Requirement of ryanodine receptors for pacemaker Ca²⁺ activity in ICC and HEK293 cells

Many missing links in the pathway that generates spontaneous rhythmicity in the gastro-intestinal tract remain, despite its importance. Like the heart beat, spontaneous movement of the gastro-intestinal tract is essential to mix, digest and transport the luminal contents. Recent studies have suggested that, as in sino-atria node cells in the heart, specialised pacemaker cells exist in the gastro-intestinal wall, and drive smooth muscle cells to achieve their functions (Suzuki, 2000). The pacemaker cells of the gastro-intestinal wall express a specific tyrosine kinase, c-Kit, and are characterised from their histological features as interstitial cells of Cajal (ICCs).

L-type Ca²⁺ channels play a central role in the excitation-contraction coupling of smooth muscle (Nakayama et al., 1996; Beech and McHugh, 1996). Electrical pacemaker activity in the form of slow waves, is, however, hardly affected by dihydropyridine (DHP) L-type Ca²⁺ channel blockers. This fact suggests that other Ca²⁺ sources play important roles in gastro-intestinal pacemaking. Several groups have suggested that Ca²⁺-activated Cl^– channels in ICCs are responsible for electrical pacemaker potentials (Tokutomi et al., 1995; Edwards et al., 1999; Huizinga et al., 2002). It can therefore be deduced that oscillations of cytosolic Ca²⁺ concentration ([Ca²⁺]_i), presumably through release of Ca²⁺ from intracellular stores, are the primary pacemaker activity. This putative underlying mechanism also accounts for the low voltage-sensitivity of pacemaker frequency (Tomita, 1981).

Two families of intracellular Ca²⁺ release channels, such as inositol 1,4,5-trisphosphate (InsP₃) receptor types 1-3 and ryanodine receptor types 1-3, are known along with several intracellular messengers and modulators (e.g. cyclic adenosine diphosphate ribose and nicotinic acid adenine dinucleotide phosphate) (Galione and Churchill, 2002; Masgrau et al., 2003). Several lines of evidence suggest that the InsP₃ receptor (IP₃R) plays a role in generating spontaneous electrical activity in gastro-intestinal pacemaker cells (Suzuki et al., 2000; Hirst and Edwards, 2001; Malysz et al., 2001; Sergeant et al., 2001), and it has been suggested that RyR is involved in NO-induced Ca²⁺ transients in colonic ICCs (Publicover et al., 1993). However, the contribution of the ryanodine receptor (RyR) family to the pacemaker activity is still unclear. Recently, we have developed a new preparation to study mechanisms underlying gastro-intestinal pacemaker activity: small cell clusters containing c-Kit-immunopositive interstitial cells (ICCs), smooth muscle cells and enteric neurones, which show stable spontaneous rhythmicity in terms of mechanical, electrical and intracellular Ca²⁺ activities (Nakayama and Torihashi, 2002; Torihashi et al., 2002). In the present study, using this preparation, we directly demonstrate that RyR as well as IP₃R contribute to intracellular Ca²⁺ oscillation in ICCs, the putative pacemaker cells.

RT-PCR reveals that ICCs predominantly express ryanodine receptor type 3 (RyR3). Among the three types of RyRs, RyR3 has been proved to be widely expressed in brain, smooth muscles, skeletal muscle and non-excitable cells (Bennett et al., 1996; Ogawa et al., 2002; Sorrentino and Volpe, 1993). Elucidation of the physiological impact of RyR3 on cellular functions has, however, been hampered by lower expression levels of RyR3 in comparison with other isoforms of RyRs co-expressed in many types of cells. The contribution to [Ca²⁺]_i oscillations in ICCs is, so far as we know, the first evidence for an obligatory role of RyR3. Furthermore, we have reconstituted pacemaker-like [Ca²⁺]_i oscillations by transfecting an expression vector for RyR3 in HEK293 cells, which originally express IP₃Rs, but not RyRs, and do not display spontaneous [Ca²⁺]_i oscillation. Interestingly, RyR or IP₃R inhibitors had essentially the same effects on [Ca²⁺]_i oscillations seen in ICCs and the HEK293 cells. Taken together, these observations suggest that RyR3 can be a key molecule to generate, or at least modulate, pacemaker [Ca²⁺]_i oscillations in both cell clusters containing ICCs and reconstituted cell systems.

The preparation of cell clusters has been described previously (Nakayama and Torihashi, 2002). The mice were treated ethically according to the Guidelines for the Care and Use of Animals approved by the Physiological Society of Japan. BALB/c mice (10-20 d after birth) were killed by cervical dislocation, and small pieces of the small intestine were incubated in Ca²⁺-free Hanks' solution containing digestive enzymes as previously reported (Nakayama and Torihashi, 2002). After being rinsed with Ca²⁺-free Hanks' solution, the muscle pieces were triturated with fire-blunted glass pipettes of decreasing tip diameter. The resultant small cell clusters were plated onto a lab-made culture dish (a sterile silicone ring approximately 20 mm in diameter on a pig collagen-coated sterile 25 mm glass cover slip). After 2-3 days of incubation, the cultured cell clusters were used for Ca²⁺ imaging.

The resultant cell suspension was incubated with `normal' solution containing phycoerythrin-conjugated anti-mouse CD117 (c-Kit) antibody (PE-ACK2, eBioscience, San Diego, CA) in 1/100 v/v for 10 minutes. About 5-10 isolated smooth muscle cells and c-Kit-immunopositive cells were separately collected with glass pipettes of 10-20 μm tip diameter under a fluorescent microscope and kept at –80°C until the use of RT-PCR. Smooth muscle cells and c-Kit-immunopositive cells (ICCs) were judged by their characteristic spindle shape and immunofluorescence.

HEK293 cells (Japanese Health Science Research Resources Bank, Tokyo, Japan) were maintained as previously reported (Imaizumi et al., 2002). Cells were transfected with DNA encoding RyR2 or RyR3, using the Ca₂PO₄ co-precipitation method. Cells were plated on cover slips about 24 hours before transfection and were transfected with 2-4 μg of DNA per dish. Control cells were treated in the same way with no DNA.

The cultured cell clusters were incubated for 3-4 hours in normal solution containing 8 μM fluo-3/AM and detergents (0.02% Pluronic F-127, Dojindo, Kumamoto, Japan; 0.02% cremophor EL, Sigma). A CCD camera system (Argus HiSCA, Hamamatsu Photonics, Hamamatsu, Japan) combined with an inverted microscope (Axiovert S100TV, Zeiss, Germany) was used to monitor continuously digital images of fluo-3 emission light. The cell clusters were illuminated at 488 nm, and emission light of 515-565 nm was detected. Digital images were normally collected at 300 millisecond intervals. Changes in fluorescence emission intensity (F) were expressed as F_t/F₀, where F₀ is the basal fluorescence intensity obtained at the start of the experiment. During Ca²⁺ imaging, the temperature of the recording chamber was kept at 35°C using a micro-warm plate system (DC-MP10DM, Kitazato Supply, Fujinomiya, Japan). When the amplitude of [Ca²⁺]_i oscillations fell below 10% of the control value or within the noise level, we judged it to cease.

Average of [Ca²⁺]_i in transfected and non-transfected HEK293 cells was measured with fura-2/AM. About 48 hours after transfection, cells were loaded with 10 μM fura-2/AM for 20 minutes. [Ca²⁺]_i was measured with a Ca²⁺ imaging system (ARGUS-50/CA, Hamamatsu, Japan) under constant flow of KRH solution at room temperature. The intensity of emission fluorescence at 500 nm was measured synchronously to the alternate excitation (F340 and F380). The ratio of fluorescence emitted at 340 nm and 380 nm was converted to Ca²⁺ concentration using methods introduced previously (Grynkiewicz et al., 1985).

Confocal Ca²⁺ images were obtained using a fast laser scanning confocal microscope (RCM 8000, Nikon, Japan) and Ratio3 software (Nikon), as reported previously (Imaizumi et al., 1998). Cells were loaded with 10 μM fluo-4/AM for 30 minutes. Images were analyzed using Ratio3 software and GLOBAL LAB image (Data Translation, Marlboro, MA).

Smooth muscle layers (including the myenteric plexus) isolated from mouse small intestine were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100 for 10 minutes. The tissue was cut into small segments (∼10 mm), and was double stained sequentially with anti-RyR antibody (clone 34C produced in mouse, Sigma, St Louis, MO) and anti-c-Kit (mouse CD117) antibody for 1.5 hours. This was followed by incubation with secondary antibodies, Alexa-conjugated anti-mouse or rat IgG (Molecular Probes, Eugene, OR) at the concentration of 15 μg/ml for 1 hour. Controls were prepared by omitting the primary antibodies. Double-stained small segments were mounted on a slide glass with an anti-fading agent (ProLong, Molecular Probes) and scanned using a confocal microscope (MRC-1024: Bio-Rad, Hercules, CA).

The level of RyR protein expression in HEK293 transfectants was examined using similar protocols. After the measurements of [Ca²⁺]_i, the coverslips were fixed, permeabilized, and treated with anti-RyR antibody specific for RyR2 or RyR3 for 2 hours. After washing, the cover slips were incubated with Alexa-conjugated anti-rabbit or mouse IgG for 60 minutes. Finally, the cover slips were mounted on slides after washing, and used for the analyses with the confocal microscope (RCM 8000; Nikon, Japan).

Total RNA extraction and reverse-transcription were performed as reported previously (Ohya et al., 1997). The resultant cDNA products were amplified by PCR with gene-specific primers (Table 1). The amplification profile was as follows: 15 seconds at 95°C and 60 seconds at 60°C. In the tissue and cell-based RT-PCR, the amplification was performed for 35 and 45 cycles, respectively. The RT-PCR products were separated by electrophoresis on a 2% agarose gel, and documented on a FluorImager 595 (Amersham Biosciences, Piscataway, NJ). The no template control (NTC: Fig. 5) was an RT-PCR product in which no sample RNA was added in order to monitor non-specific amplification and spurious primer-dimer fragments. Each amplicon was sequenced using a DSQ-1000L sequencer (Shimadzu, Kyoto, Japan).

The composition of the standard solution used for cell clusters was (mM): 125 NaCl; 5.9 KCl; 1.2 MgCl₂; 2.4 CaCl₂; 11 glucose; 11.8 Tris-HEPES (pH 7.4). The standard solution used for HEK293 cells contained (mM): 125 NaCl, 5 KCl, 1.2 KH₂PO₄, 6 glucose, 1.2 MgCl₂, 2 CaCl₂ and 25 Hepes; pH 7.4, adjusted with NaOH. Ca²⁺-free KRH solution was prepared by omitting Ca²⁺ in the preparation of standard KRH solution and the addition of 0.01 mM EGTA.

The source of pharmacological agents were as follows: nifedipine and ryanodine (Sigma); caffeine (anhydrous, Kanto Kagaku, Tokyo, Japan); xestospongin C and 2APB (2-aminoethoxydiphenyl borate) (Calbiochem, San Diego, CA); fura-2/AM, fluo-4/AM, and Alexa™-labelled secondary antidodies (Molecular Probes); fluo-3/AM (Dojindo).

Numerical data are expressed as means±s.d. in the text. Statistical significance (P<0.05) was evaluated using ANOVA or Student's t-test.

Cultured cell clusters from mouse small intestine were loaded with fluo3-AM, and fluo-3 emission light was monitored as an index of intracellular Ca²⁺ using a CCD camera system. In a `normal' solution containing 2.4 mM Ca²⁺, many cell clusters showed spontaneous contractions at a frequency of 8-20 cycles/minute at 35°C (Nakayama and Torihashi, 2002). It is known that DHP L-type Ca²⁺ channel blockers abolish spontaneous contraction in gastro-intestinal smooth muscle preparations, but have little effect on slow waves (the pacemaker potentials) (e.g. Dickens et al., 1999; Huang et al., 1999). In the present study, nifedipine was thus used to distinguish pacemaker Ca²⁺ activity in cultured cell clusters. (We also checked that spontaneous contractions in cell cluster preparations were accompanied by such pacemaker Ca²⁺ activity, and cell clusters contracted like a functional syncytium in the absence of Ca²⁺ channel blockers; S.N., unpublished.) In the presence of 1-2 μM nifedipine, [Ca²⁺]_i oscillations were observed in one or more regions (e.g. Fig. 1Aa,Ba) of the cell clusters which had shown spontaneous contraction before nifedipine application. In the majority of cell clusters in which [Ca²⁺]_i oscillations were observed in multiple regions, the periodic increases in [Ca²⁺]_i were synchronized. The mean frequency of [Ca²⁺]_i oscillations in the presence of nifedipine was 18.4±5.1 cycles/minute (n=26).

Using cell cluster preparations in the presence of a DHP Ca²⁺ channel blocker, we examined drugs that could affect intracellular Ca²⁺ release channels. Fig. 1 shows examples of the effects of ryanodine on cell clusters with single (Fig. 1A) or multiple regions of [Ca²⁺]_i oscillations (Fig. 1B). The traces in Fig. 1Ab show changes in the fluorescence intensity (F_t/F₀) measured in the square region indicated in Fig. 1Aa, panel 1. The Ca²⁺ images of Fig. 1Aa, panels 1-3 were recorded at the times indicated by lines 1-3 in Fig. 1Ab, respectively. In the presence of 2 μM nifedipine, application of 1 μM ryanodine significantly reduced the amplitude and frequency of [Ca²⁺]_i oscillations after 5 minutes. An increase in the ryanodine concentration to 10 μM completely terminated [Ca²⁺]_i oscillations after 5 minutes, and the subsequent washout of ryanodine for 5 minutes did not restore it. After examination of the effects of ryanodine, the cell cluster was stained using ACK2. c-Kit-immunoreactivity was detected in the square region, indicating that [Ca²⁺]_i oscillations in ICCs were modulated by ryanodine. Similar effects of ryanodine were observed in three other cell clusters.

The cell cluster preparation shown in Fig. 1Ba contained multiple regions of [Ca²⁺]_i oscillations. The time courses of changes in [Ca²⁺]_i in Fig. 1Bb were monitored in the boxed regions x (blue line) and y (red line). The [Ca²⁺]_i oscillations seen in the two regions of the cell cluster were well synchronised (in the presence of nifedipine). Application of 10 μM ryanodine reduced both [Ca²⁺]_i oscillation amplitude and frequency in a time-dependent manner. As shown in the second and third traces of Fig. 1Bb, [Ca²⁺]_i in the x and y regions were affected similarly in terms of amplitude and frequency. Fig. 1Bc shows the correlation between F_t/F₀ in regions x and y for the three traces shown in Fig. 1Bb. For each pair of F_t/F₀, the cross correlation function was maximal (0.93-0.95) at t=0 (Fig. 1Bd). When the [Ca²⁺]_i oscillation amplitude was reduced significantly, the frequency also tended to decrease. The time for [Ca²⁺]_i oscillations to cease ranged from 3 to 18 minutes after 10 μM ryanodine application (6.6±4.9 minutes). In the F_t/F₀ trace recorded before cessation, the mean amplitude and frequency were 54±32% and 71±21% of the control, respectively.

It is well known that caffeine modulates the properties of ryanodine receptors and amplifies Ca²⁺-induced Ca²⁺ release (Endo, 1977; Ogawa et al., 2002). In Fig. 2A, panels 1 and 2 the Ca²⁺ images were obtained at resting and peak times of [Ca²⁺]_i oscillations, respectively (indicated by lines 1 and 2 in Fig. 2Ba). In the presence of nifedipine (2 μM), application of 10 mM caffeine significantly suppressed [Ca²⁺]_i oscillations. In all clusters examined (n=5), [Ca²⁺]_i oscillations ceased within 5 minutes, while in four out of the five clusters, as the amplitude decreased, the frequency also decreased. The amplitude and frequency of [Ca²⁺]_i oscillations in the traces before cessation were 42±23% and 40±24% of the control, respectively (n=4). As shown in Fig. 2Bb, caffeine prominently reduced the [Ca²⁺]_i oscillation frequency, but the [Ca²⁺]_i oscillations seen over the pacemaker area were well synchronised (blue line from region x; red line from region y). The trace Fig. 2Bc was recorded 3 minutes after 10 mM caffeine application, and the Ca²⁺ image in Fig. 2A, panel 3 was obtained at the time indicated by line 3 in Fig. 2Bc. The inhibitory effect of caffeine on [Ca²⁺]_i oscillations was reversible as shown in Fig. 2Bd. Lower concentrations (∼1 mM) of caffeine produced similar inhibitory effects, but [Ca²⁺]_i oscillations ceased within 5 minutes in only two out of five preparations.

In order to further confirm the involvement of RyR, we examined effects of other drugs affecting RyR, (because caffeine is also known to interact with type 1 IP₃R) (Maes et al., 1999; Maes et al., 2000; Bezprozvanny et al., 1994). Tetracaine and procine are known RyR blockers (e.g. Lukyanenko et al., 2001; Kimball et al., 1996). In the presence of nifedipine (1 μM), application of 30 μM tetracaine terminated [Ca²⁺]_i oscillations within 3-5 minutes (n=5) (Fig. 2C). Also, 30 μM procaine terminated it within 5 minutes (n=4, not shown). By contrast, FK506 is known to modulate the RyR channel function by binding to the 12 kDa FK506-binding proteins (FKBP12 and FKBP12.6) (Masumiya et al., 2003; Bultynck et al., 2001). As shown in Fig. 2D, [Ca²⁺]_i oscillations very rapidly ceased (1-2 minutes) after application of 10 μM FK506 (n=5).

Spontaneous electrical activity, i.e. slow waves, is observed at irregular intervals, or not at all in the gastric antral smooth muscle of mice lacking type 1 IP₃R (Suzuki et al., 2000), suggesting that IP₃Rs play an important role. Accordingly, we have investigated the possible contribution of IP₃R Ca²⁺ release channels to [Ca²⁺]_i oscillations seen in ICCs in small intestine, by examining the effects of 2APB and xestospongin C, both of which have been reported to diffuse across the cell membrane to block IP₃R (Maruyama et al., 1997; Gafni et al., 1997). Fig. 3A shows an example of the effects of 2APB on nifedipine-resistant [Ca²⁺]_i oscillations (pacemaker [Ca²⁺]_i oscillations). The blue and red lines indicate changes in [Ca²⁺]_i in two pacemaker regions ∼50 μm apart. Application of 10 μM 2APB completely abolished [Ca²⁺]_i oscillations after 5 minutes (Ac). In all four preparations used for 10 μM 2APB experiments, [Ca²⁺]_i oscillations ceased in 5 minutes. By contrast, application of 1 μM 2APB abolished [Ca²⁺]_i oscillation within 5 minutes in only two out of six preparations. 2APB tended to reduce the amplitude of [Ca²⁺]_i oscillations, without a large change in frequency. In the F_t/F₀ traces recorded before cessation of [Ca²⁺]_i oscillations, the average amplitude and frequency were 61±32% and 98±11% of the control, respectively (n=5). The traces in Fig. 3B show an example of the effect of xestospongin C. At 10 μM this drug also completely abolished pacemaker [Ca²⁺]_i oscillation in 2 minutes (n=3). For both drugs, subsequent washout did not restore [Ca²⁺]_i oscillation within 5 minutes. Taken together, these results indicate that IP₃Rs are also involved in [Ca²⁺]_i oscillations in ICCs.

In order to confirm the existence of the ryanodine receptors (RyR) in ICCs of mouse small intestine, we stained the isolated smooth muscle (including the myenteric plexus) with an anti-c-Kit antibody (ACK2) and an anti-RyR antibody (clone 34C) simultaneously (Fig. 4). Fig. 4A (red) shows that the c-Kit-immunopositive cells are distributed in a network-like configuration within the myenteric plexus. In Fig. 4B, the fluorescence of the anti-RyR antibody (green) was detected in both the c-Kit immunopositive cells (network-like) and smooth muscle cells (running perpendicularly). The merged image (Fig. 4C) clearly indicates that most of the c-Kit-immunopositive cells also reacted to the anti-ryanodine receptor antibody (yellow).

In isolated ICCs and smooth muscle cells, we examined mRNA expression of ryanodine receptor isoforms using RT-PCR. Fig. 5A shows control amplicons of ryanodine receptor type 1 to 3 (RyR1 to 3) along with GAPDH complementary DNA obtained from brain, heart and skeletal muscles. As shown in Fig. 5B, RT-PCR examination of isolated cells revealed that RyR3 was predominantly expressed in ICCs (c-Kit-immunopositive cells), while RyR2 was the major RyR isoform in smooth muscle cells. The DNA band for c-kit was detected only in ICCs. Another 5 cycles of PCR produced no additional DNA band in either ICCs or smooth muscle cells. Furthermore, mRNA for ∼12 kDa FKBP was expressed in ICCs as well as smooth muscle cells, as suggested by pharmacological examinations. FKBP12.6 was predominant in smooth muscle, while both isoforms (FKBP12 and FKBP12.6) were detected in ICCs (Fig. 5C), being consistent with previous studies on the interaction between RyR and FKBP isoforms (Timerman et al., 1996; Bultynck et al., 2001). Also, we carried out RT-PCR to check for the expression of InsP₃ receptors. The results of this experiment suggested that the type 3 and type 2 isoforms of IP₃Rs were the major components in smooth muscle and ICCs, respectively (data not shown).

In order to confirm the contribution of RyR3 to [Ca²⁺]_i oscillations seen in ICCs, we examined the reconstitution of functional RyR3 in HEK293 cells, which originally only express IP₃R. According to RT-PCR results, mRNAs of all three IP₃R types were expressed in native HEK293 cells in the order of IP₃R3 > IP₃R2 > IP₃R1, but mRNAs of RyR subtypes were not detected after 45 cycles of amplification (data not shown).

[Ca²⁺]_i was measured with a fluorescent Ca²⁺ indicator (Fura-2) 2 days after the transfection. Approximately 10% of HEK293 cells showed spontaneous global [Ca²⁺]_i oscillations. Fig. 6Aa (black line) demonstrates spontaneous oscillations of [Ca²⁺]_i over the whole cell area and the elevation of [Ca²⁺]_i in response to 1 mM caffeine. Further addition of caffeine did not induce marked response. Approximately 30% of HEK293 cells responded to 1 mM caffeine, suggesting that the efficacy for the functional expression of RyR3 was 30%. As shown by the red line in Fig. 6Aa, the remaining HEK cells (∼70%) did not respond to caffeine in the range of 1 and 10 mM, but did respond to 1 μM ACh. Fig. 6Ab shows Ca²⁺ images acquired under resting conditions and in the presence of 1 mM caffeine or caffeine plus 1 μM ACh and an image following immunocytochemical staining with anti-RyR antibody. Application of 1 mM caffeine increased [Ca²⁺]_i in two cells indicated by white arrowheads, but not in other cells. Substantial expression of RyR3 proteins was identified in cells sensitive to caffeine without exception (n>200), indicating a direct correlation between high sensitivity to caffeine and the successful expression of RyR3 proteins in transfected cells. Application of 1 μM ACh in the presence of caffeine evoked [Ca²⁺]_i transients in caffeine-insensitive cells. The results suggest that RyR3 expressed in the HEK293 cells shares the Ca²⁺ storage sites with IP₃R.

Fig. 6B demonstrates typical spontaneous [Ca²⁺]_i oscillations in HEK293 cells transfected with RyR3. Two cells (1 and 2) in the confocal Ca²⁺ images exhibited Ca²⁺ oscillations, which occurred as Ca²⁺ waves in each cell. Ca²⁺ waves started from a particular initiation point in each cell, as indicated by white arrowheads in cell 1. The frequency of Ca²⁺ oscillations was 1.3±0.6 cycles/minute (n=39 from separate 8 groups of transfection). The average value of the peak [Ca²⁺]_i was 1.6±1.4 μM (n=24). In some experiments, the relation between RyR3 protein expression and the generation of Ca²⁺ oscillation was also examined using BODIPY FL-X ryanodine (Fig. 6C). Among 19 cells in Fig. 6C, Ca²⁺ oscillations were recorded in 12 cells, which were well stained with BODIPY FL-X ryanodine, and 4 of which are indicated (1-4).

Fig. 7 shows pharmacological suppressions of spontaneous [Ca²⁺]_i oscillations in HEK293 cells transfected with RyR3. Application of ryanodine (10 μM) terminated [Ca²⁺]_i oscillations within 3 minutes (n=3) (Fig. 7A). Furthermore, applications of either tetracaine (Fig. 7B: 50 μM, n=3), procaine (10 μM, n=3, not shown) or FK506 (Fig. 7C: 20 μM, n=4), known modulators for RyR, also terminated it. By contrast, applications of either 2APB (100 μM) (D, n=3) or xestospongin C (20 μM, n=3, data not shown), both IP₃R blockers, abolished [Ca²⁺]_i oscillations. These results agree well with the notion that both RyR and IP₃R are essential in generating [Ca²⁺]_i oscillations in ICCs, the putative intestinal pacemaker cells. In Fig. 7E and F, HEK293 cells showing [Ca²⁺]_i oscillations were treated with a Ca²⁺-free solution or SK&F96365, a known general channel blocker which blocks Ca²⁺ influx through the transient receptor potential (TRP) family of channels in the plasma membrane (e.g. Zhu et al., 1998). Either treatment completely abolished [Ca²⁺]_i oscillations, suggesting that Ca²⁺ influx from the extracellular space is required in pacemaking (Torihashi et al., 2002). Addition of 0.1 mM Ni²⁺ or Cd²⁺, natural Ca²⁺ blockers, also suppressed [Ca²⁺]_i oscillations (not shown).

Fast Ca²⁺ imaging was performed in order to analyse the local Ca²⁺ mobilization underlying spontaneous [Ca²⁺]_i oscillations. Confocal Ca²⁺ images were acquired every 33 milliseconds from HEK293 cells (1) transfected with RyR3 and (2) highly sensitive to caffeine. As shown in Fig. 7, the HEK293 cells under the two conditions are considered to be a good model for [Ca²⁺]_i oscillations in ICC. The cell shown in Fig. 8 was chosen because this cell showed spontaneous [Ca²⁺]_i oscillations (Ca²⁺ waves) with relatively low frequency, and consequently it was easy to observe elementary Ca²⁺ events during the interval between Ca²⁺ waves. The images in Fig. 8Aa show the initiation and propagation of spontaneous Ca²⁺ waves in the cell. The images were sequentially obtained at the time indicated by dashed vertical lines in Fig. 8Ab. The Ca²⁺ wave was initiated at the lower part of the cell indicated by x and propagated in the up-direction. Average fluorescence intensity was measured in the small circles (2 μm in diameter) indicated by x, y and z in Fig. 8Aa. Fig. 8Ab shows the high-time resolution changes in the fluorescent intensity (F_t/F₀) upon initiation of a Ca²⁺ wave. The [Ca²⁺]_i rise (ΔF/F₀ = 0.4) in the region x (red line in Fig. 8Ab) clearly preceded the local rises in [Ca²⁺]_i in the regions y and z.

Fig. 8Ba shows time courses of F_t/F₀ in the three circles, x, y and z, during the interval of Ca²⁺ waves. The Ca²⁺ image in Bb was acquired at the time indicated by a line in Ba. Transient small [Ca²⁺]_i rises were recorded only in the region x. It should be noted that abrupt [Ca²⁺]_i rises are observed upon generation of a Ca²⁺ wave in region x. It is deduced that propagating Ca²⁺ waves observed in this HEK293 cell transfected with RyR3 were triggered by subcellular Ca²⁺ release events, such as Ca²⁺ sparks, in an initiating site (x), as has been reported in cardiac myocytes and cardiac trabeculae (Cheng et al., 1996; Wier et al., 1997). Essentially the same results were obtained in all HEK293 cells analysed for a relationship between local Ca²⁺ events and global [Ca²⁺]_i oscillations (n=5). The elementary Ca²⁺ events reached the peak within 20 milliseconds (time to peak: 17.3±5.3 milliseconds, n=12). The duration at 50% of the maximal F/F₀ was 38.3±7.6 milliseconds (n=11). These characteristics are consistent with previous reconstitution studies dealing with subcellular Ca²⁺ release mechanisms (Ward et al., 2000; Rossi et al., 2002), suggesting that the subcellular Ca²⁺ events seen in the present study are to be referred to as Ca²⁺ sparks, although IP₃R along with RyR might also have some contribution to the elementary Ca²⁺ events. In this study, 32 sites for generating spontaneous small subcelluar [Ca²⁺]_i rises were observed in 29 cells prepared by 9 separate transfections, and all of these sites tested (n=20) showed RyR3 immunofluorescence reactivity. The size of elementary [Ca²⁺]_i rises at the peak amplitude was similar to that of punctuated staining patterns of RyR3 protein (2.1±0.3 μm vs. 1.9±0.3 μm, n=13). Small [Ca²⁺]_i rises occurred repetitively in the same spatial location of a cell with a mean frequency of 2.9±1.2 events/second (n=18). These results strongly suggest that the sites initiating [Ca²⁺]_i oscillations correspond to regions where RyR3 are expressed in a clustered fashion.

The cultured cell cluster preparation that we have recently developed (Nakayama and Torihashi, 2002) contains smooth muscle, enteric neurones and c-Kit-immunopositive interstitial cells (ICCs). The last cell member is the putative pacemaker of the gastro-intestinal tract (Suzuki, 2000; Thuneberg, 1982; Torihashi et al., 1995; Komuro, 1999). In addition, neurotransmission is well known to modulate the spontaneous rhythmicity and contractility in gastro-intestinal smooth muscle tissues (although nervous activity did not seem to contribute basal [Ca²⁺]_i oscillations in our preparation because tetrodotoxin had no effect). This cell cluster preparation is therefore considered to consist of the essential minimum cell members necessary to investigate mechanisms underlying gastro-intestinal motility and pacemaker function.

In the present study, we demonstrate that reasonably low concentrations (1-10 μM) of ryanodine suppressed and eventually terminated [Ca²⁺]_i oscillations in ICCs. This is probably because the small size of the cell cluster preparation enables a sufficient amount of ryanodine molecules to enter ICCs relatively rapidly. Further evidence is provided by the effects of other drugs affecting RyRs, intracellular Ca²⁺ release channels. As shown in Fig. 2C,D and Fig. 3, application of these drugs (i.e. tetracaine, FK506) completely abolished [Ca²⁺]_i oscillations at relatively low concentrations (∼10 μM). By contrast, IP₃R blockers also eliminated [Ca²⁺]_i oscillations in ICCs (Fig. 3), as described below. Taken together, the results are the first pharmacological evidence for involvement of both RyR3 and IP₃R in terms of pacemaker [Ca²⁺]_i oscillations in ICCs. The fact that many drugs, including ryanodine, are effective suggests that use of this preparation is advantageous in pharmacological investigations of [Ca²⁺]_i oscillations. [In previous reports, greater concentrations (>50 μM) of ryanodine for longer exposure times compared with the present experiments failed to terminate pacemaker potentials in the gastro-intestinal smooth muscle tissues (e.g. Malysz et al., 2001).]

Periodic Ca²⁺ release from the intracellular Ca²⁺ store is considered to be the fundamental event in ICC pacemaking, because Ca²⁺-dependent ion permeability, i.e. Ca²⁺-activated Cl^– channels, has been suggested to underlie electrical oscillation (Tokutomi et al., 1995; Edwards et al., 1999; Huizinga et al., 2002). The present experiments indicate that both RyR and IP₃R occur in ICCs. It has been proposed that some smooth muscles have distinct types of ER (endoplasmic reticulum) in terms of the distribution of RyR and IP₃R (Iino, 1991; Yamazawa et al., 1992), but that other smooth muscles do not. Since application of either drugs related to RyR or IP₃R terminated [Ca²⁺]_i oscillations, both receptors appear to exist in the ER responsible for the periodic Ca²⁺ release in ICCs.

2APB and xestospongin C are known to affect other mechanisms for [Ca²⁺]_i regulation (Bootman et al., 2002). For example, 100 μM xestospongin C largely inhibits Ca²⁺ influx through endoplasmic reticulum Ca²⁺ pumps (SERCA). At the concentration (10 μM) used in the present study, this drug, however, has little effect on SERCA (De Smet et al., 1999). As described above, lines of evidence have been shown for the involvement of IP₃R in spontaneous rhythmicity in ICCs (e.g. Suzuki et al., 2000). The present results with 2APB and xestospongin C appear to support the notion for the importance of IP₃R. However, we have previously shown that Ca²⁺ influx from the extracellular space (presumably via TRP homologue channels) is also required to maintain pacemaker activity in ICCs (Nakayama and Torihashi, 2002; Torihashi et al., 2002). In the light of the essential similarity between 2APB and xestospongin C (although the evidence for inhibition of store-operated Ca²⁺ entry has not yet been reported for xestospongin C) (Bootman et al., 2002), we can also speculate that IP₃R might additionally contribute to [Ca²⁺]_i oscillations in ICCs via store-operated types of Ca²⁺ entry processes.

RT-PCR examinations in the present study revealed that ICCs in murine small intestine predominantly express RyR3, while the major RyR isoform in smooth muscle is RyR2. RyR3 is known as a brain type isoform, because of its characteristic, but faint, expression in several brain regions, i.e. the corpus striatum, hippocampus and thalamus (Ogawa et al., 2002). RyR3 is also expressed in skeletal and smooth muscles, albeit at low levels. RyR3 in skeletal muscle may amplify Ca²⁺ release originally elicited by activation of RyR1 (Ogawa et al., 2002; Rossi and Sorrentino, 2002; Yang et al., 2001). Although the presence of the RyR3 isoform is recognised in numerous tissues and organs, predominant expression over RyR1 or RyR2 has been shown only in non-pregnant uterus, where RyR3 may contribute Ca²⁺ signalling only when SR is overloaded with Ca²⁺ (Mironneau et al., 2002). In arterial vascular smooth muscle cells, a contribution by RyR3, in addition to RyR1 and RyR2, to Ca²⁺ spark generation has been suggested based on results from RyR3-deficient mice (Lohn et al., 2001). A similar contribution of RyR3 to cellular Ca²⁺ signalling has been suggested in Ca²⁺-overloaded cultured rat portal vein smooth muscle cells (Mironneau et al., 2001). It has been found, however, that the RyR3 splice variant predominantly expressed in smooth muscle is not a functional Ca²⁺ release channel, but it could function in a dominant negative fashion (Jiang et al., 2003). There is no report, so far, that RyR3 subtype contributes to generating global [Ca²⁺]_i oscillation.

The predominant expression of RyR3 in ICCs contributed to pacemaker function, and could be a new molecular marker for pacemaker cells in tissues and organs which possess spontaneous rhythmicity in peripheral systems. Indeed, like the double staining of murine small intestine with anti-ryanodine and anti-c-Kit antibodies in the present study (Fig. 4), significant expression of RyR (stained with 34C) in c-Kit-immunopositive cells has been observed in several other gastro-intestinal tissues (e.g. guinea-pig stomach and small intestine; J.W., unpublished).

Pacemaker-like global [Ca²⁺]_i oscillations were observed in approximately 10% of HEK293 cells transfected with RyR3 (Fig. 6). Judging from high responsiveness to 1 mM caffeine, the efficacy of transient RyR3 expression may be ∼30%. It is first shown in this study that heterologous expression of RyR3 induced periodical Ca²⁺ oscillations. The [Ca²⁺]_i oscillations seen in the HEK293 cells and ICCs share the following features. (1) Applications of drugs affecting either RyR (ryanodine, caffeine, tetracaine, procaine, FK506) or IP₃R (2APB, xestospongin C) modulate and/or terminate [Ca²⁺]_i oscillations. In both cases, the contribution of voltage-dependent Ca²⁺ channels to [Ca²⁺]_i oscillation can be ruled out. (2) Although Ca²⁺ release from the internal stores is the fundamental event in spontaneous [Ca²⁺]_i oscillations, Ca²⁺ influx from the extracellular space is also essential. Namely, [Ca²⁺]_i oscillations are suppressed upon Ca²⁺ removal or application of SK&F96365, a TRP homologue channel blocker. Furthermore, applications of inorganic Ca²⁺ blockers, which would totally suppress Ca²⁺ influx, terminate the spontaneous electrical activity corresponding to [Ca²⁺]_i oscillation [A.Y., unpublished for HEK293 cells; for ICCs, see Torihashi et al. (Torihashi et al., 2002) and Nakayama and Torihashi (Nakayama and Torihashi, 2002)].

Ca²⁺ spark is the landmark event for RyRs (Cheng et al., 1993; Bootman et al., 1995; Imaizumi et al., 1998). From their characteristics in the duration, amplitude and width, the elementary Ca²⁺ events observed in the present study are to be referred to Ca²⁺ sparks, although co-contribution of IP₃R in the frequent `spark' sites, cannot be ruled out. Ca<sup>2+</sup> sparks have been identified in cardiac, skeletal and smooth muscle cells (Bootman et al., 1995), where RyR2, RyR1 and RyR2/3 are predominantly expressed, respectively (Ogawa et al., 2002). In oligodendrocyte progenitors, Ca<sup>2+</sup> sparks and puffs have been recorded and cross-talk between RyR3 and IP<sub>3</sub>R2 is speculated (Haak et al., 2001). It has been suggested that IP<sub>3</sub>R2 is essential to induced Ca<sup>2+</sup> puffs and waves, while RyR3 generates the Ca<sup>2+</sup> sparks. It is noteworthy that `Ca²⁺ spark' reconstituted in HEK293 cells by expression of RyR3, triggers Ca²⁺ waves repeatedly (i.e.=[Ca²⁺]_i oscillations) in association with IP₃R. As mentioned above, the [Ca²⁺]_i oscillations seen in ICCs and reconstituted HEK293 share many characteristic features. Taken together, the results of the present study imply a possibility for reconstruction of the intestinal pacemaker: spontaneous rhythmic [Ca²⁺]_i oscillations presumably generated by the co-operative activation of RyR3 and IP₃R.

In conclusion, ryanodine receptors are involved in the generation, or at least modulation, of pacemaker [Ca²⁺]_i oscillations in ICCs. The present study provides the first evidence for an obligatory role of RyR3, and is further supported by the global [Ca²⁺]_i oscillations reconstituted with RyR3 in HEK293 cells.

This work was supported by grants-in-aid for scientific research from the Japan Society for the Promotion of Science, research grants from the ministry of Health and Welfare (to S.N. and Y.I.) and by research grants from the Canadian Institutes of Health Research (to S.R.W.C.). The authors are grateful to Alison F. Brading and Jeff Bolstad for critical reading of the manuscript.