Paradoxical SR Ca²⁺ release in guinea-pig cardiac myocytes after β-adrenergic stimulation revealed by two-photon photolysis of caged Ca²⁺

Abstract

In heart muscle the amplification and shaping of Ca²⁺ signals governing contraction are orchestrated by recruiting a variable number of Ca²⁺ sparks. Sparks reflect Ca²⁺ release from the sarcoplasmic reticulum (SR) via Ca²⁺ release channels (ryanodine receptors, RyRs). RyRs are activated by Ca²⁺ influx via L-type Ca²⁺ channels with a specific probability that may depend on regulatory mechanisms (e.g. β-adrenergic stimulation) or diseased states (e.g. heart failure). Changes of RyR phosphorylation may be critical for both regulation and impaired function in disease. Using UV flash photolysis of caged Ca²⁺ and short applications of caffeine in guinea-pig ventricular myocytes, we found that Ca²⁺ release signals on the cellular level were largely governed by global SR content. During β-adrenergic stimulation resting myocytes exhibited smaller SR Ca²⁺ release signals when activated by photolysis (62.3% of control), resulting from reduced SR Ca²⁺ content under these conditions (58.6% of control). In contrast, local signals triggered with diffraction limited two-photon photolysis displayed the opposite behaviour, exhibiting a larger Ca²⁺ release (164% of control) despite reduced global and local SR Ca²⁺ content. This apparent paradox implies changes of RyR open probabilities after β-adrenergic stimulation, enhancing local regenerativity and reliability of Ca²⁺ signalling. Thus, our results underscore the importance of phosphorylation of RyRs (or of a related protein), as a regulatory physiological mechanism that may also provide new therapeutic avenues to recover impaired Ca²⁺ signalling during cardiac disease.

Depolarization of cardiac myocytes during an action potential leads to the opening of voltage-dependent L-type Ca²⁺ channels. The subsequent elevation of the Ca²⁺ concentration ([Ca²⁺]_i) in the dyadic cleft activates Ca²⁺ release channels (referred to as ryanodine receptors or RyRs) forming macromolecular complexes located in the sarcoplasmic reticulum (SR) membrane and in the dyadic cleft. This Ca²⁺-induced Ca²⁺ release (CICR) process amplifies and scales the Ca²⁺ signal via summation of discrete elementary Ca²⁺ signalling events termed Ca²⁺ sparks. In order for the Ca²⁺ signal to be transient the Ca²⁺ release from the SR has to terminate. Furthermore, cytosolic Ca²⁺ needs to be pumped back into the SR by the SR Ca²⁺ pump (SERCA) and transported out of the cell via the Na⁺–Ca²⁺ exchange (for review see Bers, 2002).

The discovery of Ca²⁺ sparks as elementary Ca²⁺ signalling events has fundamentally changed the view of excitation–contraction coupling (EC coupling) on the cellular and subcellular level (Cheng et al. 1993; Niggli, 1999). Contrary to the notion of systems in which released Ca²⁺ has access to a common cytosolic pool (Stern, 1992), amplification and shaping of the Ca²⁺ signal is now understood to be fine-tuned by recruiting a variable number of functionally independent Ca²⁺ sparks, each of which is an all-or-none event with a high degree of positive feedback. After opening of an L-type Ca²⁺ channel each SR Ca²⁺ release site capable of generating a Ca²⁺ spark is activated with a specific probability, which may depend on a variety of variables, including regulatory mechanisms and pathophysiological or diseased states. Many of these variables can thus affect EC coupling by changing the Ca²⁺ spark trigger probability. Indeed, a reduced Ca²⁺ spark trigger probability has been identified as a cause underlying impaired EC coupling in myocytes from hypertrophied and failing rat hearts (Gomez et al. 1997).

Among possible mechanisms that may affect the Ca²⁺ sensitivity of the RyRs, and thus the Ca²⁺ spark trigger probability, variations of SR Ca²⁺ content and regulatory changes due to protein phosphorylation after β-adrenergic stimulation have received much attention recently. On the basis of receptor number, the β₁-adrenergic receptor (β₁AR) is the predominant β-receptor subtype in cardiac ventricular myocytes and couples to stimulatory G_s proteins, which leads to activation of adenylyl cyclase after receptor activation. Adenylyl cyclase synthesizes the second messenger cyclic adenosine monophosphate (cAMP), which increases the activity of protein kinase A (PKA). PKA phosphorylates many substrates, several of which play an essential role in Ca²⁺ signalling: (1) phosphorylation of L-type Ca²⁺ channels enhances the Ca²⁺ current, thus increasing both the trigger signal for CICR and the extent of SR loading with Ca²⁺ (Reuter, 1983); (2) phosphorylation of phospholamban (PLB) relieves its inhibitory effect on the SERCA, subsequently stimulating the Ca²⁺ pump, again increasing the Ca²⁺ load of the SR (Lindemann et al. 1983; James et al. 1989; Shannon et al. 2001; Bers, 2002); (3) it has recently been reported that PKA could also directly phosphorylate the RyRs, possibly inducing dissociation of calstabin-2 (formerly called FKBP12.6) and increasing RyR open probability (Lu et al. 1995; Valdivia et al. 1995; Marx et al. 2000). Dissociation of calstabin-2 may subsequently disrupt coupled gating of the RyRs (Marx et al. 2001). Generally, PKA-mediated phosphorylation of the RyRs is thought to increase the Ca²⁺ sensitivity of the release channels. Conversely, several research groups found that RyRs were not phosphorylated (Jiang et al. 2002), or, when phosphorylated, the RyR open probability appeared to remain unchanged (Li et al. 2002). In addition, others have found that RyR phosphorylation by Ca²⁺-calmodulin-dependent kinase II (CaMKII) induced stimulatory or inhibitory effects in lipid bilayers and profound stimulatory effects in freshly isolated myocytes (Takasago et al. 1991; Hain et al. 1995; Lokuta et al. 1995; Li et al. 1997; Wang et al. 2004).

In intact cells changes of CICR after β-adrenergic stimulation could also result from increased SR Ca²⁺ load. Increased Ca²⁺ load leads to more Ca²⁺ release by the law of mass action (Fabiato, 1985; Bassani et al. 1995; Tripathy & Meissner, 1996; Santana et al. 1997; Satoh et al. 1997; Sitsapesan & Williams, 1997; Györke & Györke, 1998; Frank et al. 2000; Lukyanenko et al. 2001). In addition, the Ca²⁺ concentration inside the SR may have a regulatory effect on CICR by modulating the Ca²⁺ sensitivity of the RyRs, either via a Ca²⁺ receptor directly located on the luminal side of the RyRs (Györke & Györke, 1998; but see Tripathy & Meissner, 1996) or by involving a signalling pathway comprising calsequestrin as a Ca²⁺ sensor and one or more small accessory junctional SR proteins, such as triadin and junctin (Guo & Campbell, 1995; Zhang et al. 1997; Györke et al. 2004). In the failing heart, Ca²⁺ regulation is proposed to be one of several key players in the altered contractility (Houser et al. 2000; Hasenfuss & Pieske, 2002), but many other processes are affected too, such as structural changes, altered protein expression and decreased PKA-dependent phosphorylation of several proteins (Marks et al. 2002). Furthermore, heart failure may be associated with RyR hyperphosphorylation, possibly leading to dissociation of calstabin-2 from the RyRs and resulting in an increased SR Ca²⁺ leak, which may finally decrease SR Ca²⁺ content (Shou et al. 1998; Marx et al. 2000; Prestle et al. 2001). Ultimately, diminished SR Ca²⁺ content may then cause smaller Ca²⁺ transients and reduced cardiac force.

In the present study we found a strong correlation between global cellular Ca²⁺ release amplitude and SR Ca²⁺ load when the trigger signal was a spatially homogeneous photolytic Ca²⁺ transient induced by a UV flash. In contrast, highly localized Ca²⁺ release signals generated with diffraction-limited two-photon photolysis were less sensitive to SR Ca²⁺ load. In particular, larger localized Ca²⁺ release signals were recorded after β-adrenergic stimulation, even in conditions where the SR Ca²⁺ content was reduced. Based on this discrepancy between global and local Ca²⁺ signals we conclude that, in addition to the secondary effect via changes of luminal Ca²⁺, β-adrenergic stimulation may change the gating properties of the RyRs themselves, either directly or indirectly. These findings have been presented in preliminary form to the Biophysical Society (Lindegger & Niggli, 2002).

Methods

Isolation of guinea-pig myocytes

Cardiac ventricular myocytes were isolated from adult male guinea-pigs using established enzymatic methods (DelPrincipe et al. 1999). All animal handling procedures were performed with the permission of the State Veterinary Administration and according to Swiss Federal Animal handling law. Guinea-pigs were killed by cervical dislocation, the hearts rapidly removed and mounted on a Langendorff system and retrogradely perfused with a Ca²⁺-free solution at 37°C for about 5 min (Mitra & Morad, 1985). For enzymatic digestion, collagenase type 2 (0.12 mg ml⁻¹, Worthington, Switzerland) and protease type XIV (0.04 mg ml⁻¹, Sigma, Switzerland) were added to the perfusion solution for another 3–5 min. After digestion, the ventricles were cut into small pieces, placed on a gently rotating shaker in a solution containing 200 μm Ca²⁺ and kept at room temperature until use.

Solutions

For the experiments cells were transferred into a chamber mounted on the stage of an inverted microscope. The extracellular superfusion solution contained (mm): NaCl 140, KCl 5, CaCl₂ 1.8, CsCl 1, BaCl₂ 0.5, Hepes 10, glucose 10, pH 7.4 (adjusted with NaOH). Cells were continuously superfused using a custom-made rapid superfusion system (t_1/2 < 500 ms). Where indicated, isoproterenol (isoprenaline, Iso) 1 μm ([−]-N-iso-propyl-l-noradrenaline hydrochloride; Sigma) was added from a frozen stock (1 mm in 10% ascorbic acid). Cyclopiazonic acid (CPA, 10 μm; Sigma) was added to block the SERCA and to avoid loading of the SR. Short applications of 20 mm caffeine (Sigma) were used to estimate SR Ca²⁺ content. In some control experiments the SR function was inhibited using 1 μm thapsigargin and 10 μm ryanodine (both from Alamone Laboratories, Jerusalem, Israel). The pipette-filling solution contained (mm): caesium aspartate 120, TEA-Cl 20, Na₄-DM-nitrophen 2 (Calbiochem, La Jolla, CA, USA), reduced glutathione (GSH) 1, CaCl₂ 0.5, K₂-ATP 5, Hepes 10 and K₅-fluo-3 0.05 (TefLabs, Austin, TX, USA), pH 7.2 (adjusted with CsOH). All experiments were carried out at room temperature (21°C).

Voltage clamp

Electrodes were pulled from filamented borosilicate glass capillaries (GC150F, Clark Electromedical Instruments, Pangbourne, UK) on a horizontal puller (DMZ, Zeitz Instrumente, Augsburg, Germany) to a series resistance of 1–2 MΩ. Cells were voltage-clamped in the whole-cell configuration of the patch-clamp technique and held at a resting potential of −70 mV or −40 mV using an Axopatch 200 amplifier (Axon Instruments, Foster City, CA, USA). Potentials were not corrected for the junction potential, which was calculated to be ∼12 mV for our pipette filling solution. The SR and cytosolic DM-nitrophen were loaded with Ca²⁺ by a variable number of L-type Ca²⁺ currents with voltage steps to 0 mV or +10 mV lasting 200 ms. Currents were digitized at 3 kHz using an A/D converter and custom-written data acquisition software developed by us under LabView (National Instruments, Ennetbaden, Switzerland) running on an Apple Macintosh G3 computer. Membrane current and voltage data were stored on a hard disk for later analysis using IgorPro software (WaveMetrics, Lake Oswego, OR, USA).

Confocal Ca²⁺ imaging

Cells were imaged with a 40 × oil-immersion objective lens (Fluor, N.A. = 1.3; Nikon) and loaded with fluo-3 by dialysis through the recording pipette. Fluo-3 was excited with the 488 nm line of an argon-ion laser (ILT-5000, Ion Laser Technology, Salt Lake City, UT, USA) at 50–150 μW on the cell. The fluorescence was detected at 540 ± 15 nm with a confocal laser-scanning microscope (MRC 1000, Bio-Rad, Glattbrugg, Switzerland). The amplitude and the time course of cytosolic Ca²⁺ transients were computed off-line using customized versions of either NIH Image or Image SXM and expressed as normalized fluorescence (F/F₀). Resting [Ca²⁺] for experiments made in pipette filling solution (Fig. 1) was 20 nm and K_d was 400 nm (DelPrincipe et al. 1999). Mean Ca²⁺ concentration profiles were extracted from fluorescence images and calculated with IgorPro software using an established self-ratio calibration procedure (Cheng et al. 1993). Two-photon release signals were estimated as the maximal amplitude (mean of 5 points) minus the mean fluorescence during 100 ms prior to photolysis. Values were normalized to the maximal peak amplitude obtained in caffeine (Figs 5 and 6). Means of the normalized signals were plotted versus two-photon power and a sigmoidal function was fitted to the data. Under our conditions the fluo-3 fluorescence was not significantly quenched by the application of 20 mm caffeine (by 0.54 ± 0.16% of control, n = 2). Furthermore, the photolytic two-photon photolysis (TPP) Ca²⁺ signals remained unaffected by 20 mm caffeine.

Figure 1. Two-photon photolysis in droplet.

A, characteristic TPP power–response relationship obtained in pipette filling solution. By increasing the TPP power, more Ca²⁺ is released from the caged compound. B, when plotting the amplitude of the Ca²⁺ signal versus the photolytical power, data displayed a power dependence with an exponent of 1.8, suggesting a two-photon excitation process. C, horizontal spreading by diffusion of local Ca²⁺ signals of increasing amplitude. D, no strong dependence of the full width at half-maximal amplitude (FWHM) of the local Ca²⁺ signals versus power was found.

Figure 5. The amplitude of local Ca²⁺ signals is increased after β-adrenergic stimulation.

A, protocol used to obtain the power–response relationships in conditions similar to those of Fig. 4. The cells were held at −40 mV and diffraction-limited Ca²⁺ releases of increasing power were triggered in caffeine (Aa), in control (Ab), after a similar loading protocol and 2 min rest in Iso (Ac) and after 4 consecutive depolarizations in Iso (Ad ). B, line-scan images and averaged Ca²⁺ traces recorded during diffraction-limited releases. Ca²⁺ signals recorded in caffeine were used to estimate the amount of photolysis (Ba). Local Ca²⁺ signals were larger after 4 depolarizations in control (Bb) and further increased after Iso superfusion (Bc and Bd), independently of the global SR Ca²⁺ load (Fig. 4). C, normalized amplitude of Ca²⁺ signals for n = 6 cells in caffeine (•, 100% for the highest power), in control (▪, 120.4 ± 4.7%, n = 6), in Iso (▴, 144.5% ± 7.4%, n = 6) and after reloading in Iso (▾, 158.5% ± 9.0%, n = 6) are plotted versus power and fitted with sigmoidal functions. D, same data as in C but after subtraction of the photolytical component (signals in caffeine). CICR signals obtained in control (▪, 100%, n = 4, for the largest release) were significantly larger after Iso superfusion, whether cells were resting (▴, 164.6 ± 47.7%, n = 5) or being reloaded (▾, 184.2 ± 67.6%, n = 6).

Figure 6. Enlargement of local Ca²⁺ signals is not prevented by superfusion of CPA.

A, cells were held at −70 mV and a first TPP power–response relationship was triggered in caffeine to determine the amplitude of photoreleased Ca²⁺ signals (Aa). Cells were then reloaded by eight 200 ms depolarizing steps to +10 mV and after 1 min rest in control, a TPP power–response relationship was recorded (Ab). After 2 min superfusion in Iso another power–response relationship was measured (Ac), which was followed by 2 min of superfusion with Iso and CPA (10 μm), after which a last TPP power–response relationship was elicited (Ad). B, data obtained from one cell: Ba, line-scan images and traces obtained in caffeine; Bb, in control; Bc, in Iso; and Bd, in Iso and CPA. C, plot of the normalized amplitude of Ca²⁺ release versus TPP power in caffeine (•, 100% for the maximal trigger, n = 14), control (▪, 126.8 ± 10.3%, n = 14), in Iso (▴, 163.7 ± 25.7%, n = 14) and in Iso with CPA (♦, 173.7 ± 36.8%, n = 11). D, same data as in C but after subtraction of the photolytical component (control: ▪, 100%, n = 11; Iso: ▴, 226.4 ± 39.3%, n = 10; and Iso + CPA: ♦, 249.2 ± 66.1%, n = 9).

Global UV flash and local two-photon photolysis of caged Ca²⁺ compounds

Photolytic Ca²⁺ concentration jumps were elicited with UV flashes from a xenon short-arc flash lamp (duration ∼400 μs, discharged energy up to 230 J) (Kaplan & Ellis-Davies, 1988). UV flashes were applied in an epi-illumination arrangement to trigger global and spatially homogenous Ca²⁺ releases from the entire SR (for details see DelPrincipe et al. 1999). In order to generate highly localized and diffraction limited photolytic Ca²⁺ sources, we used TPP of caged Ca²⁺ (DM-nitrophen) (Lipp & Niggli, 1998; DelPrincipe et al. 1999). The beam of a mode-locked titanium sapphire laser (Mira 900, Coherent) tuned to 710 nm, with < 100 fs pulse length at 80 MHz repetition rate, was guided through the camera port of the confocal microscope to produce a stationary diffraction-limited spot within the myocyte, parfocal with the plane of fluorescence detection. The excitation point spread function was determined to extend over ∼710 nm (full width at half-maximal amplitude; FWHM) in the x–y-direction and 1200 nm in the z-direction (Lipp & Niggli, 1998). Photolysis was elicited by the opening of a mechanical shutter (Uniblitz, Vincent Associates, Rochester, NY, USA). The shutter opening was set for a duration of 60 ms for all experiments, and the interval between subsequent shutter openings was 430 ms. Both the shutter and the UV flash were synchronized to the pixel-clock of the laser scanner to coerce synchronization with the image acquisition and the voltage-clamp recording system. The power of the two-photon laser was attenuated by means of a neutral density filter and an adjustable linear polarization filter, both placed in series. In order to rapidly estimate the Ca²⁺ dependence of the CICR process and to analyse signals of comparable amplitude before and after β-adrenergic stimulation, a filter wheel holding nine neutral density filters (Lambda 10-2, Sutter Instruments, Novato, CA, USA) was placed in the optical pathway. The filter wheel was controlled by a Motorola 68HC12 microcontroller, running software written by us under the LEMPS development system (GIBB, Bern, Switzerland). Using this system, a TPP power–response relationship could be measured within 6 s, as shown in an experiment where DM-nitrophen was photolysed in a droplet of pipette filling solution (Fig. 1). Ca²⁺ signals elicited by TPP in pipette solution did not saturate and displayed a power dependence with an exponent of 1.8, suggesting a two-photon process. In addition, this relationship implies that the amplitude of the photolytic Ca²⁺ release signals is proportional to the Ca²⁺ release flux. The width (FWHM) increased only slightly, consistent with the minimal power dependence of the volume excited by TPP, as determined by recording the fluorescent point spread function (PSF) in fluoresceine (not shown). Each power–response plot includes 10 different power levels (as a percentage of full power): 16, 22, 25, 31, 42, 53, 68, 84, 93 and 100, respectively.

Statistics

Data are expressed as mean ±s.e.m, and n represents the number of analysed cells. Significance was tested with Student's t test and is denoted as * (P < 0.05) or ** (P < 0.02). For estimates of CICR (Figs 5 and 6), recordings in which local Ca²⁺ signals at maximal photolytic power were not larger in control than in the presence of caffeine (i.e. contained only photolytic Ca²⁺ release and no CICR component) were excluded from the analysis. Furthermore, data where the photolytic signal amplitude at maximal laser power was less then 2 times the noise (s.e.m.) of the resting Ca²⁺ signal were not taken into account.

Results

In initial experiments we verified that the entire β-adrenergic signalling and second messenger cascade were present and functional under the conditions of our experiments. Successful β-adrenergic stimulation was also confirmed at the end of each subsequent experiment. After 2 min of 1 μm isoproterenol (Iso) superfusion, the time needed to get a robust β-adrenergic stimulation in most cells, L-type Ca²⁺ currents elicited by depolarizations from −40 mV to +10 mV were considerably larger than in control (Fig. 2A). This resulted in an increase of the Ca²⁺ transient, as can be seen on the line-scan images and the traces averaged from the line-scans (Fig. 2B and C).

Figure 2. β-Adrenergic stimulation increases the amount of Ca²⁺ entry via L-type Ca²⁺ channels.

A, voltage protocol and recorded current. Left: cells were held at −40 mV and depolarized to +10 mV for 200 ms to activate L-type Ca²⁺ current. Right: same protocol after 2 min Iso superfusion. Entry of Ca²⁺ was larger thereby enhancing the subsequent CICR. B, line-scan images recorded during the depolarizing step. C, averages of the line-scan images in B, expressed as normalized fluorescence.

Activation of global cellular CICR with UV flash photolysis

In all subsequent experiments involving global cellular Ca²⁺ transients, we applied flash photolysis of caged Ca²⁺ to activate and examine CICR, since this is a trigger signal which itself is not affected by β-adrenergic stimulation, unlike the L-type Ca²⁺ current. To ensure a comparable and intermediate SR Ca²⁺ content, a specific loading protocol was carried out before each UV flash. It consisted of an initial SR emptying with a puff of caffeine, followed by a train of four L-type Ca²⁺ currents (depolarizing steps from −40 mV to 0 mV for 200 ms; Fig. 3A). To examine how CICR was affected by β-adrenergic stimulation while excluding the L-type Ca²⁺ current as the trigger signal, we applied UV flashes after 2 min at rest, either under control conditions or in the presence of Iso. Contrary to our expectations, the amplitude of the Ca²⁺ transient did not become larger after 2 min treatment with Iso. It even decreased to 62.3 ± 16.1% (n = 7, P < 0.05) of the control amplitude (Fig. 3Cb and D). This decline was even more surprising since loading protocols with trains of L-type Ca²⁺ currents were carefully kept identical for both, control and Iso-treated cells, to avoid any alterations of SR Ca²⁺ content (Fig. 3Bi and ii).

Figure 3. Whole-cell Ca²⁺ release signals are smaller after rest in Iso.

A, voltage protocol. Caffeine (20 mm; arrowhead), briefly applied to empty the SR, was followed by 4 consecutive depolarizations from −40 mV to 0 mV for 200 ms to moderately load the SR (Ai). After 2 min a UV flash was triggered to induce a global and homogenous Ca²⁺ release from the SR (Aa) and immediately followed by another short application of caffeine. Four similar depolarizations were again applied (Aii) and followed by superfusion of Iso. After 2 min rest another UV flash (Ab) was triggered, followed by a short application of caffeine. Finally, 4 depolarizations were applied during β-adrenergic stimulation (Aiii) and followed by a UV flash (Ac). B, the third L-type Ca²⁺ current of a train of four consecutive depolarizations: Bi, in control: Bii, in control and before Iso application for 2 min: and Biii, during β-adrenergic stimulation. Line indicates 0 pA. C, line-scan images acquired during the UV flashes. Ca, control; Cb, resting in Iso; and Cc, stimulated in Iso. D, normalized amount of extruded Ca²⁺ (∫I_NCX) in control (100%), after rest in Iso (62.3 ± 16.1%, n = 7, P < 0.05) and after reloading in Iso (160.4 ± 45.4%, n = 7). E, decays of global fluorescence signals were fitted with a monoexponential function and τ was normalized to control. As expected CPA prolonged the decay to 124 ± 12.5%, n = 3 and to 136 ± 23.7%, n = 4, in the continued presence of CPA and Iso.

Thus we wondered in what respect our flash photolytic experiments differed from the more physiological situation relying on L-type Ca²⁺ currents as triggers for CICR. First, our trigger signal was, on purpose, not increased by Iso. Second, we had chosen conditions to avoid exposing the SR to augmented Ca²⁺ currents during the loading protocol, in order to maintain the SR Ca²⁺ loads similar for both experiments. Could it be that the larger L-type Ca²⁺ currents are a prerequisite to obtain more release from the store, by either representing a more efficient trigger signal for CICR or by providing more Ca²⁺ influx for reloading the SR? To answer these questions and to distinguish between the two possibilities we slightly modified our loading protocol and started the application of Iso 2 min before performing the SR Ca²⁺ loading protocol. Or, in other words, we loaded the SR with larger Ca²⁺ currents (Fig. 3Biii) while keeping the trigger signal constant (i.e. the photolytic [Ca²⁺] jump). As it turned out, the amount of Ca²⁺ released was indeed larger than in control (160.4 ± 45.2%, n = 7; Fig. 3C and D). In addition to the increased L-type current after β-adrenergic stimulation used to reload the SR, SERCA stimulation could be seen during the decay of the signals. The maximal rate of decay increased to 137 ± 11.3% (n = 7) after 2 min rest in Iso and to 118 ± 12.6% (n = 7) after reloading in Iso. In guinea-pigs, cytosolic Ca²⁺ removal is more dependent on Na⁺–Ca²⁺ exchange than, for example, in rat (Sham et al. 1995). Thus the effect of SERCA stimulation is less pronounced. Finally, as reported recently (Ginsburg & Bers, 2004), the maximal rate of Ca²⁺ release was indeed larger when cells were reloaded in Iso (147.8 ± 53.4%, n = 7). As expected, superfusion of 10 μm CPA prolonged the decay of the Ca²⁺ signal to 124 ± 12.5% and to 136 ± 23.7% in the presence of CPA and CPA plus Iso, respectively (Fig. 3E). But how can we explain the reduced Ca²⁺ signal amplitude after β-adrenergic stimulation, despite the identical SR Ca²⁺ loading protocol? First, run-down of L-type Ca²⁺ channels, occuring within a time window of 20–30 min, could not be held responsible for the decreased loading as shown in Fig. 3Bi and ii. Smaller Ca²⁺ transients could also result from a decrease in the Ca²⁺ sensitivity of the RyRs or from an acceleration of the SR Ca²⁺ loss occurring during the 2 min rest in Iso. The first possibility seemed less likely after β-adrenergic stimulation. To evaluate changes of SR Ca²⁺ content occurring during the resting period, we used the same loading protocol as above, but instead of triggering CICR with UV flashes, short puffs of caffeine were applied to completely empty the SR (Fig. 4). The resulting membrane currents reflecting electrogenic Ca²⁺ removal via the Na⁺–Ca²⁺ exchanger (I_NCX) were integrated to estimate SR Ca²⁺ content (Trafford et al. 1998). Interestingly, the SR contained significantly less Ca²⁺ when resting for 2 min in Iso (58.6 ± 7.3%, n = 6, P < 0.02) than when resting in control solution (100%, n = 10; Fig. 4C and D). However, when the cells were Ca²⁺ loaded with L-type Ca²⁺ currents increased by β-adrenergic stimulation, the SR Ca²⁺ content was raised to 123.0 ± 11.7% (n = 10, P < 0.05). Thus, the amplitude of the global Ca²⁺ signals observed after triggering CICR with identical photolytic triggers mainly reflected the SR Ca²⁺ content prevailing under the various conditions. The reduced SR Ca²⁺ content after a resting period in Iso is an interesting finding by itself and could be the consequence of an increased SR Ca²⁺ leak, possibly resulting from an enhanced Ca²⁺ sensitivity of the RyRs after phosphorylation (Marx et al. 2000). This leak was presumably small since no significant change in resting [Ca²⁺] was measured between control and Iso (see below).

Figure 4. SR Ca²⁺ content governs the amplitude of whole-cell Ca²⁺ signals.

A, protocol used to estimate total SR Ca²⁺ content. Short puffs of caffeine were applied in control (Aa), after a similar loading protocol and 2 min rest in Iso (Ab) and after reloading during β-adrenergic stimulation (Ac). B, third L-type Ca²⁺ current of a train of four consecutive depolarizations: Bi, in control; Bii, in control and before Iso application for 2 min; and Biii, during β-adrenergic stimulation. C, caffeine-induced NCX currents recorded in the 3 different conditions described above (Ca, Cb and Cc, respectively) were integrated to estimate the SR Ca²⁺ load. D, normalized ∫I_NCX: control (100%, n = 10), resting in Iso (58.6 ± 7.3%, n = 6, P < 0.02) and stimulated in Iso (123 ± 11.7%, n = 10, P < 0.05).

Local activation of CICR with two-photon photolysis

We have previously observed that Ca²⁺ refilling of the SR occurs much faster after local activation than after global activation of CICR, presumably because a rapid redistribution of Ca²⁺ within the SR network takes place after functional Ca²⁺ depletion of a highly confined Ca²⁺ release unit only (i.e. one junctional SR structure near a dyad) (DelPrincipe et al. 1999). Thus, localized Ca²⁺ release might be expected to behave differently than global cellular signals, possibly by being less dependent on the global SR Ca²⁺ content. Thus, local signals might be more sensitive to changes of RyR gating. Therefore, we next elicited highly localized photolytic Ca²⁺ signals by TPP of DM-nitrophen (Fig. 5), while following an SR loading protocol analogous to the one used for UV flash experiments (Fig. 3). TPP power–response relationships in six cells (Fig. 5B) were recorded in order to obtain local Ca²⁺ signals of varying amplitudes under each condition (i.e. in caffeine, in the presence or absence of Iso, at low and high SR Ca²⁺ load). Signals of various amplitudes were required for later comparison of the signals and their properties, which may depend on the signal amplitude itself. As in our previous studies using two-photon photolytic activation of CICR (Lipp & Niggli, 1998; DelPrincipe et al. 1999), the first series of TPP power–response relationships was obtained in caffeine to estimate the photolytic component of the Ca²⁺ signals (normalized to the highest two-photon power applied, n = 6). Caffeine, in our hands, did not interfere with the TPP signals and the quench of fluo-3 fluorescence was negligible (−0.5 ± 0.2%, n = 2). After reloading the SR with four depolarizing steps another TPP power–response relationship was recorded in control solution (120.4 ± 4.7%, n = 6) and immediately followed by a short caffeine application to empty the SR. After SR reloading and a subsequent resting period of 2 min in Iso, another TPP power–response relationship was recorded. During these 2 min at rest, the SR Ca²⁺ content was known to decay more in Iso than in control solution, as we found above (see Fig. 4). Surprisingly however, despite reduced SR Ca²⁺ content and much unlike global Ca²⁺ signals, the local Ca²⁺ release signals were larger during β-adrenergic stimulation than in control (144.5 ± 7.4%, n = 6). When the SR was reloaded during β-adrenergic stimulation to an extent that was above control, only a small and not significant further increase of the signal amplitude was found, despite the higher SR Ca²⁺ load under these conditions (to 158.5 ± 9.0%, n = 6). The weak effect of the reloaded SR on the local signals was expected since cells were kept at low SR Ca²⁺ load to avoid the appearance of regenerative Ca²⁺ signal spreading. A more prominent and significant increase in the signal amplitude has been observed with higher SR Ca²⁺ loads, such as after 12 prepulses (Lipp & Niggli, 1998). To estimate the Iso effect on SR Ca²⁺ release the photolytical component obtained in caffeine was subtracted from the total signals obtained for each trigger in control and in Iso at low and high SR Ca²⁺ load. After normalization to the CICR amplitude obtained for the maximal trigger in control (100%, n = 4), data were plotted versus power (Fig. 5D) and fitted with a sigmoidal function. Interestingly, CICR started at lower power levels and was larger in Iso whether the SR was depleted (164.6 ± 47.7%, n = 5) or reloaded (184.2 ± 67.6%, n = 6). Taken together, these findings suggest that β-adrenergic stimulation with Iso enhanced local SR Ca²⁺ release irrespective of global SR Ca²⁺ content, much unlike the behaviour of global Ca²⁺ transients triggered with UV flashes.

What could be responsible for this discrepancy between global and local Ca²⁺ signals? Could we derive conclusions regarding the underlying mechanism from this peculiar behaviour? First we excluded possible Ca²⁺ sources other than the SR by confirming the ryanodine sensitivity of the signal amplification by Iso (data not shown). Given that the source of the Ca²⁺ seemed to be the SR, only two basic possibilities remained: (1) either the SR Ca²⁺ content was elevated above the remainder of the SR network at the site where we performed photolysis or (2) local Ca²⁺ release from the SR was larger despite the reduced SR Ca²⁺ content and regardless of the smaller Ca²⁺ gradient across the SR membrane. The latter possibility would require a change of the gating or Ca²⁺ permeability of the RyRs. Because of the slow kinetics of the TPP (Ca²⁺ source lasting for 60 ms) one could imagine that some of the photoreleased Ca²⁺ could be locally transported into the SR. The extent of this local uptake would depend on the activity of the SERCA. When more Ca²⁺ is pumped into the SR, the local SR Ca²⁺ content could increase above the remainder of the SR network, which could explain our observations of enhanced local Ca²⁺ releases from the SR in the presence of Iso. To test for the possibility of local Ca²⁺ uptake we applied CPA to inhibit the SERCA. CPA would prevent the local amplification of the TPP Ca²⁺ signals if it were mediated by the SERCA. Thus, the following experiments were carried out (Fig. 6): a first TPP power–response relationship was obtained during superfusion of caffeine. After SR reloading another series of Ca²⁺ signals was recorded, and this was repeated after treatment with Iso. Again, local Ca²⁺ signals were larger during β-adrenergic stimulation than in control and the signals obtained in the presence of functioning CICR were significantly larger than in caffeine, as expected. Finally, the SERCA was blocked by 2 min superfusion of 10 μm CPA in the presence of Iso. As it turned out, CPA could not prevent the amplification of CICR by β-adrenergic stimulation (Fig. 6B). Figure 6C presents data obtained from 14 cells and normalized to the local release signal recorded in caffeine for the maximal TPP power (100%). The Ca²⁺ signal recorded in control was 126.8 ± 10.3%, demonstrating the presence of the biological signal amplification by CICR. The signal obtained during Iso superfusion had an amplitude of 163.7 ± 25.7% (n = 14), and in the presence of both CPA and Iso, the amplitude was 173.7 ± 36.8% (n = 11). After subtraction of the photolytical Ca²⁺ release component, it became evident that CICR was increased to 226.4 ± 39.3% (n = 10) by Iso and remained unaffected at 249.2 ± 66.1% (n = 9) when the SERCA was inhibited (Fig. 6D). This set of data suggests that CPA could not suppress the amplification of local CICR by β-adrenergic stimulation. To confirm successful inhibition of the SERCA by CPA in these experiments, decay rates after UV flash-triggered Ca²⁺ transients were analysed (Fig. 3E) confirming successful inhibition of the SERCA by CPA.

Finally changes in resting Ca²⁺ were not significant between control and Iso (2.5 ± 5.3%, n = 14) or between control and Iso + CPA (7.7 ± 10.5%, n = 11), suggesting that cytosolic Ca²⁺ did not interfere with the increased local signals. In summary, these findings indicate that local loading of the SR with Ca²⁺ did not play an important role in the amplification of the TPP release signals by β-adrenergic stimulation. Thus, the increased TPP Ca²⁺ fluorescent signals were resulting from more Ca²⁺ being released from the SR despite reduced Ca²⁺ content.

Discussion

The notion of a cardiac Ca²⁺ signalling system relying on the recruitment of functionally independent elementary signalling events (Ca²⁺ sparks) has opened the door for a new concept of EC coupling incorporating a probabilistic paradigm, whereby each Ca²⁺ spark event is triggered under local control by an L-type Ca²⁺ channel with a specific non-zero probability (Cheng et al. 1993; Lipp & Niggli, 1998; Niggli, 1999). Such an arrangement not only allows for a new understanding of modulatory changes of EC coupling, but also for a novel mechanism of EC coupling failure under pathological conditions, whereby the Ca²⁺ spark trigger probability could be altered by modulatory, metabolic and pathophysiological mechanisms (Gomez et al. 1997; Marx et al. 2000; Pacher et al. 2002; Isaeva & Shirokova, 2003). Two mechanisms are thought to be very important for this type of regulation: the Ca²⁺ content of the SR (Terentyev et al. 2003) and the phosphorylation of various Ca²⁺ signalling proteins (Takasago et al. 1991; Simmerman & Jones, 1998; Marx et al. 2000). These two regulatory pathways are expected to exhibit a high degree of cross-talk because of at least two reasons: (1) functional changes of Ca²⁺ signalling proteins after phosphorylation will affect SR Ca²⁺ content (Hussain & Orchard, 1997); (2) changes of Ca²⁺ concentrations may alter the extent of phosphorylation, for example through the CaMKII pathway.

In the present study we used a combination of biophysical techniques in an attempt to dissect these two regulatory pathways, for example by using photolysis of caged compounds as a trigger for CICR to remove some variables (e.g. changes of I_Ca after phosphorylation of L-type Ca²⁺ channels). With this combination of approaches we were able to simplify the system, eliminate some variables while controlling and measuring others. By comparing features of global (i.e. whole-cell) Ca²⁺ fluorescent signals with those of subcellularly localized signals (i.e. diffraction-limited two-photon photorelease signals) we could derive conclusions about the relative significance of either mechanism for CICR (i.e. the SR Ca²⁺ load or the phosphorylation of Ca²⁺ signalling proteins) under the two conditions.

Whole-cell signals are dominated by SR Ca²⁺ load

It is known that an increase in luminal Ca²⁺ can enhance the Ca²⁺ sensitivity on the cytosolic side of RyRs reconstituted in lipid bilayers (Sitsapesan & Williams, 1997). This finding may also be related to the generation of spontaneous Ca²⁺ sparks and Ca²⁺ waves in isolated cardiac myocytes under conditions of SR Ca²⁺ overload (Wier et al. 1987; Lukyanenko et al. 1999; Terentyev et al. 2003). In contrast, a decrease of SR Ca²⁺ content is thought to control termination and refractory behaviour of CICR in cardiac myocytes (DelPrincipe et al. 1999; Sobie et al. 2002; Terentyev et al. 2002; Szentesi et al. 2004). Our data show that whole-cell Ca²⁺ signals are largely governed by the SR Ca²⁺ load, a feature which possibly masks subtle effects due to changes of RyR Ca²⁺ sensitivity arising from phosphorylation. Indeed, irrespective of how many RyRs open before or after their phosphorylation, only a specific releasable fraction of SR Ca²⁺ will be liberated into the cytosol (previously determined to be typically 57% under our conditions (DelPrincipe et al. 1999)). Thus, the number of opening RyRs will only affect the kinetics of SR Ca²⁺ release, but not increase the amount of Ca²⁺ that can be released (unless CICR would be more exhaustive after RyR phosphorylation, for which there is little evidence). Thus, with or without β-adrenergic stimulation, the extent of whole-cell SR Ca²⁺ release is dominated (i.e. limited) by SR Ca²⁺ content and by processes which terminate CICR by functional SR Ca²⁺ depletion.

Local signals reveal effects of β-adrenergic stimulation

Interestingly, the present study revealed that subcellularly localized SR Ca²⁺ signals exhibited a strikingly different behaviour than whole-cell signals. Even though local Ca²⁺ signals also depend somewhat on the SR Ca²⁺ load, particularly at more elevated SR Ca²⁺ loads (Lipp & Niggli, 1998; DelPrincipe et al. 1999), the localized Ca²⁺ releases are not expected to significantly lower the global SR Ca²⁺ content. Since SR Ca²⁺ release only occurs from one or a few functional units of the SR, we think that local depletion of intra-SR Ca²⁺ will be less pronounced, since neighbouring SR sites do not release and Ca²⁺ can, in fact, rapidly diffuse to the active sites within the SR network. Thus, under these conditions more open channels allow more Ca²⁺ to be released (Sobie et al. 2002), even at reduced SR Ca²⁺ content. Thus, the declining global SR Ca²⁺ load no longer terminates the release flux, contrary to the whole-cell activation. Thus, TPP signals might be a suitable technique to allow observation of localized SR Ca²⁺ release process modulation.

Estimates of CICR

Subtraction of fluorescent signals obtained in caffeine from signals in control, Iso or Iso and CPA was used to estimate the amount of CICR. As seen in Figs 5D and 6D, a threshold seems to appear at about 70–80% of the maximal trigger power in control, while it is shifted toward smaller triggers in Iso or in Iso and CPA. For the maximal trigger, Iso approximately doubled the caffeine-insensitive component (165 to 249%); as for the lowest trigger power range no clear separation could be made. This suggests that very low photolytic power may not detectably increase the open probability of the RyRs at the low SR Ca²⁺ load used in this study, thereby triggering no CICR. Finally, the sigmoidal function fit to the data point tends to become more shallow and saturate. In the droplet, saturation occurs at very high powers and is thought to result from depletion of DM-nitrophen within the diffraction limited excited volume. In a droplet this is expected to occur at higher photolytic power levels than in cells, because DM-nitrophen diffusion may not be entirely free in the cytosol. In addition, inside living cells many other factors may also contribute to saturation and shallow release functions, including mobile and stationary Ca²⁺ buffers and saturation of CICR.

Conclusion

In conclusion, we found that when CICR was activated in a synchronized way throughout the cell, the Ca²⁺ signals were predominantly governed by SR Ca²⁺ content, thus defining the amplitude of the resulting Ca²⁺ transient. Although β-adrenergic stimulation may lead to synchronization of SR Ca²⁺ release channels (Song et al. 2001; Viatchenko-Karpinski & Györke, 2001; Ginsburg & Bers, 2004), this will not necessarily affect the amount of global Ca²⁺ release from the SR. This observation on the global level is therefore consistent with the notion that whole-cell SR Ca²⁺ release is terminated by emptying of the Ca²⁺ store, leading to deactivation of the RyRs by reducing their Ca²⁺ sensitivity (DelPrincipe et al. 1999; Terentyev et al. 2002; Szentesi et al. 2004). However, when CICR was activated on a local level (i.e. with TPP), the system behaved in a completely different and apparently opposite way, that cannot be explained by global changes in SR Ca²⁺ content. Insensitivity of these unexpected signals to inhibitors of the SERCA also suggested that these effects were not mediated by increased local SR Ca²⁺ loading via the SERCA that could, in principle, occur before and during the photorelease process. Increased local Ca²⁺ releases were thus observed regardless of the decreased SR content, suggesting a direct modulatory effect of the β-adrenergic signalling cascade on the SR Ca²⁺ release pathway. Increased Ca²⁺ release from the SR despite reduced Ca²⁺ content can only occur when the open probability of the RyRs is somehow increased, either by activation of more channels or by changing the gating properties of the channels (or both). Based on these observations, we conclude that increased local Ca²⁺ release signals are the result of the PKA-dependent phosphorylation of a Ca²⁺ signalling protein other than the L-type Ca²⁺ channels, phospholamban and the SR Ca²⁺ pump, and possibly the ryanodine receptors themselves (or other SR proteins).