Impaired β-adrenergic responsiveness accentuates dysfunctional excitation–contraction coupling in an ovine model of tachypacing-induced heart failure

Non-technical summary

Heart failure is where the heart is unable to pump sufficient blood in order to meet the requirements of the body. Symptoms of heart failure often first present during exercise. During exercise the blood levels of a hormone, noradrenaline, increase and activate receptors on the muscle cells of the heart known as β-receptors causing the heart to contract more forcefully. We show that in heart failure the response to β-receptor stimulation is reduced and this appears to be due to a failure of the β-receptor to signal correctly to downstream targets inside the cell. However, by-passing the β-receptor and directly activating one of the downstream targets, an enzyme known as adenylyl cyclase, inside the cell restores the function of the muscle cells in failing hearts. These observations provide a number of potential targets for therapies to improve the function of the heart in patients with heart failure.

Abstract

Abstract

Reduced inotropic responsiveness is characteristic of heart failure (HF). This study determined the cellular Ca²⁺ homeostatic and molecular mechanisms causing the blunted β-adrenergic (β-AR) response in HF. We induced HF by tachypacing in sheep; intracellular Ca²⁺ concentration was measured in voltage-clamped ventricular myocytes. In HF, Ca²⁺ transient amplitude and peak L-type Ca²⁺ current (I_Ca-L) were reduced (to 70 ± 11% and 50 ± 3.7% of control, respectively, P < 0.05) whereas sarcoplasmic reticulum (SR) Ca²⁺ content was unchanged. β-AR stimulation with isoprenaline (ISO) increased Ca²⁺ transient amplitude, I_Ca-L and SR Ca²⁺ content in both cell types; however, the response of HF cells was markedly diminished (P < 0.05). Western blotting revealed an increase in protein phosphatase levels (PP1, 158 ± 17% and PP2A, 188 ± 34% of control, P < 0.05) and reduced phosphorylation of phospholamban in HF (Ser16, 30 ± 10% and Thr17, 41 ± 15% of control, P < 0.05). The β-AR receptor kinase GRK-2 was also increased in HF (173 ± 38% of control, P < 0.05). In HF, activation of adenylyl cyclase with forskolin rescued the Ca²⁺ transient, SR Ca²⁺ content and SR Ca²⁺ uptake rate to the same levels as control cells in ISO. In conclusion, the reduced responsiveness of the myocardium to β-AR agonists in HF probably arises as a consequence of impaired phosphorylation of key intracellular proteins responsible for regulating the SR Ca²⁺ content and therefore failure of the systolic Ca²⁺ transient to increase appropriately during β-AR stimulation.

Introduction

Heart failure (HF) remains a leading cause of mortality and morbidity (Lloyd-Jones et al. 2002). One of the hallmarks of HF is a reduced contractile reserve in response to a variety of inotropic manoeuvres, e.g. catecholamines (Chattopadhyay et al. 2010) or exercise-tolerance tests (Borlaug et al. 2006). The reduced response to β-AR agonists observed clinically is also present at both the level of the isolated muscle preparation (e.g. Ginsburg et al. 1983; Feldman et al. 1987; Maier et al. 2002) and in single cardiac myocytes (Sande et al. 2002; Leosco et al. 2008). Additionally, in vivo echocardiographic and haemodynamic studies have shown similar impairments in β-AR responsiveness in a number of genetic models of HF in the mouse (e.g. Cho et al. 1999; Montgomery et al. 2005).

Considerable data suggest that a key mechanism responsible for the reduced contractile reserve in HF is perturbed β-AR signalling. Alterations to the β-AR signalling pathway occur at multiple control points ranging from, for example, reductions in β-AR receptor density (Bristow et al. 1982; DiPaola et al. 2001; Leosco et al. 2008) and increased G-protein receptor kinase expression (GRK-2, alternatively βARK1) (Choi et al. 1997; Cho et al. 1999) to enhanced intracellular phosphatase activity (Reiken et al. 2003; El-Armouche et al. 2004). Indeed strategies aimed at correcting components in the β-AR signalling pathway that are altered in HF, e.g. GRK-2 activity or β-AR blocker therapy lead to improvement in, or complete restoration of, β-AR agonist responsiveness as well as increased basal cardiac contractility (Freeman et al. 2001; Kubo et al. 2001; Tachibana et al. 2005; El-Armouche et al. 2008).

Moreover, many of the intracellular downstream targets following β-AR activation are directly coupled to regulation of systolic Ca²⁺ and the function of many of these Ca²⁺ regulatory proteins is altered in HF thus contributing to the decreased contractile performance of the diseased heart (Kubo et al. 2001; Sande et al. 2002; Plank et al. 2003; Díaz et al. 2004; Desantiago et al. 2008). However, precisely how such alterations to the β-AR signalling cascade and intracellular protein phosphorylation influence the L-type Ca²⁺ current, the systolic Ca²⁺ transient, SR Ca²⁺ content and the cellular fluxes of Ca²⁺ are not completely understood. This study was therefore designed to investigate how HF influences intracellular Ca²⁺ homeostatic responses to β-AR stimulation. The major findings are that in HF myocytes the systolic Ca²⁺ transient and SR Ca²⁺ content respond minimally to the β-AR agonist isoprenaline (isoproterenol). Direct activation of adenylyl cyclase using forskolin ‘rescues’ the systolic Ca²⁺ transient, SR Ca²⁺ content and sarco-endoplasmic reticulum Ca²⁺-ATPase (SERCA) activity in HF cells. In line with these observations a number of molecular alterations are present in HF cells which would lead to β-AR desensitisation and reduced phosphorylation of Ca²⁺ homeostatic proteins. These findings elucidate a number of potential therapeutic targets to improve the performance of the failing myocardium.

Methods

All procedures accord to the United Kingdom Animals (Scientific Procedures) Act, 1986 and University of Manchester Ethical Review Process.

Induction of heart failure and isolation of single ventricular myocytes

An ovine model of HF was used in the present study. Eighteen female Welsh Mountain sheep (35.6 ± 1.7 kg and approximately 18 months of age) were subjected to right ventricular tachypacing using a minor modification of a method described in detail previously (Dibb et al. 2009). In the present study animals were anaesthetised using isoflurane (1–4% v/v in oxygen) and under fluoroscopic guidance a single IS-1 bipolar endocardial pacing lead (St Jude Medical or Medtronic Flextend II) was fixed transvenously at the right ventricular apex and connected to a cardiac pacemaker (EnPulse, Medtronic) located subcutaneously in a lateral cervical position. Peri-operative analgesia was provided with meloxicam (0.5 mg kg⁻¹) and antibiosis with enrofloxacin (2.5 mg kg⁻¹) administered subcutaneously. Animals were allowed to recover from the surgical procedure for 7–10 days before a high rate pacing patch (Medtronic) was applied telemetrically to the implanted pacemaker and the heart paced at 210 beats min⁻¹ (resting heart rate of ∼100 beats min⁻¹). Cardiac remodelling was assessed in vivo using trans-thoracic echocardiography (SonoSite MicroMaxx; 5–1 MHz phased array probe). A five-lead electrocardiogram (EMKA Technologies) was also digitised to a personal computer at a sampling rate of 1 kHz (IOX, EMKA Technologies). The QT interval was determined from the points where the derivative of the ECG signal crossed the isoelectric line at the start of the QRS complex and end of the T wave. For determination of heart rate and QT interval when HF had developed the pacemaker was switched off for 15 min before measurements were made.

Animals were killed after 34.8 ± 1.2 days of tachypacing by intravenous administration of pentobarbitone sodium (200 mg kg⁻¹) and heparin (10,000 i.u.). The atria were removed from the heart base and the left anterior descending coronary artery cannulated and single ventricular myocytes isolated from the mid-myocardial layer of the left ventricular free wall using a collagenase and protease digestion technique as described previously (Dibb et al. 2004).

Voltage clamp studies

The whole-cell voltage clamp technique was used in the present study. Electrodes (∼2–3 MΩ resistance) were filled with (in mmol l⁻¹): CsCl, 120; TEA-Cl, 20; Hepes, 10; Na₂ATP, 5; CsEGTA, 0.02; pH 7.2 with CsOH. Intracellular Ca²⁺ concentration ([Ca²⁺]_i) was measured using the moderate affinity indicator Fluo-5F pentapotassium salt (100 μmol l⁻¹) loaded via the voltage clamp pipette. The use of lower affinity indicators protects against saturation of the fluorescence signal during adrenergic stimulation of the cells. Fluorescence was excited using a 475 nm light-emitting diode (Cairn Instruments, Kent, UK) and emitted fluorescence (515–600 nm) used to calculate [Ca²⁺]_i assuming a K_d for Ca²⁺ of 1035 nmol l⁻¹ (Dibb et al. 2004). Voltage control was achieved using an Axopatch 200B and pCLAMP software (Molecular Devices, UK). Series resistance errors (control, 4.5 ± 0.15 MΩ; HF, 5.0 ± 0.2 MΩ) were compensated (∼80%) following rupture of the patch. SR Ca²⁺ content was calculated by integrating the resulting inward Na⁺–Ca²⁺ exchange (NCX) current following the rapid application of caffeine (10–20 mmol l⁻¹) (Varro et al. 1993; Dibb et al. 2004). Intracellular Ca²⁺ buffering was also calculated during the application of caffeine as previously described (Trafford et al. 1999). The total amount of Ca²⁺ added to the cytosol was derived by integrating the caffeine-evoked NCX current and plotted as a function of [Ca²⁺]_i. The data were fitted with a linear regression to obtain the buffering power. The rates of Ca²⁺ removal from the cell by the SR and NCX (k_SR and k_caff, respectively) were calculated by fitting the decay phases of the systolic and caffeine-evoked rises of [Ca²⁺]_i with single exponential functions as described in detail previously (Díaz et al. 2004; Dibb et al. 2004, 2005).

All voltage clamp recordings were made at 37°C after an initial 5 min stabilisation period and cells were perfused with (in mmol l⁻¹): NaCl, 134; glucose, 11; Hepes, 10; 4-aminopyridine, 5; KCl, 4; probenecid, 2; CaCl₂, 1.8; MgCl₂, 1; DIDS, 0.1; BaCl₂, 0.1; pH 7.4 with NaOH. Voltage-dependent activation and inactivation of I_Ca-L were determined as described previously (Dibb et al. 2004). In both cases data were normalised to the peak current obtained and fitted with a Boltzmann function:

Where f is the fit to the data, V the test potential, V_mid the potential at which current is half-maximal, V_c the slope of the fit, and C a constant offset.

Western blot analysis and protein kinase A assay

For Western blotting, samples of left ventricular wall were obtained at the time of cardiac removal and snap frozen in liquid nitrogen until use. Protein samples were prepared for SDS-PAGE as described previously (Díaz et al. 2004; Graham & Trafford, 2007; Walden et al. 2009). Following electrophoresis, samples were transferred to nitrocellulose membranes and blocked with 5% w/v bovine serum albumin or non-fat milk before being exposed to primary and secondary antibodies. Details of protein loading, and primary and secondary antibody concentrations are provided in Table 1. All secondary antibodies were horseradish peroxidase conjugated and protein reactivity detected using enhanced chemiluminescence (Pierce Supersignal West Pico) captured digitally (Syngene). An internal protein standard (internal control) was prepared from a single control sheep heart and loaded on each gel. Proteins were quantified relative to the internal standard sample since preliminary experiments demonstrated significant changes of the immunoreactivity of the so-called ‘housekeeping’ proteins GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and β-actin in heart failure. Each gel was repeated 3 times to minimise any loading or transfer errors.

Table 1.

Western blotting conditions

Protein of interest	Protein (μg)	Primary antibody	Dilution and duration	Secondary antibody	Dilution and duration
SERCA2a	5	Santa Cruz (sc73022)	1:5000 overnight 4°C	Pierce goat anti-mouse	1:5000 1 h room temp
Total phospholamban (PLN)	5	Badrilla (A010–14)	1:10,000 overnight 4°C	Pierce goat anti-mouse	1:5000 1 h room temp
Ser16 phosphorylated PLN (Ser16)	20	Badrilla (A010–12)	1:10,000 overnight 4°C	Pierce goat anti-rabbit	1:2500 90 min room temp
Thr17 phosphorylated PLN (Thr17)	20	Badrilla (A010–13)	1:5000 overnight 4°C	Pierce goat anti-rabbit	1:2500 90 min room temp
Calsequestrin (CSQ)	20	Affinity Bioreagents (PA1–913)	1:2500 overnight 4°C	Novus donkey anti-rabbit	1:25,000 1 h room temp
Protein phosphatase 1 (PP1)	20	Santa Cruz (sc7482)	1:1000 overnight 4°C	Pierce goat anti-mouse	1:2500 90 min room temp
Protein phosphatase 2A (PP2A)	20	Santa Cruz (sc14020)	1:1000 overnight 4°C	Pierce goat anti-rabbit	1:2500 90 min room temp
G-protein receptor kinase 2 (GRK-2)	20	Santa Cruz (sc562)	1:1000 overnight 4°C	Pierce goat anti-rabbit	1:5000 1 h room temp
Ryanodine receptor (RyR2)	15	Abcam (ab2827)	1:1000 1 h room temp	Pierce goat anti-mouse	1:5000 1 h room temp
Ser2808 phosphorylated RyR	10	Badrilla (A010–30)	1:5000 1 h room temp	Novus donkey anti-rabbit	1:10,000 1 h room temp
CAMKIIδ	10	Santa Cruz (sc5392)	1:1000 overnight 4°C	Santa Cruz donkey anti-goat	1:5000 1 h room temp

Summary showing primary and secondary antibody dilutions used to quantify changes in proteins of interest.

For PKA determination a non-radioactive cAMP-dependent assay (PepTag, Promega) was used. Briefly tissue samples were snap frozen at the time of cardiac excision and stored in liquid nitrogen until use. Homogenates of left ventricular tissue were prepared according to the manufacturer's instructions and 27 μg of protein incubated with the PepTag reagents for 30 min at room temperature. Following gel electrophoresis active PKA was measured under UV light and quantified relative to the manufacturer's internal standard.

Statistics

Data are presented as mean ± standard error of the mean for n experiments. Differences between groups were determined using t tests, analysis of variance (ANOVA) or two-way repeated measures ANOVA as appropriate. Where data were not normally distributed they were transformed (log₁₀ or reciprocal) or a Mann–Whitney Rank Sum Test used where transformation failed to produce a normal distribution of data. P < 0.05 was used as the cut-off point for statistical significance.

Results

Cardiac remodelling in heart failure

All of the tachypaced animals were dyspnoeic and had oedematous lungs at the time of killing indicating clinical symptoms of HF. Due to cardiac orientation in the sheep it is not possible to obtain trans-thoracic apical four chamber views; however, Table 2 summarises cardiac measurements made from the right para-sternal window. The development of HF was associated with a 64 ± 19% increase in left ventricular end-diastolic internal dimension (EDID) and 54 ± 19% and 42 ± 12% decreases in fractional shortening and short axis fractional area change, respectively, when compared to pre-pacing values (all P < 0.001 vs. pre-pacing values). In addition to gross chamber dilatation, cellular hypertrophy was also present as evidenced by an increase in cellular capacitance from 130.3 ± 4.2 pF in control cells to 152.6 ± 4.8 pF in HF cells (n = 62–71 cells, P < 0.05). The planar dimensions of isolated myocytes showed an increase in length from 129.3 ± 1.4 μm in control cells to 133.8 ± 1.6 μm in HF cells (n = 135–204, P < 0.05). Conversely, cell width decreased in HF from 25.8 ± 0.5 μm in control cells to 23.8 ± 0.4 μm (P < 0.001).

Table 2.

Echocardiographic measurements

Right para-sternal echocardiographic parameter	Pre-pacing measurements (control)	Post-pacing measurements (heart failure)	P value
End-diastolic internal dimension; m-mode (EDID, cm)	2.44 ± 0.16	3.80 ± 0.08	<10−4
End-systolic internal dimension; m-mode (ESID, cm)	0.81 ± 0.08	2.63 ± 0.12	<10−7
Fractional shortening; m-mode	0.67 ± 0.02	0.31 ± 0.02	<10−7
Fractional area change (short axis view)	0.66 ± 0.03	0.38 ± 0.02	<10−4
Free wall thickness (end-diastole, cm)	1.15 ± 0.08	0.71 ± 0.04	<0.005
Septal wall thickness (end-diastole, cm)	1.72 ± 0.10	1.23 ± 0.09	<0.005
Relative wall thickness	1.02 ± 0.13	0.37 ± 0.03	<0.005

Summary data describing changes in cardiac geometry and contractility with the development of heart failure. Relative wall thickess is calculated as (2 × left ventricular free wall thickness)/EDID.

The QT interval and heart rate were also determined from the electrocardiogram before pacing commenced and prior to the animals being killed when HF was evident. Heart rate was unaltered following pacing (measured 15 min after termination of pacing and compared to pre-pacing values) being 105 ± 7 min⁻¹ pre-pacing and 110 ± 6 min⁻¹ post pacing (P = 0.54). The QT interval increased from 277 ± 11 ms to 330 ± 10 ms (P < 0.005, paired t test).

Reduced Ca²⁺ transients in HF are not due to changes of SR Ca²⁺ content

In common with the findings in many models of HF and cardiac dysfunction (Hobai & O’Rourke, 2001; Díaz et al. 2004) the amplitude of the systolic Ca²⁺ transient is reduced in the sheep model of HF, in this case to 70 ± 11% of control values (control, 124.5 ± 10.4 nmol l⁻¹; HF, 87.1 ± 12.6 nmol l⁻¹, n = 22–28 cells, P < 0.05). Additionally, diastolic [Ca²⁺]_i was also reduced in HF cells (control, 187 ± 18 nmol l⁻¹; HF, 129 ± 20 nmol l⁻¹, P < 0.05). Figure 1Aa shows representative Ca²⁺ transients from control non-paced sheep and a tachypaced HF sheep. Despite the reduction in Ca²⁺ transient amplitude, the normalised traces in Fig. 1Ab suggest that the rate of decay of the systolic transient is not altered in HF. This is borne out by the summary data in Fig. 1Ac (control, 4.60 ± 0.19 s⁻¹; HF, 4.12 ± 0.25 s⁻¹, P = 0.09).

Figure 1. Altered intracellular Ca²⁺ homeostasis in heart failure.

Aa, representative systolic Ca²⁺ transients from control (left) and HF (right); Ab, normalised and superimposed Ca²⁺ transients from control (black) and HF (red); Ac, mean data showing unaltered rate of decay of the systolic Ca²⁺ transient in HF (n = 22–28 cells, 9–14 hearts). B, assessment of SR Ca²⁺ content through application of 20 mmol l⁻¹ caffeine as indicated showing the caffeine-induced rise of [Ca²⁺]_i (upper panel), associated NCX current (lower panel) and mean data for SR Ca²⁺ content (right panel, n = 35–52 cells, 11–17 hearts). Ca, representative I_Ca-L records normalised for cell capacitance obtained during a 100 ms test step from −40 mV to 10 mV; Cb, mean I_Ca-L current–voltage data for control (black circles) and HF (red triangles) cells.

The next series of experiments had two purposes: firstly to determine the mechanisms responsible for the smaller systolic Ca²⁺ transient in HF, and secondly to establish why the rate of decay of systolic Ca²⁺ is unaltered in HF. We will consider Ca²⁺ transient amplitude first. Given that the amplitude of the systolic Ca²⁺ transient is highly (approximately cubic; Trafford et al. 2000, 2001) dependent on SR Ca²⁺ content we sought to determine if SR Ca²⁺ content is reduced in HF. These data are summarised in Fig. 1B where caffeine (20 mmol l⁻¹) was used to discharge the SR Ca²⁺ store and the integral of the resulting NCX current used to calculate SR Ca²⁺ content. Under the conditions of these experiments SR Ca²⁺ content in myocytes from control hearts is 32.1 ± 2 μmol l⁻¹ and that from failing hearts 30.8 ± 2.6 μmol l⁻¹ (P = 0.37). Additionally, the amplitude of the caffeine-evoked rise of [Ca²⁺]_i is the same in both groups (data not shown; control, 235 ± 24 nmol l⁻¹; HF, 258 ± 45 nmol l⁻¹).

Decreased I_Ca-L is responsible for the smaller systolic Ca²⁺ transient in HF

Since SR Ca²⁺ content is unaltered, changes in Ca²⁺ transient amplitude could arise through either changes in the trigger for Ca²⁺ release from the SR or alterations to the Ca²⁺ buffering properties of the cell (Bassani et al. 1995; Trafford et al. 2001; Díaz et al. 2001). Figure 1Ca shows that under the voltage clamp conditions used to elicit the Ca²⁺ transients shown in Fig. 1A the amplitude of I_Ca-L is reduced in HF, an effect (Fig. 1Cb) observed across the activation range for I_Ca-L. At a test potential of 10 mV the peak inward Ca²⁺ current is reduced from 7.87 ± 0.36 pA pF⁻¹ in control cells to 3.97 ± 0.23 pA pF⁻¹ in HF cells (P < 0.001). Conversely, intracellular Ca²⁺ buffering was unaltered with the ratio δCa_Total:δCa_free decreasing from 135 ± 11 in control cells to 104 ± 11 in HF cells (P = 0.11, n = 16–23 cells, data not shown) suggesting that the reduction in I_Ca-L is the primary mechanism for the smaller systolic Ca²⁺ transient in HF.

We then sought to determine if the reduction of I_Ca-L observed in HF was responsible for decreasing systolic Ca²⁺. These experiments were performed in control cells by inhibiting I_Ca-L to a similar level to that observed in HF (∼50%) using nicardipine (1 μmol l⁻¹). Figure 2A demonstrates that reductions in Ca²⁺ transient amplitude parallel those of I_Ca-L during nicardipine application. The mean reduction of I_Ca-L density is to 50.4 ± 4.2% of control (Fig. 2Ba; control, 8.70 ± 0.98 pA pF⁻¹; HF, 4.56 ± 0.86 pA pF⁻¹, P < 0.001) whereas the amplitude of the systolic Ca²⁺ transient is reduced to 61.9 ± 3.5% of control (Fig. 2Bb; control, 148.7 ± 21.2 nmol l⁻¹; HF, 91.1 ± 14.2 nmol l⁻¹, P < 0.001, n = 12 cells). These experiments therefore clearly demonstrate the importance of reduced I_Ca-L in HF as the likely mechanism responsible for the generation of the smaller systolic Ca²⁺ transients observed in HF.

Figure 2. Reduction of I_Ca-L decreases Ca²⁺ transient amplitude.

A, time course showing typical response of a single ventricular myocyte to the application of 1 μmol l⁻¹ nicardipine (open horizontal bar). Ca²⁺ transient amplitude (upper) and peak I_Ca-L (lower) are reduced during nicardipine application (open symbols). The line through the nicardipine data is a single exponential fit to the data for clarity. B, mean data for the effects of nicardipine on (i) Ca²⁺ transient amplitude and (ii) peak I_Ca-L. Filled bars represent control (pre-drug) data and hashed bars the steady-state nicardipine effects. *P < 0.05 vs. control, n = 12 cells.

Increased sarcolemmal Ca²⁺ efflux maintains the rate of decay of systolic Ca²⁺ in HF

The unaltered rate of decay of the systolic Ca²⁺ transient in HF cells under ‘basal’ conditions is perhaps surprising given that SERCA function is generally reduced in HF and SERCA-mediated Ca²⁺ uptake is the principal mechanism responsible for the decay of systolic Ca²⁺ (Pieske et al. 1999; Bers, 2001; Neary et al. 2002; Díaz et al. 2004). However, this paradox could be resolved through compensatory increases in sarcolemmal-mediated Ca²⁺ efflux offsetting any reduction in SR-mediated Ca²⁺ removal from the cytosol. Whether surface membrane-mediated Ca²⁺ extrusion is increased in HF can be assessed by comparing the rates of decay of the caffeine-evoked Ca²⁺ transients (k_caff) where SERCA function is effectively by-passed (Díaz et al. 2004; Dibb et al. 2004; Walden et al. 2009). k_caff is increased from 1.58 ± 0.08 s⁻¹ in control cells to 2.29 ± 0.17 s⁻¹ in HF cells (P < 0.001, data not shown). Using a related approach, the SR-dependent rate of Ca²⁺ removal from the cytoplasm (k_SR) can be obtained through subtraction of the surface membrane-dependent rate of Ca²⁺ decay (k_caff) from the rate of decay of the systolic Ca²⁺ transient (k_SR=k_sys–k_caff). Using this approach, the SR-dependent rate of Ca²⁺ removal, k_SR, is found to be reduced from 3.11 ± 0.16 s⁻¹ in control cells to 1.88 ± 0.21 s⁻¹ in HF (P < 0.001, data not shown). Thus the increase in surface membrane-mediated Ca²⁺ efflux exactly balances the decrease in SR-mediated Ca²⁺ uptake resulting in an unaltered rate of decay of the systolic Ca²⁺ transient.

Impaired responsiveness of HF myocytes to β-AR stimulation

A major finding in the previous section was a reduction in Ca²⁺ transient amplitude in HF as a result of the smaller trigger for SR Ca²⁺ release by I_Ca-L. The next series of experiments sought to determine if Ca²⁺ transient amplitude could be normalised by β-AR stimulation. Figure 3A shows a representative response to ISO (0.1 and 1 μmol l⁻¹) obtained from a control and a HF myocyte. In control cells there is a robust (581 ± 65.8%) increase in Ca²⁺ transient amplitude (Fig. 3Ba) from 121 ± 11 nmol l⁻¹ to 681 ± 85 nmol l⁻¹ in 100 nmol l⁻¹ ISO (P < 0.001). In contrast in HF myocytes, whilst there is a response to ISO, Ca²⁺ transient amplitude only increases by 339 ± 52.6% (P < 0.02 vs. control response) from 75 ± 14 nmol l⁻¹ to 222 ± 42 nmol l⁻¹ in 100 nmol l⁻¹ ISO (Fig. 3Ba; P < 0.001 vs. control cell ISO amplitude). In both control and HF myocytes the response to 1 μmol l⁻¹ ISO was indistinguishable from that to 100 nmol l⁻¹ ISO indicating that a maximal response to ISO was achieved at the lower concentration (data not shown). A major mechanism by which β-AR stimulation increases cardiac inotropy is through increasing SR Ca²⁺ content (Hussain & Orchard, 1997; Ginsburg & Bers, 2004). In line with the observed Ca²⁺ transient response to ISO there is also a blunted increase of SR Ca²⁺ content in HF cells (Fig. 3Bb) with SR Ca²⁺ content only increasing by 86 ± 15% compared to 172 ± 18% in control myocytes (P < 0.005). To determine if Ca²⁺ pumping by SERCA also shows a blunted response to ISO the SR-dependent rate of Ca²⁺ removal (k_SR, Fig. 3Bc) was calculated as described above. The fractional increase in k_SR was markedly reduced in HF (109 ± 25%vs. 198 ± 19%, P < 0.01).

Figure 3. Impaired response to β-AR stimulation in heart failure.

A, time course showing positive inotropic effects of 100 nmol l⁻¹ and 1 μmol l⁻¹ ISO in representative control (left) and HF (right) cells (note difference in ordinate ranges). B, summary data showing the effect of 100 nmol l⁻¹ ISO on (i) systolic Ca²⁺ transient amplitude, (ii) SR Ca²⁺ content, and (iii) the SR-dependent rate of Ca²⁺ removal (k_SR, n = 13–33 cells from 6–16 hearts). Ca, representative Western blots for SERCA2a, total phospholamban and calsequestrin; Cb, summary showing protein expression for SERCA2a (SERCA), phospholamban (PLN) and calsequestrin (CSQ): c denotes control and f denotes failing data. Cc, mean data summarising the SERCA2a to phospholamban ratio in control and heart failure tissues (n = 6 control and 7 failing hearts). *P < 0.05 vs. control samples; §P < 0.05 vs. before ISO addition.

SR Ca²⁺ content is controlled by both sarcolemmal and intracellular mechanisms. Of the intracellular mechanisms likely to alter SR Ca²⁺ content Fig. 3Ca shows representative Western blots for SERCA2a, phospholamban (PLN) and the intra-SR Ca²⁺ buffer calsequestrin (CSQ). The fold change in expression relative to control hearts is given in Fig. 3Cb where it is clear that neither SERCA2a nor total PLN levels are affected by disease state. Additionally, the ratio of SERCA2a:total PLN is also not altered in this model of HF (Fig. 3Cc). However, there is a 24 ± 11% increase in CSQ protein in HF (P < 0.05). Considering sarcolemmal Ca²⁺ fluxes, the net fluxes of Ca²⁺ across the cell membrane will determine if SR Ca²⁺ content alters (Trafford et al. 2001, 2002). In this context it is of note that in HF (i) ISO increases I_Ca-L, although only to a level seen under basal conditions in control cells (Fig. 4A and B), and (ii) ISO does not increase the rate constant of decay of the caffeine-evoked Ca²⁺ transient (k_caff ISO, 2.27 ± 0.27 s⁻¹) indicating that NCX function is not increased by β-AR stimulation.

Figure 4. Impaired augmentation of I_Ca-L to β-AR stimulation in heart failure.

Aa, representative current recordings normalised to cellular capacitance obtained under basal conditions (continuous lines) and in the presence of 100 nmol l⁻¹ ISO (dashed lines) for control (upper) and heart failure cells (lower panel) during 100 ms steps from −40 to 10 mV at 0. Hz (note different ordinate scales); Ab, summary data for I_Ca-L response to ISO under conditions described in a. Ba, voltage protocol (upper panel) to determine current–voltage relationships in control cells (middle panel) and HF cells (lower panel) under basal conditions (filled symbols, n = 22–30 cells, 9–10 hearts) and in the presence of 100 nmol l⁻¹ ISO (open symbols, n = 9–12 cells, 5–7 hearts); Bb, normalised activation plots under basal conditions (filled symbols and continuous lines) and in the presence of 100 nmol l⁻¹ ISO (open symbols, dashed lines) for control (•, ○) and HF (▾, ▿) cells. Ca, voltage protocol to determine the voltage dependence of I_Ca-L inactivation (upper panel) and mean normalised inactivation plots under basal conditions and in the presence of 100 nmol l⁻¹ ISO (symbols as for panel B); Cb, calculated I_Ca-L window currents under basal conditions (con) and in the presence of 100 nmol l⁻¹ ISO for control (black) and HF cells (red, n = 4–15 cells, 3–6 hearts). *P < 0.05 vs. control cells; §P < 0.05 vs. before ISO application.

The properties of the L-type Ca²⁺ current are examined in greater detail in Fig. 4. Figure 4A shows representative ISO responses of I_Ca-L obtained at a test potential of 10 mV in both control and HF cells and mean data are summarised in Fig. 4Ab. The activation properties of I_Ca-L are examined in Fig. 4B: panel a shows the ISO responses of control and HF cells, respectively, whilst panel b summarises the voltage dependence of activation of I_Ca-L. Heart failure resulted in rightward shifts in both activation (Fig. 4Bb) and inactivation (Fig. 4Ca) of 4.7 and 5.4 mV, respectively (both P < 0.001 vs. control). However, the slopes of the activation and inactivation plots were unaltered. Isoprenaline resulted in a leftward shift (Fig. 4Bb) in activation in both control (9.2 mV) and HF (9.3 mV) cells (P < 0.01 vs. no ISO). Isoprenaline resulted in a leftward shift of the inactivation curves by 5.9 mV in control cells and 5.0 mV in HF cells (Fig. 4Ca, P < 0.05 vs. no ISO). The superimposed plots in Fig. 4Cb show the calculated I_Ca-L window current under control and ISO conditions. It is clear that there is a shift in the voltage range for the window current to more positive potentials in HF and that the window current area is increased in HF.

Molecular mechanisms for impaired β-AR responsiveness in HF

The next series of experiments investigated the potential involvement of PLN phosphorylation status in the reduced SR-dependent rates of Ca²⁺ removal (k_SR) in HF under both basal and ISO conditions. Using phospho-specific antibodies, Fig. 5A shows that both protein kinase A (PKA, Ser16 site) and Ca²⁺-dependent calmodulin kinase (CAMKII, Thr17 site)-dependent phosphorylation of PLN is reduced in HF (Ser16, 30 ± 10% of control levels; Thr17, 41 ± 15% of control levels, P < 0.05). Whilst the decreased phosphorylation of PLN is consistent with the reduced SR-dependent rate of Ca²⁺ removal (k_SR), the lower phosphorylation status of PLN, especially at the PKA-dependent Ser16 site, is inconsistent with the increased adrenergic drive and circulating catecholamine levels reported in HF (Benedict et al. 1996; Leineweber et al. 2005). However, similar decreases in PLN phosphorylation have been observed in diverse models of heart failure (El-Armouche et al. 2004; Gupta et al. 2005; Yatani et al. 2006). This discrepancy is addressed by a combination of increased intracellular protein phosphatase expression, potentially β-AR desensitisation and reduced PKA activity. Figure 5B quantifies protein phosphatase 1 (PP1, increased in HF by 58 ± 17%, P < 0.005) and protein phosphatase 2A (PP2A, increased in HF by 88 ± 34%, P < 0.05) expression. In both cases protein phosphatase expression is increased and this would result in the reduced phosphorylation of several key intracellular targets involved in regulating the systolic Ca²⁺ transient and SERCA function, most notably, PLN. A potential role for β-AR desensitisation in reducing the phosphorylation status of PLN and activation of PKA is examined in Fig. 5C. Here the levels of GRK-2, the principal G-protein receptor kinase responsible for β-AR desensitisation and internalisation in the heart (reviewed by Hata & Koch, 2003) is increased by 73 ± 38% in HF (P < 0.05). Conversely, PKA activity is reduced in HF to 88.8 ± 3.4% of control levels (Fig. 5Da, P < 0.05) whereas the expression of the CAMKIIδ isoform was unaltered in HF being 74.1 ± 22.7% of control values (P = 0.31, Fig. 5Db). We also sought to determine if changes in ryanodine receptor (RyR) expression and phosphorylation could contribute to the cellular dysfunction in HF (Marx et al. 2000). Whilst there was no change in total RyR expression in HF (86.2 ± 13.6% of control, Fig. 5Ea, P = 0.36), there was a reduction in RyR phosphorylation at Ser2808 in HF (46.0 ± 17.8% of control, Fig. 5Eb, P < 0.05).

Figure 5. Altered protein expression in heart failure.

A, representative phosphospecific Western blots (left) and summary data (right) for Ser16 (a) and Thr17 phosphorylated phospholamban (b). B, representative Western blots (upper) and summary data (lower panels) for protein phosphatase 1 (PP1, a) and protein phosphatase 2A (PP2A, b). C, representative Western blot (upper) and summary data (lower panel) for G-protein receptor kinase 2 (GRK-2). Da, representative gel image and summary data assessing PKA activity; non-P and phos denote non-phosphorylated and phosphorylated (activated) PKA substrate, respectively. Db, representative Western blot and summary data for CAMKIIδ expression. E, representative Western blots and summary data for total RyR2 (a) and Ser2808-phosphorylated RyR (b). *P < 0.05; §P < 0.005 vs. control hearts. n = 6 control and 7 failing hearts.

Activation of adenylyl cyclase restores the systolic Ca²⁺ transient in HF

The observed reduction in β-AR responsiveness in HF could also arise as a result of reduced β-AR-mediated activation of adenylyl cyclase as a result of a loss or desensitisation of the β-AR receptor. We have examined a potential role for dys-regulated β-AR receptor signalling by bypassing formation of the G_αs subunit on receptor binding and activating adenylyl cyclase directly with forskolin. Figure 6 demonstrates that in the presence of ISO forskolin leads to augmentation of the systolic Ca²⁺ transient only in cells from failing hearts. The additive effect of forskolin over ISO was to increase Ca²⁺ transient amplitude by 193 ± 15% in HF cells from 296 ± 44 to 528 ± 63 nmol l⁻¹ (Fig. 6Ba, P < 0.001, paired t test). In control cells on the other hand Ca²⁺ transient amplitude was not altered by forskolin, being 716 ± 109 nmol l⁻¹ in ISO and 688 ± 119 nmol l⁻¹ in forskolin (P = 0.86). The increase in Ca²⁺ transient amplitude in HF cells due to forskolin can be explained by increases in both I_Ca-L and SR Ca²⁺ content. I_Ca-L increased by 21.0 ± 6.2% (ISO, 13.0 ± 1.3 pA pF⁻¹; forskolin, 14.9 ± 1.3 pA pF⁻¹, P < 0.001, paired t test, n = 27 cells) whereas SR Ca²⁺ content increased by 28.9 ± 7.9% (Fig. 6Bb; ISO, 60.0 ± 5.5 μmol l⁻¹; FSK, 75.7 ± 3.7 μmol l⁻¹, P < 0.001, n = 22 cells). In addition to normalising Ca²⁺ transient amplitude, forskolin also increased SERCA activity in HF cells (Fig. 6Bc; k_SR ISO, 5.27 ± 0.56 s⁻¹; k_SR FSK, 6.6 ± 0.58 s⁻¹, P < 0.001). In control cells, however, forskolin had opposite effects reducing I_Ca-L (ISO, 21.0 ± 1.4 pA pF⁻¹; FSK, 17.8 ± 1.4 pA pF⁻¹, P < 0.05) and SR Ca²⁺ content (ISO, 75.9 ± 4.9 μmol l⁻¹; FSK, 70.3 ± 4.9 μmol l⁻¹, P < 0.05).

Figure 6. Direct activation of adenylyl cyclase restores the systolic Ca²⁺ transient in heart failure.

A, time course showing systolic Ca²⁺ transients in the presence of 100 nmol l⁻¹ isoprenaline and effect of 3 μmol l⁻¹ forskolin applied as indicated in a representative control (a) and heart failure (b) cell. B, mean data summarising the effects of isoprenaline (ISO, open bars) and 3 μmol l⁻¹ forskolin (FSK, hashed bars) on Ca²⁺ transient amplitude (a), SR Ca²⁺ content (b) and k_SR (c) in isoprenaline in control (black) and HF cells (red, n = 7–20 cells, 3–10 hearts). *P < 0.05 vs. control cell under same conditions. §P < 0.05 vs. ISO in same cell.

Discussion

The major findings in this article are that in the ovine tachypacing model of heart failure: (i) the smaller systolic Ca²⁺ transient is a result of decreased I_Ca-L, (ii) β-AR responsiveness is reduced in failing cells, and (iii) direct activation of adenylyl cyclase restores the systolic Ca²⁺ transient in failing cells to a level comparable to that observed in control cells during β-AR stimulation. At the level of molecular regulation, the changes in intracellular phosphatase and kinase expression or activity provide a mechanism for the impaired SERCA activity seen in HF whereas the increase in GRK-2 expression would contribute to the impaired β-AR responsiveness through β-AR receptor internalisation and desensitisation. Together these findings provide several potential novel therapeutic targets to improve cardiac contractility in patients with heart failure.

Altered intracellular Ca²⁺ regulation in heart failure

A reduction in the amplitude of the systolic Ca²⁺ transient is a common finding in various models of heart failure and cardiac dysfunction (Hobai & O’Rourke, 2001; Pogwizd et al. 2001; Díaz et al. 2004). In these studies a lower SR Ca²⁺ content explained the smaller systolic Ca²⁺ transient amplitude. However, the present study reveals two important aspects of how the systolic Ca²⁺ transient is reduced in this tachypacing model of heart failure. Firstly, in the absence of β-AR stimulation (control conditions) the reduction in Ca²⁺ transient amplitude can be attributed to the smaller I_Ca-L and is not due to changes in SR Ca²⁺ content. Secondly, and discussed in more detail below, during β-AR stimulation the smaller systolic Ca²⁺ transient in HF cells is at least in part due to a failure of SR Ca²⁺ content to increase adequately. Previous HF or cardiac dysfunction studies have reported either no change (e.g. Hobai & O’Rourke, 2001; Chen et al. 2002; Díaz et al. 2004) or reductions to I_Ca-L (Yao et al. 1998; Undrovinas et al. 1999; Bito et al. 2004). It is of note that the decrease in I_Ca-L observed in the present study could be an underlying mechanism maintaining the SR Ca²⁺ content under basal conditions in HF cells despite the decrease in SERCA-mediated Ca²⁺ uptake and increase in NCX-mediated Ca²⁺ removal from the cell (Trafford et al. 2001). This latter effect depends on the smaller L-type Ca²⁺ current reducing the fractional release of Ca²⁺ from the SR and thence depletion of SR Ca²⁺ content due to surface membrane (primarily NCX)-mediated Ca²⁺ removal from the cell during systole (Bassani et al. 1995; Trafford et al. 2000, 2001).

The increase in NCX-mediated Ca²⁺ removal is significant from at least three additional perspectives. Firstly, the enhanced rate of removal of Ca²⁺ from the cell across the surface membrane observed in HF compensates for the decrease in SERCA activity meaning that the rate of decay of the systolic Ca²⁺ is unaltered under basal conditions. Secondly, greater NCX activity is likely to increase the propensity for cardiac arrhythmias since any diastolic release of Ca²⁺ from the SR will lead to the generation of increased depolarising current (Pogwizd et al. 2001; Venetucci et al. 2008). Finally with respect to the rate of decay of the systolic Ca²⁺ transient, the observation that NCX is not stimulated by PKA-dependent mechanisms (this study and Ginsburg & Bers, 2005) means that there will be a reduced capacity for increasing the rate of decay of the systolic Ca²⁺ transient during β-AR stimulation thus contributing to systolic and diastolic impairment observed in response to catecholamine stimulation.

Influence of altered β-AR signalling in HF

The positive inotropic effect of β-AR stimulation in cardiac muscle is a result largely of increased SR Ca²⁺ content arising from phosphorylation of PLN and thence greater SERCA activity (Katz, 1980; Hussain & Orchard, 1997; Ginsburg & Bers, 2004; Dibb et al. 2007). The elevated circulating levels of catecholamines in HF patients (Benedict et al. 1996; Leineweber et al. 2005) are conceptually consistent with the use of this adrenergic capacity as a means to maintain cardiac contractility in a disease setting. We note, however, that in the intact tissue PLN phosphorylation is reduced and after cell isolation, both Ca²⁺ transient amplitude and SERCA activity are also reduced. A plausible explanation for the decreased SERCA activity and PLN phosphorylation appears to be increased intracellular protein phosphatase expression and reduced PKA activity. In contrast to other reports (e.g. Zhang et al. 2003; Wehrens et al. 2004) we find that the major cardiac Ca²⁺–calmodulin kinase isoform (CAMKIIδ) is not increased in this model of HF and therefore does not ‘compensate’ for the reduced PKA activity.

In addition to the smaller Ca²⁺ transients under basal conditions, we also note in HF a marked reduction in the responsiveness of the systolic Ca²⁺ transient, SR Ca²⁺ content and SERCA-mediated Ca²⁺ uptake (k_SR) to the β-AR agonist isoprenaline. An impaired responsiveness of the failing myocardium to another β-AR agonist, dobutamine, is a common clinical finding (Chattopadhyay et al. 2010) and is prognostic of poor outcome (e.g. Chaowalit et al. 2006). Given the approximately cubic dependence of the systolic Ca²⁺ transient on SR Ca²⁺ content (Trafford et al. 2001) the failure to increase SR Ca²⁺ content in HF cells to the same extent as in control cells during β-AR stimulation will have a greater effect on systolic Ca²⁺ in HF and provides further explanation for the smaller Ca²⁺ transient noted during β-AR stimulation. The increased intracellular phosphatase expression and reduced PKA activity will also probably contribute to the reduced β-AR responsiveness through impaired phosphorylation of key intracellular targets such as PLN and the L-type Ca²⁺ channel. Furthermore, β-AR receptor desensitisation and/or internalisation are likely consequences of the increased GRK-2 expression noted in this model. Consistent with an important role for β-AR kinases in HF progression, disruption of the β-AR kinase (GRK-2) signalling pathway using adenoviral or transgenic approaches results in improved cardiac function both in vivo and at the cellular level with increased β-AR reserve being noted (Esposito et al. 2000; Manning et al. 2000; Raake et al. 2008; Molina et al. 2009; Rengo et al. 2009). Importantly, however, none of the above studies undertook a detailed quantitative determination of the effect of these GRK-2 modification strategies in HF on cellular Ca²⁺ homeostasis and SR Ca²⁺ content.

Alterations to the L-type Ca²⁺ current in HF

In the present study the reduction of I_Ca-L is clearly important in reducing the amplitude of the systolic Ca²⁺ transient and is on its own sufficient to explain the smaller systolic Ca²⁺ transient observed in the absence of β-AR stimulation in HF cells. Additionally, the shift of the voltage dependence of activation and inactivation to more positive potentials may increase the likelihood of early afterdepolarisations due to the I_Ca-L window current associating more with the plateau phase of the action potential. Recent data have suggested that the L-type Ca²⁺ channel is important in HF from at least two additional standpoints: firstly, a reduction in I_Ca-L as observed in the present study has been shown to reduce infarct size and increase survival in a mouse model of myocardial infarction (Zhang et al. 2010); secondly, Völkers et al. (2011) suggest that prevention of normal GRK-2 signalling with a C-terminal peptide βARK-ct restores Ca²⁺ transients in failing myocytes by augmenting I_Ca-L. These latter effects were suggested to occur via removal of the inhibitory effects of G_βγ subunits on the L-type Ca²⁺ channel rather than through normalisation of cAMP-mediated intracellular effects. However, in this particular study these effects were observed in cultured myocytes where many morphological and functional changes are known to occur (e.g. Louch et al. 2004).

Study limitations

In common with all other studies involving the use of animal models of clinical conditions it should be borne in mind that differences in observations frequently occur. For example, I_Ca-L is decreased in the present ovine tachypacing model whereas in a canine tachypacing model I_Ca-L was unaltered (Hobai & O’Rourke, 2001). Whether this is a reflection of species-specific differences or the extent of HF progression remains unclear. Additionally, whilst our data investigate the expression of various protein phosphatases and kinases in heart failure this does not necessarily reflect similar changes to phosphatase or kinase activity. Moreover, the lack of information regarding the subcellular localisation of the various kinases, phosphatases and their endogenous regulators does not preclude the possibility that alternative sites on, for example, the RyR may be phosphorylated even though Ser2808 phosphorylation is found to be unaltered in the present study (Wittköpper et al. 2010; van Oort et al. 2010). Finally whilst the data are consistent with perturbed β-AR signalling in HF we cannot differentiate between this arising as a result of reduced β-AR receptor density as opposed to dysfunctional signalling between the receptor and adenylyl cyclase.

Conclusions

In this particular model of HF under basal conditions we provide evidence that increased surface membrane-mediated removal of Ca²⁺ from the cell completely compensates for the reduced Ca²⁺ uptake capacity of the SR Ca²⁺-ATPase. Despite the decrease in SERCA function and increase in NCX function, SR Ca²⁺ content is unaltered under these conditions and we suggest that this occurs as a result of the marked reduction in the L-type Ca²⁺ current. Furthermore, we provide unequivocal evidence that in HF dysfunctional β-AR signalling exists and impacts adversely on the cellular fluxes of Ca²⁺ resulting in reduced responsiveness of the systolic Ca²⁺ transient and SR Ca²⁺ content to β-AR agonists. This finding is consistent with clinical observations. Finally, the reported changes in expression of several kinases and phosphatases provide a likely mechanism for the observations reported here. Considering the likely subcellular compartmentalisation of the β-AR signalling cascades and responses to ISO and forskolin in HF myocytes, appropriate targeting of kinase and phosphatase activity within these cellular compartments could provide potential therapeutic strategies to increase cardiac contractility in response to catecholamines.

Glossary

Abbreviations

β-AR

β-adrenergic receptor

CAMKII

Ca²⁺-dependent calmodulin kinase II

CAMKIIδ

CAMKII δ isoform

CSQ

calsequestrin

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GRK-2

G-protein receptor kinase 2

HF

heart failure

ISO

isoprenaline

NCX

sodium–calcium exchanger

PKA

protein kinase A

PLN

phospholamban

PP1

protein phosphatase 1

PP2A

protein phosphatase 2A

RyR

ryanodine receptor

SERCA2a

sarco-endoplasmic reticulum Ca²⁺-ATPase

Ser16

serine 16 phosphorylation site of phospholamban

SR

sarcoplasmic reticulum

Thr17

threonine 17 phosphorylation site of phospholamban

Author contributions

Cellular experiments (S.J.B., A.W.T.); Western blotting (J.L.C., H.K.G., J.D.C.); animal model and in vivo measurements (A.W.T., M.A.H., S.J.B., M.C.S.H., J.D.C., D.J.G., M.A.R., H.K.G., K.M.D.); experimental concepts, direction and manuscript preparation (A.W.T., D.A.E., K.M.D., S.J.B.). All authors approved the final version of the manuscript. All experiments were performed at The University of Manchester.