HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplementary material All Versions of this Article: 548/1/271    most recent 2002.033720v1 Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Helden, D. F. v. Articles by Imtiaz, M. S. Search for Related Content PubMed PubMed Citation Articles by Helden, D. F. v. Articles by Imtiaz, M. S.

J Physiol (2003), 548.1, pp. 271-296
© Copyright 2003 The Physiological Society
DOI: 10.1113/jphysiol.2002.033720

Ca²⁺ phase waves: a basis for cellular pacemaking and long-range synchronicity in the guinea-pig gastric pylorus

Dirk F. van Helden and Mohammad S. Imtiaz

	ABSTRACT

Top Abstract Introduction Methods Results Discussion References

Ca²⁺ imaging and multiple microelectrode recording procedures were used to investigate a slow wave-like electrical rhythmicity in single bundle strips from the circular muscle layer of the guinea-pig gastric pylorus. The 'slow waves' (SWs) consisted of a pacemaker and regenerative component, with both potentials composed of more elementary events variously termed spontaneous transient depolarizations (STDs) or unitary potentials. STDs and SW pacemaker and regenerative potentials exhibited associated local and distributed Ca²⁺ transients, respectively. Ca²⁺ transients were often larger in cellular regions that exhibited higher basal Ca²⁺ indicator-associated fluorescence, typical of regions likely to contain intramuscular interstitial cells of Cajal (ICC_IM). The emergence of rhythmicity arose through entrainment of STDs resulting in pacemaker Ca²⁺ transients and potentials, events that exhibited considerable spatial synchronicity. Application of ACh to strips exhibiting weak rhythmicity caused marked enhancement of SW synchronicity. SWs and underlying Ca²⁺ increases exhibited very high 'apparent conduction velocities' ('CVs') orders of magnitude greater than for sequentially conducting Ca²⁺ waves. Central interruption of either intercellular connectivity or inositol 1,4,5-trisphosphate receptor (IP₃R)-mediated store Ca²⁺ release in strips caused SWs at the two ends to run independently of each other, consistent with a coupled oscillator-based mechanism. Central inhibition of stores required much wider regions of blockade than inhibition of connectivity indicating that stores were voltage-coupled. Simulations, made using a conventional store array model but now including depolarization coupled to IP₃R-mediated Ca²⁺ release, predicted the experimental findings. The linkage between membrane voltage and Ca²⁺ release provides a means for stores to interact as strongly coupled oscillators, resulting in the emergence of Ca²⁺ phase waves and associated pacemaker potentials. This distributed pacemaker triggers regenerative Ca²⁺ release and resultant SWs.

(Received 4 October 2002; accepted after revision 14 January 2003; first published online 7 February 2003)
Corresponding author D. F. van Helden: The Neuroscience Group, School of Biomedical Sciences, Faculty of Medicine and Health Sciences, University of Newcastle, NSW 2308, Australia. Email: dirk.vanhelden{at}newcastle.edu.au

	INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

Many cell systems including those in the brain, heart and smooth muscle exhibit rhythmical, electrical oscillations with high levels of synchronicity over large spatial separations. The heart has provided one class of pacemaker model for such rhythmical action. It is based primarily on cyclical activation of voltage-dependent channels in the cell membrane (Noble, 1984; DiFrancesco, 1991) which is reinforced by a Na⁺-Ca²⁺ exchange current induced by Ca²⁺ release from intracellular stores (Brown et al. 1984; Rigg & Terrar, 1996; Ju & Allen, 1998; Huser et al. 2000). In contrast, investigations of various smooth muscle rhythmicities have provided a very different pacemaker model. Here, pacemaking is driven by cyclical Ca²⁺ release from intracellular stores, the subsequent oscillation in cytoplasmic [Ca²⁺] generating an inward current across the plasma membrane and resultant electrical rhythmicity (Komori et al. 1993; van Helden, 1993; Liu et al. 1995; Hashitani et al. 1996; Edwards et al. 1999; Hill et al. 1999; Suzuki & Hirst, 1999; Suzuki et al. 2000; van Helden et al. 2000; Ward et al. 2000; Peng et al. 2001). The surprising aspect about these disparate mechanisms is that they produce very similar outcomes.

The ability to pace both single cells and large multicellular tissues is a formidable task for both pacemaker mechanisms. In the case of Ca²⁺ store-controlled pacemaking, the mechanism needs to overcome an enormous impedance mismatch, as the current generated by a single Ca²⁺ store is microscopic compared to that required to drive electrical pacemaking in a whole tissue. Store Ca²⁺ release events, variably referred to as Ca²⁺ puffs for IP₃Rs (Parker & Yao, 1991) or Ca²⁺ sparks for ryanodine receptors (RyRs) (Cheng et al. 1993) are very small. As a result these Ca²⁺ release events induce small outward (e.g. spontaneous transient outward currents, STOCs; Benham & Bolton, 1986) or inward currents (e.g. spontaneous transient inward currents, STICs; van Helden, 1991; Wang et al. 1992). Thus, if STICs were to act as a pacemaker current then individual STICs would be unlikely to drive single cells and certainly could not drive multicellular tissues to threshold for activation of voltage-dependent Ca²⁺ entry and associated contraction. This problem is overcome by the co-ordinated activation of larger groups of stores (van Helden, 1993; Hashitani et al. 1996; Edwards et al. 1999; Hashitani & Edwards, 1999; Hill et al. 1999; Sergeant et al. 2000; van Helden et al. 2000; Peng et al. 2001; Hirst et al. 2002; Yamazawa & Iino, 2002). However, exactly how this happens remains unclear, especially for tissues with rhythmicities such as slow waves that are likely to be generated independent of voltage-dependent channels (Liu et al. 1995; Hashitani et al. 1996; Suzuki & Hirst, 1999; van Helden et al. 2000; Ward et al. 2000; Hirst & Edwards, 2001; Yamazawa & Iino, 2002).

Ca²⁺ waves have been widely reported to occur in both single cells and multicellular tissues and provide a primary mechanism by which groups of Ca²⁺ stores can be activated (Gilkey et al. 1978; Rooney et al. 1990; Sanderson et al. 1990; Lechleiter et al. 1991; Parker & Yao, 1991; Berridge, 1993; Bootman et al. 1997). The limitation with this mechanism is that Ca²⁺ waves propagate slowly (typical range 0.005-0.1 mm s^-1) and based on the known properties of Ca²⁺ stores could only activate a relatively small population of stores at any one time. While Ca²⁺ waves could certainly drive pacemaking in single cells, it is highly unlikely that they could near synchronously activate sufficient cells to drive electrical slow waves in multicellular tissues. Furthermore, even if they somehow did locally pace slow waves, the rate of Ca²⁺ wave propagation could not explain the high degree of spatial synchronicity of slow waves, which exhibit apparent conduction velocities some three orders of magnitude faster than that for Ca²⁺ waves.

One way, by which the perceived limitations of this pacemaker mechanism might be overcome, would be if Ca²⁺ stores interacted as strongly linked coupled oscillators. In this way oscillatory Ca²⁺ release from intracellular stores could underlie both generation and 'propagation' of chemical and associated electrical rhythmicities. Coupled oscillator-based interactions were first described for an array of pendulums that, when linked (e.g. by springs) and randomly activated, entrain their activity over time. The result is a phase wave in which each oscillator has the same frequency of oscillation but with a spatial variation in phase that depends on the strength of the coupling between the oscillators. All linked oscillators can undergo such interactions, in some cases on a grand scale through long-range parallel interactions (e.g. firefly displays; Strogatz & Stewart, 1993). It is these principles that led to coupled oscillator-based models for pacemaking in the heart (van der Pol & van der Mark, 1926) and to modelling of gastrointestinal slow waves (Nelsen & Becker, 1968; Daniel et al. 1994). While, these models were mathematically based with no understanding of the underlying biological oscillators, they provided great interest and various research groups have continued to argue strongly for such mechanisms (e.g. see Daniel et al. 1994).

Ca²⁺ stores can interact as coupled oscillators, with oscillatory Ca²⁺ release from stores advancing or retarding the cycling of other stores by Ca²⁺-induced Ca²⁺ release (CICR) until entrainment is achieved (Jafri & Keizer, 1994; Roth et al. 1995; Hofer, 1999; Dupont et al. 2000; Rottingen & Iversen, 2000). However, coupling of Ca²⁺ stores by diffusion of activators is relatively weak and generates Ca²⁺ waves with low 'apparent conduction velocities' ('CVs'; e.g. < 0.1 mm s^-1 Roth et al. 1995). Such 'CVs' are orders of magnitude lower than those reported for slow waves (see Huizinga et al. 1992). This large discrepancy could be reconciled if Ca²⁺ stores proved to be much more strongly coupled in tissues exhibiting slow waves. Importantly, there are two factors that together could substantially enhance coupling and hence synchronization between Ca²⁺ stores. These are: (a) store Ca²⁺ release inducing depolarization of the plasma membrane and (b) depolarization enhancing further store Ca²⁺ release, which in the tissue of our study is mediated through IP₃Rs (Edwards et al. 1999; Suzuki & Hirst, 1999; van Helden et al. 2000; Hirst et al. 2002). Increased coupling between Ca²⁺ stores would occur because membrane voltage and the underlying current have a much greater spatial influence than chemical diffusion of store activators. Voltage coupling to store Ca²⁺ release could enable the coupled oscillator-based mechanism to produce powerful Ca²⁺ phase waves. Here, each wave reflects phase differences between Ca²⁺ release oscillations in arrays of entrained Ca²⁺ stores. Significantly, our experimental data provide evidence for the existence of such Ca²⁺ phase waves. This distributed, Ca²⁺ store-based pacemaker mechanism underlies generation of pacemaker potentials and resultant regenerative SW rhythmicities in the tissue of our study.

	METHODS

Top Abstract Introduction Methods Results Discussion References

In vitro experiments were conducted to examine the origin of SW activity recorded in single bundle strips (length 3-10 mm; width/thickness < 0.12 mm) or multi-bundle sheets of circular smooth muscle from the gastric pylorus using 5-20 day guinea-pigs. All procedures conformed with the University of Newcastle Animal Care and Ethics and Occupational Health and Safety Committees. Animals were killed by overexposure to the inhalation anaesthetic halothane (5-10 % in air) followed by exsanguination. Tissues were superfused at 1-5 ml min^-1 in a chamber of volume 0.5-1 ml with gas bubbled (95% O₂-5% CO₂), physiological saline of composition (mM): NaCl 120, KCl 5, CaCl₂ 2.5, MgCl₂ 2, NaHCO₃ 25, NaH₂PO₄ 1, glucose 10, maintained at a pH of 7.3 and temperature of 35 °C. The methods were the same as previously reported (van Helden et al. 2000) with the following changes. Except where noted, tissues were studied with L-type Ca²⁺ channels blocked using 1 µM nifedipine, a condition where Ca²⁺ release from stores is likely to be the major initiator of SW-associated increases in [Ca²⁺]_i (Suzuki & Hirst, 1999; van Helden et al. 2000; Ward et al. 2000; Hirst et al. 2002; Yamazawa & Iino, 2002). Nifedipine was used with attention to its sensitivity to light and it was always functionally active as it blocked contraction and L-type Ca²⁺ channel-based action potentials.

Intracellular recordings were made with up to four microelectrodes. 'Resting' membrane potential (V_m) was taken as the most negative potential excluding any brief transients. The characteristics of spontaneous transient depolarizations, pacemaker potentials or SWs were in some cases defined by measuring the peak amplitude, the 20-80 % rise time (t_r), or the duration at half-amplitude (t_0.5). The 'CV' of propagating waves was measured from the time difference between the 50 % rise points of the waveform at three or four sequential sites along the muscle strip. The calculations excluded the time difference between the two sites nearest the wave origin in case the event originated between these sites. The exceptions to this were measurement of the 'CVs' of sub-threshold pacemaker potentials and distributed STDs which exhibited worse signal-to-noise characteristics. Strips were found to have electrical characteristics approximating those expected for a one-dimensional core conductor. The strip was modelled as an electrical cable comprising repeating elements of set length with current flowing internally through an intercellular resistance r_i primarily set by intercellular connectivity (see Hofer, 1999) and leaking out across the cell membrane of each element through a membrane conductance (g_m; i.e. 1/resistance r_m) and capacitance c_m. Three key properties defining the characteristics of the core conductor are the input resistance (R₀), membrane time constant (_m) and the length/space constant (), where R₀ =~ 0.5(r_mr_i)^0.5, _m = r_mc_m and =~ (r_m/r_i)^0.5. These were determined during multi-electrode recordings made with three or four intracellular electrodes by fitting the voltage responses (excluding that at the source electrode) resulting from injection of a constant current through one of the electrodes near an end of the strip. Fitting was made using a one-dimensional cable model allowing for the finite length of the cable (see Jack et al. 1975), with _m and adjusted to best fit the response curves.

Some experiments were made applying a transverse stream of buffer solution, containing 18- glycyrrhetinic acid, caffeine or 2-aminoethoxydiphenyl borate (2-APB) and an inactive marker dye (phenol red; 100-240 µM), while recording membrane potentials with two to four intracellular electrodes along each strip. The transverse stream was visually monitored with a reasonably stable flow of desired width achieved centrally across the tissue by appropriately positioning the inflow tube and the use of baffle plates and/or arrays of outflow pipettes to maintain the laminar flow of the stream.

Ca²⁺ imaging was performed using a confocal microscope (Biorad MRC 600 or 1000 coupled to a Zeiss Axiovert 10 or Nikon TM 200 inverted microscope) using either oil ( 40, NA 1.3) or water ( 60, NA 1.2) immersion objective lenses and the confocal aperture fully open. Relative changes in [Ca²⁺]_i were detected using the fluophore Oregon Green/AM loaded at 1-4 µM for 60 min at 4 °C. Images were captured at rates of up to 5 Hz, with calculations of Ca²⁺ wave velocities based on the imaged tissue length and effective scan time. A least squares best-fit algorithm was used to align simultaneously obtained electrophysiological recordings. Immunohistochemistry was performed to label intramuscular interstitial cells of Cajal (ICC_IM) with a monoclonal Kit antibody (ACK2) and an anti-rat secondary antibody conjugated to Cy5 (Amersham, Little Chalfont, Bucks, UK) diluted 1:100 in PBS using the approach of Burns et al. (1997). Numerical data are presented as means ± S.E.M. with n referring to the number of tissues unless stated otherwise. Differences were evaluated by Student's paired t test, and P values less than 0.05 were taken as statistically significant.

Simulation of Ca²⁺ dynamics

Numerical simulations were made using a conventional store array model used to simulate IP₃R-mediated Ca²⁺ waves but now modified to include voltage-coupled production of IP₃. To do this, we firstly included Ca²⁺ release-generating inward current into each cell. This was simulated by activation of a Ca²⁺-activated cationic conductance (e.g. see Inoue & Isenberg, 1990), but the same outcomes would be achieved if this current were simulated by a Ca²⁺-activated Cl^- current (Wang et al. 1992; Hirst et al. 2002) or through a Na⁺-Ca²⁺ exchange current (see Ju & Allen, 1998). Second, depolarization-induced IP₃R-mediated Ca²⁺ release (Suzuki & Hirst, 1999; van Helden et al. 2000) was included in each cell, by coupling membrane potential to delayed production of IP₃ as considered in detail in a recent study (Imtiaz et al. 2002). However, we note that the production of IP₃ by voltage-dependent Ca²⁺ entry (Harootunian et al. 1991) would produce similar outcomes.

Modelling of multicellular tissue strips was simplified on the basis that intracellular conductivity is likely to be much higher than intercellular conductivity (Hofer, 1999). Therefore, each cell was assumed to be isopotential and was modelled to function with a single store (Fig. 11A). The strip was approximated by a one-dimensional array of these cells, with each cell considered to have a length of 100 µm. Stores were set to have a specific sensitivity to IP₃ according to a Gaussian distribution. Both Ca²⁺ and IP₃ were considered to pass between the cells, as was electrical current. Current flow was based on each cell having a transmembrane leakage and a Ca²⁺-dependent conductance. The model incorporated stimulus-dependent external Ca²⁺ entry pathway(s), as indicated by the rundown of pyloric SWs upon removal of extracellular Ca²⁺ (van Helden et al. 2000). Cells were modelled to be electrically and chemically interconnected by gap junctions. Electrical parameters were set to provide a length constant of 2.8 mm, a value of the same order as found for strips.

The Ca²⁺ dynamics of stores were simulated using a one-pool Ca²⁺ store array model (Dupont & Goldbeter, 1993, 1994) with the additions noted above incorporated as follows. First, a term for the store release-generated Ca²⁺-activated conductance g_Ca was included as given by:

g_Ca = g_maxZ^q/(K_Ca^q + Z^q), (1)

where g_max is the maximum Ca²⁺-activated conductance, Z the [Ca²⁺]_i, K_Ca the activation threshold of the conductance and q the Hill coefficient. The resulting membrane potential change is dependent on current spread in the tissue, with the tissue modelled as a one-dimensional electrical cable by the equation:

_m( V/t) = ² (²(V)/x²) - g_Ca(V - E_Ca) - g_i(V - E_i), (2)

where V is the membrane voltage, x is the distance along the tissue strip, _m is the membrane time constant, is the strip electrical length constant, and E_Ca and E_i are the reversal/null potentials for the Ca²⁺-dependent conductance g_Ca and the leakage conductance g_i, respectively. We consider only the case where L-type Ca²⁺ channels are blocked and the simulations do not include this or other voltage-dependent conductances. A term was added to eqn (6) of Dupont & Goldbeter (1994) to account for voltage-dependent IP₃ synthesis as follows:

(V) = _m(1 - V^r/(K_v^r + V^r)), (3)

where _m is a constant, K_v is the threshold of activation and r is a Hill coefficient for this process (parameters estimated from previously published data; van Helden et al. 2000). The third addition was to modify the term , the saturation level of the IP₃ receptor (eqn (7) of Dupont & Goldbeter, 1994) from a fixed value to one where stores had different sensitivities to IP₃. A Gaussian distribution was applied to the term K₁ (mean 0.5031 µM; S.D. 0.1325; maximum 0.6729; minimum 0.1287) in this same equation to account for the IP₃ sensitivity of each store. Simulations were carried out in Matlab/Simulink (Mathworks Inc.) on an IBM compatible desktop computer. Parameter terms and values used for the simulation were the same as reported (Dupont & Goldbeter, 1993, 1994). Parameter values for our additional terms are _m = 0.18 s, = 2.8 mm, g_max = 4 mS, E_Ca = -20 mV, K_Ca = 1.4 µM, q = 4, K_V = -58 mV, r = 8, _m = 0.8 µM min^-1, g_i = 1.12 mS, E_i = -67.2 mV or as presented in the text. Ca²⁺ and IP₃ diffusions across gap junctions were set to 1.2 µm² min^-1 and 3.6 µm² min^-1, respectively.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Slow waves

Pyloric SWs and underlying Ca²⁺ transients were recorded simultaneously. Ca²⁺ transients occurred near synchronously over the ~200 µm long imaged regions of the multicellular strips (n = 10 strips, Fig. 1). Figure 1A presents SWs (upper record) and associated Ca²⁺ transients, for the entire imaged region (sum) and for marked regions of interest (ROIs) as defined by the grid of Fig. 1C. There was a global synchronicity in onset, a finding also evident in the space-time image of Fig. 1B, which presents the temporal changes in relative intensity for all the ROIs. Based on the scan rate (~3 Hz) and image size, synchronicity to this extent indicates that the activity across the imaged cells has a 'CV' greater than 1 mm s^-1, a 'CV' much faster than for reported Ca²⁺ waves.

		View larger version [in this window] [in a new window]
Figure 1. Spatially synchronous Ca2+ transients associated with strip 'slow waves' A, SWs recorded near the middle of a strip and aligned associated transient increases in relative [Ca2+]i for the entire image (sum) and for the marked regions of interest (ROIs) shown in C. The tissue length viewed was 0.2 mm, occupied ~0.3 of the vertical image and was imaged at a rate of ~5 Hz. Vm was -55 mV. Nifedipine (1 µM) was present throughout. B, the space-time image showing the temporal changes in relative intensity for all the ROIs shown in C. This plot presents the intensity of ROIs within each row (NB: rows a, b and c-h have 2, 4 and 9 ROIs, respectively) plotted as a vertical sequence of values for each time point. Data for each ROI has been baseline corrected by subtracting the mean intensity of the ROI. C, confocal image of the Oregon Green-loaded tissue and ROI grid over which the measurements were made. D and E, difference images (i.e. image at the peak of activity minus image directly before activity) for SW * in A and for the average of 10 sequential slow waves (SWs). The insets show the corresponding SWs (blue) and associated aligned mean relative Ca2+ transients (red) (normalized scales). F, intramuscular interstitial cells labelled with Kit antibody ACK2 in a composite confocal image of a tissue strip taken over a depth of 17 µm. G, composite image (depth 8 µm) of an Oregon Green-loaded tissue taken within 15 min after loading. H, composite image (depth 3 µm) of the same tissue region as G but after fixation and immunohistochemical labelling with the Kit antibody with arrows pointing to the same cells as arrows in G (note fixation shrinkage). Intensity scale in E applies to all images except C. Scale bar in E applies to all images except for vertical axes of G and H, which are expanded 3.

Another notable property of SW-associated Ca²⁺ transients was their variable magnitude across different cellular regions of the strips (Fig. 1A). This is exemplified by comparing the image of the tissue showing basal fluorescence (Fig. 1C) with the difference images showing the increase in fluorescence during a single SW or for an average of 10 SWs (Fig. 1D and E). Regions that exhibited larger Ca²⁺ transients often, though not always, correlated with regions that exhibited more intense basal Ca²⁺ indicator-associated fluorescence. These same regions exhibited higher basal fluorescence. As such, these regions may have contained ICC_IM (see also Burns et al. 1997), which were found associated with more brightly fluorescent cellular regions (Fig. 1G and H; n = 3). More generally, ICC_IM were found to be present at a density of (3.9 ± 0.7) 10^-5 cells µm^-3 (n = 8 strips; Fig. 1F), which was ~7 % that of the cell density of the smooth muscle ((5.3 ± 1.5) 10^-4 cells µm^-3; n = 5 strips).

SW pacemaker potentials

SWs in single bundle pyloric strips consist of two components, a 'pacemaker component' and a triggered 'regenerative component' (van Helden et al. 2000). These components, as for SWs in gastric antral strips, each comprise more elementary STDs (also known as unitary potentials) (Edwards et al. 1999; Suzuki & Hirst, 1999; van Helden et al. 2000). They differ primarily in the basis for the recruitment of these more elementary events; the resultant pacemaker component establishes timing and when the pacemaker component is super-threshold, it triggers the regenerative component. Measurement of Ca²⁺ transients associated with the pacemaker component, as studied for sub-threshold events, proved difficult to measure because the increases were small and we only occasionally resolved such activity (n = 4 strips; Fig. 2). The electrical recording of Fig. 2A demonstrates both sub-threshold pacemaker potentials and full SWs. The two events were readily distinguishable because the sub-threshold pacemaker potentials were smaller in amplitude (e.g. < 15 mV compared with 20-30 mV) and slower in rise time (e.g. t_r > 2 s compared with t_r < 1 s) than full SWs (see also van Helden et al. 2000). Pacemaker potentials were associated with Ca²⁺ transients that occurred more intensely in specific sub-regions of the imaged area of tissue. Two instances of this are shown in Fig. 2A with marked pacemaker potentials PP1 and PP2 exhibiting the largest Ca²⁺ transients in ROIs a3 and b3. These regions also exhibited the largest Ca²⁺ transients during SWs. Smaller Ca²⁺ transients could be discerned over other areas of the imaged tissue. Comparison of the space-time image (Fig. 2B), allows the same comparison but now with the Ca²⁺ changes in each ROI normalized as described for Fig. 1B. This plot suggests that near synchronous Ca²⁺ transients occurred across much of the imaged tissue for both SW sub-threshold pacemaker potentials and SWs.

		View larger version [in this window] [in a new window]
Figure 2. Spatially synchronous Ca2+ transients associated with sub-threshold pacemaker potentials and SWs A, SW pacemaker and regenerative potentials aligned with the associated transient increases in mean relative [Ca2+]i for the entire image (sum) and for the marked ROIs shown in image E. B, the space-time plot (see Fig. 1 legend). C and D, difference images for the marked pacemaker potential (PP2) and SW (SW2). E, Oregon-Green loaded tissue showing the confocal image and ROI grid over which intensity measurements were made. F and G, difference images for the mean of 5 sequential pacemaker and SW events. H, difference image for the mean of 5 SW events recorded before application of nifedipine. Insets show the corresponding pacemaker potentials or SWs (blue) with aligned associated relative increases in [Ca2+]i (red, normalized scales). Nifedipine (1 µM) present throughout except for H. Intensity scale in H applies to all images except E. Vm -56 mV.

Further insight into the spatial characteristics of the Ca²⁺ transients can be gleaned from Fig. 2C, D, F and G, which show respective difference images for the marked pacemaker potential (PP2) and the SW (SW2) of Fig. 2A and for the mean of five pacemaker potentials and five SWs. The mean of the pacemaker difference images had an average intensity of 0.25 ± 0.02 intensity units (Fig. 2F). This was significantly larger (P < 0.05) than the mean intensity of baseline difference images of 0.13 ± 0.03 intensity units (n = 5 randomly selected baseline difference images ~2 s apart). By contrast SWs showed a much larger increase in [Ca²⁺]_i over the tissue with a mean intensity difference of 0.83 ± 0.07 intensity units (n = 5; significance compared to baseline P < 0.01). Shown also is a mean difference image for five SWs, which was derived from data recorded before addition of the L-type Ca²⁺ channel blocker nifedipine, immediately after the onset of rhythmicity (Fig. 2H). [Ca²⁺]_i was now markedly increased more generally over the tissue, with the mean intensity of the associated difference images 2.18 ± 0.13 intensity units (n = 5; significance compared to baseline P < 0.01). The ~5-fold difference in the intensity of the SW-associated Ca²⁺ transients with and without L-type Ca²⁺ channel blockade is consistent with data from studies on other gastrointestinal slow waves (Hirst et al. 2002; Yamazawa & Iino, 2002).

Spontaneous transient depolarizations

Local Ca²⁺ transients that are likely to underlie spontaneous transient inward currents and resultant STDs/unitary potentials have recently been reported in ileal smooth muscle syncytia using a high sensitivity CCD camera recording system (Yamazawa & Iino, 2002). In contrast, the sensitivity of our confocal microscope system was more limiting, making it difficult to image these more elementary events. However, simultaneous imaging and intracellular recording from freshly dissected strips that exhibited STDs but were not yet rhythmically active (see Suzuki & Hirst, 1999; van Helden et al. 2000), did provide evidence for larger STD-associated Ca²⁺ release (n = 3 strips; without nifedipine). Figure 3 presents recordings of membrane potential and relative [Ca²⁺]_i from such a strip, made simultaneously before and during application of the L-type Ca²⁺ channel blocker nifedipine (Fig. 3B and C). Analysis of the recording before nifedipine indicates the strip exhibited STDs and a triggered action potential (*). The Ca²⁺ transients generally aligned with STDs, though there were cases where STDs occurred without measurable Ca²⁺ transients and vice versa. The former probably occurred because the Ca²⁺ release events were either too small to measure or they occurred outside the field of view. By contrast, the latter finding indicates that Ca²⁺ release events do not always activate measurable plasmalemmal current. The Ca²⁺ transients were at times near periodic (e.g. Fig. 3B; mean interval of eight sequential events 5.3 ± 1.2 s). The measurable STD-associated Ca²⁺ transients occurred as near synchronous distributed events across specific sub-regions of the image as indicated by the ROI intensity-time plots and by the difference image (Fig. 3D and E). The mean voltage and Ca²⁺ transients for the average of nine events and the corresponding mean difference image are shown (Fig. 3F and G). The increase in [Ca²⁺]_i, while distributed, occurred along a specific cellular region of the strip, a region that exhibited higher basal Ca²⁺ indicator-associated fluorescence (Fig. 3A), suggesting that these transients may have been associated with cellular regions containing ICC_IM (see Fig. 1). Larger STDs, triggered L-type Ca²⁺ channel-mediated action potentials (record * Fig. 3B; see also van Helden et al. 2000). Action potentials produced a much larger Ca²⁺ transient with a time course that, unlike STDs, no longer paralleled that of the voltage change (Fig. 3H and J). The action potential-associated Ca²⁺ transients occurred in the same general territory as the STD-associated Ca²⁺ transients (Fig. 3H-K). Recordings of STD-associated Ca²⁺ transients when made in the presence of nifedipine to block L-type Ca²⁺ channels were much smaller, occurring as distributed activity within the same band of cells as observed before addition of nifedipine (Fig. 3L-O).

		View larger version [in this window] [in a new window]
Figure 3. Ca2+ transients associated with spontaneous transient depolarizations A, imaged region of Oregon Green-loaded strip with ROI grid superimposed. B and C, recording of strip membrane potential (blue) aligned with the associated recording of relative [Ca2+]i from an active ROI (a2) before and during application of nifedipine (1 µM). D, STD aligned with the sum of the Ca2+ traces for the active regions (on a normalized scale) and the relative Ca2+ traces for the specific ROIs before nifedipine. E, the corresponding difference image. F, the mean of 9 STDs aligned with the associated Ca2+ transients (the latter on a normalized scale); G, the mean of the difference images for these same events. H, I, J and K, data sequence as for D, E, F and G showing activity associated with STD-triggered L-type Ca2+ channel-mediated action potentials, one of which is marked * in B. L, M, N and O, data sequence as for D, E, F and G now recorded during nifedipine (1 µM). Intensities of the ROIs in D, H and L are plotted on a relative scale compared to the largest Ca2+ transient recorded. Scale bars in D apply to corresponding records. Mean records based on 9, 3 and 7 events for F, J and N, respectively. Time and voltage scale bars in F apply to all traces except B and C. Intensity scale in G applies to all images except A. Vm -53 mV.

The spatial extent of STD activity was also examined electrophysiologically with multiple electrode recording. Examples are presented in Fig. 4A from a strip, where recordings were made before the onset of SWs. STDs in the control showed considerable variability ranging in size from within the baseline noise (e.g. < 0.5 mV) to up to ~3 mV. STDs were found to be either focally generated (i.e. 'local events') being much larger at one or other of the recording sites (Fig. 4A1 and A2), but there were cases where STDs appeared as 'distributed events' showing little decrement (Fig. 4A3). These latter events could not have arisen through electrotonic conduction as potential changes caused by steady state currents showed substantial spatial decrement (Fig. 4Ca and b). Detailed analysis of 50 s of this recording revealed 32 STDs of amplitude > 1 mV (i.e. frequency 38 min^-1). Of these, 23 were local events, with 16, 0 and 7 generated nearer sites 1, 2 and 3, respectively. Events recorded near the active sites (sites 1 and 3) differed only in respective inter-event intervals (2.6 ± 0.5 s, n = 16 and 6.5 ± 1.4 s, n = 7), with mean event amplitude, 20/80 % rise time (t_r) and duration at half-peak amplitude (t_0.5) not significantly different. The respective means of these parameters at all sites were 1.6 ± 0.1 mV, 62 ± 5 ms and 148 ± 9 ms (n = 23). All local events showed steep spatial decrement, which measured as the ratio of amplitudes of the most distal recording site to the middle site, a distance of 1.5 mm, provided a ratio of 0.32 ± 0.04 (n = 23). This decrement was most likely to have occurred through electrotonic conduction, as the decrement was at least that observed in response to focal current injection (Fig. 4C), this tissue exhibiting a length constant of 2.0 ± 0.1 mm (n = 15 pulses, 8 through electrode 1, 7 through electrode 3; calculations took into account the finite cable length; see Methods). Analysis of local STDs of amplitude > 1 mV provided a mean amplitude, t_r and t_0.5 of 2.2 ± 0.3 mV, 60 ± 5 ms and 155 ± 6 ms, respectively (n = 5 strips; 8-23 events strip^-1).

		View larger version [in this window] [in a new window]
Figure 4. Local and distributed spontaneous transient depolarizations STD activity recorded from a strip at 3 sites according to the schema (note electrode colour codes), under control conditions (A) and just after application of 50 nM ACh (B) in a strip where recordings were made before the onset of SWs. Simultaneous multiple electrode recording indicated that STDs arose as either 'local events' which showed rapid decrement or 'distributed events' which underwent little decrement between recording sites. C, injection of a -3 nA current (duration 2 s) through electrodes 1 (a) and 3 (b), respectively (NB: recordings through current passing electrodes made in bridge recording mode; data segments taken from the recording of Fig. 5).

A total of nine distributed events of amplitude > 1 mV was also recorded during this same period. These showed an average inter-event interval of 5.9 ± 1.5 s and no significant decrement in amplitude (e.g. ratio of amplitudes of the most distal recording site to middle site: 0.94 ± 0.07). The mean amplitude, t_r and t_0.5 of the nine distributed events were not significantly different at the three recording sites with mean values for the nine events of 2.0 ± 0.2 mV, 128 ± 27 ms and 329 ± 31 ms, respectively. This indicates an approximate doubling in t_r and t_0.5 compared with local STDs. The average time delay between the outer electrodes (separation 3 mm) was 59 ± 12 ms (n = 9 events), an interval ~1/850th of the 50 s recording period. Distributed events may arise through near synchronous activation of STDs at two or more sites. However, this is most unlikely to happen by chance alone, as the probability of two occurring within 60 ms of each other in the 50 s record is ~1 in 26. Yet, of the 32 measurable STDs that occurred in this 50 s time period, nine were found to be distributed events, a probability of ~1 in 10¹². These data indicate that STD activity can be initiated near synchronously at multiple sites at a frequency much greater than is probabilistic. Analysis of distributed events of measurable amplitude averaged across the recording sites for each of four strips using 8-9 events strip^-1 provided a mean amplitude, t_r, t_0.5 and 'CV' of 2.2 ± 0.2 mV, 140 ± 16 ms, 348 ± 47 ms and 34 ± 6 mm s^-1, respectively.

Pacemaker emergence

SW pacemaker activity and resultant SWs, composed of both pacemaker and triggered components, are known to emerge in freshly dissected single bundle strips some 1-3 h after dissection (Suzuki & Hirst, 1999; van Helden et al. 2000). Application of ACh, an agonist that induces synthesis of IP₃, accelerates this process (van Helden et al. 2000). Analysis of clearly measurable STDs (i.e. amplitude > 1 mV) in 50 s record segments, taken before and immediately after application of 50 nM ACh (Fig. 4 and Fig. 5), indicated that ACh caused an initial increase in event size with the largest 10 events increasing from 2.5 ± 0.1 to 4.6 ± 0.5 mV. There was an associated increase in frequency from 38 min^-1 before to 72 min^-1 immediately after application of 50 nM ACh, though this was biased by events becoming larger and hence exceeding the acceptance threshold amplitude of 1 mV. The increase in amplitude was presumably due to a direct effect on Ca²⁺ release and not changes in strip electrical properties, as while ACh could cause substantial increases in strip input resistance (R_in) this was not the case for concentrations as low as 50 nM, which only had a small effect, increasing R_in to 110 ± 5 % of control (n = 4 strips). Analysis of the spatial characteristics of STDs upon application of ACh indicated that there were 44 local and 16 distributed events > 1 mV in the 50 s recording. Local and distributed events occurred at average inter-event intervals of 1.3 ± 0.2 s (n = 44) and 3.0 ± 0.6 s (n = 16), respectively. The mean amplitude, t_r and t_0.5 of the 44 local and 16 distributed events were 2.3 ± 0.2 and 2.0 ± 0.2 mV, 75 ± 5 and 119 ± 13 ms and 218 ± 14 and 320 ± 24 ms, respectively. The mean 'CV' for the distributed events was 44 ± 8 mm s^-1. These data parallel those obtained for local and distributed STDs before application of ACh, differing primarily in that more events of larger amplitude were detectable upon application of ACh.

		View larger version [in this window] [in a new window]
Figure 5. Spatio-temporal investigation of the emergence of rhythmicity A, electrical recordings at 3 sites along a strip according to the schema (note electrode colour codes) made before and during the emergence of rhythmical electrical activity induced by application of ACh (50 nM). Application of this IP3 mobilizing agonist first increased the frequency of STDs. Larger depolarizing activity then emerged. B, marked regions in A shown on an expanded scale (note different vertical scales for records 4-6). C, power spectral density plots showing spectral peaks for 50 s recordings commencing at a, b or c as marked in A (* vertically scaled 10). Nifedipine (1 µM) present throughout. Vm -56 mV (NB: data of Fig. 4 taken from this recording).

The continued application of ACh then caused a transformation in the electrical activity of the strip with the onset of pacemaker potentials, some appearing to exhibit regenerative components (Fig. 5A and B). The approximate synchronicity of the activity at the three electrodes was not the result of electrotonic conduction, as the magnitude and time course of the activity was not predictable by cable theory. For example the pacemaker activity of record Fig. 5B6 showed no decrement between electrodes, yet current injection into the strip demonstrated substantial spatial decrement, the strip exhibiting a length constant of 2.0 mm (Fig. 4C). This indicates that the pacemaker is a distributed activity as a focal pacemaker would have exhibited a decrement of ~45-70 % between outer electrodes (separation 3 mm) depending on the origin of the current.

Power spectral density analysis of the data was also undertaken to examine periodicities associated with emergence of SWs. Figure 5C presents such analysis applied to the data set of Fig. 5A using respective 50 s record segments taken before ACh (a), immediately after application of ACh (b) and after onset of large amplitude pacemaker potentials (c). Peaks that occurred at very low frequencies 0.04 Hz were at the limit of resolution of the analysis and were ignored. The spectra showed a small peak immediately after application of ACh and a large peak during maintained application of ACh after the emergence of SW activity. Power spectral density analysis of a total of seven strips, in which ACh (50-100 nM) was used to elicit SWs, demonstrated spectral peaks of mean amplitudes 4 ± 1, 7 ± 1 and 960 ± 210 mV² Hz^-1 at 0.10 ± 0.01 0.13 ± 0.01 and 0.10 ± 0.01 Hz (n = 7) before, immediately after application (P < 0.05) and during maintained application (P < 0.01) of ACh. These data suggest that the tissue strips exhibited small but measurable basal pacemaker rhythmicity even before the application of ACh, with the peaks becoming significantly larger immediately after application (P < 0.05) and during maintained application (P < 0.01) of ACh.

Another way to investigate the emergence of pacemaker activity is to examine the actions of agents that reversibly inhibit Ca²⁺ stores. Previous investigations indicate that such inhibition sequentially inhibits SWs, underlying pacemaker potentials and then residual STDs, with this sequence reversed upon recovery from blockade (van Helden et al. 2000). Multi-electrode recording showed an important feature of this process, namely that there was a high degree of synchronicity of sub-threshold pacemaker potentials at wide separations, as observed during inhibition or upon reconstitution of store Ca²⁺ release. For example application of cyclopiazonic acid (CPA, 10 µM) revealed that the underlying pacemaker potentials were near synchronous at widely separated recording sites, even when these events were small (n = 2; Fig. 6A and B). Power spectral density analysis indicated that CPA effectively abolished all pacemaker rhythmicity (Fig. 6C). Pacemaker rhythmicity and regenerative SWs re-emerged on return to control. This same behaviour was also observed for application of other inhibitors of IP₃R-mediated Ca²⁺ release, such as reversible inhibition by caffeine (n = 2).

		View larger version [in this window] [in a new window]
Figure 6. Effect of store Ca2+-ATPase inhibition on SWs recorded at two sites A, effect of CPA (10 µM) on SWs recorded in a strip at 2 sites 4.2 mm apart. CPA inhibited SWs at both recording sites to transiently reveal near synchronous underlying pacemaker potentials before complete suppression of all rhythmicity. This sequence reversed during recovery from the CPA. B, marked records in A shown on expanded time scales. C, power spectra showing presence or absence of spectral peaks for 50 s record segments commencing at the alphabetically marked positions in record A (* vertically scaled 10). Nifedipine (1 µM) present throughout. Vm -62 mV.

Agonist-induced increases in IP₃ enhance spatial synchronicity

SWs, once established, exhibited significant variability in strength. For example, recordings with 3-4 electrodes revealed that some strips showed weak rhythmicities with irregular regenerative SWs that either did not propagate or exhibited low mean 'CVs' (e.g. 1.2 ± 0.4 mm s^-1; n = 5 strips). By comparison, others were found to exhibit strong regular regenerative SWs with relatively high mean 'CVs' (e.g. 26 ± 6 mm s^-1; n = 5). Importantly, application of ACh could markedly increase the spatial extent, periodicity, frequency and 'CVs' of SWs in strips where such activity was weak. For example, application of 100 nM ACh to the strip of Fig. 7 caused SWs to change from infrequent arrhythmic events (average frequency 0.6 waves min^-1) to rhythmic activity occurring at a frequency of 6.0 SWs min^-1 (n = 10-28 SWs). The spatial extent of SWs was markedly increased with the probability that SWs were recorded at both electrodes increasing from 50 % in control to 100 % in 100 nM ACh. The corresponding time delays between the onset of SWs at the two sites decreased from 1.9 ± 0.2 s (n = 10 SWs; Fig. 7B1) to 0.26 ± 0.06 s (n = 28; * P < 0.03; Fig. 7B2), without substantial change in the SW shape. Experiments on three other weakly rhythmic strips made with simultaneous three- or four-electrode recordings confirmed the ACh-induced increase in synchronicity, with 100 nM ACh causing the mean SW 'CV' to increase by 430 ± 70 % of control. ACh, applied at 100 nM, did have some effect on the strip electrical properties. For example, finite cable analysis of the voltage responses at 2-3 sites to injection of a constant current indicated that 100 nM ACh increased strip input resistance, electrical time constant and length constant by 147 ± 16, 154 ± 11 and 133 ± 4 % of control (n = 3 strips), respectively. This indicates an increase in membrane resistance (r_m) with little if any change in membrane capacitance (c_m) or internal resistance (r_i). Increases in r_m would have little if any effect on action potential conduction, which is primarily dependent on c_m and r_i (Hodgkin & Huxley, 1952), indicating that SWs do not conduct in an action potential-like manner.

		View larger version [in this window] [in a new window]
Figure 7. Agonist-induced enhancement of synchronicity A, SWs recorded in a strip at 2 sites 4.5 mm apart before and during application of ACh (100 nM). B, marked regions in A presented in expanded form. Nifedipine (1 µM) present throughout. Vm -58 mV.

Spatio-temporal characteristics of SWs and underlying pacemaker potentials

Multiple electrode recordings provided a means to examine the spatio-temporal characteristics of SWs and underlying pacemaker potentials over electrically long distances (n = 5; Fig. 8A). The regenerative component of the SW was triggered by underlying distributed pacemaker activity, evident as relatively slowly rising, fluctuating depolarizations along the strip (Fig. 8B). The site where regenerative SWs were first triggered varied between events. For example measurement based on the half-rise point of the SWs indicated that 78, 8 and 33 of 119 SWs from the strip of Fig. 8 were first generated nearer sites 1, 2 and 3, respectively, with these sites frequently corresponding to the larger pacemaker potentials (Fig. 8B). The variability in the sequence of activation of SWs along the strip was probably a result of local differences in the underlying pacemaker activity. Such site-related variability in pacemaker activity was also likely to be the cause for some SWs being triggered at the two outside electrodes before the central electrode, as was the case for 14 of the 119 SWs of the strip (Fig. 8B2), though SWs more typically activated sequentially (e.g. 105 of 119 SWs; Fig. 8B1 and B3). Another property of the SWs was a large variability in 'CVs', with SWs showing very large differences in this parameter (Fig. 8E). This was likely to be at least in part due to event-to-event variability in the relative synchronicity of underlying pacemaker activity along the strip.

		View larger version [in this window] [in a new window]
Figure 8. Spatio-temporal characteristics of pacemaker potentials and regenerative SWs A, continuous recording of SWs recorded at 3 sites along a strip according to the schema (note electrode colour codes; electrodes are labelled el1, el2 and el3). B, expanded views of the SWs marked in A. C, expanded view of the sub-threshold pacemaker potential marked as PP1 in A. D, the average voltage responses to injection of a -4 nA current pulse through eI3 (n = 3) with superimposed numerical fits for a one dimensional core conductor using a finite cable model with m = 0.20 s and = 2.8 mm. E, histogram demonstrating that SWs exhibited variable 'CVs' (for calculation of this parameter, see Methods) and sites of origination as shown for sequentially recorded events first appearing near site 1 (red) or site 3 (blue), the latter plotted as negative values. Nifedipine (1 µM) present throughout. Vm -61 mV.

Sub-threshold pacemaker potentials were also observed to occur in this strip (Fig. 8C). The onset of these paralleled that of super-threshold pacemaker potentials, differing primarily in that the underlying activity was smaller and didn't trigger regenerative responses. Importantly, sub-threshold pacemaker potentials, measured between the outer electrodes from this same strip, while small and noisy, also exhibited considerable synchronicity. Sub-threshold pacemaker potentials were unlikely to arise through action potential-like conduction, as they were not obviously regenerative. Furthermore, they did not arise through sequential passive conduction, as 11 of 16 of these depolarizations were either similar in amplitude at the three recording sites or relatively larger at the two outside sites. By comparison, passively conducting electrical events would exhibit substantial electrotonic decrement (Fig. 8D). This would be between ~50-80 % depending on the origin of the activity, as the strip exhibited the properties of a one-dimensional core conductor with a length constant () determined using a finite cable model of 2.2 mm. These data do not fit with SWs arising through electrotonic or active sequential conduction but support the hypothesis that regenerative SWs are triggered by an underlying distributed pacemaker activity that establishes SW frequency and synchronicity.

Central interruption of SWs in single bundle strips

The nature of SW propagation was further investigated by inhibiting SW rhythmicities in the centre of strips, while electrically recording residual SW activity at the two ends. An action potential-like event (e.g. a regenerative Ca²⁺ wave markedly accelerated through the positive feedback of membrane depolarization causing IP₃R-mediated Ca²⁺ release; see Suzuki & Hirst, 1999; van Helden et al. 2000) would either not be able to transmit across the blocked region or, if it did jump this region, would remain in phase. By comparison, a coupled oscillator-based mechanism would predict that rhythmicities decouple, persisting at the two tissue ends but losing all phase relationships. We tested this in two ways. The first test was to block intercellular connectivity. We found that global application of 18- glycyrrhetinic acid (40-60 µM) blocked intercellular electrical connectivity and SWs (n = 3 strips). Electrical connectivity was determined by recording the voltage response downstream to current injection. As exemplified in Fig. 9A, application of the 18- glycyrrhetinic acid blocked the voltage response of -2.3 ± 0.3 mV (n = 4 pulses), which became immeasurable (-0.2 ± 0.3 mV; n = 4). This action was not due to shunting caused by a leaky cell membrane, as the power spectral density of the voltage noise between SWs, which itself was some 100-fold greater than the electrode noise, did not decrease but increased in the glycyrrhetinic acid (data not shown). The non-specific effect of 18- glycyrrhetinic acid to cause some depolarization (e.g. from -63 to -54 mV) was not a basis for SW blockade, as depolarization of this magnitude is known to increase SW activity in these pyloric strips (van Helden et al. 2000).

		View larger version [in this window] [in a new window]
Figure 9. Central interruption of intercellular connectivity decouples SWs A, dual electrode recording at ~1 mm separation from a 3 mm strip during global application of the gap junction blocker 18- glycyrrhetinic acid (18- GA; 40 µM). 18- GA blocked both intercellular conductivity (as measured by injection of current -2 nA through eI2) and SWs. B, SWs simultaneously recorded at 2 sites along a strip before, during and after central application of 60 µM 18- GA (see schema). C, expanded record segments marked in B. Nifedipine (1 µM) present throughout. Vm: A -63 mV, B -59 mV.

Transverse application of this agent to a localized central region (width < 1 mm) of strips caused decoupling of rhythmicities at each end (Fig. 9B and C; n = 4 strips). Decoupling commenced ~1.5 min after application of the blocker and was not phase locked, as more SWs eventuated at site 2 compared to site 1. For example, upon commencement of decoupling approximately three SWs occurred at site 1 over the next minute compared to approximately four at site 2, with delays between the SWs (site 2 - site 1) of 0.8, 3.2, 7.9 and 9.5 s for the first five sequential SWs. Thus, while rhythmicities persisted at the two strip ends, they lost all phase relationships, now exhibiting frequencies differing by about 30 %.

A second procedure was to inhibit store release while maintaining electrical connectivity. This was first achieved using caffeine (0.5-5 mM), as it rapidly and reversibly abolished SWs, most likely by inhibiting underlying Ca²⁺ increases (Fig. 10A; n = 5 strips) with minimal effects on the passive electrical properties of the strips (van Helden et al. 2000). In contrast to the experiment of Fig. 9 where connectivity was interrupted, application of caffeine (1-5 mM) to a localized central region only decoupled SWs when applied over relatively wide central regions (e.g. 1.5-5 mm; n = 3 strips) but never with narrow regions (e.g. 1 mm; n = 6 strips). This is demonstrated in Fig. 10B. Application of a 3 mm caffeine-containing stream to this strip caused substantial delays between the SWs, but a one-to-one correspondence between waves at the two ends was maintained. In contrast application of this stream to the same strip at a width of 5 mm fully decoupled the SWs. Decoupling commenced ~1 min after application of the blocker and was not phase locked, with SWs at the two recording sites now occurring at significantly different frequencies (P < 0.05; frequency 3.7 ± 0.1 and 4.4 ± 0.1 min^-1 at electrodes 1 and 2, respectively; n = 10 SWs). SWs fully resynchronized upon cessation of the caffeine-containing stream.

		View larger version [in this window] [in a new window]
Figure 10. Central interruption of store Ca2+ release decouples SWs A, caffeine (0.5 mM), applied to an Oregon Green-loaded strip, blocked SWs (upper trace) and associated Ca2+ transients (lower trace). B, SWs recorded at 2 sites 6 mm apart along a strip before, during and after central application of 1 mM caffeine (see schema) applied at stream widths of 3 and 5 mm. SWs fully resynchronized upon cessation of the caffeine stream. C, SWs recorded at 2 sites 7 mm apart along a strip before and after central application of 50 µM 2-aminoethoxydiphenyl borate (2-APB) applied at stream widths of 1, 3 and 6 mm. ACh (50 nM) present throughout in C to increase SW frequency. Nifedipine (1 µM) present throughout. Vm: A -56 mV, B -67 mV, C -64 mV.

Decoupling was also effected by central application of 2-APB, a cell-permeant inhibitor of specific IP₃ receptors and of store-operated Ca²⁺ channels (Kukkonen et al. 2001) which blocks antral slow waves (Hirst & Edwards, 2001). Application of 2-APB (40-75 µM) to a wide central region (range 2-6 mm; n = 3 strips), but not a narrow region (e.g. 1 mm; n = 4 strips), decoupled the rhythmicities, which persisted at the strip ends but lost all phase relationships. This strong dependence on the width of interruption of store function is exemplified in Fig. 10C where 50 µM 2-APB was streamed across the centre of the strip at widths of 1, 3 and 6 mm. The 1 mm stream had no effect. The 3 mm, stream markedly increased jitter between the delays of the waves at the two strip ends, but did not fully decouple the rhythmicity. In contrast the 6 mm stream decoupled the SWs. Decoupling commenced ~1.5 min after application of the blocker and was not phase locked, with SWs at the two recording sites now occurring at significantly different frequencies (P < 0.01; frequency 3.6 ± 0.1 and 5.9 ± 0.9 min^-1 at sites 1 and 2, respectively; n = 10 SWs). These data indicate that the SWs in these strips do not sequentially propagate but arise through coupled oscillator-based interactions between Ca²⁺ stores.

Store array model with voltage-dependent IP₃R-mediated Ca²⁺ release

Simulations were made using a conventional store array model used to generate IP₃R-mediated Ca²⁺ waves (Dupont & Goldbeter, 1993, 1994) but now including membrane voltage coupled to IP₃R-mediated Ca²⁺ release (Fig. 11A; Methods). A gradual increase of [IP₃]_i caused the emergence of SWs (Fig. 11B). IP₃ first activated the most sensitive stores causing these to undergo oscillatory Ca²⁺ release and associated depolarization (Fig. 11C1). The increased recruitment of stores during further application of IP₃ caused entrainment of stores such that Ca²⁺ release events at spatially disparate locations tended to synchronize (Fig. 11C2), this entrainment continuing until sub-threshold pacemaker activity became evident (Fig. 11C3 and C4). A further increase in [IP₃]_i led to the pacemaker Ca²⁺ release events becoming super-threshold such that regenerative Ca²⁺ release and associated SWs were triggered (Fig. 11C5 and C6). This occurred through both regenerative activation of unspent IP₃R-mediated Ca²⁺ stores by Ca²⁺-induced Ca²⁺ release (CICR_IP3; see Iino, 1990) and by the depolarization recruiting stores through by IP₃-induced Ca²⁺ release (IICR). Importantly, a further increase in [IP₃]_i caused enhanced spatial synchronicity of the SWs and underlying Ca²⁺ transients (Fig. 11C6). These properties parallel our experimental findings.

		View larger version [in this window] [in a new window]
Figure 11. Global stimulation of a store array model simulates the emergence of rhythmicity A, schema showing a single cell coupled to its neighbours by gap junctions. The interplay of Ca2+ (blue), IP3 (red) and membrane potential (green) are indicated. See text for further details. B, simulation of the response to a ramp increase of [IP3] (stimulus increased from 0.0 to 0.51 µM at a rate of 0.0036 µM min-1) in a 10 mm one-dimensional model strip comprised of 100 cells. The records present the average changes in membrane potential, cytosolic [Ca2+] and [IP3] for the 3 colour-coded regions shown in the schema. C, expanded records of [Ca2+] marked in B show the initial store activity (C1), the onset of entrainment of store-related activity (C2), the evolution of sub-threshold pacemaker events (C3 and 4) and then super-threshold pacemaker events triggering regenerative store release (C5 and 6). See Methods for details of the modelling and parameter values. An animation for a similar simulation (but for a two-dimensional tissue) can be seen on:

The voltage-coupled store array model exhibited rhythmicities that responded similarly to various interventions applied in our experiments. For example, inhibition of store Ca²⁺ refill blocked SWs to reveal underlying spatially synchronous SW pacemaker potentials with this activity then also blocked (Fig. 12A). Global blockade of intercellular connectivity led to complete abolition of synchronized rhythmicity leaving only asynchronous store release across the cells of the strip (Fig. 12B). Another parallel was the increase in the frequency of store release events between SWs (Fig. 12C), which is similar to reported experimental findings (Hirst & Edwards, 2001; Yamazawa & Iino, 2002).

		View larger version [in this window] [in a new window]
Figure 12. Some properties of the model-based rhythmicity A, gradual reduction of store refill (dashed line) to 70 % of control (continuous line) inhibited rhythmicity. Rhythmicity re-emerged when store refill was returned to control levels. Higher levels of refill inhibition achieved the same result but there was proportionally larger residual depolarization (not shown). B, gradual reduction of gap junction coupling (dashed line) to no coupling (continuous line) abolished rhythmicity. Global coupling re-emerged upon return to control conditions. C, control record segment from A and histogram, taken at fixed intervals between the arrows, demonstrating that the number of store release events increased between SWs. Records in A and B show the average [Ca2+]i for the 3 colour-coded regions in the schema. Marked record segments in A and B are shown on expanded scales. The stimulus was held constant at = 0.9 µM during these simulations. See Methods for details of the modelling and parameter values.

Simulations where intercellular connectivity was interrupted in the mid region of the model strip caused decoupling of the rhythmicities at the two strip ends (Fig. 13A). Furthermore, simulations where store function but not intercellular connectivity was inhibited predicted the experimental data, as inhibition of store Ca²⁺ release in a mid-region of the model strip caused decoupling of rhythmicities for wide but not narrow regions of inhibition (Fig. 13B and C), confirming the fundamental importance of voltage coupling. The luxury of being able to control specific parameters allowed further demonstration of the importance of voltage coupling to stores. Store-based rhythmicity persisted when there was electrical coupling but no chemical diffusion between cells, but failed to entrain in the 10 mm model strip when there was only chemical diffusion of Ca²⁺ and IP₃ between cells (Fig. 13D).

		View larger version [in this window] [in a new window]
Figure 13. Role of voltage coupling in entrainment of Ca2+ stores A, interruption of connectivity in the middle of the model strip decoupled the rhythmicities on the two sides of the strip (note colour coding). The stronger residual rhythmicity on one side of the strip arises through the relative distribution of store sensitivities, which have been randomly assigned. The simulated activity also exhibited sub-threshold pacemaker potentials (e.g. PP1). B, as for A, but now before and during central inhibition of store Ca2+ release leaving chemical and electrical connectivity intact. Inhibition of 30 cells (i.e. 3 mm) in the centre of the strip resulted in weakened decoupled rhythmicities persisting at the two strip ends at different frequencies, dependent on the distribution of IP3 sensitivities. C, as for B, but for narrower central inhibition (i.e. 1.5 mm). There was now no decoupling irrespective of duration of the central inhibition. D, rhythmical Ca2+ release and associated depolarizations present in a model strip (a) persisted when there was no diffusion of Ca2+ and IP3 between stores (b) but failed to emerge when there was Ca2+ and IP3 diffusion, but no voltage coupling to IP3 synthesis (c). The stimulus was held constant at = 0.72 µM during all simulations. See Methods for details of the modelling and parameter values.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Data presented here support the hypothesis that the pacemaker mechanism underlying the gastric SW rhythmicity of our study is mediated by Ca²⁺ phase waves. Our findings build on data that smooth muscle cells and/or interstitial cells exhibit localized Ca²⁺ release events that generate spontaneous transient depolarizations (also known as unitary potentials) (van Helden, 1991; Janssen & Sims, 1992; Wang et al. 1992; Hashitani et al. 1996; ZhuGe et al. 1998; Edwards et al. 1999; Suzuki & Hirst, 1999; Sergeant et al. 2000; van Helden et al. 2000; Hirst et al. 2002; Yamazawa & Iino, 2002). STDs are the first events recorded in gastric tissues before the onset of global rhythmicities (Edwards et al. 1999; Suzuki & Hirst, 1999; van Helden et al. 2000). Importantly, STDs can exhibit a high degree of spatial synchronicity as first seen during the initial stage of emergence of SWs. Such synchronicity is much greater than expected by their random occurrence or through electrotonic conduction (Figs 3-5). We interpret these data as representing the first stage of the entrainment of store Ca²⁺ release over distributed regions during the onset of global rhythmicity. Stores with a similar sensitivity to IP₃ and/or Ca²⁺ oscillate at the same frequency (Berridge, 1993) and hence such stores will most strongly entrain through coupled oscillator-based interactions. This entrainment became larger during application of ACh to generate near synchronous pacemaker potentials, with larger pacemaker events triggering regenerative responses (Fig. 5).

The emergence of such rhythmicity is mimicked by simulation using a store array model conventionally used for generating CICR-based Ca²⁺ waves. The novel feature of the model is that it accounts for the known interaction of membrane depolarization enhancing store Ca²⁺ release. A description of the response of the store array model to globally increasing [IP₃]_i follows. Firstly, stores with the highest sensitivity to IP₃ become active producing local Ca²⁺ release events and associated depolarizations. A further increase in [IP₃]_i then causes the onset of long range coupled oscillator-based interactions, where groups of stores with similar sensitivities to IP₃ couple across the array of stores. Sub-threshold pacemaker events comprised of Ca²⁺ release events and associated pacemaker potentials then emerge. While, the envelope of such activity and the associated pacemaker potentials arise near synchronously across the model tissue, there is regional variability in the composition of local Ca²⁺ release events. The resultant Ca²⁺ phase wave, when of sufficient synchronicity and magnitude, subserves as a distributed pacemaker to trigger regenerative Ca²⁺ release and associated depolarization. Such regenerative Ca²⁺ release results from recruitment of the large residual pool of unspent stores which now become activated by CICR_IP3 and also by the associated SW depolarization causing an increase in [IP₃]_i. These model-based outcomes, presented in Fig. 11, closely parallel the experimental findings, predicting local STDs, distributed STDs, the emergence of SW pacemaker potentials through entrainment of STDs that when super-threshold trigger the regenerative component resulting in full SWs (see Figs 1-5).

Some strips demonstrated weak SWs that occurred infrequently, did not always propagate and when they did propagate exhibited relatively low synchronicity (i.e. low 'CVs'). However, application of ACh, an agonist known to enhance synthesis of IP₃, caused marked enhancement of 'CVs', an increase, which cannot be explained by mechanisms involving sequential conduction (Fig. 7). Enhancement of synchronicity was also evident in the model-based outcomes with the SW-associated Ca²⁺ transients attaining higher levels of synchronicity with higher [IP₃]_i (Fig. 11C6). Simulations also predicted other characteristics of SWs including interventions where store refill was globally inhibited (cf. Fig. 12A and Fig. 6A). Global inhibition of cell connectivity caused desynchronization of store release. The residual activity represents continuing uncoordinated Ca²⁺ release across the array of stores (Fig. 12B). These findings paralleled those obtained experimentally as rhythmicity was abolished by inhibition of intercellular connectivity (Fig. 9A). The likelihood that entrainment occurred through coupled oscillator-based interactions between stores was supported by experiments where intercellular connectivity was impeded centrally in strips. In this case SWs at the two strip ends persisted but decoupled, now operating at different frequencies (Fig. 9B and C). The same observation was made by simulation (Fig. 13A).

A striking feature of both the experimental and simulated data is the high degree of synchronicity of the Ca²⁺ store-mediated pacemaker mechanism, which based on the SW 'CV' was ~20 mm s^-1 in strongly rhythmic strips. Such 'CVs' are orders of magnitude greater than those reported for Ca²⁺ waves, irrespective of whether the Ca²⁺ waves are generated by sequential conduction (Sanderson et al. 1990) or by coupled oscillator-based interactions between stores linked only by diffusion of Ca²⁺ (Roth et al. 1995). One possible explanation for this is that stores in the tissue of our study are very strongly linked through membrane depolarization enhancing IP₃R-mediated Ca²⁺ release (Edwards et al. 1999; Suzuki & Hirst, 1999; van Helden et al. 2000). This would markedly strengthen interactions between stores because membrane current and voltage have a spatial influence orders of magnitude greater than for ionic diffusion. Ca²⁺ stores might now interact as strongly coupled oscillators producing phase waves of relatively high synchronicity and hence high 'CVs'. Support for this hypothesis was provided from experiments where store function was inhibited in the centre of strips, without impeding intercellular connectivity. It was found that rhythmicity persisted but decoupled at both ends of the strips only when stores in wide central regions of the strips were inhibited; no decoupling occurred when stores were inhibited in narrow regions (e.g. < 1 mm; Fig. 10B and C). Simulation predicted this same result (Fig. 13B and C). This finding cannot readily be explained on the basis that stores were dominantly coupled by diffusion of chemical activators, as the effective diffusion range of Ca²⁺ or IP₃ is very short (range ~0.006 and ~0.025 mm, respectively; Allbritton et al. 1992). In this case, narrow central regions of store inhibition (e.g. < 1 mm) should have decoupled rhythmicities at the two strip ends. This did not occur. In contrast, the results indicate that stores are voltage coupled through depolarization-induced IP₃R-mediated Ca²⁺ release, as membrane voltage has much greater spatial influence, strips exhibiting electrical length constants of 2-3 mm (Fig. 4C and Fig. 8D). These data support the hypothesis that Ca²⁺ stores interact as strongly coupled oscillators to produce Ca²⁺ phase waves. This rhythmical pacemaker Ca²⁺ release and associated depolarization acts as a distributed pacemaker triggering regenerative Ca²⁺ release and resultant SWs across the tissue.

IP₃ phase waves

Simulations with the store array model were based on modelling depolarization-induced enhancement of Ca²⁺ release, as occurring through modulation of synthesis of IP₃. This was a convenient assumption used to model the known relationship between membrane depolarization and enhanced synthesis of IP₃R-mediated Ca²⁺ release (see Methods). While this relationship must be factored into simulations, the assumption that it occurs through synthesis of IP₃ per se does not influence the simulations. However, should this assumption prove true then it heralds an important outcome, namely the existence of IP₃ phase waves, which will occur concurrently with the Ca²⁺ phase waves (Fig. 11 and Fig. 13). Importantly, there are several lines of evidence indicating that membrane potential can be coupled to synthesis of IP₃. First, there is a considerable body of evidence in a range of tissues that membrane voltage modulates production of IP₃ either directly or through voltage-dependent entry of Ca²⁺ (Best & Bolton, 1986; Harootunian et al. 1991; Itoh et al. 1992; Ganitkevich & Isenberg, 1993; Mahaut-Smith et al. 1999). Second, studies on a cultured cell line provide direct evidence that IP₃ and Ca²⁺ oscillations can coexist within cells (Hirose et al. 1999). Therefore, it seems likely that IP₃ phase waves coexist with the Ca²⁺ phase waves.

Voltage-accelerated Ca²⁺ waves

In a previous paper, it was concluded that SWs were either generated by Ca²⁺ stores interacting as strongly coupled oscillators or through voltage-accelerated Ca²⁺ waves (van Helden et al. 2000). While the data of the present study indicate that the pacemaker arises through the former mechanism (i.e. Ca²⁺ phase waves), there remains a high likelihood that voltage-accelerated Ca²⁺ waves also play a role. We predict that this occurs at a more local level especially in tissues where pacemaker activity is weak, such that the coupled oscillator-based pacemaker Ca²⁺ release is super-threshold at some but not all sites. In such circumstances, sequentially conducting regenerative Ca²⁺ release and resultant SWs are likely to occur. Such propagation will not only occur through diffusion of Ca²⁺ and sequential triggering of resultant regenerative Ca²⁺ release, but is likely to be markedly accelerated through the positive feedback of membrane depolarization causing IP₃R-mediated Ca²⁺ release (see van Helden et al. 2000). These voltage-accelerated Ca²⁺ waves should conduct rapidly, the speed of conduction following the same principles of a conventional action potential (Hodgkin & Huxley, 1952) with the caveat that there will be slowing due to the known delay between membrane depolarization and IP₃R-mediated Ca²⁺ release (Suzuki & Hirst, 1999; van Helden et al. 2000; Hirst et al. 2002). This phenomenon may underlie the observation that some SWs appeared to arise without evidence of an obvious preceding pacemaker event (e.g. Fig. 8B1, middle trace).

The relative interplay between the two mechanisms is likely to be important. For example, a weak Ca²⁺ phase wave that is super-threshold at only one site will trigger a regenerative event that thereafter can only spread sequentially. A key factor, which will determine the speed of conduction and possibly the extent of spread of this Ca²⁺ wave, will be the magnitude, spatial extent and relative synchronicity of the weak Ca²⁺ phase wave. This will more globally pre-depolarize the tissue and in so doing alter the amount of current required to activate regenerative Ca²⁺ release ahead of the voltage-accelerated Ca²⁺ wave. Indeed, event-to-event variability in the magnitude/synchronicity of the Ca²⁺ phase wave could explain the observation that regenerative SWs in weakly rhythmic tissues did not always propagate (Fig. 7). The relative roles of the two mechanisms would be expected to shift when store activity and associated coupling was high, as the Ca²⁺ phase waves would now be largely super-threshold over the entire strip, with regenerative Ca²⁺ release in each region now driven by pacemaker Ca²⁺ increases in these same regions. Spatial synchronicity (i.e. 'CV') would now be established by the strength of the Ca²⁺ phase waves as determined by the level of Ca²⁺ store activity and the strength of coupling between stores, with higher agonist levels enhancing coupling and increasing 'CVs' (e.g. Fig. 11C6). Interestingly, very high levels of stimulation cause disruption of rhythmicity (data not shown).

Intramuscular interstitial cells of Cajal

There is considerable data indicating that the extramuscular network of ICC has a key pacemaker role in generation of slow waves (Thunenberg, 1982; Ward et al. 1994; Huizinga et al. 1995; Dickens et al. 1999). However, single bundle strips such as used in this study are devoid of such ICC exhibiting only ICC_IM that are interconnected to other ICC_IM and smooth muscle by gap junctions (Daniel et al. 1998). ICC_IM have not been implicated in slow wave pacemaking (Burns et al. 1997), though recent evidence from the circular muscle layer in intact mouse gastric antrum preparations indicate ICC_IM subserve a key role in the regenerative component of the slow wave (Dickens et al. 2001). Our imaging studies are consistent with this finding, as regions that showed larger SW-associated Ca²⁺ transients were indicated to be associated with cellular regions containing ICC_IM. However, whether the enhanced Ca²⁺ release occurs in ICC_IM and/or in adjacent ICC_IM-influenced smooth muscle or both has yet to be established.

The imaging data of this paper also suggests an association of ICC_IM-related regions with pacemaking. With our model, pacemaker regions are regions containing more sensitive IP₃R-operated Ca²⁺ stores and it is these regions that first show synchronized Ca²⁺ release. The imaging in Fig. 3 indicates that more brightly fluorescent Ca²⁺ indicator labelled regions were the first to exhibit such activity before the onset of SW activity. Examination of the Ca²⁺ transients associated with sub-threshold pacemaker potentials also indicates a dominant role of such cellular regions, as pacemaker potential-associated increases in [Ca²⁺]_i were observed in more brightly fluorescent regions (Fig. 2). The spatial increases in [Ca²⁺]_i associated with regenerative SWs paralleled those for the pacemaker potential but were now proportionally greater across the tissue (Fig. 2). These data indicate that more intensely fluorescent cellular regions and hence putative ICC_IM-containing regions have a dominant role in both pacemaker and regenerative events.

Intact tissues

The concept that Ca²⁺ stores interact as coupled oscillators to generate phase waves provides a key missing element to longstanding coupled oscillator-based models of gastrointestinal slow waves (see Daniel et al. 1994). These models have raised stiff debate in part because the underlying oscillator was not delineated (see Publicover & Sanders, 1989), with an alternative cardiac-like pacemaker model proposed (Publicover, 1995). The relevant aspect of this latter model is that the electrical characteristics of the cellular syncytium are indeed fundamental to pacemaking, at least in the tissue of our study. Thus, while a pacemaker model based on cyclic activation of voltage-dependent channels in the cell plasmalemma, does not apply to the smooth muscle of our study, electrical connectivity is a key factor.

The store array model predicts key experimental observations made in isolated strips and/or intact tissues. First, evidence has been presented that the regenerative component of SWs is composed of summations of more elementary events (Edwards et al. 1999). This is predicted by simulation, these being unspent IP₃R-operated Ca²⁺ stores that are regeneratively activated consequent to pacemaker-associated increases in [IP₃]_i and [Ca²⁺]_i. Second, it has been shown that the frequency of local Ca²⁺ release events or STDs proportionally increases between slow waves from very low levels just after each slow wave (Hirst & Edwards, 2001; Yamazawa & Iino, 2002). Analysis of the relative activity of local Ca²⁺ release events and associated STDs between simulated slow waves demonstrates a similar increase in the frequency of these events (Fig. 12C). Third, it has been shown that slow waves in intact gastric antrum are 'stochastic events arising at intervals which show much larger variation than would be expected for a pacemaker model when pacemaker activity involves sequential activation of voltage-dependent ion channels' (Hirst & Edwards, 2001). This is also a characteristic