Effects of temperature on intracellular [Ca²⁺] in trout atrial myocytes

Cardiac contractility is profoundly influenced by temperature in all vertebrates. This is not surprising as the transient rise in intracellular Ca²⁺ concentration ([Ca²⁺]_i), which underlies excitation—contraction coupling, represents the culmination of many temperature-sensitive cellular processes; i.e. action potential configuration(Møller-Nielsen and Gesser,1992; Puglisi et al.,1999; Coyne et al.,2000), myofilament Ca²⁺ sensitivity(Harrison and Bers, 1990; Churcott et al., 1994; Gillis et al., 2000) and Q₁₀ effects on protein pumps and ion channels(Sitsapesan et al., 1991; Herve et al., 1992; Xue et al., 1999; Kim et al., 2000; Shiels et al., 2000). In most vertebrates, the transient change in [Ca²⁺]_i in response to stimulation is the product of both extracellular Ca²⁺ influx across the sarcolemmal membrane (SL) and intracellular Ca²⁺ release from the sarcoplasmic reticulum (SR), although the relative Ca²⁺contribution from each source may vary(Fabiato and Fabiato, 1978; Driedzic and Gesser, 1985;Bers, 1985, 1989; Tanaka and Shigenobu, 1989; Hove-Madsen, 1992; Bassani et al., 1994).

In mammals, the temperature sensitivity of extracellular and intracellular Ca²⁺ fluxes are well characterized. Transsarcolemmal Ca²⁺ influx via the L-type Ca²⁺ channel current(I_Ca) is strongly influenced by temperature, with a Q₁₀ranging from 2 to 3.5 (Cavalie et al.,1985; Herve et al.,1992). Cold temperatures reduce SR function in many mammalian hearts, impairing both release (Bers,1987; Sitsapesan et al.,1991) and uptake (Liu et al., 1991, 1997; Wang et al., 2000) of SR Ca²⁺ stores. Under most conditions, excitation—contraction coupling in rainbow trout heart is thought to be driven primarily by extracellular Ca²⁺ influx via I_Ca, with SR Ca²⁺ release playing a secondary role(Tibbits et al., 1992; Hove-Madsen, 1992; Keen et al., 1994; Shiels and Farrell, 1997). Recent studies of I_Ca in trout myocytes indicate that the temperature sensitivity of peak current amplitude is not very different from that of mammals (Q₁₀ of approximately 1.7-2.1; Kim et al., 2000; Shiels et al., 2000) but that the total transsarcolemmal Ca²⁺ influx is fairly well maintained during temperature change (Shiels et al.,2000). Furthermore, the fish cardiac SR has been shown to uptake and release Ca²⁺ at temperatures well below those known to inhibit SR function in most mammals (Aho and Vornanen, 1999; Hove-Madsen et al., 2001; Shiels et al.,2002a; Tiitu and Vornanen,2002).

Despite the fact that the individual Ca²⁺ flux pathways in trout hearts seem to demonstrate some tolerance to temperature change, in situ heart and in vitro cardiac muscle strip studies show both inotropic and chronotropic responses to temperature change(Keen et al., 1994; Shiels and Farrell, 1997; Aho and Vornanen, 1999). We wanted to understand to what extent these responses might be mediated by changes in cellular Ca²⁺. The transient change in[Ca²⁺]_i during excitation—contraction coupling represents the sum of both extracellular and intracellular Ca²⁺flux, and its amplitude is directly related to the strength of cardiac contraction (Yue, 1987). Thus,the focus of the present study was to examine [Ca²⁺]_i at different temperatures in rainbow trout atrial myocytes. Previous studies have shown that trout experience acute temperature fluctuations of as much as±10°C either while transversing thermoclines or as a result of diurnal changes in shallow streams (Reid et al., 1997; Matthews and Berg, 1997). Thus, we conducted our studies on 14°C-acclimated rainbow trout at 7°C, 14°C and 21°C. By simultaneously quantifying[Ca²⁺]_i and I_Ca at each of these temperatures, we were able to assess the effects of temperature on[Ca²⁺]_i and indirectly separate the effect of temperature on extracellular and intracellular Ca²⁺ fluxes.

In fish hearts, ambient temperature also directly modulates heart rate(Farrell and Jones, 1992). In turn, changes in contraction frequency affect many of the cellular processes that underlie the transient change in [Ca²⁺]_i(Møller-Nielsen and Gesser,1992; Hove-Madsen and Tort,1998). For example, increasing the contraction frequency from 0.2 Hz to 1.4 Hz causes significant reductions in the amplitude of I_Ca,as well as reductions in both the amplitude and duration of the AP in trout ventricular myocytes (Harwood et al.,2000). Because the frequency of stimulation will modulate cellular Ca²⁺ cycling, it is critical to consider the physiologically relevant contraction frequency associated with a particular temperature when trying to understand the implications of temperature change on[Ca²⁺]_iin vivo. Thus, we examined[Ca²⁺]_i and I_Ca in trout atrial myocytes that were stimulated with physiological APs at physiologically relevant contraction frequencies for the applied temperature. We also examined the effect of temperature on [Ca²⁺]_i and I_Ca in response to conventional square (SQ) voltage clamp pulses at slow (0.2 Hz) and physiologically relevant frequencies to gain mechanistic insight and to facilitate comparison of our results with earlier studies.

Rainbow trout Oncorhynchus mykiss Walbaum (mean mass 416.7±17.2 g, N=12) were obtained from a local fish farm(Kontiolahti, Finland) and held in the laboratory at 14±1°C for a minimum of four weeks prior to experimentation. They were fed daily on commercial trout pellets (Biomar, Brande, Denmark) and exposed to a 15 h:9 h L:D cycle. All procedures were in accordance with local animal handling protocols.

A detailed description of the rainbow trout myocyte preparation has been published previously (Vornanen,1998; Shiels et al.,2000). Briefly, myocytes were obtained by retrograde perfusion of the heart, first with an isolating solution for 8-10 min and then with a proteolytic enzyme solution for 20 min. After enzymatic treatment, the atrium was removed from the ventricle, cut into small pieces and then triturated through the opening of a Pasteur pipette. The isolated cells were stored for up to 8 h in fresh isolating solution at 14°C.

The isolating solution contained: NaCl, 100 mmol l^-1; KCl, 10 mmol l^-1; KH₂PO₄, 1.2 mmol l^-1;MgSO₄, 4 mmol l^-1; taurine, 50 mmol l^-1;glucose, 20 mmol l^-1; and Hepes, 10 mmol l^-1; adjusted to pH 6.9 with KOH. For enzymatic digestion, collagenase (Type IA), trypsin(Type III) and fatty-acid-free bovine serum albumin (BSA) were added to the isolating solution. Standard internal and external solutions were used to eliminate contaminating Na⁺ and K⁺ currents. Tetrodotoxin (TTX, 1 μmol l^-1) was included in all external solutions to effectively eliminate fast Na⁺ currents(Vornanen, 1998). Cs⁺-based internal and external solutions with 15 mmol l^-1 tetraethylammonium chloride (TEA) in the pipette were used to block K⁺ currents (Hove-Madsen and Tort, 1998). The extracellular solution for recording I_Ca contained: NaCl, 150 mmol l^-1; CsCl, 5.4 mmol l^-1; MgSO₄, 1.2 mmol l^-1;NaH₂PO₄, 0.4 mmol l^-1; CaCl₂, 1.8 mmol l^-1; glucose, 10 mmol l^-1; and Hepes, 10 mmol l^-1; adjusted to pH 7.7 with CsOH. Additionally, 5 nmol l^-1 adrenaline was added to emulate the tonic level of adrenaline circulating in the blood of resting rainbow trout in vivo(Milligan et al., 1989). The pipette solution contained: CsCl, 130 mmol l^-1; MgATP, 5 mmol l^-1; TEA, 15 mmol l^-1; MgCl₂, 1 mmol l^-1; oxaloacetate, 5 mmol l^-1; Hepes, 10 mmol l^-1; EGTA, 0.025 mmol l^-1; Na₂GTP, 0.03 mmol l^-1; and K₅-Fura-2, 0.1 mmol l^-1; pH adjusted to 7.2 with CsOH. All drugs, with the exception of TTX (Tocris, Bristol, UK),were purchased from Sigma (St Louis, MO, USA).

Whole-cell voltage-clamp experiments were performed using an Axopatch 1D amplifier with a CV-4 1/100 headstage (Axon Instruments, Foster City, CA,USA). Pipettes had a resistance of 2.2±0.30 MΩ when filled with pipette solution. Junction potentials were zeroed prior to seal formation. Pipette capacitance (9.3±0.2 pF) was compensated after formation of a GΩ seal. Mean series resistance was 6.5±0.3 MΩ. Membrane capacitance (31.2±0.8 pF, N=76) was measured using the calibrated capacity compensation circuit of the Axopatch amplifier. Signals were low-pass filtered using the 4-pole lowpass Bessel filter on the Axopatch-1D amplifier at a frequency of 2 kHz and were then analyzed off-line using pClamp 6.0 software (Axon Instruments).

In a separate series of experiments, we recorded APs from rainbow trout atrial myocytes (Fig. 1). Cells were stimulated to elicit APs at 7°C, 14°C or 21°C and at a frequency that corresponds to the resting heart rate of rainbow trout in vivo at each of these temperatures. The temperature/frequency pairs were as follows: 0.6 Hz at 7°C, 1.0 Hz at 14°C and 1.4 Hz at 21°C(Tuurala et al., 1982; Farrell et al., 1996; Aho and Vornanen, 2001). APs were elicited by the minimum voltage pulse able to trigger a self-sustained AP(approximately 1 ms, 0.8 nA). The resultant AP waveforms were subsequently used to provide a more physiologically relevant stimulus under which to evaluate I_Ca and [Ca²⁺]_i during temperature change. In all experiments, the capacitive transients resulting from the AP waveform were compensated for by the amplifier.

Peak current amplitude of I_Ca was calculated as the difference between the peak inward current and the current recorded at the end of the depolarizing pulse. The charge (pC) carried by I_Ca was calculated from the time integral of the current and then normalized to cell capacitance(pC pF^-1).

For each cell, after formation of the GΩ seal, and before rupture of the cell membrane, background fluorescence (approximately 40% and 25% of the total signal at 340 nm and 380 nm, respectively) was recorded and then subtracted from all subsequent measurements. Temperature change had no effect on background fluorescence. Myocytes were alternately illuminated (optical chopper, 0.2 kHz, 75 W Xenon arc lamp) at 340 nm and 380 nm with a high-speed dual-wavelength filter-based illuminator (PowerFilter, Photon Technology International, Brunswick, NJ, USA). Emitted light was filtered at 510 nm and detected by a photon-counting photomultiplier tube, the output of which was connected to a computer via an I/O board and saved to disk for later analysis using the FeliX acquisition and analysis software (7-point smoothing and trace averaging, Photon Technology International). The emission aperture window was kept as small as possible and was focused close to, but out of view of, the pipette tip to minimize any effect of cell movement during contraction(cells typically contracted approximately 10-15% of the resting length).

The ratio of emitted fluorescence at 340 nm/380 nm (R) was calculated to give an index of [Ca²⁺]_i.[Ca²⁺]_i was calculated using the acquired fluorescent ratios and calibration parameters derived both experimentally and through calculation. To measure R_max and R_min,cells were patched-clamped and perfused with an extracellular solution containing metabolic inhibitors (rotenone, 2 μmol l^-1; carbonyl cyanide m-chlorophenyl-hydrazone (CCCP), 5 μmol l^-1;jodo acetic acid, 5 mmol l^-1). Pipette and extracellular solutions were the same as those described above except that nucleosides and sulphates were omitted, respectively. To estimate R_max, 10 μmol l^-1 Ca²⁺ ionophore (4-bromo A23187) and 1.8 mmol l^-1 CaCl₂ were added to the perfusate. To estimate R_min, cells were then perfused with a solution containing 10 mmol l^-1 EGTA. The values obtained are as follows: R_max=9.80±0.67, R_min=0.99±0.01, β (signal fluorescence at 380 nm/background fluorescence at 380 nm)=4.62±0.44; N=17 (A. Ryökkynen and M. Vornanen, unpublished data). The affinity constant K_d of Fura-2 has been measured in vivo over a range of temperatures (5-37°C) in enterocytes from the Atlantic cod(Gadus morhua; Larsson et al.,1999). We adjusted the K_d values obtained in that study to match our experimental conditions of pH, buffer and ionic strength by calculating the effects of pH, buffer and ionic strength on 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA)(Harrison and Bers, 1987)using the MaxChelator program (Bers et al.,1994). The calculated K_d values were as follows: 395 nmol l^-1 at 7°C, 366 nmol l^-1 at 14°C and 336 nmol l^-1 at 21°C. The fluorescence ratios obtained during the experiments were converted into[Ca²⁺]_i using these K_d values and the above measurements of R_max, R_minand β, as described by Grynkiewicz et al.(1985).

Atrial myocytes were superfused at a rate of 2 ml min^-1 with extracellular solution at 7°C, 14°C or 21°C. Each cell was tested at one experimental temperature only. The extracellular solution was heated or chilled by a water bath circuit before emptying into the recording chamber(RC-26, Warner Instrument Corp., Hamden, CT, USA; volume 150 μl). A thermocouple was placed inside the recording chamber and positioned no less than 5 mm from the cell to ensure it was experiencing the desired temperature. At the onset of each experiment, cells were stimulated with 25 physiological APs at a frequency that corresponded to the in vivo heart rate of trout at each temperature (see Fig. 1) to ensure that SR Ca²⁺ content was similar in all cells (Shiels et al., 2002a). However, the subsequent stimulation protocols may alter SR Ca²⁺content as the experiment proceeds, which is a limitation of our approach.

Cells were first stimulated to contract with five SQ depolarizing pulses(from -80 mV to +10 mV, 500 ms duration) at a frequency of 0.2 Hz(SQ_0.2Hz), and I_Ca and [Ca²⁺]_iwere simultaneously recorded. The cells were then stimulated to contract with five SQ depolarizing pulses at a frequency that corresponded to the in vivo heart rate for rainbow trout at the given test temperature(SQ_physiol). SQ voltage-clamp protocols are vital for characterizing channel kinetics and simplify interpretation of mechanistic relationships, but they may have only limited relevance to the condition in vivo (Linz and Meyer,1998; Puglisi et al.,1999). Therefore, after stimulation under SQ_physiolconditions, the cells were stimulated with an AP waveform (see Fig. 1) appropriate for the test temperature and at a frequency appropriate for the test temperature(AP_physiol). I_Ca and [Ca²⁺]_i were simultaneously recorded throughout (see Fig. 4). Although the SQ_physiol part of the protocol is of little physiological significance, we believe that comparing the results obtained at physiological stimulation frequencies from both a 500 ms SQ (SQ_physiol) and a physiological AP (AP_physiol) allows us to evaluate the importance of the AP and the possible limitations of the SQ pulse in assessing physiological processes. The duration of each protocol was approximately 1 min and we did not observe any changes in intracellular buffering by Fura-2 or run-down of I_Ca over this time period.

One-way repeated measures analysis of variance (RM ANOVA) were used to compare the effects of changing the frequency and the stimulus waveform on I_Ca and [Ca²⁺]_i. One-way ANOVAs were used to test the effects of temperature on I_Ca and[Ca²⁺]_i. Differences were considered significant at P<0.05, as assessed by Student—Newman—Keuls (SNK) post-hoc analysis. The mean values ± S.E.M. and the statistical analysis for [Ca²⁺]_i appear in the tables. Five transients were recorded under each experimental pulse condition (i.e. SQ_0.2Hz, SQ_physiol and AP_physiol; see Fig. 4). The diastolic[Ca²⁺]_i at the end of the 4th transient and the peak[Ca²⁺]_i of the 5th transient were used to represent a steady-state value for each condition (Tables 1,2,3). For future experiments, it would, however, be prudent to provide a longer train of stimulation pulses, especially at warm temperatures, to ensure a steady-state level had been attained. On the other hand, excursions by fish across thermoclines can be very rapid and transitory in nature and therefore I_Ca may not reach a steady-state in vivo before temperature (and thus heart rate) change again.

For both SQ and AP pulses, the amplitude of the stimulus-induced change in[Ca²⁺]_i decreased with increasing temperature(Fig. 2, Table 1). The time required to reach peak [Ca²⁺]_i also decreased with increasing temperature (Table 2), probably due to the reduced amplitude because the rate of rise did not change significantly (0.625±0.33 μmol l^-1 s^-1 at 7°C and 0.460±0.07 μmol l^-1 s^-1 at 21°C with SQ pulses at 0.2 Hz).

The peak amplitude of I_Ca increased with temperature (Figs 2B, 3A). The charge carried across the SL via I_Ca was unchanged by temperature during SQ pulse stimulation and decreased with temperature during AP stimulation(Fig. 3B).

During SQ pulse stimulation, we observed a temperature-dependent change in the amplitude of [Ca²⁺]_i but no change in the charge density calculated by integrating I_Ca. This suggests that, in addition to Ca²⁺ flux across the SL, other Ca²⁺ flux pathways, most probably Ca²⁺ release from the SR, must be changing with respect to temperature in these cells. During AP stimulation, the reduced charge density of I_Ca at warm temperatures(Fig. 3B) may underlie the large percentage reduction in [Ca²⁺]_i observed under these conditions (Fig. 2, Table 1).

The frequency and shape of the depolarizing stimulus profoundly altered cellular Ca²⁺ dynamics. As a detailed example, Fig. 4 illustrates the frequency dependence and the influence of pulse shape on I_Ca and[Ca²⁺]_i for an experiment conducted at 21°C. During SQ_0.2Hz, diastolic Ca²⁺ concentrations returned to rest between depolarizing pulses (Fig. 4A). When stimulation frequency was increased from SQ_0.2Hz to SQ_1.4Hz, diastolic Ca²⁺concentration increased significantly due to the long (500 ms) and high-frequency (1.4 Hz) depolarization of the membrane (also see Table 3). In addition,[Ca²⁺]_i increased by approximately 30%(Fig. 4, Table 1). The rate of rise also increased significantly (from 0.73±0.11 μmol l^-1s^-1 to 1.16±0.84 μmol l^-1 s^-1), and the time required to reach the peak of the transient decreased during SQ_1.4Hz compared with that during SQ_0.2Hz(Table 2). When the stimulus waveform was changed from SQ_1.4Hz to AP_1.4Hz,[Ca²⁺]_i and diastolic Ca²⁺ levels both decreased significantly (Tables 1, 2), and the rate of rise of the transient decreased significantly to a value (0.59±0.17 μmol l^-1 s^-1) that was not significantly different from that for the SQ_0.2Hz pulse. Qualitatively similar effects of frequency and stimulus shape were observed at 14°C and 7°C (Tables 1,2,3). Thus, independent of temperature, an increase in stimulation frequency from 0.2 Hz to a physiologically relevant frequency for the experimental temperature (SQ_physiol) produced a large increase in both[Ca²⁺]_i and in diastolic Ca²⁺ levels with 500 ms SQ stimulation pulses. By contrast, the AP stimulus, which is much shorter in duration, reduced [Ca²⁺]_i and diastolic Ca²⁺ levels compared with SQ pulses at the same frequency.

The amplitude of I_Ca was significantly smaller with an AP stimulus compared with a SQ stimulus at the same frequency (Figs 2,3,4; Table 1). The charge carried by I_Ca did not differ significantly at 14°C and 21°C but was greater at 7°C (Fig. 3B). This situation probably arose as a result of sustained Ca²⁺ influx during the AP plateau and because of Ca²⁺ influx occurring via the window current (see Fig. 2B and below).

This is the first study we are aware of to measure cytosolic free Ca²⁺ in trout atrial myocytes and examine how it changes with temperature, stimulus shape and frequency. Furthermore, an effort was made to simulate conditions that were more physiologically relevant with respect to temperature, as well as stimulation waveform and frequency. There were three principle findings: (1) the transient change in [Ca²⁺]_iincreases with decreasing temperature, (2) the transient change in[Ca²⁺]_i is smaller in response to physiological AP stimulation than to 500 ms SQ stimulus pulses, and (3) diastolic Ca²⁺ levels are lower at physiological frequencies when contraction is stimulated with APs compared with 500 ms SQ pulses. Our results indicate that temperature-dependent changes in I_Ca alone cannot account for the temperature-dependent changes in [Ca²⁺]_i. Thus,although we made no direct measurements of SR Ca²⁺ flux, our results suggest temperature may modulate SR Ca²⁺ cycling in trout atrial myocytes.

With both SQ and AP pulses, [Ca²⁺]_i was slowest and largest at 7°C compared with at 14°C and 21°C. This observation agrees with results from multicellular cardiac muscle studies from rainbow trout, where isometric contraction is slower and stronger as temperature is decreased (Driedzic and Gesser,1994; Shiels and Farrell,1997; Aho and Vornanen,2001). These findings also agree with mammalian studies that demonstrate an increase in the amplitude of the transient change in[Ca²⁺]_i with decreases in temperature(Wang et al., 2000). The small amplitude of [Ca²⁺]_i and the elevated diastolic Ca²⁺ levels during physiological AP pulses at 21°C is striking and is consistent with the finding that the maximum power output of trout heart preparations in situ falls off at temperatures above 18°C(Farrell et al., 1996). Indeed, the poor maximum cardiac performance at temperatures approaching the upper incipient lethal temperature for salmonids (23-25°C; Black, 1953), coupled with the observed reduction in maximum swimming capabilities of salmonids as temperatures approach 21°C in both lab(Brett, 1971) and field(Tierney, 2000) studies, has led to the suggestion (Farrell,1997) that cardiac failure may be a critical factor when salmon are exposed to temperatures that exceed their preferred temperature(approximately 15°C; Brett,1971). The implication of the present study is that the small amplitude of [Ca²⁺]_i in the cardiac myocyte may be critical to setting the upper thermal regime in fishes. Thus, it follows that data obtained from trout cardiac myocytes at room temperature should be regarded with some caution.

Previous studies with trout atrial(Shiels et al., 2000) and ventricular (Kim et al., 2000)myocytes have shown a reduction in the peak current amplitude of I_Ca with decreasing temperature, and our present study supports these earlier findings. The similar Ca²⁺ influx viaI_Ca at 7°C, 14°C and 21°C probably resulted from a combination of higher and faster peak I_Ca, offset by more rapid I_Ca inactivation as temperatures increased (see Fig. 2 and Shiels et al., 2000). A similar lack of effect of temperature on the charge density of I_Cahas been reported for rabbit myocytes at 25°C and 35°C(Puglisi et al., 1999). We observed an increase in the outward current at the end of the stimulus pulse at 21°C compared with at 14°C and 7°C (see Fig. 2B), which is probably attributable to increased Ca²⁺ influx via reverse-mode sarcolemmal Na⁺/Ca²⁺ exchange (NCX; Hove-Madsen et al., 2000). During AP stimulation, the charge density of I_Ca was reduced at warm temperatures (Fig. 3B),probably due to the shorter AP duration. This will have contributed to the reduction in the transient change in [Ca²⁺]_i observed during stimulation at 21°C. However, certainly during the SQ pulse stimulus protocol, and also during the AP stimulus protocol, the `mismatch'between temperature-induced changes in SL Ca²⁺ influx and[Ca²⁺]_i suggests that temperature must modulate another Ca²⁺ flux pathway in trout atrial myocytes.

Increasing SQ pulse stimulation frequency (from SQ_0.2Hz to SQ_physiol) caused an approximately 30% increase in[Ca²⁺]_i and an approximately 75% increase in the diastolic Ca²⁺ level (Figs 2, 4; Tables 1, 3), indicating that more Ca²⁺ enters the cytosol than exits via SR uptake or SL efflux. The high-frequency application of long SQ pulses (500 ms) increased Ca²⁺ influx via reverse-mode NCX, as evidenced by the increased outward current at the end of the stimulation pulse (see Fig. 4B, SQ_1.4Hz). In avian (Lee and Clusin,1987) and mammalian cardiomyocytes(Frampton et al., 1991), the frequency-dependent increase in [Ca²⁺]_i has been attributed, in part, to increased SR Ca²⁺ load and thus an increased SR Ca²⁺ release. Increased Ca²⁺ influx via NCX may have resulted in greater SR Ca²⁺ loading and thus greater Ca²⁺-induced Ca²⁺ release, contributing further to the increase in the amplitude of the transient change in[Ca²⁺]_i in the present study. An increase in the frequency of field stimulation from 0.6 Hz to 1.0 Hz at 14°C increased diastolic Ca²⁺ (approximately 25-50%) without changing the amplitude of [Ca²⁺]_i in trout ventricular myocytes(Harwood et al., 2000). In the same study, the authors found that application of caffeine revealed a greater SR Ca²⁺ content at 1.0 Hz compared with at 0.6 Hz, indicating that SR Ca²⁺ content increased with pacing frequency but was not releasable during a field-stimulated twitch(Harwood et al., 2000). The differences between studies may indicate that a SQ pulse is a better trigger for SR Ca²⁺ release than is field stimulation, where the membrane is depolarized with an AP (see below). Indeed, Hove-Madsen et al.(2001) have demonstrated a similar degree of CICR at 7°C and 21°C with SQ pulses in trout atrial myocytes.

[Ca²⁺]_i and diastolic Ca²⁺ levels decreased dramatically when the SQ pulse was changed to an AP pulse at the same frequency. This reflects a reduction in SL Ca²⁺ influx and also possibly a reduction in CICR with AP pulses, as the smaller amplitude of I_Ca during AP pulses (see Figs 2, 3) may be a less effective trigger for CICR.

The myocytes were better able to manage diastolic Ca²⁺ levels at physiological frequencies with AP pulses(Table 3) but not as effectively as at 0.2 Hz. This suggests that the Ca²⁺ efflux mechanisms are not able to remove all of the Ca²⁺ effectively during diastole at physiological pacing frequencies. However, it is possible that resting diastolic Ca²⁺ concentration in the absence of stimulation and at 0.2 Hz is sub-physiological. The true `diastolic Ca²⁺ concentration' for fish hearts remains unknown and is worthy of future investigation. In any case, visual observations indicate that the cells were relaxing between stimulation pulses at physiological frequencies with AP pulses at all temperatures.

It has been estimated from trout ventricular myocytes at room temperature that, when operating maximally (i.e. at or near V_max), the trout SR Ca²⁺-ATPase should be able to remove the total intracellular Ca²⁺ transient between depolarizations at physiological heart rates (Hove-Madsen et al., 1998). However, our study reveals that, at room temperature,the trout intracellular Ca²⁺ transient is small compared with cooler temperatures and suggests that the SR Ca²⁺-ATPase is unlikely to be operating near V_max. Furthermore, estimates of the maximal Ca²⁺ efflux rate of the rainbow trout NCX at room temperature (Hove-Madsen and Tort,2001) suggest that the NCX can remove a physiological Ca²⁺ transient from the cytosol in a few hundred milliseconds. Based on those estimations, the NCX should have been able to remove all of the cytosolic Ca²⁺ during the 700 ms between depolarizations at 1.4 Hz in the present study at 21°C. The observed increase in the diastolic Ca²⁺ levels in the present study are in contrast to these estimates of Ca²⁺ efflux and may reflect exogenous buffering by Fura-2. The in vivo buffering capacity of the fish myoplasm and its modulation by temperature is unknown. However, because both contractility and the time course of I_Ca were maintained in the presence of Fura-2, it is unlikely that exogenous buffering was large relative to the measurements being made. A comparison of the time to 50% relaxation of[Ca²⁺]_i with AP pulses in our study at 14°C(approximately 250 ms) with the ratiometric results from the only other published study of [Ca²⁺]_i in fish hearts, which examined field-stimulated trout ventricular myocytes loaded with Fura-2-AM at 15°C (approximately 230 ms; Harwood et al., 2000), suggests comparable time courses of the transients. Additional studies are needed to resolve the physiological efficacy of trout Ca²⁺ efflux pathways and the effect of exogenous buffering and temperature change therein.

The frequency-dependent decrease in I_Ca at 21°C reported here agrees well with previous studies that report a 15-20% decrease in peak current amplitude over a similar frequency range at 21°C in trout myocytes(Hove-Madsen and Tort, 1998; Harwood et al., 2000). These responses are akin to the negative force-frequency response observed independent of temperature in the myocardium of a number of fish species(Driedzic and Gesser, 1985; Keen et al., 1994; Shiels and Farrell, 1997; see Shiels et al., 2002b for a recent review). The smaller amplitude of the peak current during an AP pulse compared with during a SQ pulse is qualitatively similar to what we have found previously in trout atrial myocytes(Shiels et al., 2000) and similar to what has been described in mammals(Linz and Meyer, 1998; Puglisi et al., 1999).

The shape of the trout AP, in combination with the relatively depolarized resting membrane potential (-50 mV), may have led to an overestimation of SL Ca²⁺ flux via the I_Ca window current in the present study (Fig. 2B). This is because the late phase depolarization after the initial hyperpolarization(Fig. 1) brings the membrane potential close to the window current voltage range for I_Ca in trout atrial myocytes (-40 mV to +30 mV, with a peak at -10 mV; see Shiels et al., 2000). Recent measurements in vivo suggest that the resting membrane potential in rainbow trout atrial cells is approximately -65 mV(Vornanen et al., 2002). At this membrane potential, the late phase depolarization would not enter the window voltage range and thus I_Ca inactivation would be maintained. The relatively depolarized membrane potential in isolated myocytes from our study results from the loss of cholinergic tonus and also reflects the lower density of inward rectifier K⁺ channel current (I_K1) in atrial compared with ventricular cells(Vornanen et al., 2002). Future studies with isolated trout myocytes could consider including tonic levels of acetylcholine when measuring APs to account for this.

Collectively, we have characterized some of the temperature-, frequency-and waveform-dependent changes in [Ca²⁺]_i and I_Ca that exist in rainbow trout atrial myocytes. By calibrating the Fura-2 signal, we were able to make meaningful comparisons of[Ca²⁺]_i at physiologically relevant temperatures for rainbow trout. We have shown that increased temperature causes a decrease in[Ca²⁺]_i. This is the first cellular observation of a temperature-dependent change in [Ca²⁺]_i in fish myocytes and may help to explain the well-known negative inotropic effect of warm temperatures on trout heart function. Our results suggest that the temperature-dependent changes in [Ca²⁺]_i are not solely dependent on temperature-dependent changes in I_Ca. Therefore,temperature and frequency modulation of SR Ca²⁺ accumulation and release may be involved, and future studies should be directed at understanding these mechanisms. Finally, we have shown that the amplitude of[Ca²⁺]_i and the level of the diastolic Ca²⁺concentration is dependent on the shape and the rate of stimulation. At all temperatures, cells were better able to maintain diastolic Ca²⁺levels at physiological frequencies with AP pulses compared with 500 ms SQ pulses. This suggests that temperature-dependent modulation of the shape of the AP is important for cellular Ca²⁺ regulation during temperature and frequency changes in rainbow trout heart.

We acknowledge Kontiolahti fish farm for donating the trout and Ari Ryökkynen for his efforts with the Fura-2 calibrations. We also thank Dr Kerry Delaney for his technical advice and Dr Leif Hove-Madsen for his constructive comments on earlier versions of this manuscript. The study was supported by NSERC to A.P.F., Simon Fraser University Steele Memorial Fellowship to H.A.S., and the Academy of Finland (project #63090) to M.V.