Platelet-activating factor stimulates cytoplasmic alkalinization and granule acidification in human eosinophils

The effects of platelet-activating factor (PAF) and IL-5 on intracellular pH were investigated in human eosinophils. Purified peripheral blood eosinophils were loaded with the ratiometric fluorescent pH indicator BCECF-AM ester. Stimulation of eosinophils with PAF produced time-dependent alkalinization of the cytoplasm from an initial pH of 7.1±0.04 to 7.5±0.05. A similar alkalinization response was produced by the calcium ionophore, ionomycin and by the calcium ATPase inhibitor, thapsigargin. These compounds as well as PAF produce significant increases in cytoplasmic calcium ([Ca2+]i). In contrast, IL-5 and the protein kinase C (PKC) activator phorbol myristate acetate (PMA) did not produce cytoplasmic alkalinization and had no effect on [Ca2+]i in eosinophils. PAF-stimulated alkalinization was not inhibited under conditions that blocked plasma membrane Na+-H+ exchange, proton channel or plasma membrane H+-ATPase activities. Measurements of intragranule pH with a cell permeant pH indicator (LysoSensor Yellow/Blue DND-160), which partitions into intracellular acidic compartments, revealed that PAF-stimulated cytosolic alkalinization correlated with intragranule acidification. These results suggest that the increase in [Ca2+]i after PAF stimulation activates a H+-ATPase present in the granule membranes, leading to enhanced granule acidification and cytoplasmic alkalinization. We propose that granule acidification is an important step in solubilization of major basic protein crystals, which are stored within the granule core, in preparation for degranulation and release of these proteins.

with electron transport (Rotstein et al., 1987). Several studies have shown that a voltage-dependent, Zn 2+ -sensitive plasma membrane proton channel is activated in parallel with the increase in NADPH oxidase activity Cherny et al., 2003;DeCoursey et al., 2001;DeCoursey et al., 2003;Gordienko et al., 1996;Petheo et al., 2003). This channel is regulated by mediators that increase PKCδ activity and, following PKC activation, the cytosolic pH (pH i ) does not significantly change in response to increases in superoxide production (Bankers-Fulbright et al., 2001). Thus, proton channels play a major role in the control of pH i in activated human eosinophils.
Other mechanisms responsible for pH i regulation in resting and activated eosinophils have not been investigated in detail, but it is likely that they share some similarity with neutrophils (Coakley et al., 2002;DeCoursey and Cherny, 1994;Demaurex et al., 1996;Fukushima et al., 1996;Grinstein et al., 1986;Grinstein et al., 1991). Early investigations of pH i regulation in neutrophils during the oxidative burst revealed that Na + -H + exchange (NHE-1) activity as well as the activation of proton channels was essential to prevent intracellular acidification (Coakley et al., 2002;DeCoursey and Cherny, 1994;Demaurex et al, 1996;Fukushima et al., 1996). In addition, a 5749 The effects of platelet-activating factor (PAF) and IL-5 on intracellular pH were investigated in human eosinophils. Purified peripheral blood eosinophils were loaded with the ratiometric fluorescent pH indicator BCECF-AM ester. Stimulation of eosinophils with PAF produced timedependent alkalinization of the cytoplasm from an initial pH of 7.1±0.04 to 7.5±0.05. A similar alkalinization response was produced by the calcium ionophore, ionomycin and by the calcium ATPase inhibitor, thapsigargin. These compounds as well as PAF produce significant increases in cytoplasmic calcium ([Ca 2+ ] i ). In contrast, IL-5 and the protein kinase C (PKC) activator phorbol myristate acetate (PMA) did not produce cytoplasmic alkalinization and had no effect on [Ca 2+ ] i in eosinophils. PAF-stimulated alkalinization was not inhibited under conditions that blocked plasma membrane V-type H + -ATPase was also identified in the plasma membrane of human neutrophils (Coakley et al., 2002;Nanda et al., 1992). This proton pump did not contribute to the regulation of basal pH i , but it did play a role in preventing intracellular acidification following stimulation of superoxide production. Previous studies of neutrophil phagocytosis revealed that exposure of cells to opsonized zymosan induced a transient decrease in pH i followed by a sustained alkalinization that was dependent on Na + -H + exchange activity (Fukushima et al., 1996). Tyrosine kinase inhibitors were shown to block the activation of NHE-1 following exposure to opsonized zymosan, suggesting a role for tyrosine phosphorylation in the activation of the exchanger. Similar mechanisms for acid extrusion (Na + -H + exchange and V-type H + -ATPases) have been identified in other phagocytic leukocytes (including macrophages) and are essential for recovery from acid loading (Bidani et al., 1996).
Conditions that produce cytoplasmic alkalinization have been shown to involve Na-independent Cl-HCO 3 exchange activity in human neutrophils (Simchowitz et al., 1991). The kinetic properties of this exchanger appear to be similar to that of the Cl-HCO 3 exchanger present in erythrocytes. In macrophages however, both Na-dependent and Naindependent Cl-HCO 3 exchangers have been identified and play distinct roles in pH i regulation (Tapper and Sundler, 1992). Acid extrusion by means of Na/Cl/HCO 3 exchange activity (NCBE exchanger) was demonstrated in mouse macrophages by the dependence of pH i on extracellular sodium and bicarbonate concentrations, intracellular chloride concentration and the alkalinizing effects of 4-acetamido-4′isothiocyanatostilbene-2,2′-disulphonic acid (SITS) and 4,4′di-isothiocyanatostilbene-2,2′-disulphonic acid (DIDS). In contrast, cytoplasmic acidification results from the activity of Na-independent Cl-HCO 3 exchange activity. More recently, studies on mouse peritoneal macrophages have shown that dexamethasone treatment increases the Na/Cl/HCO 3 exchange and V-type H + -ATPase activity, leading to intracellular alkalinization (Naucler et al., 2000). This increase in cytoplasmic pH was suggested to be one mechanism by which glucocorticoids could exert their anti-inflammatory effects.
We now report the effects of calcium mobilizing and PKC activating agents on pH i in isolated human eosinophils. Compounds that increase [Ca 2+ ] i produce a dramatic cytoplasmic alkalinization, whereas agents that increase PKC activity independent of changes in [Ca 2+ ] i have no significant effect on pH i . The alkalinization response was not dependent on the activities of plasma membrane acid or base transport mechanisms and directly correlated with a decrease in intragranule pH. These results suggest that cytoplasmic alkalinization is a consequence of calcium-dependent activation of H + -ATPases localized in the granule membranes.

Materials and Methods
Chemicals and reagents Unless otherwise noted below or in the text, all chemicals and reagents were obtained from Sigma (St Louis, MO). BCECF-AM and LysoSensor Yellow/Blue DND-160 were purchased from Molecular Probes (Eugene, OR). Phorbol myristate acetate (PMA), Fura-PE3/AM, thapsigargin, ionomycin, bafilomycin and folimycin were purchased from Calbiochem (La Jolla, CA). Hank's Buffered Salt Solution (HBSS) and NuPage Bis/Tris gels and buffers were purchased from Invitrogen (Carlsbad, CA). Mouse brain lysates, goat anti-human V-ATPase subunit E, normal goat IgG, and secondary anti-goat HRP conjugated antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-CD18 antibodies were purchased from BD Biosciences Pharmingen (San Diego, CA) and anti-CD16 magnetic beads were purchased from Miltenyi Biotech (Auburn, CA). Platelet-activating factor (PAF) C-18 was purchased from BioMol (Plymouth Meeting, PA) and human recombinant Interleukin-5 (IL-5) was a generous gift from Schering Plough (Kenilworth, N.J.). Alpha calf serum was purchased from HyClone Laboratories (Logan, UT).

Eosinophil isolation
Human eosinophils were isolated from venous human blood as previously described (Hansel et al., 1991;Ide et al., 1994). Briefly, heparinized blood was collected from atopic and nonatopic volunteers, an equal volume of 1× PIPES was added (25 mM PIPES, 50 mM NaCl, 5 mM KCl, 25 mM NaOH, 5.4 mM glucose, pH 7.4) and the diluted blood was layered onto Percoll (density 1.085 g/ml). After centrifugation at 1000 g for 30 minutes at 4°C, the plasma and Percoll layers were removed by aspiration. Tubes were wiped to remove contaminating leukocytes and red cells were lysed by osmotic shock with water. The remaining pellet, containing neutrophils and eosinophils, was incubated with an equal volume of anti-CD16 magnetic beads on ice for 30 minutes. After incubation, the cell mixture was diluted with 1× PIPES and 1% α calf serum and eluted through a steel wool column suspended in a strong magnet. Column eluate (14 ml) was collected and the number and purity of eosinophils was determined by staining with Randolph's stain. Eosinophil purity was always greater than 95% and the major contaminating cells were neutrophils.

Measurement of intracellular calcium
Purified eosinophils were resuspended to 0.5×10 6 cells/ml in PSS and 1% alpha calf serum and loaded with 5 µM Fura-PE3/AM for 30 minutes at 37°C. Eosinophils were then washed in HBSS (with 10 mM PIPES, 0.1% BSA, pH 7.4) and resuspended to 0.25×10 6 cells/ml. Cells were seeded at low density on coverslip chamber slides for 10-15 minutes to allow attachment, but not adhesion, to the glass.
The slide was mounted onto the stage of a Nikon Diaphot inverted microscope with an epifluorescence attachment. Fluorescence in single cells was visualized using a Nikon UV-fluor 40× oil-immersion objective. The fluorescence excitation, image acquisition and realtime data analyses were controlled by Image-1 Metamorph software (Universal Imaging, Westchester, PA). [Ca 2+ ] i was measured as the ratio of fluorescence emitted at 510 nm when the cells were alternately excited at 340 nm and 380 nm (F 340 /F 380 ).

Granule pH measurements
Purified human eosinophils were resuspended to 2×10 6 cells/ml in HBSS (with 10 mM PIPES, 0.03% gelatin, pH 7.4) and added to Lab-Tek eight-well chambered cover glass slides (Nalge Nunc International Corporation, Naperville, IL). Eosinophils were incubated with 10 µM LysoSensor Yellow/Blue DND-160 for 10 minutes at 37°C, washed, and then resuspended in HBSS/PIPES/ gelatin. Eosinophils were then stimulated as indicated and changes in fluorescence were documented using a 63× water immersion objective on a Zeiss LSM 510 confocal laser-scanning microscope (Carl Zeiss, Oberkochen, Germany) with excitation at 364 nm and emission measured at 385-470 nm (blue) and 505-550 nm (green-yellow). Data are shown as overlays of blue and green-yellow emission images for each stimulus.
Microarray data and analysis Isolated eosinophils were cultured at 37°C and 5% CO 2 overnight in RPMI-1640 (Celox Laboratories, St. Paul, MN) containing 10 mM HEPES and 10% heat-inactivated alpha calf serum. RNA was isolated from eosinophils using a phenol/guanidine isothiocyanate lysis followed by chloroform extraction and isopropanol precipitation. The RNA was then washed with 75% ethanol and finally resuspended in distilled water. To ensure sufficient quality for microarray analysis, the RNA was treated with Qiagen RNeasy (Qiagen Inc., Valencia, CA). Control eosinophils were lysed immediately after isolation and stored overnight at -20°C in phenol/guanidine isothiocyanate. To reduce donor-to-donor variability, RNA extracts from seven different eosinophil donors were pooled (eosinophils per sample ranged from 5-12×10 6 /sample, depending on recovery). cDNA transcription and hybridization to Affymetrix Human Genome U95A GeneChips ® (Affymetrix Inc., Santa Clara, CA) was performed by the Mayo Clinic Advanced Genomics Technology Center Microarray Shared Resource; GeneChip ® expression data were analyzed using Affymetrix ® Microarray Suite version 5.0 (statistical algorithms). All transcripts in Table 1 were defined as 'present' (P<0.04) by software analysis. The P-value associated with this test reflects the confidence of the detection call of 'present'.

Western blotting
Purified eosinophils were lysed with SDS sample buffer (62.5 mM Tris/HCl, pH 6.8, 10% glycerol, 2% SDS, 0.1% bromophenol blue, 50 mM DTT) at 20×10 6 eosinophils/ml. Lysates were boiled for 5 minutes and the equivalent of 4×10 5 eosinophils was loaded onto a 10% bis/tris gel with MOPS buffer. After protein transfer, the nitrocellulose membrane was cut, blocked in milk buffer and sections were incubated with either normal goat IgG (control antibody) or anti-V-ATPase subunit E antibody (each at 1 µg/ml final volume). A secondary anti-goat HRP-conjugated antibody (1:10,000) was used for detection and the blot was developed using an ECL kit (Amersham Pharmacia Biotech, Piscataway, NJ) followed by exposure to Kodak Biomax film.

Immunohistochemistry
Eosinophil cytospins (1×10 5 cells/slide) were fixed in 1% paraformaldehyde in PBS for 30 minutes on ice. Slides were washed in PBS and incubated overnight in 10% normal rabbit serum in PBS (Pel-Freez Biologicals, Rogers, AR). The next day, slides were washed and overlaid with equal concentrations of either affinitypurified goat anti-human vATPase E (D-20) (Santa Cruz Biotechnology, Santa Cruz, CA) or normal goat IgG (Sigma, St. Louis, MO) for 30 minutes at 37°C. The slides were washed, incubated in 1% Chromotrope 2R (J. T. Baker, Phillipsburg, NJ) for 30 minutes at room temperature, washed and overlaid with FITClabeled affinity-purified rabbit anti-goat IgG (50 µg/ml) (Jackson ImmunoResearch Laboratories, West Grove, PA) for 30 minutes at 37°C. After a final wash, the slides were mounted with glycerol/PBS containing 0.1% paraphenylenediamine (Aldrich, Milwaukee, WI) and fitted with a coverslip. The slides were viewed using a 63× water immersion objective on a confocal laser-scanning microscope (LSM 510, Carl Zeiss, Inc., Oberkochen, Germany) at 488-nm excitation wavelength.

Statistical analysis
Statistical significance was determined (where indicated) by the twotailed paired Student's t-test. Significance was defined as P<0.05.

Results
To measure the pH i of resting and activated eosinophils, we used the ratiometric pH indicator BCECF/AM (Boyer and Hedley, 1994). In the absence of stimulation, eosinophil pH i was stable, at approximately 7.1. However, PAF (3 µM) stimulation of eosinophils induced cytosolic alkalinization within 10 minutes (Fig. 1A). This alkalinization response was maintained for more than 90 minutes with a mean time constant (τ; the time required to reach 63% of the maximum pH i ) of 20±3 minutes. Under the same conditions, IL-5 (0.2 µM) and PMA (2 nM) had no significant effect on pH i (Fig.  1B). Although all three stimuli induce robust eosinophil activation, including NADPH oxidase activation, only PAF produced an increase in pH i .
To determine whether the alkalinization response following PAF stimulation was due to changes in [Ca 2+ ] i , the effects of the calcium ionophore, ionomycin and the Ca 2+ -ATPase inhibitor thapsigargin were examined (Fig. 2). Ionomycin and thapsigargin stimulated alkalinization similarly to PAF. However, the rate of alkalinization was slower for thapsigargin (τ=73±18 minutes) compared to PAF (Fig. 1A) or ionomycin (τ=29±2 minutes). Extracellular Ca 2+ chelation with EGTA To determine the effects of PAF and IL-5 on [Ca 2+ ] i , we loaded human eosinophils with the fluorescent Ca 2+ indicator, Fura-PE3/AM and measured changes in [Ca 2+ ] i in single attached eosinophils (Fig. 3). Under these conditions, PAF (3 µM) produced a rapid increase in [Ca 2+ ] i that exhibited periodic oscillations over more than 200 seconds (Fig. 3A). In contrast, treatment of cells with IL-5 (0.2 µM) did not affect [Ca 2+ ] i , but subsequent treatment with 3 µM PAF again produced an increase in [Ca 2+ ] i (Fig. 3B). Comparison of the kinetics of the calcium response in the presence and absence of IL-5 suggested that IL-5 pre-treatment modulated the duration of the PAF-induced increase in [Ca 2+ ] i , but the initial onset and peak calcium responses were not significantly affected by pretreatment with IL-5.
To determine if PAF-stimulated proton efflux is occurring  through known plasma membrane transport mechanisms, eosinophils were treated with PAF in the presence of several inhibitors (Fig. 5). To establish whether plasma membrane Na + -H + exchange activity was responsible for the PAFstimulated alkalinization response, Na was replaced with the impermeant cation, NMDG, in the extracellular solution as a means of inhibiting Na + -H + exchange activity. In addition, a known Na + -H + exchange blocker, EIPA, was tested in a separate series of experiments. Inhibition of plasma membrane Na + -H + exchange (NHE) activity by either Na + replacement (pH i =7.9±0.3; τ=30±4 minutes) or treatment with ethyl isopropyl amiloride (EIPA) (pH i =7.5±0.04; τ=27±5 minutes) failed to significantly alter the time course or magnitude of the PAF response (τ=26±3 minutes). Inhibition of the NADPH oxidase-associated proton channel with Zn 2+ slowed the time course of alkalinization (pH i =7.5±0.08; τ=50±10 minutes), but the relative increase in pH was not significantly different to that observed for PAF under control conditions (Fig. 5A). Inhibition of NADPH oxidase activity with 1 µM diphenyleneiodonium chloride (DPI) had no effect on PAFstimulated eosinophil alkalinization (data not shown). Treatment with the H + -ATPase inhibitors bafilomycin (pH i =7.5±0.02; τ=22±4 minutes) or folimycin (pH i =7.5±0.05; τ=38±12 minutes) also did not significantly reduce the magnitude or affect the time course of PAF-stimulated alkalinization (pH i =7.5±0.04; τ=29±5 minutes) (Fig. 5B). Thus PAF-induced alkalinization could not be explained by plasma membrane activity of the NADPH oxidase associated proton channel, Na + -H + exchange, or plasma membrane H + -ATPase activity. Eosinophils are characterized by their distinctive secondary (specific) granules that contain preformed toxic proteins and are relatively acidic (pH ~5.1) compared to the cytosol (Kurashima et al., 1996). To examine whether PAF stimulates active proton transport into the granule compartment, eosinophils were loaded with a fluorescent, pH-sensitive dye, LysoSensor Yellow/Blue DND-160. This dye preferentially partitions into acidic compartments (excluding the nucleus) and appears blue when the pH is neutral or slightly acidic and green/yellow when the pH falls below approximately 4.5. Confocal imaging of eosinophils treated with LysoSensor Yellow/Blue showed staining consistent with dye localization to the secondary granules. Unstimulated eosinophils exhibited an overall blue, granular appearance that indicated a granule pH >4.5 (Fig. 6A, far left). Following stimulation with PAF (3 µM), a shift in fluorescence was observed from blue to green, indicating a drop in granule pH. The time course of this effect correlated with the cytosolic alkalinization response produced by PAF, suggesting that the two events were temporally linked. IL-5 stimulated eosinophils also showed fewer distinct green compartments 40 minutes after stimulation (Fig. 6B, left panel) when compared to PAF-stimulated eosinophils (right panel), consistent with PAF-stimulated proton transport into the secondary granules.
Previous studies have indicated that eosinophil secondary granules are physiologically similar to lysosomes and contain an H + -ATPase that functions to acidify the granule compartment (Kurashima et al., 1996;Persson et al., 2002).
Using microarray analysis to screen for gene expression, we found that eosinophils constitutively express several H + -ATPase subunits at the mRNA level (Table 1). To confirm that Journal of Cell Science 117 (24) n.s. (p=0.06) * * * * * * Fig. 5. PAF-stimulated cytosolic alkalinization is not blocked by inhibiting known proton-efflux mechanisms. Purified human eosinophils were loaded with BCECF/AM and stimulated with media control or PAF (3 µM) in the presence or absence of EIPA (100 µM), Zn 2+ (250 µM), bafilomycin (2.5 µM), or folimycin (2.5 µM) at 37°C. Fluorescence was measured over 90 minutes and was converted to pH i using a standard curve as described in Materials and Methods. Data are shown as mean±s.e.m. (A) Eosinophils stimulated with media alone (n=5) showed no significant changes in pH i compared with pH i at time 0. Eosinophils stimulated with PAF alone (n=3) or in the presence of EIPA (n=3), Na + -free media (n=3) or Zn 2+ (n=3) as indicated, all show significant alkalinization compared to pH i at time 0. Initial pH i values for EIPA, Na + -free media, and Zn 2+ data sets were 6.8±0.0, 6.9±0.1 and 6.7±0.0, respectively. None of the inhibitors resulted in a final pH i that was significantly different from that induced by PAF alone. (B) Eosinophils stimulated with media control (n=6) showed no significant changes in pH i compared with pH i values at time 0. Eosinophils stimulated with PAF alone (n=7) or in the presence of bafilomycin (n=7) or folimycin (n=5) showed similar levels of alkalinization. Initial pH i values for media control, PAF, PAF + bafilomycin and PAF + folimycin data sets were all 7.0±0.1. There were no significant differences in final pH i in when cells were stimulated with PAF, PAF + bafilomycin or PAF + folimycin. *P<0.01 (compared to pH i at time 0); n.s., not statistically different. at least one of these subunits was also translated and expressed in the peripheral blood eosinophils, we performed western blot analysis on H + -ATPase subunit E in whole cell eosinophil lysates (Fig. 7). Additionally, we stained eosinophils prepared by cytospin to document the location of the H + -ATPase subunit in the cells (Fig. 8). As expected, the staining had a granular pattern, consistent with our LysoSensor Yellow/Blue staining and with previous reports of H + -ATPase expression in human eosinophils (Kurashima et al., 1996;Persson et al., 2002).

Discussion
The results of this study demonstrate that PAF, a signaling molecule that increases intracellular calcium, produces significant and persistent intracellular alkalinization in human eosinophils. Similar effects were observed with other agents that increase [Ca 2+ ] i , including ionomycin and thapsigargin. However, alkalinization was not produced by the PKC activating compound PMA or by IL-5, which activate eosinophils without increasing [Ca 2+ ] i . Although increases in [Ca 2+ ] i appear to play a primary role in cytosolic alkalinization, our finding that the magnitude of ionomycin-induced alkalinization is reduced when eosinophil adhesion is blocked suggests that a secondary signal is required for maximum effect. Because PAF stimulates cytosolic alkalinization equally well regardless of cell adhesion, we propose that PAF-induced signaling alone is sufficient to maximally stimulate alkalinization, whereas ionomycin provides only partial stimulation and appears to require β2 integrin ligation for full effect.
We explored the mechanism of PAF-induced cytosolic alkalinization using several blockers of plasma membrane proton transporters. Neither replacement of Na + nor treatment with EIPA blocked PAF-stimulated alkalinization, suggesting that PAF does not increase cytoplasmic pH by stimulating Na + -H + exchange or by activating other acid/base transport mechanisms that depend on extracellular Na + . To determine whether plasma membrane proton channels were involved, eosinophils were treated with Zn 2+ , which inhibits proton channel activity, and subsequently stimulated with PAF. The alkalinization effect of PAF was not inhibited. Furthermore, activation of proton channels following PMA addition had no effect on pH i , indicating that the alkalinization effects of PAF were not mediated by plasma membrane proton channels. Pre-treatment of eosinophils with bafilomycin or folimycin, blockers of plasma membrane H + -ATPase activity, had no significant effect on PAF-stimulated alkalinization suggesting that proton transport across the plasma membrane was not dependent on H + -ATPase activity.
Although inhibitors or Na + replacement did not significantly affect the total magnitude of PAF-induced eosinophil alkalinization, the baseline cytosolic pH i was affected. Conditions designed to inhibit Na + -H + exchange activity (NMDG or EIPA) produced relatively small decreases in basal pH presumably owing to the dependence of intracellular pH upon basal Na + -H + exchange activity. The effect of EIPA appeared to be greater than Na + replacement, but the reasons for Representative images of eosinophils stimulated with IL-5 or PAF for 40 minutes at room temperature. Note the granular appearance of the dye, consistent with accumulation of the dye within the characteristic secondary granules of the eosinophil. Fig. 7. Eosinophils express V-ATPase subunit E in whole cell lysates. Eosinophil whole cell lysates were prepared and examined for V-ATPase subunit E by western blotting as described in Materials and Methods. Commercially available mouse brain extract was used as a positive control (+). Shown are representative control (nIgG) and test (V-ATPase E) blots from three independent experiments using eosinophils. Each Eo Lysates lane represents an eosinophil extract from a separate donor. A band specific to the V-ATPase E blot is seen around 30 kDa in the positive control and in each eosinophil extract lane (arrow). The non-specific band around 48 kDa in the eosinophil extracts is probably eosinophil peroxidase (EPO). this are not clear. Pre-treatment with ZnCl 2 produced a decrease in eosinophil resting pH i to a greater extent than was observed with either EIPA or Na + replacement. Although Zn 2+ inhibits plasma membrane proton channels, these channels are only minimally active in unstimulated eosinophils and thus inhibition of their activity by Zn 2+ would probably not account for the lower initial pH i (Bankers-Fulbright et al., 2001).
Negative results obtained from experiments designed to identify plasma membrane transport pathways involved in cytoplasmic alkalinization suggested that activation of a proton transport mechanism associated with intracellular organelles could explain the PAF-induced increase in pH i . Because eosinophils contain many acidic intracellular granules, we investigated the effects of PAF on intragranule pH using a fluorescent pH indicator that preferentially partitions into acidic compartments within the cell. PAF stimulation produced a timedependent decrease in intragranule pH that correlated with cytoplasmic alkalinization. Although we could not quantitatively determine the rate of acidification, these results did provide qualitative evidence for PAF-stimulated proton transport into eosinophil granules. Because ionomycin and thapsigargin mimic the actions of PAF, an increase in [Ca 2+ ] i is presumably necessary to activate proton transport into the granules.
Previous studies on human eosinophils and the present study have demonstrated the presence and functional activity of H + -ATPase proteins associated with granule membranes (Kurashima et al., 1996;Persson et al., 2002). Furthermore, increases in [Ca 2+ ] i have been previously shown to play a role in regulating H + -ATPase activity in other cell types. For example, when porcine oocytes are stimulated with the calcium ionophore A23187, they show a large, time-dependent increase in pH i that is inhibited by bafilomycin (Ruddock et al., 2000). Increases in H + -ATPase activity in neutrophils have also been documented following stimulation with chemo-attractant peptides that produce increases in [Ca 2+ ] i (Nanda and Grinstein, 1995), and inhibition of [Ca 2+ ] mobilization significantly reduced the degree of pump activation. More recently, studies of rabbit non-pigmented ciliary epithelium revealed that elevation of [Ca 2+ ] i produces significant cytoplasmic alkalinization that is inhibited by bafilomycin (Hou et al., 2001). Inhibition of calcium channel activity with verapamil or nifedipine blocks the effects of membrane depolarization on Ca 2+ influx and inhibits the alkalinization response. Calcium-dependent stimulation of H + -ATPase activity has been most recently reported in MDCK cells following stimulation with arginine vasopressin (AVP). AVP activates both V 1 and V 2 vasopressin receptors to produce increases in cAMP and [Ca 2+ ] i . However, increases in [Ca 2+ ] i alone were not sufficient for H + -ATPase activation, indicating that both cAMP and increases in [Ca 2+ ] i are necessary. The results of these previous studies demonstrate that Ca 2+dependent cytoplasmic alkalinization mediated by H + -ATPase activation is not unique to eosinophils. However, the localization of proton pumps in the granule membranes of eosinophils provides a mechanism for proton transport from the cytoplasm into the granules that can be stimulated by signaling molecules that increase [Ca 2+ ] i .
One approach to confirm the role of a granule membrane Vtype H + -ATPase in cytosolic alkalinization could be to treat the cell with an H + -ATPase blocker, such as bafilomycin. However, although it is established that bafilomycin and other macrolide antibiotics can accumulate in the plasma membrane, it is unclear whether or not they subsequently partition into intracellular organelle membranes in intact cells (Drose and Altendorf, 1997). In the present study, bafilomycin did not inhibit PAF-stimulated cytosolic alkalinization, suggesting that bafilomycin does not partition readily into eosinophil granules in intact cells. However, previous reports examining the effect of bafilomycin or other macrolide antibiotics on human eosinophils indicate that these blockers can cause relative neutralization of the secondary granules in unstimulated or IL-5 stimulated eosinophils, with no effect on cytosolic pH (Kurashima et al., 1996;Persson et al., 2002). Thus, the Ca 2+ -regulated V-type H + -ATPase proposed in our model may differ from that previously reported in accessibility and/or sensitivity to bafilomycin.
Eosinophil granules contain several toxic proteins that are released in response to agonist activation. These proteins include EDN, eosinophil cationic protein (ECP) and eosinophil peroxidase (EPO), each stored in soluble form in the granule matrix and major basic protein (MBP) stored in a crystalline form on a negatively charged backbone protein (Egesten et al., 2001). Of these proteins, MBP is thought to be an important mediator of eosinophil-induced tissue damage. MBP tissue deposition can be clearly detected in damaged areas of lungs from patients with asthma, as well as in skin lesions from atopic dermatitis patients and on helminths in the gut . Furthermore, treatment of guinea pig tracheal rings with purified MBP results in epithelial desquamation indistinguishable from that seen in human asthma (Motojima et al., 1989), suggesting that this granule protein in particular plays a role in asthma pathogenesis. Although virtually all eosinophil stimuli, including PMA, PAF, IL-5, GM-CSF, IFNγ and secretory IgA (Giembycz and Lindsay, 1999), induce at least some EDN release, only Ca 2+ -mobilizing stimuli appear to induce substantial MBP release within a short time period following stimulation . Interestingly, purification of MBP in vitro requires the use of solutions at very low pH (~2) to solubilize the protein fully . Thus, a potential function of granule acidification could be to solubilize Journal of Cell Science 117 (24) Fig. 8. The V-ATPase subunit E is localized to granular compartments in eosinophils. Eosinophils were isolated and stained for the V-ATPase subunit E and were analyzed by confocal microscopy as described in Materials and Methods. Shown are representative images from four independent experiments. Original images were taken at a magnification of ϫ630. (A) Eosinophils stained with an isotype control antibody followed by FITC-conjugated secondary antibody. (B) Eosinophils stained with anti-V-ATPase subunit E followed by FITC-conjugated secondary antibody show a granular staining. the MBP core in preparation for degranulation. The need for granule acidification for MBP solubilization and optimal degranulation could explain the long-standing observation that NH 4 + -induced granule neutralization inhibits degranulation. In conclusion, activation of human eosinophils with agonists that increase [Ca 2+ ] i results in a large, persistent increase in pH i . Attempts to identify plasma membrane transport pathway(s) responsible for this effect were unsuccessful. However, measurements of intragranule pH following stimulation with PAF revealed an increase in acidification that was temporally correlated with alkalinization of the cytoplasm. We speculate that intragranule acidification may be an important step in the solubilization of MBP crystals that are stored within the granule and that calcium-mobilizing stimuli effectively initiate the release of soluble MBP. Finally, our results do not exclude the possibility that some portion of the alkalinization response observed following treatment with PAF, ionomycin or thapsigargin might result from activation of a plasma membrane transport mechanism. However, the experimental conditions that block the function of previously identified acid transport pathways in granulocytes do not inhibit calcium-dependent, cytoplasmic alkalinization in human eosinophils.