Pacing-induced calcineurin activation controls cardiac Ca²⁺ signalling and gene expression

Abstract

Calcineurin, a Ca²⁺–calmodulin-dependent protein phosphatase (PP2B) is one of the links between Ca²⁺ signals and regulation of gene transcription in cardiac muscle. We studied the Ca²⁺ signal specificity of calcineurin activation experimentally and with modelling. In the rat atrial preparation, an increase in pacing frequency increased nuclear activity of the calcineurin-sensitive transcription factor, nuclear factor of activated T-cells (NFAT), 2-fold in a cyclosporin A (CsA)-sensitive manner. In line with this, modelling results predicted that the frequency of cardiac Ca²⁺ transients encodes the stimulus for calcineurin activation. We further observed experimentally that calcineurin inhibition by CsA modulated Ca²⁺ release in a Ca²⁺-dependent manner. CsA had no effect on [Ca²⁺]_i at a pacing frequency of 1 Hz but it significantly suppressed the amplitude of Ca²⁺ transients, systolic [Ca²⁺]_i and time averaged [Ca²⁺]_i at 6 Hz. Calcineurin had a differential role in the expression of immediate-early genes B-type natriuretic peptide (BNP) and c-fos. CsA inhibited the pacing-induced BNP gene expression, whereas pacing alone had no effect on the expression of c-fos. However, in the presence of CsA, c-fos mRNA levels were significantly augmented by increased pacing frequency. These results show that frequency-dependent calcineurin activation has a specific role in [Ca²⁺]_i regulation and gene expression, constantly recruited by varying cardiac Ca²⁺ signals.

Muscle cells are capable of changing their structure and function in response to altered activity. Research in recent years shows that Ca²⁺ has a central role in this adaptation process (for review see Molkentin, 2000; Wilkins & Molkentin, 2002; Frey & Olson, 2003). One suggested link between the free myoplasmic Ca²⁺ concentration ([Ca²⁺]_i) and altered gene expression is the Ca²⁺–calmodulin-dependent protein phosphatase-2B, calcineurin (Timmerman et al. 1996; for review see Rusnak & Mertz, 2000). Activated calcineurin mediates nuclear translocation of cytosolic nuclear factor of activated T-cells (NFAT), which controls transcription co-operatively with other transcription factors, e.g. activator protein-1 (AP-1), myocyte enhancer factor and GATA-4 (Molkentin et al. 1998; Wu et al. 2000; Macian et al. 2000; for review see Rao et al. 1997).

Originally it was shown that transgenic mice with constitutively active calcineurin develop cardiac hypertrophy and express fetal cardiac genes (Molkentin et al. 1998), making the calcineurin–NFAT pathway an attractive candidate for coupling of Ca²⁺ signals to cardiac gene expression. Moreover, overexpression of an endogenous calcineurin inhibitory protein, modulatory calcineurin-interacting protein-1 (MCIP1), inhibits hypertrophy induced by overexpression of constitutively active calcineurin or chronic administration of β-adrenoreceptor agonist (Rothermel et al. 2001), indicating that activation of calcineurin is critical for the development of the hypertrophy. However, pharmacological inhibition of calcineurin in a variety of rodent models of heart disease has produced controversial results. Calcineurin inhibitors (CsA or FK506) have been reported to inhibit the load-induced hypertrophy (Shimoyama et al. 1999; Zou et al. 2000) or to have no effect (Ding et al. 1999). CsA has even been found to escalate the development of the mouse cardiomyopathy induced by a myosin heavy chain mutation (Fatkin et al. 2000). Furthermore, calcineurin inhibition is not selective to the pathological hypertrophy, since both CsA (Eto et al. 2000) and MCIP1 (Rothermel et al. 2001) also suppress the favourable hypertrophic adaptation to exercise. One possible explanation for these conflicting results could be that the effect of calcineurin inhibition depends on the hypertrophic model used. For example, lack of MCIP1 in mice heart results in an increased hypertrophic response to overexpression of constitutively active calcineurin, but a reduced hypertrophic response to pressure overload and to adrenergic stimulation (Vega et al. 2003), suggesting that calcineurin may have a different role depending on the stimulus inducing the hypertrophy.

While there is strong evidence from genetic mouse models to support the involvement of calcineurin in the development of various forms of hypertrophy (for reviews see Molkentin, 2000; Wilkins & Molkentin, 2002; Frey & Olson, 2003), very little is known about the normal Ca²⁺ activation of calcineurin and its immediate functional implications. Because the frequency of Ca²⁺ transients encodes an adequate stimulus for calcineurin activation in skeletal muscle (Liu et al. 2001; Kubis et al. 2002), we first examined if cardiac calcineurin is activated by pacing-induced [Ca²⁺]_i changes in rat atrium. Secondly, since it was reported that calcineurin inhibitors (McCall et al. 1996; Janssen et al. 2000) and/or activated calcineurin (Bandyopadhyay et al. 2000; Münch et al. 2002) regulate the [Ca²⁺]_i balance in cardiac myocytes, we studied the Ca²⁺-dependent effects of CsA on the myocyte Ca²⁺ signalling. Thirdly, the findings that in several cell types the calcineurin–NFAT cascade has been implicated in both activation (Rao et al. 1997; Molkentin et al. 1998; Macian et al. 2000) and suppression of the expression of immediate genes (Su et al. 1996; Bito et al. 1996; Schaefer et al. 1998) led us to examine if the calcineurin-dependent transcriptional pathway shows dual effects with respect to the expression of immediate-early genes.

Methods

Preparation and perfusion of rat atrial appendix

The rat atrial appendix preparation was set up and perfused as previously described (Tavi et al. 1999). The Animal Use and Care Committee of the University of Oulu approved the experimental design. Male Sprague-Dawley rats weighing 290–400 g were used. The rats were decapitated, and the hearts were rapidly removed and placed in oxygenated (ca 10°C) Tyrode buffer solution (mm): 113.8 NaCl; 17.6 NaHCO₃; 4.7 KCl; 2.0 CaCl₂; 1.1 MgSO₄; 1.2 KH₂PO₄; 11.0 glucose; and 10 μIU ml⁻¹ insulin; pH was 7.4 when bubbled with 5% CO₂–95% O₂ gas. The same solution was used for superfusion of the atrium (2.5 ml min⁻¹) at 37°C. The atria were paced with a field stimulus (1 ms, 50% over threshold) by two platinum electrodes located inside the perfusion chamber. CsA (1 μm, Sigma-Aldrich) was applied to the perfusion medium 30 min before the change of the pacing frequency. For mRNA measurements the atria were preincubated for 50 min with 1 Hz pacing, followed by 30 min pacing at 1, 4, 5, or 6 Hz. The tissue was frozen in liquid nitrogen and stored at –70°C.

[Ca²⁺]_i measurements

For indo-1AM loading, the atria were superfused for 25–40 min at 25–32°C (flow 7 ml min⁻¹) with Tyrode solution (volume 4 ml) containing (mm): 103.8 NaCl; 17.6 NaHCO₃; 4.7 KCl; 2.0 CaCl₂; 1.1 MgSO₄; 1.2 KH₂PO₄; 10 glucose; 9.1 sodium pyryvate. Indo-1 AM (Molecular Probes, Europe) was dissolved in DMSO with 20% Pluronic. The final concentration of indo-1 AM was 20 μm and the amount of DMSO/Pluronic was 25 μl ml⁻¹ Tyrode solution. The solution was bubbled with 5% CO₂–95% O₂ (pH 7.4). The Ca²⁺ measurement setup has been previously described (Tavi et al. 1998, 1999). The Indo-1 emission ratio (405 nm/485 nm) was acquired at 200 Hz with two photomultiplier tubes (Hamamatsu, Japan) and filtered at 100 Hz. Both excitation and emission were guided to and from the tissue with a quartz fibre optic cable (diam. 1 mm). The distance between the tissue and the cable head was constant (∼3 mm) and the illuminated area covered 100% of the tissue area facing the cable. Therefore the amount of the Indo-1 molecules within the illuminated area was constant. For calibration of the measured signal, R_min and R_max were determined at the end of each experiment. To get R_min the atria were perfused with Ca²⁺-free buffer (0 Ca²⁺, 0.5 Mg²⁺, 5 mm EGTA) containing 10 μm of the Ca²⁺ ionophore 4-Br-A23187 (Molecular Probes) until a steady state fluorescence signal was obtained (10–12 min). This was followed by perfusion with high Ca²⁺ solution (15 mm) with 10 μm Ca²⁺ ionophore. The atria was then exposed to high frequency stimulation (8–10 Hz), resulting in a tetanic-like contraction within 10 min and a steady-state value of R_max. To evaluate the degree of loading of the non-cytosolic compartments, 1 mm of Mn²⁺ was applied at the end of calibration procedure. The residual fluorescence after Mn²⁺ was 6 ± 2% (n = 6) representing the maximal error caused by non-cytosolic fluorescence. To estimate the non-myocyte fluorescence, we induced Ca²⁺ release in endothelial cells with 10 μm bradykinin (Field et al. 1994), which caused no detectable change in the fluorescence (n = 4), indicating that either the tissue does not contain endothelial cells or these cells were not loaded with indo-1. The [Ca²⁺]_i was calculated by using Grynkiewicz formalism:

(1)

where K_d is the dissociation constant for indo-1 and β is the ratio of the free/bound indo-fluorescence at 485 nm from measured R_min and R_max. We used a K_d value of 844 nm determined in cellular environment of cardiac myocytes (Bassani et al. 1995).

Isolation of mRNA and quantitative PCR

Total RNA was prepared from rat atria and used as a template for the cDNA first strand synthesized by M-MuLV reverse transcriptase. The quantitative PCR reactions were performed with an ABI 7700 Sequence Detection System using the TaqMan chemistry. The forward and reverse primers for rat b-type natriuretic peptide (BNP) mRNA detection were TGGGCAGAAGATAGACCGGA and ACAACCTCAGCCCGTCACAG, and for c-fos GGCTGAACCCTTTGATGACTTC and GGGCAGTCTCCGAGCCA, respectively. The bifunctional fluorogenic probes for BNP and c-fos were 5′-Fam-CCAAGCGACTGACTGCGCCG-Tamra-3′ and 5′-Fam-TGTTTCCGGCATCATCTAGGC-Tamra-3′, respectively. The results were normalized to 18S RNA quantified from the same samples as previously described (Majalahti-Palviainen et al. 2000).

Oligonucleotides and EMSA

All oligonucleotides were purchased from Sigma Chemical Co. For electrophoretic mobility shift assay (EMSA), the NFAT binding element located at –927 bp BNP promoter (NFAT-BNP, Molkentin et al. 1998) was used as probe and intact and mutated NFAT binding elements of the interleukin-2 promoter (NFAT-IL-2 and NFATmut-IL-2, respectively, Northrop et al. 1994) were used as unlabelled competitor oligonucleotides (coding strand shown, point mutations in bold and 5′-overhangs in italics): NFAT-BNP, 5′-AGAGCTATCCTTTTGTTTTCCATCCTGGCCC-3′; NFAT-IL-2, 5′-AGAGCGCCCAAAGAGGAAAATTTGTTTCATAGCCC-3′ and NFATmut-IL-2: 5′-AGAGCGCCCAAAGCTTAAAATTTGTTTCATAGCCC-3′. The sense and corresponding antisense oligonucleotides were annealed to generate double-stranded oligonucleotides. NFAT-BNP was sticky-end-labelled with [³²P]-dCTP by Klenow enzyme. Nuclear extracts from frozen auricular tissue were prepared (Hautala et al. 2001) and protein concentration from each sample was colourimetrically determined (Bio-Rad Laboratories). For each reaction mixture, 12 μg of nuclear protein and 3 μg of poly(dI-dC) were used in a buffer containing 10 mm Hepes (pH 7.9), 1 mm MgCl₂, 50 mm KCl, 1 mm DTT, 1 mm EDTA, 10% glycerol, 0.025% NP-40, 0.25 mm PMSF and 1 μg ml⁻¹ each of leupeptin, pepstatin and aprotinin. Reaction mixtures were incubated with a labelled probe for 15 min followed by non-denaturating gel electrophoresis on 5% polyacrylamide gel. Subsequently, gels were dried and exposed in a PhosphorImager screen and analysed with ImageQuant (Molecular Dynamics, Amersham Biosciences, CA, USA). To confirm DNA sequence specificity of the protein–DNA complex formation, competition experiments with 100 m excesses of unlabelled oligonucleotides with intact or mutated NFAT binding sites were performed. Competitor oligonucleotides were added to the reaction mixture 15 min before the labelled probe.

Model for calcineurin Ca²⁺ activation

We utilized a previously published reaction scheme for Ca²⁺- and calmodulin-dependent activation of calcineurin (CaN) (Bhalla & Iyengar, 1999), where parameters have been adjusted to better meet the characteristics of cardiac myocytes. The reactions and the corresponding rate constants of the model are shown in Fig. 1A. All the reactions were incorporated into Mathlab (Mathworks, USA) as time-dependent differential equations of concentration and they were solved using the well-stirred assumption, which states that each molecule has equal access to each other in a single compartment. As initial concentrations we have used 6 μm for calmodulin ([CaM]) (Shannon et al. 2000) and 1 μm for calcineurin ([CaN]) (Bhalla & Iyengar, 1999; Crabtree, 1999) while the other concentrations being zero at the beginning. The reaction rates were taken initially from the previous model (Bhalla & Iyengar, 1999) and the equilibrated CaN-activation was calculated at different Ca²⁺ concentrations, which allowed plotting of the CaN activation against [Ca²⁺] (Fig. 1B) and data points were fitted to the Hill equation:

(2)

We compared the parameters of our fitted Hill equation with the published experiments (n = 3, K_d= 0.6 μm, Stemmer & Klee, 1994) and adjusted the reaction rate constants (see Fig. 1A) by re-running the simulations until the Hill coefficients matched. After adjusting the parameters, the responses to three Ca²⁺ concentration steps of different magnitude were simulated and the time constants of each calcineurin activation/inactivation were determined (Fig. 1C and D). Thereafter, experimentally determined Ca²⁺ signals (sampling interval 0.005 s) were used as stimuli for the model. The signals were first smoothed with a three-point moving average filter, after which linear interpolation was used to provide continuous signals as input. To solve the differential equations, we used the variable-step Mathlab routine ‘ode45’, based on a fourth order explicit Runge-Kutta formula (Dormand & Prince, 1980) with an average time step of 0.003 s.

Figure 1. Characteristics of the calcineurin model.

A, reactions with forward and reverse rates included into the model. CaNAB represents calcineurin A and B subunits, which are always assumed to be bound under physiological conditions. The four-Ca²⁺-bound form of calcineurin (CaNAB-Ca₄) has a small activity, but it is negligible compared to the CaM-bound forms. For clarity, Ca²⁺ is represented multiple times in the scheme, but it is modelled only as a single reactant. All the CaM-bound forms of calcineurin have been assumed to have the same enzymatic activity, thus the CaN activity can be obtained by simply summing the concentrations of each form. B, steady-state activation of the model. The Hill equation (continuous line) has been fitted to simulated steady-state levels of active calcineurin at different [Ca²⁺] (○). C, activation time constants of calcineurin in the model upon steps (from 0.2 μm to 0.4, 0.6 or 0.80 μm [Ca²⁺]. D, deactivation time constants by corresponding Ca²⁺ stimuli (as in C).

Artificial Ca²⁺ signals were constructed to study the frequency and base-line dependence of the CaN-activation. The experimentally determined 1, 4, 5 and 6 Hz signals were used to determine the shape of an average Ca²⁺ transient for each frequency. These average transients were then repeated a number of times to form oscillatory signals, which were finally scaled to the same amplitude.

Statistical testing

Results are expressed as mean ± s.e.m. The statistical testing was done with one-way and two-way anova. Pairwise comparisons between different groups were done by Student–Newman-Keuls method (SigmaStat, USA). For comparison between multiple groups, Bonferroni correction was applied. P values smaller than 0.05 were considered statistically significant. The data analysis was done with Origin 6.1 (OriginLab) and equations were solved with Mathlab.

Results

Pacing-induced changes in [Ca²⁺]_i

Increase of the pacing frequency in different types of cardiac preparations causes a substantial Ca²⁺ accumulation, manifested as an elevation of diastolic [Ca²⁺]_i. This has been reported in rat ventricular trebeculae (Layland & Kentish, 1999; Brandes & Bers, 2002) and in isolated cardiac myocytes of mouse (Antoons et al. 2002; Knollmann et al. 2003), rabbit (Chudin et al. 1999) and cat (Wang et al. 2001). Because both the sustained component of the Ca²⁺ signal, like the diastolic Ca²⁺ accumulation, and the frequency of Ca²⁺ transients activate calcineurin (Timmerman et al. 1996; Rusnak & Mertz, 2000; Kubis et al. 2003), we hypothesized that in cardiac myocytes pacing might serve as a physiological stimulus for calcineurin activation. To quantify this, we measured [Ca²⁺]_i levels from atria paced at different frequencies. At 1 Hz atrial myocytes generated Ca²⁺ transients from a stable diastolic [Ca²⁺]_i of 200 ± 7 nm to a systolic [Ca²⁺]_i of 802 ± 18 nm, resulting in an amplitude of the transients of 600 ± 21 nm and an average [Ca²⁺]_i of 300 ± 8 nm (n = 24). An increase in the pacing frequency rapidly shifted the diastolic Ca²⁺ to higher levels as shown by the representative recording in Fig. 2A. The diastolic [Ca²⁺]_i was dependent on the frequency of the pacing and increased up to 511 ± 40 nm after 10 min at 6 Hz (n = 6, Fig. 2B). The increase in diastolic [Ca²⁺]_i was accompanied by a corresponding increase in the systolic and average [Ca²⁺]_i, but there was no change in the amplitude of the Ca²⁺ transients.

Figure 2. Pacing increases diastolic, systolic and average [Ca²⁺]_i but does not change the amplitude of Ca²⁺ transients.

A, [Ca²⁺]_i during change in pacing frequency from 1 to 4 Hz. Arrows indicate systolic, diastolic and average Ca²⁺ (1 s moving average). B, pacing-induced changes at the steady-state in different parameters of [Ca²⁺]_i at different pacing frequencies (1, 4, 5 and 6 Hz, n = 6 each). Note that diastolic, systolic and average [Ca²⁺]_i increase as function of the pacing frequency, whereas the amplitude of Ca²⁺ transients remains virtually constant.

Simulation of calcineurin activation with cardiac Ca²⁺ signals

To explore the dynamics of calcineurin activation induced by pacing, we used a mathematical model to simulate calcineurin activation using the measured [Ca²⁺]_i signals as input to the model. A data set for each pacing group was selected where [Ca²⁺]_i was closest to the mean values of systolic, diastolic and average as calculated from six identical experiments. The simulations showed that calcineurin activity is ∼10% of maximum when the pacing frequency is 1 Hz. This baseline activity is brought about by the diastolic [Ca²⁺]_i since Ca²⁺ transients at 1 Hz do not cause any accumulative activation of calcineurin in the simulations. By increasing the pacing frequency, the calcineurin activity is increased to ∼50% at 4 Hz, to ∼60% at 5 Hz and to ∼70% at 6 Hz (Fig. 3A). From the simulation with measured [Ca²⁺]_i as an input it cannot be judged how much of the calcineurin activity is caused by diastolic [Ca²⁺]_i increase and how much is due to increase of the frequency of the Ca²⁺ transients. To study this we modelled artificial data mimicking the normal Ca²⁺ signals at different frequencies (1, 4, 5 and 6 Hz, Fig. 3B), but with fixed baseline (0.2 or 0.4 μm) and amplitude (0.8 μm). With baseline [Ca²⁺]_i of 0.4 μm, the frequency-dependent calcineurin activity increases from 37% at 1 Hz to 59% at 6 Hz and with baseline [Ca²⁺]_i of 0.2 μm the corresponding values were 13% and 36% (Fig. 3B), indicating that calcineurin is activated by the frequency of cardiac Ca²⁺ transients.

Figure 3. Simulated calcineurin (CaN) activation by pacing-induced Ca²⁺ changes.

A, simulated calcineurin activity produced by representative [Ca²⁺]_i measurements from rat atrium. Left panel shows samples from the data used at 4, 5 and 6 Hz pacing. B, the steady state CaN activity from simulations with artificial [Ca²⁺] data with 0.4 μm (○) and 0.2 μm (□) baseline [Ca²⁺], respectively, compared with the simulated CaN activity produced by measured [Ca²⁺]_i changes (•).

Effect of cyclosporin on the frequency-dependent Ca²⁺ changes in the rat atrium

Cyclosporin A has been reported to cause a calcineurin-independent suppression of Ca²⁺ transients with a corresponding negative inotropy in cardiac muscle by acting on the sarcoplasmic reticulum (SR) (Janssen et al. 2000). However, active calcineurin may modulate the Ca²⁺ release in cardiac myocytes by regulating the ryanodine receptors (RyRs) (Bandyopadhyay et al. 2000) and reduce the SR Ca²⁺ uptake by dephosphorylating phopholamban, the regulator of cardiac SR Ca²⁺-ATPase (Münch et al. 2002). Measuring the Ca²⁺ dependence of the CsA effect can be used to distinguish between the calcineurin-dependent and calcineurin-independent effects of CsA. When we superfused rat atria paced at 1 Hz with 1 μm CsA for 15 min, no statistically significant differences in any of the measured parameters of [Ca²⁺]_i were observed compared to control experiments without CsA (Fig. 4). However, when CsA was applied prior to 6 Hz pacing, [Ca²⁺]_i was decreased. In six experiments the systolic [Ca²⁺]_i (P < 0.05), the average [Ca²⁺]_i (P < 0.01) and the amplitude of the Ca²⁺ transients (P < 0.05) were all significantly suppressed compared to control. It should be noted that CsA affected the Ca²⁺ transients in a biphasic manner. After the onset of 6 Hz pacing the amplitude was transiently increased (Fig. 5A), which was then followed by decay to a lower level than in the control (Fig. 5B). This suggests that calcineurin has a role in the maintenance of Ca²⁺ release in the face of high levels of [Ca²⁺]_i.

Figure 4. Cyclosporin A suppresses the pacing-induced [Ca²⁺]_i changes in rat atrium.

A, effect of CsA (1 μm) and pacing on the diastolic, systolic, average and amplitude of the Ca²⁺ transients in rat atrium. Each bar represents mean ± s.e.m. of 6 separate experiments. *P < 0.05 and **P < 0.01.

Figure 5. Cyclosporin A induces biphasic changes in the Ca²⁺ transient amplitude after onset of the pacing.

A, after onset of the 6 Hz pacing in rat atrium [Ca²⁺]_i reaches a steady state within 1 min (left). Pre-incubation of rat atrium with of CsA (1 μm) promotes biphasic behaviour of [Ca²⁺]_i when pacing is switched from 1 to 6 Hz (right). B, mean change in the amplitude of Ca²⁺-transient in the absence (▪) and presence (○) of CsA (1 μm). Note the biphasic behaviour of transient amplitude with CsA characterized by transient increase of the Ca²⁺ transients before decaying to lower level after 4 min.

Effects of calcineurin activation on immediate-early gene expression

To study the immediate effects of calcineurin activation on gene expression, we measured the mRNA levels of two rapidly inducible genes, BNP and c-fos. First we measured pacing sensitivity of the expressions of these genes. Atrial tissues were subjected to different pacing frequencies (1, 4, 6 and 8 Hz) for 30 min and mRNA levels measured. BNP mRNA level showed near-linear pacing dependence, whereas pacing did not affect the c-fos mRNA level (Fig. 6A). From the frequencies used here, 8 Hz pacing for 30 min gave the maximal BNP mRNA response of ∼2-fold compared to the baseline expression in 1 Hz. Next we studied the role of calcineurin in these pacing-induced changes in gene expression by exposing atrial tissues to one of the following stimuli for 30 min: 1 Hz pacing, 1 Hz pacing after preincubation (30 min) with 1 μm CsA, 8 Hz pacing, and 8 Hz pacing with preincubation (30 min) with 1 μm CsA. Pacing at 8 Hz induced a significant increase in the BNP mRNA levels (P < 0.05), which was abolished by pre-exposure of CsA (Fig. 6B). It was a surprise to see that while the mRNA levels of c-fos were not sensitive to 8 Hz pacing or CsA alone, pacing at 8 Hz increased the c-fos mRNA levels when the atria were pre-exposed to CsA (n = 6, P < 0.01, Fig. 6C).

Figure 6. Cyclosporin A inhibits pacing-induced BNP gene expression while it activates c-fos gene expression during pacing.

A, relative (1 Hz equals 1) changes in the mRNA (n = 6 each) levels of BNP and c-fos in rat atria exposed to different pacing frequencies at 30 min as function of given pacing frequency. Apparent slope of the linear fit shows clear pacing sensitivity of BNP mRNA but no pacing sensitivity of the c-fos mRNA. The mRNA levels of BNP (B) and c-fos (C) genes from atrial tissues exposed to low pacing 1 Hz (−) or high pacing 8 Hz (+) for 30 min. When indicated, cyclosporin A (CsA, 1 μm) was included in perfusion medium from start of preincubation BNP and c-fos mRNA levels are expressed relative to 18S RNA levels, and each bar represents mean ± s.e.m. of 6 separate experiments. *P < 0.05 and **P < 0.01.

Pacing-induced increase in NFAT activity

We isolated nuclear fractions from atria exposed to different pacing and/or CsA and from these fractions we measured the NFAT binding activity on the BNP promoter. This step was taken to answer two questions: first, to determine if pacing-induced Ca²⁺ changes are sufficient to activate calcineurin and resulting NFAT nuclear translocation, and second, to study if cyclosporin A inhibits the NFAT activation. This indirect method was chosen because the fast inactivation of calcineurin after Ca²⁺ removal (Stemmer & Klee, 1994) might complicate direct activity measurements from tissue samples, while the calcineurin-induced NFAT nuclear accumulation persists for several minutes (Timmerman et al. 1996; Liu et al. 2001; Kubis et al. 2002). The specificity of NFAT complex formation on BNP promoter was confirmed by the competition analysis for nuclear protein binding with NFAT-BNP as a radioactively labelled probe exposed to 100 m excess of unlabelled competitor oligonucleotides of NFATmut-IL-2, NFAT-BNP or NFAT-IL-2 (Fig. 7A). In rat atrium, 8 Hz pacing for 30 min induced a 2-fold increase in NFAT nuclear activation (n = 6, P < 0.01, Fig. 7B). Application of 1 μm of CsA did not change the NFAT activation at 1 Hz pacing but totally inhibited the activation induced by pacing at 8 Hz.

Figure 7. Pacing activates NFAT in rat atrium with a cyclosporin A-dependent manner.

A, specific NFAT binding on BNP promoter (NFAT-BNP). Competition analysis for nuclear protein binding was carried out with NFAT-BNP as a radioactively labelled probe and when designated with 100 m excess of unlabelled competitor oligonucleotides of NFATmut-IL-2, NFAT-BNP and NFAT-IL-2. Nuclear extracts from 8 Hz paced atria were used for competition experiments. B, pacing-induced NFAT activation in rat atrium. For binding reaction, 12 μg of a nuclear protein extract from atrial tissue were used. After preincubation (50 min, 1 Hz), atrial pacing was continued at 1 (−) or 8 Hz (+) for 30 min. When indicated, cyclosporin A (CsA, 1 μm) was included in perfusion medium from start of preincubation. Each bar represents mean ± s.e.m. of 6 separate experiments. **P < 0.01.

Discussion

This is the first report that establishes a relationship between cardiac Ca²⁺ signals and calcineurin activation. Our results show that (1) frequency of cardiac Ca²⁺ transients forms an adequate stimulus for calcineurin activation and corresponding NFAT activation, (2) calcineurin activation regulates Ca²⁺ release in the myocytes, and (3) calcineurin activation results in a dual effect on the immediate gene expression with both augmentation (BNP) and suppression (c-fos).

Calcineurin activation by pacing

The measurements showed a clear CsA-dependent increase in the nuclear NFAT activity when the frequency of atrial pacing was increased. This is in line with the modelling results that predicted that calcineurin activation depends on the frequency of cardiac Ca²⁺ transients. Similarly, in skeletal muscle cells NFAT translocation has been shown to be sensitive to the frequency of stimulation and to CsA (Lui et al. 2001; Kubis et al. 2002). Although the NFAT translocation is induced by calcineurin activation, the time courses of these two processes are different. The nuclear import of NFAT requires a minimum stimulation time of 1.5–5 min and is complete after 20–30 min (Liu et al. 2001; Kubis et al. 2002), whereas calcineurin in vitro responds to Ca²⁺ on a time scale of seconds (Stemmer & Klee, 1994). This was also seen in the present study where the effects of calcineurin inhibition on [Ca²⁺]_i were apparent within seconds after increasing the pacing frequency. In our simulations, the activation of calcineurin was fast enough (τ_act= 0.7–1.6 s) to respond to a brief rise in [Ca²⁺]_i such as that during the cardiac Ca²⁺ transient. The kinetics of calcineurin deactivation (τ_deact∼1.5 s) ensures that activity triggered by a single Ca²⁺ transient decays if the frequency is low enough. Upon an increase of frequency, the deactivation of calcineurin becomes incomplete before the next transient and calcineurin activation increases. The frequency-dependent calcineurin activation appears to be prominent at frequencies corresponding to the normal rat heart rate (∼5–8 beats s⁻¹) and thus calcineurin is likely to exhibit a significant level of activity even at rest, with rapid responses to both increase and decreases of heart rate. It is also interesting to note that although pacing produces an increase in the diastolic [Ca²⁺]_i that activates calcineurin effectively, our modelling experiments show that changes in the frequency of Ca²⁺ transients alone are sufficient to alter calcineurin activity (see Fig. 3B). This supports the idea that calcineurin is part of the normal adaptation mechanism, and as such participates in the maintenance of the normal cardiac phenotype (Eto et al. 2000). Analogously, it has been suggested that in slow type skeletal myocytes, calcineurin maintains the slow phenotype because the continuous activity in these cells upholds the calcineurin activity (Chin et al. 1998).

Suppression of pacing-induced Ca²⁺ increase by cyclosporin A

In the present study, we found that that all of the effects of CsA on [Ca²⁺]_i were present only with high pacing frequency (see Fig. 4). This indicates that the changes were caused by inhibition of Ca²⁺-activated calcineurin. In cardiac myocytes, three different calcineurin targets may be involved in Ca²⁺ regulation. First, the inhibitory regulator of SR Ca²⁺-ATPase, phospholamban, is inhibited by calcineurin in ventricular myocytes (Münch et al. 2002) and therefore calcineurin inhibition would have a stimulatory effect on the pump. This sort of Ca²⁺-ATPase stimulation would cause faster Ca²⁺ transient decay, possibly a reduced diastolic [Ca²⁺]_i and increased SR Ca²⁺ content resulting in an increased Ca²⁺ release and subsequently sustained augmentation of Ca²⁺ transients (Song et al. 2003). This is opposite to what we found to be the effect of CsA during pacing where the endpoint was a significant reduction in Ca²⁺ transient amplitude. This finding is not surprising since atrial muscle contains much less phospholamban than ventricular muscle (Koss et al. 1995). Second, calcineurin inhibition was recently found to stimulate the l-type calcium current and thereby cause a sustained increase of calcium release in ventricular myocytes (Santana et al. 2002). This mechanism had either a transient or a small effect in our atrial cells where the endpoint was a significant reduction in calcium release. Third, calcineurin may act on the Ca²⁺ release channels, i.e. the RyRs (Bandyopadhyay et al. 2000) or the IP₃ receptors (Cameron et al. 1995). The contribution of IP₃ receptors to the normal calcium release is small in rat atrial myocytes (Mackenzie et al. 2002). Calcineurin regulates cardiac ryanodine receptor via FKBP12.6, which de-sensitizes the channel to Ca²⁺, thereby reducing the Ca²⁺ leak during diastole (Bandyopadhyay et al. 2000; Marks, 2003). Calcineurin inhibitors make RyRs more sensitive to Ca²⁺ leading to increased RyR openings at lower levels of Ca²⁺ manifested as spontaneous Ca²⁺ release events (McCall et al. 1996; Bandyopadhyay et al. 2000). This type of RyR sensitization leads first to an increase in the Ca²⁺ release, which in turn leads to a reduction of SR Ca²⁺ content and Ca²⁺ release ensuring that the effect is transient (Trafford et al. 2000; Eisner et al. 2000). If this is accompanied by even a small increase in the passive Ca²⁺ leak like that with ryanodine (Bers et al. 1987) or with CsA (Bandyopadhyay et al. 2000), a sustained decrease of Ca²⁺ release can be expected. Altogether this is exactly what we found in the present study (see Fig. 4), which supports the idea that RyR-mediated SR Ca²⁺ release was the main target of CsA.

Effect of calcineurin inhibition on the BNP and c-fos gene expression

Part of the controversy concerning calcineurin signalling in cardiac muscle gene expression and hypertrophy is related to the difficulty in isolating the calcineurin pathway from other hypertrophic pathways. In vivo hypertrophic stimuli consist not only of activation of Ca²⁺-dependent processes, but also of hormonal, mechanical and neuronal adaptations with corresponding activation of different hypertrophic signalling cascades (for review see Tavi et al. 2001). To study the earliest calcineurin-induced transcriptional events, we measured the effects of altered [Ca²⁺]_i on the expression of the immediate-early genes c-fos and BNP, two rapidly inducible marker genes for load-induced hypertrophy (see, e.g. Tavi et al. 2001). We found that the Ca²⁺–calcineurin pathway has opposite effects on the expressions of c-fos and BNP. While the BNP expression is part of normal adaptation of heart to increased load (Tavi et al. 2001; Tokola et al. 2001) and a direct target of calcineurin–NFAT-mediated transcription (Molkentin et al. 1998), the c-fos induction is additionally associated with pathological developments such as induction of fetal gene expression by mechanical unloading of the heart (Depre et al. 1998). More importantly, induction of c-fos expression is likely to regulate the impending expression of other genes because it is a part of the heterodimeric transcription factor AP-1. We have shown here that in rat cardiac myocytes, CsA increases the c-fos mRNA levels in a Ca²⁺-dependent manner. This seems to be a common feature of many cell types since stimulation of c-fos expression by CsA has been demonstrated in T-cell lymphoma cells when exposed to ionomycin (Su et al. 1996) and in murine erythroleukaemia cell line CsA stimulates c-fos expression in a Ca²⁺-dependent manner in correlation with the inhibition of CaN activity (Schaefer et al. 1998). These phenomena are probably not caused by unspecific actions of CsA since in hippocampal neurones FK506 enhances the c-fos expression after short bouts of electrical stimulation (Bito et al. 1996). These results suggest that the same calcium signal promotes both activation and inhibition of the c-fos expression. To explain a similar phenomenon in hippocampal neurones, Bito et al. (1996) suggested that calcineurin acts indirectly to de-phosphorylate the CaM kinase-phosphorylated cyclic AMP-responsive element binding protein (CREB) transcription factor, thereby suppressing the CREB-induced c-fos expression.

Supporting such a dual role of calcineurin in controlling cardiac gene expression, MCIP1^-/- mice have an increased hypertrophic response to constitutively active calcineurin expression but a reduced hypertophic response to pressure overload and to adrenergic stimulation (Vega et al. 2003). If the primary role of calcineurin activation is suppression of some genes, like c-fos, the inhibition of calcineurin would allow these genes to be expressed and would secondarily change the whole pattern of gene expression. This dualism in the calcineurin-dependent immediate-early gene transcription together with the apparent activity of calcineurin, maintained already by moderate heart rates, suggests that calcineurin may orchestrate the transcriptional signals during both normal adaptation and pathological developments. This is not unexpected since the output of calcineurin-dependent transcription has been shown to depend on other accompanying signal cascades (e.g. Rac, Ras or protein kinase C, for review see Crabtree, 2001) also in cardiac myocytes (Bueno et al. 2002; for review see Wilkins & Molkentin, 2002).

Physiological and pathophysiological implications

On the basis of the modelling in the present study it can be predicted that, in addition to the frequency, other factors that modulate the shape or amplitude of cardiac Ca²⁺ transients and the diastolic [Ca²⁺]_i may have a substantial impact on calcineurin activity. Increased amplitude of the Ca²⁺ transients, induced by increased SR Ca²⁺ uptake (for example, by phosphorylation of phospholamban), may actually reduce calcineurin activity by shortening the duration of transients and decreasing the diastolic [Ca²⁺]_i. On the other hand, factors that act to lengthen the Ca²⁺ transients or augment Ca²⁺ transients without increasing the rate of Ca²⁺ removal may elevate calcineurin activity as shown previously with endothelin-1 (Zhu et al. 2000). This mechanism would provide a novel pathway for communication between different hypertrophic pathways.

This study shows that calcineurin influences excitation–contraction coupling in the heart by regulating the Ca²⁺ release. This extends the spectrum of calcineurin effects in the heart, but also points to the likely consequences of calcineurin inhibition. The vast majority of the reported effects of calcineurin inhibition on the hypertrophy development appear to be due to inhibition of calcineurin-induced transcription activity. Nevertheless, part of the effects can be due to concomitant suppression of cardiac Ca²⁺ signals, which in turn suppresses many other Ca²⁺-mediated processes in the myocytes.

The extent to which frequency-dependent calcineurin activation, as described in the present study, contributes to the development of left ventricular hypertrophy and cardiac failure cannot be precisely estimated. However, given that the increased sympathetic outflow with associated tachycardia constitutes one of the major risk factors of chronic heart failure mortality (Poole-Wilson et al. 2003) and calcineurin activity has been reported to increase in patients with cardiac hypertrophy and failure (Haq et al. 2001; Ritter et al. 2002), the drugs that reduce the activation of the sympathetic nervous system like β-receptor blockers may be effective in control of cardiac growth and remodelling (Packer et al. 1996).