Localised calcium release events in cells from the muscle of guinea-pig gastric fundus

Abstract

After enzymatic dispersion of the muscle of the guinea-pig gastric fundus, single elongated cells were observed which differed from archetypal smooth muscle cells due to their knurled, tuberose or otherwise irregular surface morphology. These, but not archetypal smooth muscle cells, consistently displayed spontaneous localized (i.e. non-propagating) intracellular calcium ([Ca²⁺]_i) release events. Such calcium events were novel in their magnitude and kinetic profiles. They included short transient events, plateau events and events which coalesced spatially or temporally (compound events). Quantitative analysis of the events with an automatic detection programme showed that their spatio-temporal characteristics (full width and full duration at half-maximum amplitude) were approximately exponentially distributed. Their amplitude distribution suggested the presence of two release modes. Carbachol application caused an initial cell-wide calcium transient followed by an increase in localized calcium release events. Pharmacological analysis suggested that localized calcium release was largely dependent on external calcium entry acting on both inositol trisphosphate receptors (IP₃Rs) and ryanodine receptors (RyRs) to release stored calcium. Nominally calcium-free external solution immediately and reversibly abolished all localized calcium release without blocking the initial transient calcium release response to carbachol. This was inhibited by 2-APB (100 μm), ryanodine (10 or 50 μm) or U-73122 (1 μm). 2-APB (100 μm), xestospongin C (XeC, 10 μm) or U-73122 (1 μm) blocked both spontaneous localized calcium release and localized release stimulated by 10 μm carbachol. Ryanodine (50 μm) also inhibited spontaneous release, but enhanced localized release in response to carbachol. This study represents the first characterization of localized calcium release events in cells from the gastric fundus.

The availability of fluorescent Ca²⁺ indicators, coupled with advances in fluorescence microscopy, has allowed the detailed characterization of patterns of intracellular Ca²⁺ ([Ca²⁺]_i) release in isolated cells (Tsien, 1992). These patterns include propagating waves of [Ca²⁺]_i and non-propagating, spatially restricted transients in [Ca²⁺]_i or ‘localized events’. The latter have different functions in different cell types; in smooth muscles the cell-wide increases in [Ca²⁺]_i associated with contraction are initiated by, and perhaps composed of, such localized or elementary events (Gordienko et al. 1998). Additionally in smooth muscle and certain neurones they activate calcium-dependent conductances at the plasma membrane, thus generating transient currents (under voltage clamp), called variously STICs, STOCs (spontaneous transient inward or outward currents, respectively) or SMOCs (spontaneous miniature outward currents). These currents may modulate the steady-state (resting) potential of the cell and thereby the tissue or electrically coupled syncytia. Such a view has had support from studies of tone in cerebral arteries (e.g. Nelson et al. 1995; Porter et al. 1998; Alioua et al. 2002). However possibly more widespread is the involvement of localized events in regenerative or non-linear responses to external stimuli or during repetitive spontaneous activity (Parker & Ivorra, 1990; Klink & Alonso, 1997; Edwards et al. 1999; Laer et al. 2001, Shalinsky et al. 2002). Usually stochastic localized events can be temporally and spatially coordinated in response to agonists or electrical stimuli (e.g. Callamaras et al. 1998; Cannell et al. 1995; Kockskamper et al. 2001). This may lead to more complex patterns of discrete release, a propagating all-or-none response (a [Ca²⁺]_i wave) or even apparently homogeneous responses.

Hypotheses about the mechanism and coordination of localized events, and thereby the interpretation of experimental data, have been directed by two assertions. Firstly that localized events are ‘discrete’ due to the spatial segregation of discrete ‘clusters’ of calcium release channels on an otherwise continuous sarcoplasmic reticulum. Physical evidence for this has been provided by immunochemistry and electron microscopy (Protasi et al. 2000; Yin & Lai, 2000). However the quantitative interpretation of the spatio-temporal properties of localized events, in terms of channel clustering, has been severely constrained by the method used to characterize them – confocal line scanning – which samples four dimensional events in only two dimensions. The second assertion is that these clusters are not mixed: they consist entirely of either inositol trisphosphate receptors (IP₃Rs) or ryanodine receptors (RyRs), but not both together. This view is reflected in the predominate dichotomy between ‘sparks’ (inhibited by ryanodine) and ‘puffs’ (stimulated by IP₃ or IP₃-generating agonists). Despite this there have been reports of events with a mixed IP₃–ryanodine receptor pharmacology (Koizumi et al. 1999; Haak et al. 2001; Gordienko & Bolton, 2002).

In smooth muscle, localized events when they have been seen, have predominantly been characterized as sparks (Gordienko et al. 1998; Jaggar et al. 2000). There are only a few reports of puffs, namely in colonic myocytes (Bayguinov et al. 2000, 2001a,b) and uteric myocytes (Boittin et al. 2000). In this study we describe novel and heterogeneous localized calcium release events in hitherto undescribed single cells from the muscle of guinea-pig fundus. These events were characterized in terms of their pharmacology and quantitative spatio-temporal dynamics. Some of these results have been presented in abstract form (Parsons & Bolton, 2001, 2002).

Methods

Cell isolation

Guinea-pigs were killed humanely by instantaneous stunning, followed by exanguination, in accordance with UK guidelines. Both ventral and dorsal aspects of the gastric fundus were removed in situ from the mucosa as single pieces, using scapel and forceps. These were kept in warmed, HEPES-buffered saline (HBS; see below) and cut into smaller pieces, before enzymatic digestion in HBS with 0.5 mm Ca²⁺, 0.25 mm Mg²⁺ and 1 mg ml⁻¹ each of collagenase type 1A (clostridiopeptidase A), protease type XIV, type II-S soybean trypsin inhibitor and bovine serum albumin. Digestion was for 20 min at 36°C. The tissue was then washed with warmed HBS and dispersed by tituration with a blunt-ended, fired glass Pasteur pipette. Dispersion was in warmed HBS with 0.5 mm Ca²⁺, 0.25 mm Mg²⁺ and the resulting suspension was pipetted on to glass coverslips, attached to experimental chambers. After allowing the cells to settle and attach for 1 h, the dispersion solution was replaced with HBS containing 2.0 mm Ca²⁺ and 1.0 mm Mg²⁺. All experiments were conducted in this solution at room temperature.

Confocal microscopy

Confocal microscopy was performed using a Zeiss Axiovert 100m microscope with the LSM 510 laser scanning module. Unless otherwise stated a × 40 oil immersion plan-neofluar objective (NA = 1.3) was used with a 66 μm pinhole (giving an optical slice of < 0.9 μm). To image [Ca²⁺]_i, 5 μm fluo-3AM was included in the dispersion solution and a 488 nm argon laser was used for excitation, with a 505 nm bandpass emission filter. Temporal changes in cell [Ca²⁺]_i were imaged by repeated scanning of either a whole confocal field (an xyt series) or just a single line positioned along the cell's long axis (line scan or xt series). Unless otherwise stated, xt series were taken at 6.59 Hz (3.8 ms to scan a line, with a 148 ms interline delay). Time series were analysed with programmes personally written in the IDL software suite (v5.3). To ratio a time series, the value of each scanned pixel (F) was divided by the mean of its value (F₀) over an quiescent time period (i.e. representative of the baseline fluorescence).

Automatic event detection

A computer program was written with IDL software (version 5.3) to analyse the spatio-temporal properties of individual calcium release events in xt scans. There were two main points to this program: firstly to perform event detection automatically, to avoid the subjective bias imposed by human selection (Song et al. 1997; Cheng et al. 1999); secondly to separate out individual peaks (apparent events), which had coalesced spatially or temporally into ‘compound events’. In such a situation the definition of an event was the area around a peak at its half-maximum amplitude (amplitude at half-maximum or AHM). The AHM was defined as follows:

where m= the maximum amplitude of the peak and b= baseline amplitude (1.0 F/F₀).

The fundamental element of the automatic event detection programme was the IDL routine, ‘label_region.pro’. This routine allows for the definition (separation) of regions within an image (xt scan) at any arbitrary threshold (α), a ‘region’ being a contiguous area of an xt series/ image of value greater than α (analogous to the area enclosed by a single contour on a map). In the program this process (region definition) is applied iteratively, beginning at a rather low value of α (the initial threshold or α_i) and then continuing at successively higher values of α, increasing by steps of β (the ‘amplitude resolution’). At each successive value of α (α_i, α_i+β, α_i+ 2β, α_i+ 3β….α_i+nβ), certain rules can be applied to each region individually, to determine whether it may be considered as an event. The result of these rules and their operation (as follows; see Fig. 1) is that an ‘event’ is defined as the contiguous region around a peak (the maximum of the region), at the peak's AHM.

Figure 1. A program for automatic event selection.

α is the threshold value for region definition and β is the value by which α is stepped with each successive iteration (the loop in the figure). See Methods for explanation.

Step 1.Define regions greater than α_i (say α_i= 1.25F/F₀).

Step 2.For each region found by steps 1 or 3 (at a threshold of α):

1. if its AHM is less than α, then this region is not considered as an event and is discarded;

2. if its AHM is between α and α+β (say β= 0.2F/F₀), then the region is considered as an event and is passed to step 4;

3. if its AHM is greater than α+β then the region is passed to step 3.

Step 3.The process of region definition is repeated within the region found in step 2c) (which has been defined at a threshold of α), at a threshold of α+β. This step in effect splits compound events (regions bound at α) into regions separated at α+β. These regions are then passed back through step 2 again, with α increased by β.

Steps 2 and 3 are repeated iteratively (with α increasing by β with each iteration) until no regions are found to have an AHM > α+β (step 2c above).

Step 4.The maximum amplitude (MA), full width and full duration at half-maximum (FWHM and FDHM) of all the events found by step 2b are calculated. The FWHM and FDHM are defined as the two orthogonal lengths which pass through the event's maximum and are bounded by the event's AHM.

As an additional note, at various stages in the program defined regions are checked to see if they surpass a minimum area (the ‘xt resolution’ or XTR). This prevents the consideration of noise. For all the data shown in the Results section, the XTR was set to 3 μm. s, α_i was 1.25F/F₀ and β was 0.2F/F₀.

Analysis of total activity

To give a single quantitative estimate of overall [Ca²⁺]_i release activity (in the presence of antagonists, etc), 90 s periods of xt images were averaged, over the full spatial axis, to give a mean pixel value (MPV; see Fig. 6). Also small rectangular areas of the xt image were selected by eye as representative of the basal [Ca²⁺]_i (Fig. 6). The mean values of these boxes (the ‘baseline’) are compared to the MPV (Figs 6–11).

Figure 6. Effect of carbachol on Ca²⁺ release: control protocol.

A, ratioed xt series of fluo-3 fluorescence in a fundus cell. The line was scanned twice for 4 min, separated by an interval of 10 min. Midway during the first scan (i) vehicle (0.33% v/v DMSO) was added and midway during the second scan (ii) this was supplemented by 10 μm carbachol. This separated the protocol into four periods each of which was quantitatively analysed (B: period 1, before addition of vehicle; period 2, in the presence of vehicle; period 3, 12 min in the presence of vehicle; period 4, in the presence of vehicle and carbachol). The baseline F/F₀ (hatched columns in B) was determined from the mean value of the red-dashed rectangular area for each period. The mean pixel value or MPV (open columns in B) was determined for the last 90 s of each period (indicated by dashed black rectangles in i and ii). All values have been normalized to the baseline of period 1. For each of the last three periods paired t test significances (with respect to the first period) were determined for each parameter (baseline and MPV) and these are indicated above the relevant columns, where significant (*P < 0.05; **P < 0.02; ***P < 0.01; ****P < 0.001). Also paired t test significances for the difference between baseline and MPV for each period are shown (indicated by brackets; P values indicated as before).

Figure 11. Effect of PLC antagonists on Ca²⁺ release.

A, a ratioed xt series of fluo-3 fluorescence in a fundus cell. 4 μm U-73122 was added 2 min into the scan and 3 min later this was supplemented by 10 μm carbachol. The F/F₀ scale bar shown applies to all figures. B and C, ratioed xt series of fluo-3 fluorescence in two fundus cells. Scanning protocols were the same as in Fig. 6A (control) except vehicle was replaced by 1 μm U-73122 or 20 μm neomycin, respectively. D and E, quantitative parameters determined as in the control (see legend of Fig. 6) for 1 μm U-73122 and 20 μm neomycin, respectively. Statistical tests and significance indicated as before (*P < 0.05; **P < 0.02; ***P < 0.01; ****P < 0.001).

In some protocols two scans of the same line, which had been ratioed separately, had to be compared. To allow a meaningful comparison of the two scans the second scan was normalized to the first scan by multiplying each ratioed pixel in it by the ratio of the mean of the F₀ of the two scans, thus:

where p_n= the normalized pixel value of the second scan, p_o= the original pixel value of the second scan, m₂= the mean of the F₀ of the second scan and m₁= the mean of the F₀ of the first scan. As this operation was performed in the same way on every pixel, quantitative parameters (such as MPV and the baseline) that are simply the mean pixel value of a certain area can be normalized in the same manner. This method was generally used (as opposed to normalizing the actual image) except where images are presented (any second scans in a presented series have been normalized).

Peaks of initial Ca²⁺ transients in response to carbachol or caffeine (Fig. 12) were calculated as the maximum value obtained by averaging F/F₀ values in all pixels along the scan line, less the baseline value for the period preceding the transient.

Figure 12. Initial [Ca²⁺]_i transients in response to carbachol and caffeine.

Peak amplitudes of the initial scan-wide transients in response to 5 mm caffeine (first column), 10 μm carbachol (second column) or 10 μm carbachol in the presence of drugs or in nominally calcium-free solution (all other columns). The data were obtained from the same xt series analysed in Figs 6–11 and the peak is calculated as the peak of the spatially averaged transient, less the baseline value preceding the addition of carbachol/caffeine. The fractions of cells in which the response to 10 μm carbachol was completely blocked are indicated in parentheses above each column (where appropriate) and are included in the analysis as zero values. Unpooled, unpaired t test significances, with respect to the control response to carbachol, are also indicated where appropriate (*P < 0.05; **P < 0.02, ***P < 0.01).

Solutions and drugs

HBS had the following composition (mm): 119 NaCl, 6 KCl, 12 glucose, 10 HEPES, adjusted to pH 7.35 with NaOH. Calcium chloride or magnesium chloride was added to this as indicated elsewhere. Ryanodine, U-73122, neomycin, carbachol and caffeine were diluted from aqueous stock solutions. The following were diluted from DMSO stock solutions: 2 aminoethyldiphenyl borate (2-APB, 60 mm), xestospongin C (XeC, 5 mm) and fluo-3AM (0.88 mm). Drug solutions were applied as 1 ml syringe volumes connected by thin tubing to a chamber of ∼300 μl volume. Ryanodine, xestospongin C, 2-APB and U-73122 were from Calbiochem (Beeston, UK). Fluo-3AM was from Molecular Probes (Eugene, OR, USA). All other chemicals were from Sigma Chemical Co. (Poole, UK).

Analysis and statistics

Data have been summarized as the mean ± standard error of the mean (s.e.m.), except where indicated. Comparisons were conducted by either paired or unpaired (unpooled) Student's t tests.

Results

The vast majority of cells from the dispersed fundus had a morphology consistent with an archetypal smooth muscle cell. That is they were long (>50 μm), thin (5–10 μm), bipolar, often serpentine or sinuous in shape (in the relaxed state) and had a smooth/uniform membrane. These cells did not show spontaneous [Ca²⁺]_i events, but were responsive to caffeine and carbachol which caused a wave of [Ca²⁺]_i and contraction (not shown).

In contrast a second class of cell, which represented 2–3% of the dispersed population, had a membrane characterized by bumps, ridges or knurls (Fig. 2), and showed spontaneous localized [Ca²⁺]_i events in >95% of cases. It is only these cells which are considered throughout the rest of this study. Inspection of xyt series indicated that the spatial and temporal characteristics of these localized events varied considerably. Figure 3A shows a typical example of this. In at least four regions of the cell (arrows) there were sustained increases in or plateaux of [Ca²⁺]_i, lasting over tens of frames (20–40 s). These events seemed to originate at a point source close to the cell membrane and then spread out over a period of 1–2 s to fill the full width, and over 20 μm of the length, of the cell (Fig. 3B). In addition smaller, more localized and transient rises in [Ca²⁺]_i occurred (arrowheads in Fig. 3A).

Figure 2. Archetypal smooth muscle cells and myocytes with a knurled appearance.

A montage of transmitted light images (phase-contrast) of fundus myocytes with an archetypal smooth muscle cell (ASMC) appearance or with irregular or knurled membranes (all other cells). All of the latter displayed spontaneous localized [Ca²⁺]_i events but ASMCs did not.

Figure 3. Calcium release events: frame or xyt series.

A, a ratioed xyt series of fluo-3 fluorescence in the cell shown at the top right corner of Fig. 2 (laterally inverted here; series runs horizontally from left to right). The series was taken at 2.53 Hz (394 ms frame⁻¹) and every 20th frame is shown. Arrows point to four cell regions with sustained rises in [Ca²⁺]_i. Arrowheads point to shorter [Ca²⁺]_i events. B, part of the series in A, at full temporal resolution and showing the area of the cell indicated by the box in the 6th frame of A (this is the 101st frame of the series and B shows frames 106–117). The arrow points to the origin of a sustained [Ca²⁺]_i release event (corresponding to region indicated by the arrow marked x, in A).

Line or xt scanning of localized events and their analysis

In addition to xyt (frame) series, line scan or xt series of localized [Ca²⁺]_i events were acquired by positioning the scan line (typically 40–70 μm long) parallel to the long axis of the cell in an active region. A typical example of such a scan is shown in Fig. 4A. As in an xyt scan (Fig. 3), there is a mixture of shorter and longer localized [Ca²⁺]_i events. Of the shorter, transient events, most consisted of a rapidly rising phase followed by an approximately exponential decay (Fig. 4Bi). Of the longer events, some seemed to represent true plateaux in [Ca²⁺]_i (Fig. 4Bii), whilst others were compound events made up of doublets or triplets of shorter transients (e.g. event 05 in Fig. 4A).

Figure 4. Automatic detection of [Ca²⁺]_i events.

A, a ratioed xt series of fluo-3 fluorescence in a fundus cell, annotated with the output of the automatic event detection program. The intersection of horizontal and vertical black lines (FOHM and FWHM, respectively) indicate the position of the event maxima and the black contours indicate the event AHMs. Bi and ii, examples of events from A (04 and 10, respectively), plotted along the axes used to define their FDHM (a) and FWHM (b).

Spontaneous events in xt images were analysed quantitatively with an automatic event detection programme (see Methods). Detected events were analysed for three parameters commonly used to quantify [Ca²⁺]_i events; the maximum amplitude (MA), the FWHM and the FDHM (full width and full duration at half-maximum amplitude, respectively; Sun et al. 1998; Thomas et al. 1998; Boittin et al. 2000; Jaggar et al. 2000, Haak et al. 2001; Shuai & Jung, 2002, Gordienko & Bolton, 2002). The results of this analysis for 392 events, from 31 line scans of 31 cells are shown in Fig. 5. This analysis must be considered within the constraints imposed on the detection of events by the temporal resolution of the scan (152 ms line⁻¹; see Song et al. 1997 for theoretical consideration of this matter). Also it must be noted that there was a lower limit set on the amplitude of events that were detected of 1.5F/F₀ (α_i= 1.25F/F₀; see Methods). The frequency distributions of MA (mean = 3.67F/F₀, s.d.= 2.0F/F₀), FDHM (mean = 5.6 s, s.d.= 5.8 s) and FWHM (mean = 6.7 μm, s.d.= 3.9 μm) all decayed in a roughly exponential manner (Fig. 5Ai, Bi and Ci, respectively). This would be expected from the random sampling of four-dimensional (x, y, z, t) events by a ‘two-dimensional’ method (the xt scan). For amplitude at least, the distribution should be hyperbolic, i.e. of the form n ≈ (a – B)⁻¹, where n is the amplitude frequency, a is the amplitude and B is a constant of the observation function (which describes the statistical sampling of an event's maximum amplitude at varying distances from its actual origin; Izu et al. 1998). Consequently if the inverse of the amplitude frequency (1/n) is plotted against amplitude, a linear relationship should be obtained. In fact the inverse distribution of MA was marked by two linear trends of different slope (Fig. 5Bii). This change in slope (an ‘inflection’) reflects the apparent ‘break’ in the normal distribution (Fig. 5Bi) at the same value of MA.

Figure 5. Distributions of event parameters.

A–C, normal (i) and inverse (ii) distributions of events, according to FDHM (A), maxiumum amplitude (B) and FWHM (C). All 1/n values of 1 (i.e. where n= 1) have been excluded. Aii is fitted with an exponential, 1/n= e^FDHM/25.1.Bii is fitted with two lines of slope 0.0053 (left) and 0.042 (right). Cii is fitted with an exponential, 1/n= e^FWHM/3.75.

Pharmacology: protocols, parameters and controls

The pharmacology of the localized [Ca²⁺]_i events was also addressed with xt series, taken in the same manner as before (see Methods). For most drugs, which took some time to act fully, a standard protocol was adopted consisting of two 4 min scans separated by an interval of 10 min. Midway through the first scan the drug was added and midway through the second scan this was supplemented by 10 μm carbachol, to assess the effects on muscrinic receptor-induced Ca²⁺ release. Thus the protocol consisted of four periods (Fig. 6). In cases where the action of the drug was faster, shorter protocols were adopted consisting of one scan divided into three periods (before drug, in the presence of drug, in the presence of drug and 10 μm carbachol. Whatever protocol was used, each period was at least 2 min long. This allowed for quantitative parameters to be determined for a 90 s span at the end of each period. Two different parameters were calculated (see Methods): the mean pixel value (MPV) and baseline F/F₀. Also the peak of the initial carbachol-induced Ca²⁺ transient, which spanned the whole width of the line scan, was calculated (see Methods).

In control experiments with the four period protocol (n= 6), where drug vehicle (0.33% v/v DMSO) was added alone in the first scan, there was no significant change in any of the parameters over the first three periods (Fig. 6B). The vehicle caused no significant inhibition of Ca²⁺ release as the difference between the baseline and MPV remained significant (by paired t test) over all four periods. Carbachol caused a significant increase in both the MPV (P < 0.01) and baseline (P < 0.05) after the initial scan-wide transient which averaged 4.69 ± 0.85F/F₀ (Fig. 12). Caffeine (5 mm) produced a scan-wide transient of similar amplitude (n= 6; Fig. 12) but had no effect on localized release (Fig. 7): there was no significant change in baseline or MPV immediately after the initial transient and the difference between the MPV and baseline remained significant (statistical tests as before; not shown).

Figure 7. Effect of caffeine on Ca²⁺ release.

A ratioed xt series of fluo-3 fluorescence in a fundus cell. 5 mm caffeine was added 2.5 min into the 5 min scan.

Pharmacology: ryanodine

The effect of ryanodine on Ca²⁺ release was tested with the four period protocol at two different concentrations (Fig. 8). At 10 μm, ryanodine (n= 10) had limited effects. As in the control there was only a significant change in the baseline or MPV after the addition of carbachol. Throughout the protocol there was a significant difference between baseline and MPV. However the initial carbachol-induced transient was significantly (P < 0.02) reduced in amplitude, being abolished in half of the cells tested (Fig. 12).

Figure 8. Effect of ryanodine on Ca²⁺ release.

A, ratioed xt series of fluo-3 fluorescence in a fundus cell. The scanning protocol was the same as in Fig. 6A (control) except that the vehicle was replaced by 50 μm ryanodine. B and C, quantitative parameters determined as in the control (see legend of Fig. 6 and Results), for 10 and 50 μm ryanodine, respectively. Statistical tests and significance indicated as before (see Fig. 6B: *P < 0.05; **P < 0.02; ***P < 0.01; ****P < 0.001).

At 50 μm, ryanodine (n= 6) caused a stronger inhibition of spontaneous Ca²⁺ release. After 12 min (period 3) there was only a slight, but nevertheless significant (P < 0.02), difference between the baseline and MPV. This inhibition was brought about by a rise in the baseline, rather than a decrease in MPV, suggesting that ryanodine caused a gradual increase in steady-state Ca²⁺ release from the SR (made the SR ‘leaky’). When carbachol was applied to cells, this ‘leakage’ was reflected by a much more robust Ca²⁺ release than was apparent under control conditions (compare Aii and B in Fig. 6 and Aii and C in Fig. 8). Often this was to the extent that it became difficult to define the ‘baseline’. In two cells (not included in the analysis) 50 μm ryanodine alone increased the baseline so much over the 10 min interval, that the fluorescence signal was almost saturated by the third period and carbachol had little further effect on the MPV. At 50 μm, ryanodine significantly (P < 0.05) reduced the amplitude of the initial carbachol-induced transient (Fig. 12).

Pharmacology: IP₃R inhibitors

Two different antagonists of IP₃R-dependent Ca²⁺ release were tested for their effect on Ca²⁺ release (Fig. 9). 2 Aminoethoxydiphenylborate (2-APB) has an IC₅₀ of 42 μm on IP₃-induced Ca²⁺ release from rat cerebellar microsomes, but has no effect on caffeine-induced release from skeletal or cardiac microsomal preparations (Maruyama et al. 1997). Recently it has been employed as an IP₃R inhibitor in a number of studies of isolated smooth muscle cells up to concentrations of 100 μm (e.g. Sergeant et al. 2001, 100 μm; Gordienko & Bolton, 2002, 30 μm; White & McGeown 2003 100 μm; Lee et al. 2003, 100 μm). Xestospongin C has also gained popularity as a membrane-permeable IP₃R inhibitor and has been used at concentrations up to 10 μm with isolated cells (Bayguinov et al. 2000, 5 μm; Gordienko & Bolton, 2002, 10 μm; White & McGeown 2003, 2 μm; Ozaki et al. 2002, 3 μm; Lee et al. 2003, 1 μm). Xestospongin C (XeC) is a ‘macrocyclic’ bis-1-oxaquinolizidine, isolated from Australian sponges of the Xestospongia genus. It has an IC₅₀ of 358 nm on IP₃-induced Ca²⁺ release from rabbit cerebellar microsomes but does not affect the binding of IP₃ itself to the receptor (Gafni et al. 1997). Despite their popularity both 2-APB and XeC may have some limitations in regard to their specificity. 2-APB inhibits store-operated channels (SOCs) in hepatocytes and B cells, seemingly independently of any affect on Ca²⁺ stores (see Discussion). XeC blocks Ca²⁺ uptake by permeabilized A7r5 smooth muscle cells (De Smet et al. 1999) and may open the IP₃Rs at concentrations above 20 μm (Schaloske et al. 2000).

Figure 9. Effect of IP₃R antagonists on Ca²⁺ release.

A, ratioed xt series of fluo-3 fluorescence in a fundus cell. The scanning protocol was the same as in Fig. 6A (control) except that vehicle was replaced by 100 μm 2-APB. B, a ratioed xt series of fluo-3 fluorescence in another fundus cell using the same scanning parameters but scanned once for 6 min. 10 μm xestospongin C was added after 2 min and this was supplemented by 10 μm carbachol after a further 2 min (splitting the protocol into three periods: period 1, before addition of XeC; period 2, in the presence of XeC; period 3, in the presence of XeC and carbachol). C and D, quantitative parameters determined as in the control (Fig. 6), for 100 μm 2-APB and 10 μm XeC, respectively. Statistical tests and significance indicated as before (*P < 0.05; **P < 0.02; ***P < 0.01; ****P < 0.001).

2-APB (100 μm,n= 6) had no immediate effect on Ca²⁺ release and therefore a four period protocol was adopted as for ryanodine (Fig. 9A and C). Over all four periods there was no significant change in either the baseline or MPV. However the baseline did gradually increase and MPV gradually decrease, with the result that the difference between the two was insignificant after 12 min (at the third period). After addition of carbachol both baseline and MPV increased but not as much as in the controls or with ryanodine, and the difference between them remained insignificant. The initial carbachol-induced transient was significantly (P < 0.01) reduced in amplitude, being abolished in half of the cells tested (Fig. 12).

XeC (10 μm,n= 6) acted much more quickly than 100 μm 2-APB. This probably reflects the greater membrane permeability of XeC. Therefore a three period, single scan protocol was used: XeC was added after 2 min and after a further 2 min this was supplemented with carbachol, splitting the protocol into three periods (Fig. 9B and D). By the second period the MPV was reduced significantly (P < 0.01) and there was no longer any significant difference between it and the baseline. Unlike ryanodine and 2-APB, XeC alone caused no increase in the baseline. Carbachol increased MPV and the baseline slightly but the difference between them remained insignificant. With the three period protocol XeC had no significant effect on the initial carbachol-induced transient (Fig. 12). This was not due to the limited incubation period of XeC, as there was still no significant inhibition of the transient by 10 μm XeC using a four period protocol as for ryanodine and 2-APB (n= 3, mean transient peak minus baseline = 5.33 ± 0.25F/F₀).

Effect of lowering external calcium

Changing the external Ca²⁺ concentration from the standard 2 mm to nominally Ca²⁺-free immediately abolished all Ca²⁺ events (n= 10; Fig. 10). MPV was significantly reduced (P < 0.001) and no significant difference remained between it and the baseline. This was fully reversible with readmission of Ca²⁺ (n= 5; Fig. 10B and C). At 10 μm, carbachol caused a calcium transient in nominally Ca²⁺-free solution, which was not significantly different in amplitude from controls (n= 5; Figs 10A and 12). However carbachol had no stimulatory effect on localized release (Fig 10A and D).

Figure 10. Effect of lowering external Ca²⁺ on Ca²⁺ release.

A and B, ratioed xt series of fluo-3 fluorescence in two fundus cells. In A the line was scanned for 8 min and after 3 min the bath [Ca²⁺] was reduced from 2 mm to nominally free (i.e. no calcium added). 10 μm carbachol was added after a further 3 min. In B the line was scanned for 9 min and again after 3 min the bath [Ca²⁺] was reduced from 2 mm to nominal (i.e. none added). After a further 2.5 min the bath [Ca²⁺] was restored to 2 mm. Both protocols are therefore split into three periods (period 1, 2 mm Ca²⁺; period 2, nominally Ca²⁺-free; period 3, 2 mm Ca²⁺ or 10 μm carbachol). C and D, the last 90 s of each period was analysed quantitatively as in Fig. 6. Statistical tests and significance indicated as before (*P < 0.05; **P < 0.02; ***P < 0.01; ****P < 0.001).

Pharmacology: PLC inhibitors

Two different antagonists of phospholipase C (PLC) were tested for their effect on Ca²⁺ release. U-73122 is an aminosteroid which inhibits phosphatidylinositol phosphate (PIP) metabolism in permeabilized platelets and diacylglycerol production in mononucleocytes, with an IC₅₀ of 2–40 μm, depending on the metabolite (Bleasdale et al. 1990). U-73122 has been used for some time as a PLC inhibitor for isolated cells and is used at concentrations of up to 3 μm (Bayguinov et al. 2000, 1 μm; Gordienko & Bolton, 2002, 2.5 μm; Lee et al. 2003, 1 μm). As with XeC, at relatively high concentrations U-73122 also inhibits Ca²⁺ uptake by microsomes (IC₅₀ of 9 μm in rat liver microsomes; De Moel et al. 1995; see also Taylor & Broad, 1998). Neomycin is an aminoglycoside which binds to phosphatidylinositol phosphates, thus preventing their metabolism by PLC (Hildebrandt et al. 1997). It has a K_i of 4 mm for PIP hydrolysis in permeabilized platelets (Bleasdale et al. 1990) but is not very membrane permeable.

At a concentration of 4 μm U-73122 caused a gradual increase in global Ca²⁺, consistent with an inhibition of SR uptake (Fig. 11A). At 1 μm, this did not occur (n= 6), but also there was no immediate inhibition. Therefore the four period protocol was used (Fig. 11B) as with ryanodine and 2-APB. After 12 min (at the third period) the baseline and MPV had both fallen significantly (P < 0.001 and P < 0.01, respectively; Fig. 11B and D). There was no significant difference between baseline and MPV, which was not altered by the addition of carbachol. In fact the amplitude of the initial carbachol-induced transient was significantly reduced (P < 0.02), being abolished in all but two of the cells tested (Fig. 12). Instead there was usually a shallow rise in global calcium, similar to that seen with 4 μm U-73122. The effects of U-73343, a weaker analogue of U-73122, were tested with the same protocol. U-73343 (1 μm) caused no significant change in the MPV or baseline over the last three periods (with respect to the first period), except for a decrease in baseline in period 2 (P < 0.05; results not shown). Values of MPV for the four periods were 1.47, 1.12, 1.32 and 2.35F/F₀, respectively, and values of baseline were 0.96, 0.77, 1.05 and 1.32F/F₀, respectively. However the difference between the MPV and baseline was only significant for the first two periods (P < 0.05 for both; MPV minus baseline = 0.51, 0.34, 0.28 and 1.02F/F₀ for the four periods, respectively). There was no significant inhibition of the initial carbachol-induced transient (Fig. 12).

A four period protocol was also used with neomycin at a concentration of 20 μm (n= 7). Both baseline and MPV were significantly reduced by the second period (immediately after admission of neomycin; Fig. 11E). However it was only after 12 min (in the third period) that there was no significant difference remaining between baseline and MPV. Unlike U-73122 there was no inhibition of the initial carbachol-induced transient, which if anything was actually enhanced (Fig. 12). Also there was no inhibition of localized Ca²⁺ release events after the initial carbachol-induced transient (period 4): the baseline and MPV both increased and there was a significant difference between them (Fig. 11E).

Contractile properties

The cells studied above had no or a weak contractile response to carbachol. Without this a comparison of localized release before and after carbachol would not have been satisfactory. Of those cells, analysed above, which responded at all to carbachol (with a initial Ca²⁺ transient; Fig. 12) only 6 out of 45 cells contracted at all (1 cell under control conditions, 2 in the presense of 10 μm ryanodine, 1 in the presence of 50 μm ryanodine and 2 in nominally Ca²⁺ free). Further, 2/6 cells contracted in response to 5 mm caffeine. In all these cases the degree of contraction was limited to less than 15% of the length of the line scan, and no correlation was noticed between the amplitude of the Ca²⁺ transient and the strength of contraction.

Discussion

The external calcium dependence of fundus tone, both resting and in response to agonists is well established (Kuriyama et al. 1975; Parekh & Brading, 1991; Parekh & Brading, 1992; Cox & Cohen, 1995; Korolkiewicz et al. 2000; Shimamura et al. 2000). It is also thought that many contractile agonists (e.g. carbachol, serotonin and galanin) work through IP₃R-dependent pathways in the fundus (Parekh & Brading, 1991; Foguet et al. 1992; Cox & Cohen, 1995). In addition, recent microelectrode studies on the guinea-pig antrum and fundus have demonstrated the IP₃R dependence of ‘unitary noise’, which is believed to reflect activation of calcium-dependent membrane conductances by intracellular calcium release, possibly localized (Suzuki et al. 2000; Dickens et al. 2001; Hirst & Ward, 2003 for review). Unitary noise underlies the cholinergic excitatory junction potential of the fundus (Ward et al. 2000; Beckett et al. 2002) and is believed to originate within intramuscular interstitial cells of Cajal (ICC_IM). The fundus cells, which are the subject of this study, resemble ICC_IM described in electron microscopic and immunohistological studies of the intact fundus of human and mouse, respectively (Faussone-Pellegrini et al. 1989; Epperson et al. 2000). In the mouse fundus, ICC lose their ability to bind c-kit antibodies after acute enzyme dispersion (Epperson et al. 2000), and we also found this to be the case in guinea-pig, preventing definitive identification of the cells as ICC. The ability of the cells in this study to regulate their [Ca²⁺]_i and to maintain a low level of resting [Ca²⁺]_i argues against their being simply archetypal but damaged smooth muscle cells. As regards isolated fundus cells, a number of studies have been published on the relaxing effects of VIP and NO (all from the group of Lefebvre; see Dick et al. 2002), and two studies have been made of their electrophysiology (Lammel et al. 1991; Duridanova et al. 1996). It is significant that Lammel et al. (1991) found that STOCs could only be activated in myocytes at holding potentials positive to +50 mV, which would fit with the apparent quiescence of archetypal SMCs in this study.

Localized calcium release was reduced both by ryanodine and by blockers of IP₃Rs/PLC, suggesting that it may involve calcium store release through both RyRs and IP₃Rs. Despite its dependence on store calcium, localized calcium release was acutely dependent on calcium entry. This has a precedent from studies of puffs in colonic myocytes (Kong et al. 2000; Koh et al. 2001; Bayguinov et al. 2001a) and PLC/IP₃-linked agonist-induced [Ca²⁺]_i oscillations in a variety of cell types (e.g. Felder et al. 1992; Hashii et al. 1993; Yao et al. 1994; Thorn, 1995; Wu et al. 1995; Komori et al. 1996; Kohda et al. 1998). This entry could regulate Ca²⁺ release (1) via regulation of the peripheral SR at some subplasmalemmal space (2) via stimulation of a calcium-sensitive PLC (i.e. PLCβ) or (3) via direct activation of the IP₃Rs and/or RyRs (i.e. calcium-induced calcium release, CICR) (Fig. 13).

Figure 13. A model for calcium entry and release in fundus cells.

Localized release events and the initial carbachol-induced calcium transient appear to be associated with separate mechanisms (dashed and full lines, respectively). The inhibition of localized release by ryanodine and its (ryanodine-insensitive) stimulation by carbachol can be explained by the direct modulation of IP₃Rs and/or RyRs by calcium in the immediate vicinity of the receptors. The calcium entry pathway associated with localized release is inhibited in nominally Ca²⁺-free solution. To explain the insensitivity of the initial carbachol-induced calcium release to XeC it is necessary to suppose that the copious production of IP₃ overwhelms the antagonism by this agent. However, 2-APB and U-73122 are effective blockers; neomycin blocks transient calcium release events but, perhaps because it is poorly membrane permeable, was ineffective at blocking the initial carbachol-induced calcium transient. Caffeine is not shown, but in high enough concentrations would act on RyRs to deplete stored calcium.

However, it is this dependence on calcium entry which makes the interpretation of the effects of the IP₃R and PLC inhibitors uncertain, as a number of recent studies have found that these can inhibit ‘store-operated’, non-specific cation, voltage-dependent or other inward calcium currents (see below and Bootman et al. 2002 for review).

The concentrations of 2-APB, XeC and other drugs used in this study were similar to those used by others in studies of calcium release in various cell types. Ryanodine has been used at concentrations from 10 to 100 μm (Bayguinov et al. 2000; Gordienko & Bolton, 2002; Hollywood et al. 2003; White & McGeown, 2003), 2-APB at concentrations from 30 to 100 μm (Ma et al. 2001; White & McGeown, 2002; Gordienko & Bolton, 2002; Hollywood et al. 2003; Kim et al. 2003; Lee et al. 2003), XeC at concentrations from 0.3 to 20 μm (Bayguinov et al. 2000; Ma et al. 2000; Schaloske et al. 2000; Gordienko & Bolton, 2002; Hollywood et al. 2003; Lee et al. 2003) and U-73122 up to 3 μm (Kim et al. 2003; Lee et al. 2003). Ma et al. (2000) found that 75 μm 2-APB blocked thapsigargin-stimulated (i.e. store-operated) calcium entry in HEK293 cells, with less than a minute's pretreatment. In contrast 20 μm XeC, applied for 20 min before thapsigargin, only reduced entry by half. Similar results were obtained by Bishara et al. (2002), studying calcium entry in aortic endothelial cells; either thapsigargin, ATP or ionomycin caused a initial calcium transient in the absence of external calcium (indicating SR calcium release), followed by a second slower transient upon calcium readmission (to 2.5 mm), indicative of calcium entry. Whilst 100 μm 2-APB blocked both transients with less then 10 s pretreatment, a much longer pretreatment with 10 μm XeC blocked the initial transient but only inhibited the second transient by half. Lee et al. (2003) found that in murine antral myocytes, 100 μm 2-APB blocked the transient inward current in response to 50 μm carbachol after only a minute's pretreatment. In contrast Ozaki et al. (2002) found that in guinea-pig ileal myocytes 3 μm XeC had no effect on the inward current stimulated by 10 μm carbachol (although it did inhibit a voltage-dependent barium inward current). White & McGeown (2003) found that in guinea-pig vas deferens myocytes calcium transients in response to noradrenaline were blocked by 100 μm 2-APB and U-73122, but not by 2 μm XeC. Hildebrandt et al. (1997) found that in NG108-15 cells, the [Ca²⁺]_i transient and inward current, in response to 1 μm bradykinin were blocked by 5 μm U-73122, but neomycin had no effect on either phenomenon at concentrations up to 3 mm.

These studies suggest a difference between the actions of 2-APB and U-73122 on the one hand and XeC and neomycin (which is not very membrane permeant) on the other. Whilst the former are clearly good inhibitors of receptor- (especially muscarinic-) or store-operated calcium entry/currents, the latter are not (XeC acts weakly), although they may very well inhibit voltage-dependent currents in common with U-73122 (see Hildebrandt et al. 1997 for discussion; Ozaki et al. 2003). This dichotomy was reflected in the pharmacology of calcium release in fundus cells: on the one hand 2-APB and U-73122 inhibited both localized release and the initial carbachol-induced calcium transient, while XeC and neomycin on the other hand only inhibited localized calcium release events. It might be suggested from the above that the initial carbachol-induced transient is more dependent on calcium entry as it was only inhibited by 2-APB and U-73122, whereas localized release is dependent on IP₃R/PLC as it was inhibited by all four drugs.

However, a more likely explanation is that, since the initial carbachol response was not abolished in nominally calcium-free solution (Fig. 10A), activation of muscarinic receptors by 10 μm carbachol produces a large increase in IP₃ production which overwhelms XeC antagonism, and XeC was ineffective at blocking carbachol-induced calcium store release due to copious IP₃ production which still releases substantial stored calcium (Figs 9, 12 and 13). Neomycin is poorly membrane permeant and, possibly for this reason, was similarly ineffective, despite being an effective blocker of localized calcium release events. Nevertheless, XeC had a quick effect on localized release, relative to 2-APB, perhaps because (1) 2-APB is an amine which is a cation (protonated) at physiological pH (Bootman et al. 2002) and so enters the cell more slowly, and (2) XeC was used at 28 times its IC₅₀ for release from cerebellar microsomes (IC₅₀= 358 nm with stimulation by 5 μm IP₃, Gafni et al. 1997) as opposed to ∼2 times in the case of 2-APB (IC₅₀= 42 μm with stimulation by 100 nm IP₃, Maruyama et al. 1997). Also, it should be noted that the weak inhibition of localized release by U-73343, relative to U-73122, would fit with its correspondingly smaller IC₅₀ for inhibition of phosphatidylinositol 4,5-bisphosphate (PIP₂) hydrolysis by PLC (Bleasdale et al. 1990).

The initial carbachol-induced transient was reduced by ryanodine but not significantly by XeC, so it might appear that RyRs rather than IP₃Rs, are involved. However, U-73122 also blocked the carbachol-induced transient and ryanodine releases calcium from stores, so that after several minutes of ryanodine treatment, store calcium may be reduced. However, the inhibition of localized release by ryanodine was alleviated by carbachol, suggesting that, if IP₃ production is increased, the residual store calcium released through IP₃Rs was sufficient to support localized calcium release events.

The mechanisms proposed for calcium release in fundus cells provide several possible explanations for the heterogeneity of calcium release event dynamics. RyRs within clusters could be ‘recruited’ stochastically by calcium release through IP₃Rs to generate events of varying magnitude by either stochastic variations in IP₃ levels (activity of PLC) and/or calcium entry (filling the peripheral SR or activating PLCβ (see Swillens et al. 1999; Shuai & Jung, 2002 and Falcke, 2003 for theoretical treatments of cluster recruitment) (Fig. 13). Despite this the greater differences in event magnitude seem to reflect release from spatially separate sites which generate relatively stereotyped events, i.e. some sites (‘frequent discharge sites’; Gordienko & Bolton, 2002) consistently generate large events that fill the width of the cell and lasted tens of seconds, whilst other sites consistently generate smaller transient events (Fig. 2).

In relation to the dynamics of channel clusters and how these relate to the visible event kinetics, one cannot disregard the recent theoretical treatments of amplitude distributions (Pratusevich & Balke, 1996; Smith et al. 1998; Izu et al. 1998; Cheng et al. 1999; Swillens et al. 1999; Shuai & Jung, 2002). Izu et al. (1998) showed that an inflection in the inverse plot of a hyperbolically decaying amplitude distribution, would indicate the presence of two modes in the actual distribution of event amplitudes (i.e. when all events were scanned at their origin). The authors modelled these two modes as two event populations with different mean calcium release channel currents (1 and 2 pA). However, Shuai & Jung (2002) suggested that multiple modes in the amplitude distribution could result from increases in IP₃ concentration.

In addition to the general properties of elementary release, it is necessary to consider the actual spatio-temporal profiles of individual events. Most of the smaller events (<3 s FDHM) may be described as typical ‘shot’ events, as for puffs and sparks – that is they consisted of a rapid onset followed immediately by a diffusive (exponential) decay. In this respect they are unremarkable, other than by their size, which probably scales with the number of channels open at their inception. In contrast the larger events (>3 s FDHM) displayed some rather unusual characteristics. In general they did not ‘switch off’ after the initial release, suggesting that there must be a mechanism to prevent autoinhibition of the channel by the calcium it releases (traditionally considered as the limiting/terminating factor in CICR). This might involve some protein accessory to the channel itself. ‘Plateau’ events have been noticed on occasion with sparks in skeletal muscle (Shtifman et al. 2000; Gonzalez et al. 2000; Kirsch et al. 2001), ileal mycoytes (Gordienko et al. 1998), vas deferens myocytes (White & McGeown, 2003) and puffs in Xenopus oocytes (Marchant & Parker, 2002). The lack of multiple levels in the fundus plateau events would argue against the presence of release channel subconductance states shown to operate with plateau events in frog skeletal muscle (Gonzalez et al. 2000).

This is the first study to demonstrate the role of calcium signalling at the cellular level in the fundus. The presence of spontaneous calcium release events in the cells studied, and the absence of such events in the archetypal smooth muscle cells, is consistent with the role of the former as drivers of contractile activity in the fundus but support for this possibility will require further investigation.