Ca²⁺ dynamics in salivary acinar cells: distinct morphology of the acinar lumen underlies near-synchronous global Ca²⁺ responses

The importance of intracellular Ca²⁺ in stimulus-secretion coupling in secretory epithelial cells has been long recognized (Matthews et al., 1973; Hunter et al., 1983; Petersen, 1992). The Ca²⁺ response is primarily the release of Ca²⁺ from intracellular stores and shows complex variations in time and space. Ca²⁺ oscillations in epithelia were first shown in parotid acinar cells (Gray, 1988) and since Kasai and Augustine's seminal paper on pancreatic acinar cells (Kasai and Augustine, 1990) an apical initiation and basal propagation of Ca²⁺ waves has become dogma in the field (Thorn et al., 1993; Toescu et al., 1992; Elliot et al., 1992; Lee et al., 1997). However, it is not clear if this is the case for salivary acinar cells where the patterns of Ca²⁺ signals in are still not resolved. Some report apical-to-basal Ca²⁺ waves are either not seen [parotid (Dissing et al., 1990)] or are very rapid, giving a near simultaneous Ca²⁺ response across the cell (Giovannucci et al., 2002; Takemura et al., 1999; Liu et al., 1998). By contrast, others report Ca²⁺ waves [submandibular (Lee et al., 1997; Harmer et al., 2005), parotid (Tojyo et al., 1997)] apparently similar to those seen in pancreatic acinar cells. A resolution of these different observations is essential to our general understanding of how Ca²⁺ signals are generated and how they regulate secretion.

There are two major problems with previous imaging studies of Ca²⁺ in polarized epithelia. Firstly, most use preparations of isolated cells, and small clusters of cells (<10 cells), where the characteristic morphology of polarized acinar cells is compromised because of the severing of tight junctions during cell isolation (Park et al., 2004). Secondly, the acinar lumen is not normally visible with light microscopy, so the apical domain is often identified simply as the region containing secretory granules. This distinction is functionally important; the luminal plasma membrane is the only region where exocytosis takes place and, in the generation of Ca²⁺ signals, inositol trisphosphate receptors (InsP₃Rs) are highly enriched, specifically in the endoplasmic reticulum immediately adjacent to the acinar lumen (Lee et al., 1997; Yule et al., 1997; Zhang et al., 1999).

We have overcome both of these problems using live-cell two-photon microscopy. Ca²⁺ responses from single cells, within large fragments of exocrine tissue that retain the morphological features of the intact gland (Park et al., 2004), were recorded with Fura-2. We visualized the ducts and acinar lumen by the simultaneous imaging of an extracellular fluorescent dye (Thorn et al., 2004).

Our experiments compare salivary and pancreatic tissue and show that in submandibular acinar cells agonists evoke a near-synchronous Ca²⁺ rise across the cell comparable to the apical to basal Ca²⁺ waves seen pancreatic acinar cells. We show the acinar lumen in submandibular tissue is significantly more extensive than in pancreas, often leading to an encircling of single acinar cells with a band of InsP₃Rs. This hitherto unrecognized three-dimensional organization of InsP₃Rs explains the near-synchronous global Ca²⁺ signals in submandibular acinar cells, and the extensive luminal area is probably an adaptation in these cells that secrete copious amounts of fluid.

Lobules and fragments (∼50-100 cells) of mouse submandibular gland were prepared by collagenase digestion in normal Na⁺-rich extracellular solution using the method of Thorn et al. (Thorn et al., 1993) modified to reduce the time in collagenase and limit mechanical trituration and cells were then plated onto poly-l-lysine-coated glass coverslips.

Na⁺-rich extracellular solution contained (mM): 135 NaCl, 5 KCl, 1 MgCl₂, 10 Hepes, 10 glucose, 2 CaCl₂; pH 7.4. K⁺-rich solution (mM): 5 NaCl, 135 KCl, 1 MgCl₂, 10 Hepes, 10 glucose, 2 CaCl₂; pH 7.4.

Cells attached to glass coverslips were washed in PBS, fixed in methanol for 10 minutes at –20°C. After 1 hour in 2% donkey serum plus 2% fish skin gelatine in PBS, cells were incubated in primary antibody for 1 hour and then secondary antibodies for 30 minutes. The antibody dilutions were as follows: InsP₃R2 (pAb; Chemicon, Temecula, CA, USA), 1:20; InsP₃R3 (mAb; BD Transduction Laboratories, San Jose, CA, USA) 1:100; ZO-1 3 μg/ml (Zymed Laboratories Inc., San Francisco, CA, USA); AQP5 3 μg/ml (Alpha Diagnostic International Inc., San Antonio, TX, USA). The secondary antibodies used were Cy3-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories Inc., USA; 1:200 dilution) and FITC-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories Inc., USA; 1:75 dilution), which showed no significant staining when applied alone. For the InsP₃R3 antibody, staining was abolished by preincubation with HeLa cell lysate containing the antigen.

Images were obtained on a Zeiss Axiovert LSM510 confocal microscope (Zeiss, Welwyn Garden City, UK), with a 63× oil immersion 1.4 NA objective. FITC was excited at 488 nm and emitted light captured through a 505-530 nm filter. Cy3 was excited at 543 nm and the emitted light captured through a 560 nm long-pass filter. Confocal image sections (z ∼1 μm) were processed using Metamorph (Universal Imaging Corporation, Downington, PA, USA). The images in Figs 3 and 4 were obtained by a maximal projection of a stack of z sections taken through the acinus. Three-dimensional-rendered movies were created with Imaris (Bitplane Inc., St Paul, MN, USA).

A Leica TCS MP, two-photon microscope with a 63× water immersion 1.2 NA objective, was used to collect images (resolution of 8-19 pixels/μm), which were analysed with Metamorph. Fura-2 regional kinetics were measured from regions of interest (16.9 μm²). We observed no significant photobleaching; traces were rejected if movement was observed.

Ducts and the acinar lumen were imaged using either sulphorhodamine B (SRB; 500 μM) or Oregon Green 488 BAPTA-1 (OG; 100 μM) as membrane-impermeant extracellular markers. Dyes were excited at 740-780 nm, with fluorescence emission detected at 550-650 nm and <550 nm, respectively. Intracellular Ca²⁺ was monitored (emission of 490-540 nm) in cells loaded with Fura-2AM for 30 minutes at room temperature (∼20°C). To normalize for differential Fura-2 dye distribution across the cell we obtained a pseudo-ratio (we termed self-ratio) by averaging the fluorescence for 20 frames prior to the response (F₀) and normalizing the subsequent fluorescence response according to the formula: F(pseudo ratio)=(F(₀)-F)/F(₀). The Fura-2 images were all median filtered and displayed as pseudocolour images on an arbitrary blue-red scale using the programme ImageJ. The signal from the extracellular SRB fluorescence was binarized and shown as white on these images.

Fluorescent probes were from Molecular Probes Inc. (Eugene, OR, USA); all other compounds were from Sigma (Poole, UK).

Cell fragments, loaded with iIndo-1 AM for 30 minutes were washed, and plated onto coverslips. Single cells were illuminated (Nikon Diaphot inverted microscope, 40× oil immersion NA 1.3 objective) at 360 nm and emitted light collected at 410-430 nm and 430-600nm. The ratio of the two intensities was calibrated and expressed as Ca²⁺ concentration (Phocal, Strathclyde, UK).

Student's t-tests were used to test for significance and the probability (P) is quoted.

Lobules and fragments of exocrine tissue, loaded with Fura-2AM, were imaged on a two-photon microscope. With two photon excitation at 810 nm, elevated Ca²⁺ decreases Fura-2 emission measured at 490-540 nm (see Nemoto et al., 2001). Our experiments with pancreatic tissue showed large Fura-2 fluorescence decreases in response to acetylcholine (ACh), consistent with a Ca²⁺ rise. However, experiments with submandibular tissue often failed to show a decrease in Fura-2 fluorescence. We postulated that a reduction in cell volume, and subsequent concentration of dye, might be a major problem in imaging submandibular acinar cells. To test this idea we used the two-photon microscope to image extracellular fluorescent dye (OG or SRB) and define the outside edge of cells (Fig. 1A). Apparent changes in cell volume were then determined assuming symmetrical cell dimensions. Stimulation of submandibular cells with 10 μM ACh (Fig. 1B) significantly decreased cell volume by 28.69±5.87% (mean±s.e.m., n=17, P<0.01). By contrast, ACh stimulation of pancreatic acinar cells did not produce a cell volume change (–0.24±0.84%, n=12, P=0.85; data not shown). To limit volume changes in submandibular cells we applied a K⁺-rich extracellular solution; a condition expected to depolarize the cell and limit ion movements across the plasma membrane. Under these conditions the ACh-induced cell volume decrease in submandibular acinar cells was considerably attenuated; only a 6.54±1.71% decrease was observed (n=15).

The effect of K⁺-rich extracellular solution on the Ca²⁺ signal was then determined in cells loaded with the ratiometric Ca²⁺-sensitive dye Indo-1, a method chosen because the ratio is independent of changes in cell volume. We observed no significant difference in the peak Ca²⁺ response to 10 μM ACh (Fig. 1C, 0.69±0.15 μM n=10, in control; 0.75±0.25 μM, n=6 in K⁺-rich solution, P=0.4). However, the plateau Ca²⁺, measured at 300 seconds after the peak, was significantly smaller (0.21±0.03 μM, n=10 control; 0.10±0.01 μM in K⁺-rich solution, n=9, P<0.01) consistent with cell depolarization reducing the driving force for Ca²⁺ entry. Since the initial Ca²⁺ response was unaffected, all subsequent experiments were carried out in K⁺-rich solutions.

Release of Ca²⁺ through InsP₃Rs is the major trigger for Ca²⁺ waves in acinar tissue. InsP₃Rs are found throughout the cell (Fogarty et al., 2000) but highly enriched in the endoplasmic reticulum adjacent to the luminal plasma membrane (Lee et al., 1997; Yule et al., 1997; Zhang et al., 1999). To identify the luminal region in living cells we added the fluorescent probe SRB to the extracellular media. The effective optical slice (∼1 μm depth) of two-photon excitation enabled visualization of SRB fluorescence in the ducts and acinar lumen, simultaneously with measurements of Fura-2 fluorescence inside acinar cells (Fig. 2A).

Within pancreatic fragments the acinar lumen, filled with fluorescent dye, appeared as a discrete region, usually close to the narrowest (apical) pole of the cell (Fig. 2A). Application of 10 μM ACh invariably led to a Ca²⁺ wave that was measured as the spread of the Ca²⁺ signal from a region of interest (ROI) placed adjacent to the acinar lumen to a ROI placed in a distant region in the basal pole (Fig. 2B, apparent wave velocity of 9.53±1.67 μm/second, n=30). By contrast, in submandibular fragments the acinar lumen often appeared as elongated regions not especially close to the narrow pole of the cell (Fig. 2C). ACh always elicited a Ca²⁺ response but obvious Ca²⁺ waves were not observed. Instead the Ca²⁺ signal in ROIs placed close to the apparent acinar lumen was nearly synchronous with those in ROIs placed on the opposite side of the cell (Fig. 2D, apparent wave velocity of 31.43±2.98 μm/second, n=39; significantly faster than the apparent wave velocity in pancreas P<0.01).

It has previously been shown in parotid acinar cells that the Ca²⁺ wave velocity slows at lower concentrations of agonist (Tojyo et al., 1997). This is probably due to a decrease in excitability when lower concentrations of InsP₃ are generated. It is therefore possible that submandibular acinar cells might generate more InsP₃ than pancreatic cells, through possession of either more cell-surface receptors or a more efficient signal cascade and this in turn might generate faster Ca²⁺ waves. We therefore tested the effects of lowered agonist concentrations in submandibular cells. In our hands 100 nM and 200 nM ACh rarely induced Ca²⁺ waves (2/15 cell clusters and 2/10 cell clusters, respectively). Although some cells did show noisy fluorescence signals, which might be indicative of localized Ca²⁺ responses as recently reported in submandibular cells (Harmer et al., 2005). At 300 nM ACh, Ca²⁺ waves were seen (9/15 cell clusters) and analysis of the responses showed an apparent Ca²⁺ wave velocity of 18.56±2.51 μm/second (n=29 cells, Fig. 2E,F). We conclude that even at a concentration of ACh that is threshold for inducing a Ca²⁺ wave, the apparent wave velocity in submandibular cells is still much faster (nearly double) than that in pancreatic acinar cells, consistent with a fundamental difference in the Ca²⁺ signalling machinery in the salivary cells.

SRB labelling of the acinar lumen suggested that submandibular tissue might have a much more extensive, luminal structure than that of pancreatic tissue.

Indeed early studies on salivary glands describe extensions to the main acinar lumen as intercellular canaliculi running between the cells with a morphology similar to the main acinar lumen (Tamarin and Sreebny, 1965). In subsequent studies intercellular regions were immunostained positive for InsP₃Rs and aquaporins (Lee et al., 1997; Matsuzaki et al., 1999; Takemura at al., 1999). However, these studies employed thin tissue sections in which three-dimensional structure is lost. To directly compare the luminal structures in pancreatic and submandibular tissue we therefore used tissue fragments that retain the three-dimensional structure of the intact gland. To identify the acinar lumen we immunostained for the tight junction protein zona occludens (ZO-1). In tissue fragments matched for the number of nuclei (stained with Hoechst 33258) we found a clear difference in the staining pattern of ZO-1 between the pancreas and the submandibular gland. In submandibular glands ZO-1 was much more extensively arborized compared with the simple branching structure found in the pancreas (Fig. 3A). Comparing magnified images of secretory endpieces, ZO-1 staining is apparent as a thin band (presumably, in reality, a tube) that in pancreatic tissue ends at a discrete region in one pole of the cells (Fig. 3B) but in submandibular tissue wraps around and almost completely encircles individual acinar cells (Fig. 3C,D). We quantified these differences, measuring the length of the terminal branches stained with ZO-1 (i.e. the last branch onto individual acinar cells). In pancreatic tissue branches were 6.44±0.35 μm in length (n=28), significantly shorter than branches in submandibular tissue, which were 11.67±0.64 μm (n=62) in length (P<0.01; see Movies 1 and 2 in supplementary material).

Our use of tissue fragments and three-dimensional reconstruction methods contrast the complexity of the elaborated lumens in submandibular tissue with the relatively simple structures found in the pancreas.

To test for differences in InsP₃R distribution we immunostained tissue fragments for InsP₃R-type 3 (InsP₃R3) and counter immunostained for ZO-1. Once again, submandibular cells (Fig. 4B) showed elaborate ZO-1 staining compared with that in pancreatic acinar cells (Fig. 4A). In both tissues immunostaining for InsP₃R3 localized closely with ZO-1 staining. InsP₃R2 is also present in salivary acinar cells (Lee et al., 1997; Takemura et al., 1999) and here we show that immunostaining for InsP₃R2 directly corresponds to immunostaining for InsP₃R3 (Fig. 4C).

This indicates that the whole length of the ZO-1 structures in both tissues is associated with InsP₃Rs. To determine if other luminal proteins are associated with ZO-1, we also immunostained for aquaporin-5 (AQP-5) (Matsuzaki et al., 1999), which was also found to follow the elaborated acinar lumens in submandibular cells (data not shown).

Since a band of InsP₃Rs nearly encircle submandibular acinar cells then this would provide an explanation for the near-synchronous Ca²⁺ increases across the cell. However, other possible explanations might include a role for Ca²⁺ influx. Ca²⁺ entry might be a larger component of the response in submandibular cells than in pancreatic acinar cells and as a consequence the Ca²⁺ wave might appear faster. However, two-photon experiments under the same conditions as shown in Fig. 2, but in the absence of extracellular Ca²⁺, still showed a near-synchronous Ca²⁺ rise across the cell (Fig. 5A, Ca²⁺ wave velocity 24.30±3.85 μm/second, n=23).

Another possible difference is that ryanodine receptors (RYRs), thought to be present in acinar cells [submandibular (Lee et al., 1997), parotid (Zhang et al., 1999), pancreatic (Thorn et al., 1994)] may be differentially recruited in the two tissues. However, after pre-incubation of the cells in 20 μM ryanodine (RY), a concentration known to block RYRs, we still observed near-synchronous Ca²⁺ increases across the cell induced by the application of 10 μM ACh (Fig. 5B, apparent Ca²⁺ wave velocity 21.20±2.04 μm/second, n=52).

To further investigate a potential role for RYRs we conducted experiments with 200 mM ryanodine (observed apparent wave velocity 29.92±3.21 μm/second, n=18) and with ryanodine from another source (Tocris) at 20 μM (observed apparent wave velocity 30.94±2.81 μm/second, n=26) and 200 μM (observed apparent wave velocity 28.27±3.48 μm/second, n=16). In none of these experiments was there evidence for an action of ryanodine on the response induced by 10 μM ACh.

These experiments support the idea that it is the distribution of InsP₃Rs that underlies the faster wave velocities seen in submandibular cells.

Our data is in contradiction to previous work on submandibular cells, in which slower Ca²⁺ wave velocities were recorded (Lee et al., 1997). One possible explanation is that previous studies often use single cells and clusters of small numbers of cells. To test for this possibility we measured Ca²⁺ responses from single submandibular cells. Our results showed Ca²⁺ responses to 10 μM ACh with an apparent Ca²⁺ wave velocity of 18.80±4.17 μm/second (n=15, Fig. 6). This is significantly slower than that recorded in tissue fragments, suggesting that the procedure used to isolate single cells may result in damage to the signal transduction machinery.

Our imaging methods have allowed us to identify the acinar cell lumen in living exocrine tissue fragments and to simultaneously record acinar cell Ca²⁺ signal characteristics. In pancreatic tissue the lumen abuts the apical region of the cell and single acinar cells showed the typical agonist-evoked apical to basal Ca²⁺ waves. By contrast, in submandibular tissue the lumen was more extensive and was even found close to the apparent basal pole. In submandibular cells, agonists evoked near-synchronous global Ca²⁺ signals.

To investigate the basis of the differences in the Ca²⁺ signal between the two tissues we immunostained fixed tissue fragments. Immunostaining for the tight junction protein, ZO-1, defined the acinar lumen in pancreatic tissue as a branching system that terminates abruptly at the narrow apical pole of acinar cells. By contrast, in submandibular tissue ZO-1 immunostaining was much more extensive and single acinar cells were nearly encircled by a band of ZO-1 staining. These patterns of staining in the two tissues are entirely consistent with our live-cell two-photon imaging of the exocrine lumen with extracellular dyes. Immunostaining for InsP₃R2 and InsP₃R3 showed that both colocalize with ZO-1, as does AQP-5, indicating that the elaborated luminal structures of the submandibular acinar cells are probably of functional importance in generating the Ca²⁺ signal and in fluid secretion.

We show that the fast apparent Ca²⁺ wave velocity in submandibular acinar cells is retained in the absence of extracellular Ca²⁺ and also in the presence of ryanodine. These observations show that neither Ca²⁺ influx nor RYRs are important in the generation of the rapid global Ca²⁺ response and strongly implicate the more extensive distribution of InsP₃Rs in submandibular acinar cells as the fundamental factor in generating the fast Ca²⁺ wave. We conclude that the elaborated lumen and the near-encircling of the submandibular acinar cell with InsP₃Rs initiates the Ca²⁺ response at multiple sites around the cell, giving rise to a much higher apparent Ca²⁺ wave velocity than in the pancreas.

Two independent lines of evidence, imaging of extracellular dyes and immunostaining of ZO-1, show a much more extensive acinar lumen in submandibular than in pancreatic tissue. This data is absolutely consistent with observation of intercellular canaliculi in salivary acinar cells that have been reported previously (Tamarin and Sreebny, 1965; Matsuzaki et al., 1999). However, to our knowledge, no direct comparative studies across different exocrine tissues has previously been performed, and it is this side-by-side comparison of submandibular with pancreatic tissue that highlights the contrasting anatomy of the two exocrine glands. These differences are further highlighted in our study through our use of 3D reconstruction techniques (see Fig. 4 and Movies 1 and 2 in supplementary material) where the near-encircling of submandibular acinar cells with InsP₃Rs is immediately apparent in a way not obvious in single thin sections.

The dramatic tissue-type differences we see in Ca²⁺ wave velocity (9.5 μm/second in pancreatic and 31.4 μm/second in submandibular tissue) are consistent with most previous reports (Thorn et al., 1992; Giovannucci et al., 2002; Takemura et al., 1999; Liu et al., 1998; Tojyo et al., 1997). In other reports, slower Ca²⁺ waves were seen in submandibular tissue (Lee et al., 1997; Harmer et al., 2005). In Harmer et al.'s (Harmer et al., 2005) study this may be due to the relatively high imposed Ca²⁺ buffering in their patch-clamp experiments (500 μm EGTA plus 100 μm Fura-2) which would tend to slow wave velocities (see Kidd et al., 1999) and in Lee et al.'s (Lee et al., 1997) study it may be due to the use of small clusters of acinar cells, which we show does slow Ca²⁺ wave speeds (Fig. 6).

Giovannucci et al. (Giovannucci et al., 2002) have explicitly investigated tissue differences between pancreatic and parotid glands. They showed that the parotid gland acinar cell Ca²⁺ response was more sensitive than the pancreas cells to small increases in InsP₃ concentration: low concentrations of InsP₃ that induced local Ca²⁺ responses in pancreatic acinar cells induced global Ca²⁺ responses in the parotid acinar cells. They suggested that these differences might be due to different levels of InsP₃R expression and showed that the parotid gland contained fourfold more InsP₃Rs, as measured by radioligand binding and western blotting experiments than the pancreas (Giovannucci et al., 2002). This data is entirely consistent with our own observations. It is likely that the extended distribution of InsP₃Rs we report in salivary glands, distributed along the extended acinar lumen, requires the higher levels of receptor expression seen by Giovannucci et al. (Giovannucci et al., 2002).

However, alone, these lines of evidence do not directly link InsP₃R number and distribution with the differences in the Ca²⁺ response between pancreatic and salivary acinar cells. What we have done is to directly demonstrate that neither Ca²⁺ influx nor possible Ca²⁺ release from RYR-sensitive Ca²⁺ stores can explain the fast Ca²⁺ wave in submandibular cells. Since RYRs are known to be present in salivary acinar cells (Lee et al., 1997; Zhang et al., 1999) it might be expected that they do play a role. However, we have previously shown in pancreatic acinar cells that ryanodine is most effective in blocking the Ca²⁺ responses induced by low agonist concentrations (Thorn et al., 1994). This presumably is a circumstance where relatively modest levels of InsP₃ only partially recruit the InsP₃R pool and where the total Ca²⁺ response may depend on a parallel recruitment of RYRs. By contrast, at high agonist concentrations, sufficient InsP₃ is generated to recruit the total InsP₃R pool and the majority of the Ca²⁺ response will therefore be derived from the InsP₃Rs. Whatever the explanation, ryanodine, even at the extremely high concentration of 200 μM, failed to affect the Ca²⁺ response of submandibular cells (Fig. 5) and we can conclude that the differences in the Ca²⁺ signal between the pancreas and submandibular acinar cells are therefore most probably dependent on the different distributions of the InsP₃Rs.

What we are actually observing when we measure Ca²⁺ wave velocities is not clear. The measured Ca²⁺ wave velocity is much higher than would be expected by simple Ca²⁺ diffusion. These high velocities are therefore likely to be a reflection of either the regenerative spread of Ca²⁺ along InsP₃Rs by the positive feedback process of Ca²⁺-induced Ca²⁺-release (CICR active on InsP₃Rs) or the temporally coordinated release of Ca²⁺ from InsP₃Rs. It is not clear which mechanism dominates the response in submandibular cells, but in either case it is the distribution of InsP₃Rs that would be critical to shaping of the Ca²⁺ signal.

In terms of the physiological response, the rodent pancreatic acinar cell is thought to have a relatively poor fluid secretory output (Argent et al., 1986) compared with the submandibular acinar cell (Foskett et al., 1989). This is shown in our measurements of cell volume changes, which directly reflect ion movements during the first stages of fluid secretion. Here we show significant changes in volume in submandibular cells but not in pancreatic cells, indicating a greater ion flux in the salivary tissue [see Fig. 1, and also Foskett (Foskett, 1990)]. Fluid secretion is, in the first instance, instigated by the opening of apical Cl^– channels but activation of a K⁺ channel is required to maintain cell hyperpolarization and to provide a route for K⁺ exit (Petersen, 1992; Hayashi et al., 1995; Takeo et al., 1998). Salivary acinar cells have a robust Ca²⁺-dependent K⁺ conductance (Maruyama et al., 1983; Smith and Gallacher, 1992) and the fact that this is missing in rodent pancreatic acinar cells (Kidd and Thorn, 2001) may explain their weaker secretory response. The identity of the Ca²⁺-dependent K⁺ channel, recruited during fluid secretion in salivary acinar cells, is not known (Hayashi et al., 1995) but most models of secretion would place the K⁺ channel on the basal plasma membrane. This positioning is supported by the work of Harmer et al. (Harmer et al., 2005), which does show evidence that the Cl^– current can, at low levels of cell stimulation, be activated alone. However, much work on salivary acinar cells shows a nearly simultaneous activation of K⁺ and Cl^– current on cell stimulation (Foskett et al., 1989; Takeo et al., 1998). This might arise because of an apical location of K⁺ channels, but our work suggests an alternative hypothesis. Unlike the classical view, derived from the pancreatic acinar cells, where the basal plasma membrane can be some distance from the discrete apical region, we show that in submandibular acinar cells the luminal region encircles the cells and thus the basal plasma membrane is actually close to the apical domain. We would therefore expect that the extended luminal region associated with an extended area of InsP₃Rs would rapidly active luminal Cl^– channels but the near-synchronous global Ca²⁺ responses that arise from this distribution of InsP₃Rs would also trigger the rapid activation of Ca²⁺-dependent K⁺ channels even if they were exclusively in the basal plasma membrane. Further work will be required to directly compare fluid secretion in pancreatic and submandibular tissues and test these ideas.

In conclusion, our data show an extended acinar lumen in submandibular glands nearly encircles individual acinar cells and is associated with a characteristic near-synchronous global Ca²⁺ response. We speculate that these are important adaptations to support copious fluid secretion in these cells.

This work was funded by a Wellcome Trust Overseas Fellowship to O.L. and Medical Research Council (UK) project grants (G0000214, G0400669 to P.T.). We thank Ian Parker for help with the movies.