionized cations in equilibrium with a pool of cations bound to the granule polyanionic matrix that functions as a H⁺/K⁺ and Ca²⁺/K⁺ ion exchange network (Verdugo, 1994; Marszalek et al., 1997). The model further assumes that vesicular Ca²⁺ uptake is driven by an undefined Ca²⁺-ATPase (Mitchell et al., 2001), and that the activity of V-type H⁺-ATPases is responsible for H⁺ transport to maintain a steady-state intralumenal pH (Demaurex, 2002). Oscillations of intralumenal H⁺ and Ca²⁺ result from the interaction of the granule polyanionic matrix and two Ca²⁺-sensitive ion channels located in close proximity on the granular membrane: an ASK_Ca channel to mediate K⁺ flux into the granule, and an InsP₃-R channel to release Ca²⁺ to the cytosol (Nguyen et al., 1998; Quesada et al., 2001). See text for further details.

Although the profusion of intracellular Ca²⁺-sensitive proteins explains the broad capacity of this cation to relay changes of functionality to intracellular sensor/effector molecules, it is insufficient to explain the specificity of the Ca²⁺ message. Here we show that InsP₃ can turn the secretory vesicles of mast cells into a double ionic oscillator that broadcasts both Ca²⁺ and H⁺ signals, thereby constraining the granule's message exclusively to sensor/effector proteins that are sensitive to both Ca²⁺ and pH.

MATERIALS AND METHODS

Mast cell granule isolation and dye loading

Motion artifacts can be a critical problem when performing thin optical sections of secretory granules in intact cells. The advantages of the isolated mast cell granule preparation we used in these experiments are that, because of their large size, granules can be easily resolved by optical microscopy (Quesada et al., 2001; Monck et al., 1992), and they can be securely immobilized on poly-l-lysine coated glass. In these experiments, mast cells of beige mice (Bg^j/Bg^j) (Jackson Laboratory, Bar Harbor, ME) were isolated by peritoneal lavage (Quesada et al., 2001; Marszalek et al., 1997). Granules were labeled as previously described (Nguyen et al., 1998; Quesada et al., 2001). Briefly, cells were washed twice in a Ca²⁺-free Hanks' solution (pH = 7.2) and loaded for 30 min at 37°C with either 2 μM of LysoSensor Green DND-189 (LS) (pK_a = 5.2; λ_exc = 443 nm, λ_em = 505 nm) to monitor pH_G changes, or with 5 μM of Calcium Orange-5N (CO-5N) (K_d = 20 μM; λ_exc = 545 nm, λ_em = 580 nm) (Molecular Probes, Eugene, OR) for 45 min at 37°C, to monitor [Ca²⁺]_IL (see Fig. 2). To remove any excess dye these two pools of cells were then washed and resuspended in an intracellular buffer solution containing 140 mM K⁺ glutamate, 20 mM HEPES, 5 mM MgSO₄, 2 mM ATP, and 100 nM Ca²⁺ buffered with ethylene glycol bis(β-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA), pH = 7.2. Secretory granules were extracted by sonication and separated by centrifugation at 10,000 rpm for 5 min. To detect extralumenal Ca²⁺ (Nguyen et al., 1998; Quesada et al., 2001; Belan et al., 1996), the granules were resuspended in intracellular buffer containing 10 μg ml⁻¹ of low-diffusivity nonpermeant dextran-conjugated Calcium Green-1 (K_d = 190 nM; λ_exc = 506 nm, λ_em = 531 nm) or Calcium Crimson (K_d = 185 nM; λ_exc = 570 nm, λ_em = 610 nm). Changes of extralumenal pH were reported by 10 μg ml⁻¹ of dextran-conjugated SNARF-1 (SN) (pK_a = 7.5; λ_exc = 488 nm, λ_em = 587 nm) (Molecular Probes) diluted in intracellular buffer. Granule suspensions were then allowed to attach to poly-l-lysine-coated glass chambers for 5 min. The chambers were mounted and kept at 37°C on the thermoregulated stage of a Nikon inverted fluorescence microscope. Notice that our set-up allows detection of only one emission at a time. We can monitor two ions simultaneously if their fluorescent probes have similar spectral characteristics but are localized in different compartments. Our results report simultaneous measurements of fluctuations of intra- and extralumenal Ca²⁺, or intra- and extralumenal H⁺, or intralumenal H⁺ and extralumenal Ca²⁺. In all these cases we used probes that segregate in these two compartments.

FIGURE 2.

InsP₃-induced pH_G oscillations. (A) Phase contrast and fluorescent images of intact mast cells (top panel) and isolated granules (bottom panel) loaded with the pH- and Ca²⁺-sensitive fluorescent probes LS (green) and CO-5N (orange), respectively. The large size of beige mast cell granules allows unequivocal intralumenal and extralumenal fluorescence measurements (Quesada et al., 2001). Scale bars: 5 μm. (B) Application of 3 μM InsP₃ induced pH_G oscillations of ∼0.1–0.12 Hz (n = 8). Notice that LS fluorescence increases with [H⁺], i.e., decreasing with pH. Inset displays a line-scan from a deconvoluted image of an isolated secretory granule (Nguyen et al., 1998; Quesada et al., 2001), showing intralumenal fluorescence changes resulting from pH_G oscillations after exposure to InsP₃. Scale bar: 3 μm. (C) InsP₃-induced pH_G oscillations were observed in isolated granules exposed to 0.5–1 μM bafilomycin (n = 5) or in the absence of ATP in the medium (not shown), ruling out the involvement of H⁺-ATPases in these oscillations. (D) Removal of ATP (filled circles, left axis; n = 3) or application of 500 nM bafilomycin (open circles; n = 5) caused pH alkalinization in isolated granules.

H⁺/K⁺ ion exchange

To investigate the H⁺/K⁺ ion exchange properties of the vesicular matrix, granules loaded with LS were equilibrated in ATP-free intracellular buffer containing heparin (100 μg ml⁻¹) and apamin (100 nM). Under these conditions, resting [Ca²⁺]_IL remains stable (∼25 μM), suggesting that the InsP₃-R and the ASK_Ca channel were rendered inoperative (Nguyen et al., 1998; Quesada et al., 2001). To titrate the intralumenal [K⁺], the granules were exposed to the K⁺ ionophore valinomycin (20 μM) while [K⁺] in the intracellular buffer was increased from 0 to 140 mM. Ionic strength and osmolarity were kept constant by adjusting the concentration of monovalent organic cation NMG⁺.

Calibration of extralumenal pH

The pH/photon-count transfer function for SN emission was obtained by measuring the fluorescence in thin optical sections of solutions of SN similar to those used in experiments but in which the pH buffered with MES, HEPES, or Tris (20 mM) was progressively increased from 6 to 6.8, 7.2, 7.6, 8, and 9, yielding a pK_a of 7.4.

Although the uncertainty of the quantum yield of LS in the intralumenal milieu prevented us from conducting absolute measurement of pH inside the granule, oscillations of intralumenal pH were readily reported by relative variations of LS photon count emission.

Optical sectioning

Granules were imaged with a Nikon Diaphot inverted fluorescence microscope using a 100 W mercury vapor epifluorescence source and a 100×, 1.4 NA oil-immersion objective. Images were formed on the 336 × 243 charge-coupled-device array of a thermoelectrically cooled, low dark noise (1.3 photoelectrons s⁻¹ at −36°C) frame transfer digital camera with 16-bit resolution and 10⁵ pixel s⁻¹ maximum readout rate (Spectra Source Model 400, Westlake Village, CA). The camera was attached to the photoport of the microscope using a 20× relay lens, yielding a final resolution of 10 pixels μm⁻¹. To avoid aliasing, we acquired three-line scans at a time, instead of the whole image, yielding a sampling rate of 3 scans s⁻¹ with 300-ms exposure time and ∼25 samples/period of [Ca²⁺]_IL or pH_G oscillation. Scans sampled an area of 0.3 μm × 24 μm containing one or more granules and were accumulated in a memory buffer forming 50- to 60-s long sequential scan stacks (inset in Fig. 2 B). Optical sections of ∼0.2 μm for Ca²⁺ changes and extralumenal pH measurements and ∼2 μm for pH_G were performed using a no-neighbors deconvolution algorithm (Nguyen et al., 1998; Quesada et al., 2001; Monck et al., 1992). Validation of the optical sectioning method has been published elsewhere (Nguyen et al., 1998; Quesada et al., 2001). The time course of average fluorescence intensity in photoelectron-counts per pixel s⁻¹ inside and outside the secretory granules was measured from the line scans. Free [Ca²⁺] was calculated from the readouts of the line scans following published methods (Nguyen et al., 1998; Quesada et al., 2001; Kao, 1994).

RESULTS

InsP₃ induces intralumenal pH oscillations in secretory granules

We performed experiments in isolated mast cell granules using pH-sensitive and Ca²⁺-sensitive fluorescent probes combined with digital sectioning (Nguyen et al., 1998; Quesada et al., 2001; Monck et al., 1992). Optical sections of isolated and in situ mast cell secretory granules loaded with the fluorescent probes LS and CO-5N are shown in Fig. 2 A. Isolated mast cell granules exposed to intracellular buffer containing 3 μM InsP₃ exhibit periodic oscillations of pH_G with a frequency of ∼0.12 Hz (Fig. 2 B). InsP₃-induced pH_G oscillations were blocked by exposure to intracellular buffer containing heparin (100 μg ml⁻¹) (a blocker of InsP₃-R channels) or apamin (100 nM) (a blocker of ASK_Ca channels) (Nguyen et al., 1998; Quesada et al., 2001), or by replacement of K⁺ by NMG⁺ (not shown), suggesting that activation of the InsP₃-R and inflow of K⁺ into the granule are required for pH_G oscillations. Notwithstanding the important role of H⁺ pumps in the control of pH_G (Demaurex, 2002), removal of ATP from the intracellular solution or exposure of the granules to the H⁺ V-ATPase inhibitor bafilomycin (0.5 μM), failed to abolish the pH_G oscillations, implying that they must result from a mechanism other than H⁺-pump activity (Fig. 2 C). However, in granules not treated with InsP₃, the removal of ATP or exposure of isolated granules to bafilomycin (0.5 μM) resulted in intralumenal alkalinization (Fig. 2 D). This outcome is consistent with previous reports that secretory granules have a resting H⁺ permeability leading to continuous efflux of H⁺ to the cytosol (Demaurex, 2002; Wu et al., 2001). Replacement of K⁺ glutamate by equimolar concentrations of KCl rendered similar results (not shown).

H⁺/K⁺ exchange in the intralumenal matrix mediates pH_G oscillations and oscillatory H⁺ efflux from the granule

The experimental validation that pH_G oscillations can result from H⁺/K⁺ exchange was conducted in situ, in isolated granules loaded with LS, and in vitro, by titration of H⁺/K⁺ exchange in solutions of heparin. In valinomycin (20 μM) treated granules—in which both InsP₃-R and ASK_Ca channels were blocked by heparin (100 μg ml⁻¹) and apamin (100 nM), respectively—the increase of intralumenal K⁺ led to a concomitant acidification (Fig. 3). Heparin—the major constituent of the mast cell intralumenal matrix—had been shown to work as a histamine/K⁺ exchanger (Uvnas et al., 1989), and we found that it can function as a H⁺/K⁺ exchanger as well. In dilute solutions of heparin (6 mg ml⁻¹), increasing [K⁺] decreased the pH (Fig. 3 B).

FIGURE 3.

Oscillatory efflux of H⁺ is driven by oscillations of Δμ_H⁺ resulting from intralumenal H⁺/K⁺ exchange. (A) H⁺/K⁺ exchange was evaluated by equilibrating LS-loaded isolated granules in an intracellular medium containing 100 μg ml⁻¹ heparin, 100 nM apamin to block InsP₃–R and ASK_Ca channels, respectively, and 20 μM of the K⁺ ionophore valynomicin—to short-circuit the granular membrane to K⁺. Increasing the extralumenal [K⁺] led to corresponding increase of LS photon counts resulting from granule acidification (n = 6). Data are shown as the percentage of maximal fluorescence increase—in photon counts sec⁻¹— with respect to the initial value. (B) Increasing the [K⁺] from 0 to 140 mM in a 6 mg ml⁻¹ heparin solution results in an exponential decrease of pH (n = 6). (C) The time course of [H⁺]_IL and [Ca²⁺]_EL was studied in LS-loaded granules equilibrated in an intracellular medium containing 10 μg ml⁻¹ of dextran-conjugated Calcium Green-1 and 100 nM apamin, to prevent K⁺ influx. InsP₃-induced release of Ca²⁺ (open circles) was accompanied by a concomitant intralumenal alkalinization (filled circles; n = 4). (D) Intralumenal and extralumenal pH was simultaneously measured by equilibrating LS-loaded granules in an intracellular solution containing 10 μg ml⁻¹ dextran-conjugated SN and 2 mM HEPES (pH = 7.2). SN fluorescence was captured at 587 nm. Addition of InsP₃ (3 μM) led to extralumenal pH oscillations in the immediate vicinity of the granule (open circles, right axis). Notice that these extralumenal pH oscillations exhibit the same frequency (∼0.12 Hz) and are in phase with pH_G changes (filled circles) (n = 6). In separate preparations in which granules were not loaded with LS, SN reported similar extralumenal pH oscillations in the perigranular space upon InsP₃ application (not shown).

These observations suggest that K⁺ influx into the granule must drive both a Ca²⁺/K⁺ exchange process—responsible for [Ca²⁺]_IL oscillations (Nguyen et al., 1998; Quesada et al., 2001)—and a H⁺/K⁺ exchange, that accounts for the periodic acidification of the granule during pH_G oscillations (Fig. 1). The corresponding periodic alkalinization phases during pH_G oscillations probably result from the release of Ca²⁺ through InsP₃-R channels or from efflux of H⁺ from the granule. Since free Ca²⁺ and H⁺ are in equilibrium with their respective bound forms in the matrix, the release of Ca²⁺ through InsP₃-R channels and the concomitant decrease of [Ca²⁺]_IL may displace bound Ca²⁺ from the polyanionic network to restore the equilibrium with free Ca²⁺, leaving free negative sites which H⁺ could occupy, causing alkalinization. A similar competition for binding sites—in this case, cytosolic binding sites—between Ca²⁺ and H⁺ has been suggested to explain the formation of a secondary H⁺ signal in melanotrophs (Beatty et al., 1993). In fact, Fig. 3 C shows that InsP₃–induced release of Ca²⁺ from granules in the presence of apamine, which prevents K⁺ influx, led to slight alkalinization. However, a more likely mechanism for intralumenal alkalinization is that the periodic increases of transmembrane pH gradient (Δμ_H⁺) can result in higher efflux of H⁺, with periodic intralumenal alkalinization, and corresponding periodic acidification in the extralumenal side. This outcome is supported by our results and by several reports that have indicated that the major contributor to H⁺ export from the granule is an endogenous H⁺ permeability—or “leak”—driven by the transmembrane pH gradient (Δμ_H⁺) (Demaurex, 2002; Wu et al., 2001; Schapiro and Grinstein, 2000; Farinas and Verkman, 1999). Although vesicular H⁺ “leakage” has been thought to probably take place via voltage-gated H⁺ channels (Demaurex, 2002; Schapiro and Grinstein, 2000), the specific mechanism of H⁺ efflux from the granule remains controversial (Wu et al., 2001; Schapiro and Grinstein, 2000). To test if secretory vesicles can produce extralumenal oscillations of pH, we equilibrated granules in an intracellular solution containing 10 μg ml⁻¹ of dextran-conjugated SN, a nonpermeant, low diffusivity fluorescent pH sensor. When these granules were exposed to InsP₃, the pH in the immediate periphery of the granule started to oscillate at the same frequency (∼0.12 Hz) and in phase with intralumenal pH oscillations (Fig. 3 D). Therefore, the intralumenal alkalinization we observed during pH_G oscillations must result from H⁺ efflux to the cytosol.

Temporal relationship between intralumenal and extralumenal dynamics of Ca²⁺ and H⁺

To investigate the relationship between Ca²⁺ release from the granule and pH_G, we equilibrated granules loaded with LS in an intracellular bathing solution (see Methods) containing 10 μg ml⁻¹ of Calcium Crimson, a dextran-conjugated Ca²⁺ probe, to monitor [Ca²⁺]_EL. The pH of the bathing solution was buffered at 7.2 by 40 mM of HEPES to prevent artifacts resulting from pH-dependent changes of quantum yield of Calcium Crimson. In agreement with previous results (Nguyen et al., 1998; Quesada et al., 2001), Fig. 4 shows that exposure of the granules to 3 μM InsP₃ induced a train of [Ca²⁺]_IL oscillations by triggering the release of Ca²⁺ with the corresponding rise of [Ca²⁺]_EL and decrease of [Ca²⁺]_IL. Similarly, InsP₃ produced oscillations of [H⁺]_IL of the same frequency but out of phase with the oscillations of [Ca²⁺]_EL (Fig. 4 B), i.e., decreases of [H⁺]_IL are accompanied by corresponding increases of [Ca²⁺] outside the granule. In isolated granules exposed to heparin (100 μg ml⁻¹) and apamin (100 nM), [H⁺]_IL was unaffected by raising the extralumenal [Ca²⁺] to 1 mM (not shown), ruling out the potential involvement of Ca²⁺/H⁺ exchangers on the granular membrane, in agreement with previous reports (Mitchell et al., 2001; Schapiro and Grinstein, 2000).

FIGURE 4.

Relationship between intralumenal and extralumenal H⁺ and Ca²⁺ oscillations. (A) The intralumenal and extralumenal changes of [Ca²⁺] were monitored in granules loaded with CO-5N and equilibrated in an intracellular solution containing 10 μg ml⁻¹ Calcium Crimson. Application of 3 μM InsP₃ provoked oscillations of [Ca²⁺]_IL (filled circles, left axis) and [Ca²⁺]_EL (open circles) of ∼0.12 Hz, which were ∼180° out of phase (n = 6). Periodic release of Ca²⁺ from the granules results in a corresponding increase of [Ca²⁺] outside the granule (Nguyen et al., 1998; Quesada et al., 2001). (B) Simultaneous monitoring of pH_G and [Ca²⁺]_EL was performed in granules loaded with LS and CG as in Fig. 3 C. InsP₃ provoked oscillations of [H⁺]_IL (filled circles, left axis) with the same frequency but ∼180° out of phase with the [Ca²⁺]_EL oscillations (open circles; n = 6). The results in Fig. 3 D and Fig. 4, A and B, indicate that the release of Ca²⁺ and the efflux of H⁺ from the granule are 180° out of phase.

Notice that while the intralumenal and extralumenal oscillations of [Ca²⁺] are phase-shifted (Fig. 4 A), the oscillations of [H⁺]_IL and [H⁺]_EL are in phase (Fig. 3 D). To explain this outcome we need to consider that, although the intralumenal [Ca²⁺] and [H⁺] oscillations are both coupled to K⁺ influx, the oscillations of [Ca²⁺]_IL are modulated by the open/close dynamics of both the InsP₃–R and the ASK_Ca channels, while the oscillations of [H⁺]_IL depend on the open/close dynamics of only the ASK_Ca channel and the leakage of this ion from the granule. In the case of Ca²⁺ (see model in Fig. 1), the InsP₃-mediated Ca²⁺ efflux results in a transient decrease in [Ca²⁺]_IL and an increase of [Ca²⁺]_EL. The rise of [Ca²⁺]_EL in the vicinity of the granule both closes the InsP₃–R channel and turns on the ASK_Ca channel, activating the influx of K⁺ that results in Ca²⁺/K⁺ exchange and rebound of [Ca²⁺]_IL. As Ca²⁺ around the granule dissipates by diffusion and buffering, the InsP₃–R channel opens again and the cycle repeats for as long as the InsP₃ remains bound to its receptor. In the case of H⁺ (see model in Fig. 1), the H⁺/K⁺ exchange in the matrix that increase [H⁺]_IL steps in when ASK_Ca channels open and influx of K⁺ takes place. Since H⁺ efflux is driven by its intralumenal concentration, the oscillations of [H⁺] outside the granule are in phase with [H⁺]_IL changes. During the closed time of the ASK_Ca channel, the H⁺/K⁺ exchange ceases but H⁺ still leaks out and [H⁺]_IL decreases. An implication of these results is that the extralumenal [Ca²⁺] and [H⁺] must be out of phase. In addition, the rate of H⁺/K⁺ exchange from the heparin matrix must exceed the efflux of H⁺ leakage, otherwise efflux of H⁺ should result in increased [H⁺]_EL but decreased [H⁺]_IL. In the case of Ca²⁺, the conductance of the InsP₃ channel in the open conformation must be higher than the rate of Ca²⁺ unbinding from the matrix as [Ca²⁺]_IL rebounds only when the InsP₃ channel closes and the influx of K⁺ exchanges for a new batch of Ca²⁺ from the matrix. We can also infer that oscillations of [Ca²⁺]_IL and [H⁺]_IL are probably in phase because: 1) [Ca²⁺]_IL and [Ca²⁺]_EL oscillations are out of phase, 2) [H⁺]_IL and [H⁺]_EL are in phase, and 3) [H⁺]_IL and [Ca²⁺]_EL are out phase (Fig. 4 A, Fig. 3 D, and Fig. 4 B, respectively).

Discussion

The polymer matrix found inside subcellular organelles—including the secretory granule—holds the answer to a highly significant set of questions in cell biology. From the polymer phase transition properties of the secretory matrix that allows the remarkable payload and efficient discharge of hormones and small molecules to the ion exchange properties of the intravesicular polymer networks, the granule offers one of the most elegant systems designed by evolution. The granule stores and releases material and signals its departure to the export machinery of the cell. Whereas the discovery of phase transitions of the granular matrix brought attention to storage and release in secretion (Verdugo, 1994; Marszalek et al., 1997), the study of the ion exchange properties of the matrix is shifting the focus to questions of signal transduction in secretory cells (Nguyen et al., 1998; Quesada et al., 2001). The H⁺ source/sink properties of the heparin matrix, and probably other secretory matrices, have a broad range of important implications, including pH regulation in subcellular organelles, phagosomal maturation, enzyme activation, protein packing, and sorting in the trans-Golgi network (Bell-Parikh et al., 2001; Reeves et al., 2002). However, the association of H⁺ release with the InsP₃–induced Ca²⁺ signal from the granule, their oscillatory nature, and the presence of exocytic proteins sensitive to the joint action of Ca²⁺ and pH strongly suggest that Ca²⁺/H⁺ release from the secretory granule might encode a combined intracellular signal. According to our working model (Fig. 1), the activation of an extracellular receptor is relayed to the intracellular network by production of InsP₃ (Berridge et al., 2000). The InsP₃ signal is received by InsP₃-R channels of nearby secretory granules, turning them into double ion oscillators that respond with two spatially and temporally constrained frequency-encoded signals of Ca²⁺ and H⁺. These oscillations are independent of ATP-mediated active uptake of Ca²⁺ or H⁺. Instead, they are brought about by the interaction of InsP₃-R and ASK_Ca channels of the granule (Nguyen et al., 1998; Quesada et al., 2001; Gerasimenko et al., 1996; Yoo, 2000; Thevenod, 2002), with opposite gating sensitivities to Ca²⁺; the H⁺ “leakage” properties of the granular membrane (Demaurex, 2002; Wu et al., 2001); and the unique Ca²⁺/K⁺ and H⁺/K⁺ ion exchange properties of the heparin granular matrix (Uvnas and Aborg, 1977, 1989; Verdugo, 1994; Nguyen et al., 1998; Quesada et al., 2001; Nanavati and Fernandez, 1993; Marszalek et al., 1997; Chin et al., 2002).

In the space domain, the release of Ca²⁺ and H⁺ affects an exceedingly small cytosolic volume that probably scales to intermolecular distances not much farther than the local Debye potential field present in the cleft between plasma and granular membranes before membrane fusion. With these boundary conditions, diffusional distances become irrelevant, and the local concentration of Ca²⁺ and H⁺ in the cleft could very well mirror the intravesicular concentration of these ions. Because of the buffering properties of the cytosol, these signals should be time and space limited, reaching strictly confined domains in the cleft and preventing undesired cross talk with other receptor proteins not involved in membrane fusion.

In the time domain, the observed 0.1 Hz frequency of oscillation of Ca²⁺ and pH signals allows scanning of a broad range of cytosolic [Ca²⁺] and pH in 5-s periods. Diffusional delays are unlikely to occur because the sensor-effector proteins are already present in the cleft either in free form or anchored to the granule or plasma membranes (Sudhof, 1995), and the diffusion distance for Ca²⁺ and H⁺ to reach their targets across the cleft is extremely short. Thus, considering the typical μs-ms relaxation timescale of molecular conformational changes, effector proteins would have enough time to switch configuration (Subramaniam and Henderson, 2000; Rami and Udgaonkar, 2001). The preexocytic oscillations of Ca²⁺ and H⁺ in the narrow cleft existing between the two membranes exhibit broad overlapping. They scan a wide combination of concentrations of Ca²⁺ and H⁺ that could create multiple yet unique conditions, attuned to the specific optimal Ca²⁺/pH dependency of the different exocytic proteins, perhaps triggering their individual fusogenic properties in a well programmed sequence.

Several proteins implicated in exocytosis including calmodulin, syncollin, or Rab3a exhibit high interdependent sensitivity to Ca²⁺ and pH (An et al., 2000; Kiss and Korn, 1999; Kajio et al., 2001; Hudmon et al., 1996; Kennedy et al., 1983). The interaction of calmodulin with different substrates requires not only changes of pH and [Ca²⁺] but frequency-encoded signals of [Ca²⁺]_C as well (De Koninck and Schulman, 1998). Protein kinase C is another protein involved in secretion that can also work as a decoder of oscillatory signals (Oancea and Meyer, 1998). However, the family of annexins gives the most striking case of combined Ca²⁺/pH dependence. These proteins are important mediators of exocytosis by means of their collective ability to fuse membranes in a Ca²⁺-dependent manner (Caohuy and Pollard, 2001; Konig et al., 1998). Remarkably, recent studies have demonstrated that the fusogenic efficiency of these proteins exhibits a critical sensitivity to pH, requiring an acidic environment of lower pH than the one found in the bulk cytosol. Depending on each specific annexin, different acidic pH values are required with slight variations of the synergy between Ca²⁺ and H⁺ (Langen et al., 1998; Isas et al., 2000; Caohuy and Pollard, 2002). Since the requirements of these proteins for both ions are much higher than those found in the bulk cytosol, several groups have proposed that membrane fusion induced by annexins is possible because of local signals that generate confined areas of high concentration of both Ca²⁺ and H⁺ (Langen et al., 1998; Isas et al., 2000; Caohuy and Pollard, 2002).

The present results are in agreement with observations in intact cells. Several groups have seen preexocytotic granular pH changes in pancreatic β-cells, mast cells, and neurons, postulating an active role of pH in priming granules for release (Williams and Webb, 2000; Barg et al., 2001; Han et al., 1999; Renstrom et al., 2002). The idea of a Ca²⁺/H⁺ signaling system is consistent with observations that both lumenal Ca²⁺ efflux and the maintenance of granular Δμ_H⁺ are needed for vacuole and granule fusion (Peters and Mayer, 1998; Ungermann et al., 1999; Peters et al., 2001; Scheenen et al., 1998; Mundorf et al., 2000; Yang et al., 2002). Although the mechanisms of acidification remain uncertain, the idea that pH changes may facilitate secretion by affecting exocytotic proteins, making them more fusogenic, has also been considered (Barg et al., 2001, 2002; Yang et al., 2002; Renstrom et al., 2002).

The search for how specificity is encoded in intracellular signal transduction remains one of the most interesting and challenging topics in cell biology. Instances of built-in conditional arguments are present in the intracellular web of information (Beatty et al., 1993; Berridge et al., 2000; Susini et al., 2000). However, the formalization of simple principles of information theory in this field remains virtually unexplored. Although both Ca²⁺ and H⁺ can readily induce conformational changes, switching on/off functional conformations in proteins or other polyions present in the cell, the broad effect of these cations can decrease their specificity. The assignment of their combination in signaling could represent a heuristic model of Boolean conditional signaling whereby the granule can target a specific group of sensor/effector proteins involved in implementing exocytosis.