Inositol 1,4,5-trisphosphate-dependent oscillations of luminal [Ca2+] in permeabilized HSY cells.

Oscillations in intracellular Ca2+ concentration ([Ca2+]i) are thought to play an important role in phosphoinositide-linked Ca2+ signaling events. We demonstrate corresponding inositol 1,4,5-trisphosphate (IP3)-dependent oscillations of Ca2+ concentration within the lumen of the IP3-sensitive stores ([Ca2+]L) of saponin-permeabilized HSY cells by monitoring [Ca2+]L with the fluorescent Ca2+ indicator Mag-fura-2. The associated openings and closings of the IP3-sensitive Ca2+ release channel were detected via quenching of Mag-fura-2 fluorescence due to the entry of Mn2+, a Ca2+ surrogate, into the stores. Evidence for complimentary Ca2+ oscillations at the external surface of the stores was provided by the membrane-bound Ca2+ probe Calcium Green C18. The permeabilization of saponin-treated HSY cells to macromolecules was confirmed by demonstrating that permeabilized cells readily took up and lost (t1/2 ≈ 46 s) a fluorescently tagged 70-kDa dextran. Our results impose a number of constraints on possible mechanisms for [Ca2+]i oscillations. In addition, they support recent proposals that [Ca2+]i oscillations arise directly from the (biphasic) effects of Ca2+ itself on IP3-sensitive Ca2+ channel activity.

In many cell types, stimulation of receptors linked to phosphoinositide signaling pathways results in oscillations of intracellular calcium concentration ([Ca 2ϩ ] i ) due to the periodic release and reuptake of Ca 2ϩ by intracellular stores (1)(2)(3)(4). These oscillations often take the form of "base-line spikes" in which periodic spikes of [Ca 2ϩ ] i rise from a base-line [Ca 2ϩ ] i which is close to resting levels. It has been suggested that [Ca 2ϩ ] i oscillations provide a sort of digital encoding of the receptor-mediated signal that may have certain advantages, such as noise reduction and decreased exposure to high intracellular Ca 2ϩ levels, over a continuously graded increase in [Ca 2ϩ ] i (1)(2)(3)(4).
Although a number of explanations have been proposed to account for [Ca 2ϩ ] i oscillations (1)(2)(3)(4), at present there is no clear consensus regarding the intracellular events responsible for their generation. Early suggestions included models in which these oscillations were linked to oscillations in intracellular IP 3 1 concentration, and models involving interactions between Ca 2ϩ release from two (or more) intracellular Ca 2ϩ pools. In the former case oscillations in intracellular IP 3 levels were proposed to arise from feedback by protein kinase C or [Ca 2ϩ ] i itself on early steps in phosphoinositide signaling. In the latter case it was suggested that IP 3 -dependent Ca 2ϩ release from one pool induces an explosive Ca 2ϩ release from another Ca 2ϩ -sensitive or less IP 3 -sensitive pool, thereby generating a spike in [Ca 2ϩ ] i followed by a period of reloading then a retriggering of explosive Ca 2ϩ release. More recently, models incorporating the experimentally observed modulatory effects of Ca 2ϩ itself on the IP 3 -sensitive Ca 2ϩ channel (5, 6) have received considerable attention (1-3, 7, 8). These experiments indicated that the open probability of the IP 3 -sensitive channel had a bell-shaped dependence on [Ca 2ϩ ] i such that small increases in [Ca 2ϩ ] i above resting levels would be expected to activate the channel, but large increases would be inhibitory. Several detailed mathematical analyses (9 -11)  To date experimental studies of [Ca 2ϩ ] i oscillations have utilized intact cells, making it difficult to definitively confirm or exclude the involvement of many intracellular processes in this phenomenon, and consequently complicating tests of proposed mechanisms. In the work presented here we extend these studies to permeabilized cells. For our experiments we have employed the HSY cells, a human salivary epithelial cell line (12) that has been found to be a useful model system for investigations of intracellular Ca 2ϩ signaling events (13)(14)(15). We have previously shown (15) that the luminal calcium concentration ([Ca 2ϩ ] L ) within the IP 3 -sensitive intracellular Ca 2ϩ stores of saponin-permeabilized HSY cells can be monitored with the fluorescent Ca 2ϩ indicator Mag-fura-2. Using this permeabi-* 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 all correspondence should be addressed: National Institute of Dental Research, Bldg. 10

MATERIALS AND METHODS
Media-Intracellular-like medium (ICM) contained 125 mM KCl, 19 mM NaCl, 10 mM HEPES (pH 7.3 with KOH), 3 mM ATP, 1.4 mM MgCl 2 , 330 M CaCl 2 , and 1 mM EGTA (calculated free Ca 2ϩ and Mg 2ϩ concentrations 50 nM and 0.1 mM, respectively; the former was confirmed fluorescently using fura-2). Calcium sponge-treated medium (CaS) was made from a solution containing 125 mM KCl, 19 mM NaCl, 10 mM HEPES (pH 7.3 with KOH) and 1 mM ATP. To remove Ca 2ϩ , approximately 60 ml of this solution was passed twice over a column containing 0.3 g of Calcium Sponge S (Molecular Probes, Eugene OR), regenerating the column after each passage by washing with the same solution (without ATP) adjusted to pH 4.0. Following Ca 2ϩ removal, 1.4 mM MgCl 2 was added. The final Ca 2ϩ concentration of CaS was Ϸ50 nM (determined fluorescently using fura-2). In experiments utilizing confocal microscopy an additional 10 mM glucose was added to all media.
Mag-fura-2 Loading and Cell Permeabilization-HSY cells were cultured as described previously (14). Experiments were carried out on single cells growing in glass sample chambers precoated with fibronectin (15). Cells were loaded with Mag-fura-2 and permeabilized essentially as described previously (15). Briefly, cells were incubated with 6 -8 M Mag-fura-2/AM (Molecular Probes, Eugene OR) for 25 min at 37°C in a physiological salt solution, rested for 10 -60 min, then permeabilized by exposure to 70 g/ml (w/v) saponin (Calbiochem, San Diego CA) in Mg 2ϩ /ATP-free ICM for ϳ1 min at room temperature. Following permeabilization cells were washed with Mg 2ϩ /ATP-free ICM then switched to complete ICM and left for at least 5 min to allow for filling of their intracellular Ca 2ϩ stores. Cells were then switched to CaS for the oscillation experiments.
When intact HSY cells loaded with Mag-fura-2 were examined under the fluorescence microscope the dye appeared to be uniformly distributed throughout the cytoplasm. After exposure to saponin ϳ75% of the intracellular fluorescence was lost, presumably as a result of permeabilization of the plasma membrane and release of cytosolic dye (ϳ50% of the cells were permeabilized in this way under the experimental conditions described above). The remaining fluorescence appeared as a diffuse ring encircling the (darker) cell nucleus (not shown). Extensive washing had little effect on the Mag-fura-2 retained by permeabilized HSY cells (15), consistent with the hypothesis that it was trapped in a membrane-bound compartment.
Calcium Green C 18 Loading-In experiments utilizing Calcium Green C 18 , cells were permeabilized as usual, then just before the fluorescence measurements they were incubated with 5-10 M Calcium Green C 18 (Molecular Probes) in ICM for 2-3 min, switched to ICM and incubated for 5 min, then switched to CaS. Cells labeled in this way exhibited a stable Calcium Green C 18 fluorescent signal that was unaffected by subsequent washing.
Microfluorometry-Mag-fura-2 and Calcium Green C 18 fluorescence from saponin-permeabilized HSY cells was monitored using an ARCM-MIC microfluorometer (Spex Industries, Edison, NJ) coupled to a Nikon Diaphot microscope equipped with a Nikon CF Fluor 40ϫ objective. Fluorescence emission was measured with a photomultiplier tube attached to the side port of the microscope. A pinhole aperture placed in front of the photomultiplier tube restricted its view to a field 50 m in diameter containing a single permeabilized HSY cell. A vacuum line was placed in the sample chamber and adjusted so that the volume was maintained at ϳ50 l. Solution changes were accomplished by the addition of 600-1200 l of the new solution to the sample chamber. All experiments were carried out at room temperature. Permeabilized cells loaded with Mag-fura-2 alone were alternately excited with light at 344 and 360 nm at 1-s intervals, and fluorescence emission was monitored at 510 nm. Mag-fura-2 fluorescence at 344 nm excitation (the isosbestic wavelength) was unaffected by IP 3 . In cells loaded with Calcium Green C 18 and Mag-fura-2, the dyes were excited at wavelengths of 495 and 360 nm, respectively, and fluorescence emission from both was collected at 530 nm.
Confocal Microscopy-Confocal microscopy was carried out using a Leica TCS 4D system (Leica, Heidelberg, Germany) equipped with a 40 ϫ PL Fluotar objective. Confocal images (256 ϫ 256 pixels) of Calcium Green C 18 and rhodamine B-dextran (70 kDa; Molecular Probes) fluorescence included in the paper were obtained at 488 nm excitation/530 nm emission and 568 nm excitation/590 nm emission, respectively, using a 170-m pinhole. For measurements of rhodamine B-dextran loss from permeabilized cells (see below), frames were scanned every 3-4 s over the first minute after extracellular dextran removal and every minute for 9 min thereafter. For measurements of IP 3 -dependent oscillations in Calcium Green C 18 fluorescence the excitation beam (488 nm) was attenuated (ϳ98%) by passage through a 520 nm cutoff filter. In these experiments frames (128 ϫ 128 pixels) were scanned every 2.5 s using a 400 -500 M pinhole. The Calcium Green C 18 and rhodamine B-dextran fluorescent signals from single cells were obtained from the confocal images using software supplied with the instrument.

RESULTS AND DISCUSSION
In order to observe [Ca 2ϩ ] L in permeabilized cells we have employed the low affinity Ca 2ϩ -sensitive fluorescent dye Magfura-2 (K d for Ca 2ϩ Ϸ 53 M) (16) which, under appropriate loading conditions, accumulates in the IP 3 -sensitive stores of HSY (15) and other cells (17,18) and can be used to monitor changes in [Ca 2ϩ ] L . Preliminary experiments in our laboratory have established that single intact HSY cells loaded with the high affinity [Ca 2ϩ ] i indicator fura-2 exhibit typical base-line spike-type [Ca 2ϩ ] i oscillations following stimulation with IP 3 mobilizing agonists (not shown). In Fig. 1 we show a complimentary experiment in a single Mag-fura-2-loaded, saponinpermeabilized HSY cell exposed to successively greater concentrations of IP 3 . The ratio of Mag-fura-2 fluorescence at excitation wavelengths 344 and 360 nm, which is a direct measure of [Ca 2ϩ ] L (15)(16)(17), is seen to oscillate several times when [IP 3 ] is increased from 0.3 to 0.6 M, and to oscillate regularly when [IP 3 ] Ն 1 M. In experiments of this type we have found that oscillations in [Ca 2ϩ ] L occur in response to IP 3 in approximately 50% of HSY cells treated as described here, with some variation in sensitivity to IP 3  The oscillations in [Ca 2ϩ ] L observed in Fig. 1 have a characteristic shape that was seen in virtually all of our experiments. Each oscillation consists of an abrupt fall in the Mag-fura-2 ratio followed by a (usually incomplete) recovery, then a pause or refractory period before the beginning of the next oscillation. A similar pattern has recently been observed in measurements of [Ca 2ϩ ] L during agonist-induced [Ca 2ϩ ] i oscillations in intact cells (18,19). In our experiments the half times for Ca 2ϩ re- In order to investigate the fate of the Ca 2ϩ lost from intracellular stores during [Ca 2ϩ ] L oscillations we briefly exposed Mag-fura-2-loaded, permeabilized HSY cells to the lipophilic Ca 2ϩ indicator Calcium Green C 18 (20). This indicator consists of a highly hydrophilic calcium-green molecule conjugated to a lipophilic alkyl chain that will insert into biological membranes. Thus indicators of this type can be used to monitor ion concentrations (in this case free [Ca 2ϩ ]) immediately adjacent to cell membranes (20). Using confocal microscopy we have confirmed that Calcium Green C 18 strongly labels intracellular sites in permeabilized HSY cells (see below); in contrast, in intact cells the dye is excluded from the cell interior and only labels the plasma membrane. These results are consistent with the incorporation of Calcium Green C 18 into only those cell membranes with which it has direct contact. In Fig. 2A we illustrate the results of an experiment in which Mag-fura-2 fluorescence (upper trace) and Calcium Green C 18 fluorescence (lower trace) were monitored simultaneously in the same permeabilized HSY cell. This experiment demonstrates that each oscillation in [Ca 2ϩ ] L reported by Mag-fura-2 is accompanied by a complimentary oscillation in [Ca 2ϩ ] reported by Calcium Green C 18 . In particular, the Calcium Green C 18 signal rises as [Ca 2ϩ ] L falls, peaking close to the point of the [Ca 2ϩ ] L mini-mum, then falls to base-line levels as [Ca 2ϩ ] L recovers. Given these strong temporal correlations and the known properties of the two fluorescent probes the obvious explanation of this result is that the Ca 2ϩ released from the IP 3 -sensitive intracellular store causes an increase in [Ca 2ϩ ] close to the surface of the store ([Ca 2ϩ ] S ) which is, in turn, detected by the Calcium Green C 18 incorporated into these membranes.
In Fig. 2B we show that a similar correlation between [Ca 2ϩ ] L and [Ca 2ϩ ] S is seen even in the presence of 100 M EGTA. Thus the quantity of Ca 2ϩ released during each oscillation is sufficiently large that it cannot be buffered by this concentration of EGTA, suggesting that the increase in [Ca 2ϩ ] S near the IP 3 -sensitive channel is also quite large. In numerous previous experiments carried out in the presence of 1 mM EGTA, the application of IP 3 to permeabilized HSY cells consistently resulted in sustained Ca 2ϩ release rather than oscillations of [Ca 2ϩ ] L (15) (experiments were carried out essentially as described here but in ICM). This sustained release was accompanied by little or no increase in [Ca 2ϩ ] S as monitored by Calcium Green C 18 (not shown).
The loss of oscillatory behavior with increased Ca 2ϩ buffering by EGTA is consistent with the hypothesis that modulatory effects of [Ca 2ϩ ] i itself underlie the generation of [Ca 2ϩ ] i oscillations. It has been suggested that Ca 2ϩ -dependent phosphorylation/dephosphorylation events may be involved in these effects (21). However, in our experiments we have found that [Ca 2ϩ ] L oscillations persist in permeabilized HSY cells in the presence of either the relatively nonspecific protein kinase inhibitor staurosporine (1 M; n ϭ 3) or the Ca 2ϩ /calmodulindependent kinase inhibitor compound 5 (2 M; n ϭ 3) arguing against a role of protein phosphorylation in oscillatory behavior in this system.
In order to monitor the opening of the IP 3 -sensitive Ca 2ϩ channel directly we took advantage of earlier demonstrations that Mn 2ϩ is a substrate for the channel (15,(22)(23)(24) and of the fact that Mn 2ϩ entry can be detected by its quenching of Magfura-2 fluorescence. A typical experiment is shown in Fig. 3. Mag-fura-2 fluorescence measured at 344 nm excitation (the isosbestic wavelength) and the 344/360 fluorescence ratio are shown. Thus the former is an indicator of Mag-fura-2 fluores- cence quenching while the latter provides a measure of [Ca 2ϩ ] L . At the onset of each oscillation in [Ca 2ϩ ] L a relatively rapid increase in the rate of quenching of Mag-fura-2 is observed, indicating increased Mn 2ϩ entry into the Mag-fura-2 compartment. Note, however, that there is a significant delay between the point of maximal Ca 2ϩ release and the closing of the channel as indicated by termination of Mn 2ϩ -induced fluorescence quenching (11.7 Ϯ 2.1 s; n ϭ 3). This observation is initially puzzling since it argues against the intuitive notion that [Ca 2ϩ ] L recovery occurs as a consequence of channel closure. But as discussed below this behavior is, in fact, predicted by recently proposed models of [Ca 2ϩ ] i oscillations.
An obvious concern in the interpretation of the data presented above is the degree of permeabilization resulting from the procedures employed here. This issue was explored in the experiments illustrated in Fig. 4. In the left-hand panel of Fig.  4 we show five images of the same field of HSY cells obtained as follows. A is a bright field image of seven HSY cells treated with saponin, then exposed to Calcium Green C 18 as described under "Materials and Methods." B is a confocal image of Calcium Green C 18 fluorescence. Note that in two of the cells there is considerable fluorescent labeling of internal sites indicating that they are permeabilized to Calcium Green C 18 , while in the other five cells the dye has been largely excluded from the cell interior and a clearly delineated ring of plasma membrane labeling is seen. Next the cells were exposed to rhodamine B-dextran (70 kDa; 0.1 mg/ml). C is a confocal image of rhodamine B-dextran fluorescence taken 5 min later. The image shows incorporation of the dextran into the same two cells that were permeabilized to Calcium Green C 18 in image B and exclusion from the five others. D and E are also confocal images of rhodamine B-dextran fluorescence taken 30 s and 10 min, respectively, after washing away extracellular dextran. It is clear from images C, D, and E that the two cells that are permeabilized to Calcium Green C 18 also readily take up and lose rhodamine B-dextran and thus are likewise permeabilized to macromolecules (70 kDa). In experiments of this type we found that, of 119 cells showing intracellular staining with Calcium Green C 18 (cf. image B), only two excluded rhodamine B-dextran. Of these 117 cells permeabilized to rhodamine Bdextran, 50% (58 cells) exhibited oscillations in Calcium Green C 18 fluorescence after exposure to IP 3 (right-hand panel of Fig.  4).
The degree of permeabilization to rhodamine-dextran was estimated in a subset of the cells examined above by determining the t1 ⁄2 for rhodamine-dextran fluorescence loss following the removal of extracellular dye (see "Materials and Methods"). In cells showing oscillations in Calcium Green C 18 fluorescence in response to IP 3 this t1 ⁄2 was 46.1 Ϯ 7.0 s (n ϭ 33), while in nonoscillating cells t1 ⁄2 was 46.3 Ϯ 8.1 s (n ϭ 37). Thus it is clear that there is no significant difference in dextran permeability between oscillating and nonoscillating cells.
The above experiments impose a number of constraints on possible explanations of [Ca 2ϩ ] i oscillations in HSY cells. The observation that oscillations of [Ca 2ϩ ] L can be observed in permeabilized cells where the containment of cytosol and the permeability barrier of the plasma membrane are largely lost argues strongly against the participation of any cellular enzymes or structures that are not a part of, or localized very close to, the IP 3 -sensitive Ca 2ϩ stores. In addition, it indicates that the involvement of any diffusible cytosolic messengers must occur over relatively short distances and/or the localized concentrations of these messengers must be large. The possibility that oscillations in [Ca 2ϩ ] i are linked to oscillations in [IP 3 ] in HSY cells is essentially excluded. Furthermore the fact that pulsatile Ca 2ϩ release persists as the Ca 2ϩ content of the stores gradually falls (Fig. 1) cannot be accounted for by models that require the complete refilling of intracellular stores between oscillations. These latter observations also argue strongly against a significant role of [Ca 2ϩ ] L itself in the mechanism by which oscillatory behavior is generated.
Our observations cannot definitively exclude the possibility that [Ca 2ϩ ] i oscillations may involve interactions between closely associated intracellular Ca 2ϩ pools. But as indicated above they do considerably limit the way in which these proposed mechanisms can operate. It is particularly interesting to note, however, that our results are quite consistent with recently proposed mechanisms for [Ca 2ϩ ] i oscillations (7,8) based on the well documented biphasic effects of Ca 2ϩ itself on the IP 3 -sensitive Ca 2ϩ channel (5,6). Since these schemes propose that [Ca 2ϩ ] i oscillations arise directly from the properties of the channels themselves, they have as a corollary that this phe- FIG. 4. Permeabilization of saponin-treated HSY cells to rhodamine B-dextran (70 kDa). Left-hand panel, following treatment with saponin and exposure to Calcium Green C 18 (see "Materials and Methods"), HSY cells were incubated with rhodamine B-dextran (0.1 mg/ml in ICM) for 5 min, after which the dextran was washed away. Calcium Green C 18 and rhodamine B-dextran fluorescences were monitored via confocal microscopy as described under "Materials and Methods." Images A-E show the same field (140 m ϫ 140 m) of seven HSY cells. A is a bright field image; B is a confocal image of Calcium Green C 18 fluorescence; C, D, and E are confocal images of rhodamine B-dextran fluorescence taken before washing away the dextran (C), 30 s after washing away extracellular dextran (D) and 10 min after dextran removal (E). Right-hand panel, IP 3 -induced oscillations in Calcium Green C 18 fluorescence in a single saponin-permeabilized HSY cell which was also confirmed to be permeabilized to rhodamine B-dextran. Intracellular Calcium Green C 18 fluorescence (in arbitrary units) has been smoothed by averaging data points over a 10-s window (in order to confirm the dextran permeability of the cell the data were collected on the confocal microscope, which is significantly noisier than microfluorimetry for measurements of this type; cf. Fig. 2). nomenon could also occur in permeabilized cells, as confirmed here. In addition, mathematical modeling of these mechanisms, based on experimentally derived parameters for the feed forward and feedback effects of Ca 2ϩ on the IP 3 -sensitive Ca 2ϩ channel, yields theoretical predictions for the oscillatory behavior of [Ca 2ϩ ] L , [Ca 2ϩ ] i , and IP 3 -sensitive Ca 2ϩ channel activity that are remarkably similar to our experimental results (cf. Ref. 8, Fig. 6). In particular, these calculations reproduce the shape of the oscillations in [Ca 2ϩ ] L described here and, as alluded to above, predict that Ca 2ϩ channels remain open during the initial phase of [Ca 2ϩ ] L recovery. In these models this delay in channel closure is due to the slow time course of Ca 2ϩ feedback inhibition. The recovery of [Ca 2ϩ ] L , in spite of the fact that the channels remain open, occurs because of the high [Ca 2ϩ ] i generated during oscillations. This high [Ca 2ϩ ] i increases Ca 2ϩ influx via the SERCA pump while simultaneously decreasing the driving force for Ca 2ϩ loss via the IP 3sensitive channels. The net result is that Ca 2ϩ uptake exceeds loss even before the channels close. In our experiments [Ca 2ϩ ] S is apparently large enough to duplicate this effect, but because Ca 2ϩ is able to easily diffuse away from the surface of the stores of permeabilized cells, [Ca 2ϩ ] L recovery is typically incomplete (cf. Figs. 1-3).
Our results provide the first demonstration to date that Ca 2ϩ oscillations can occur in permeabilized cells and suggest that the permeabilized HSY cell preparation described here offers many interesting possibilities for future studies of Ca 2ϩ oscillations.