Cardiac function and electrical remodeling of the calcineurin-overexpressed transgenic mouse

Abstract

Objective: To study the specificity of contractile phenotype and electrophysiological remodeling in transgenic (Tg) mice with cardiac directed calcineurin (phosphatase 2B) overexpression and evaluate a possible negative role of chronically activated calcineurin in β-adrenergic mediated contractile response. Methods: The patch-clamp technique was used to characterize electrophysiological properties of action potentials and inward rectifier (I_K1), and transient outward potassium currents (I_to). The analysis of the contractile performance was carried out on isolated retrograde perfused hearts at constant aortic pressure. Results: Tg mice demonstrated a hypercontractile phenotype characterized by a profound β-adrenergic hypo-responsiveness at 2.0 mM [Ca²⁺]_o. Transgenic cardiomyocytes showed marked action potential prolongation (209% in APD₉₀) with increased I_to,peak and I_sus and decreased protein expression level of Kv1.5 and Kv2.1. Lowering [Ca²⁺]_o to 0.75 mM restored the β-adrenergic response, indicating that the calcineurin/calmodulin/adenylyl cyclase (AC) pathway may not be directly responsible for the blunted β-adrenoreceptor mediated inotropism. Conclusions: Calcineurin overexpression leads to development of a hyperdynamic phenotype with a cellular profile of increased calcium influx. This type of functional hypertrophic remodeling is accompanied by a negative feedback regulation between increased calcium handling and β-adrenergic contractile activation.

Adrenergic (ant)agonists, Contractile function, Hypertrophy, K-channel, Membrane currents

Time for primary review 32 days.

1 Introduction

Calcineurin is a Ca²⁺/calmodulin activated serine/threonine phosphatase, which plays a significant role in cardiac hypertrophy as a sensing molecule that links alterations in calcium handling and the genetic program of hypertrophic growth. Calcineurin, as other calmodulin-dependent enzymes, is activated by prolonged, low amplitude increases in basal concentration of calcium and signals hypertrophy by dephosphorylation of transcription factors of the NFAT (nuclear factor of activated T cells) family. It is presumed that translocation of activated NF-AT3 to the nuclei leads to genetic reprogramming and initiation of the hypertrophic transcriptional response [1]. In analogy with activated T-cells in which calcineurin initiates an immune response, NF-AT3 may upregulate cytokines of the interleukin-6 family such as cardiotropin-1 and leukemia inhibitory factor, which are strong inducers of cardiomyocyte hypertrophy [2]. Cardiac specific overexpression of the catalytic subunit of calcineurin in transgenic mice was shown to induce profound hypertrophy characterized by a two- to four-fold increase in the heart size, which rapidly progresses to dilated heart failure [1]. The contractile phenotype of calcineurin overexpressors was not studied and the specific role of this pathway in functional hypertrophic remodeling remains to be investigated. Calcineurin plays an important role in diverse signal transduction pathways, as an opposing effector of kinase actions involving transcriptional as well as nontranscriptional mechanisms. Suppression of cyclic adenosine monophosphate (cAMP) signaling is required for T-cell activation by calcineurin [3]. Several transcription factors, including inducible cAMP early repressor, NF-KB, c-Jun N-terminal kinase and NFAT, have been identified as potential mediators of negative regulations of cAMP [4–6]. The mechanism of inhibition of cAMP signaling in T cells appears to be mediated by a transcriptional induction of phosphodiesterase 7 [7]. In addition, in excitable cells, initiation of the cAMP signal by adenylyl cyclase (AC) may be dynamically controlled by Ca²⁺/calcineurin and thus provide evidence for a novel mode of tuning the cAMP-dependent cellular response by protein phosphorylation/dephosphorylation cascades [8]. The effect of calcineurin on cAMP synthesis appears to be associated with the expression of a novel AC isotype IX. Human AC IX is expressed in vital organs, including the heart. In rodents AC IX is expressed in a wide variety of tissues including brain, heart, skeletal muscle and pancreas but its role in physiological control remains to be explored [9,10].

Using AC as a dynamic target of signal interaction, calcineurin has been shown to act as a Ca²⁺/calmodulin operated feedback inhibitor of the β-adrenergic receptor-evoked cAMP response in different cell types (as reviewed in Ref. [11]).

Very little is known whether the latter represents a physiologically relevant mechanism that might contribute to the well-known blunted β-adrenergic responsiveness in animal models of cardiac hypertrophy and failure as well as in human pathological conditions [12–14]. The functional versatility of structurally different ACs allows for simultaneous activation/or inhibition by multiple mechanisms. Calcium itself can provide a negative regulation of cAMP-dependent contractile responses by interacting with Mg²⁺/Ca²⁺ sites on the enzyme [15].

The aim of the present study was to evaluate the contractile phenotype, cellular electrophysiological profile of this hypertrophic mouse model and determine whether the chronically activated phosphatase 2B could be the possible cause leading to β-adrenergic hyporesponsiveness. Lowering the extracellular calcium concentration was chosen as a means to dissect the role of activated calcineurin and Ca²⁺ mobilized from the extracellular pool as different signals capable of providing putative feedback inhibition of cAMP-dependent contractile response.

2 Methods

2.1 Animals

All experiments were carried out under a protocol approved by the Institutional Animal care and Use Committee. The animals, overexpressing the catalytic subunit of calcineurin, were generated as previously described [1]. Mice analyzed in this study were heterozygous for the activated calcineurin transgene. One heterozygote transgene positive mouse is bred with a strain-matched wild-type mouse. In this manner, every litter contains approximately 50% heterozygous transgenic mice and 50% wild-type mice for direct comparison. As has been shown previously, calcineurin transgenic mice demonstrate a dramatic increase in heart rate and dilation of the ventricular chambers with increasing age (8–10 weeks) [1]. We used 4–7-week-old mice displaying significant hypertrophy, but they have not yet transitioned into heart failure. We chose this age group to avoid the characterization of secondary effects associated more with overt heart failure than the more casual effect related to chronic calcineurin activation.

2.2 In vitro study on cardiomyocytes

2.2.1 Isolation of ventricular myocytes

Cardiomyocytes were dissociated from the ventricles of 6–7-week-old nontransgenic and calcineurin-overexpressed transgenic mice. Briefly, the heart was perfused with Ca²⁺-free Tyrode's solution for 10 min, and with Ca²⁺-free Tyrode's solution containing collagenase type I (Worthington, 0.5 mg/ml) and BSA (fatty acid free 1 mg/ml) for 5–7 min by the Langendorff method at 37 °C. At the end of the perfusion the heart was removed and the ventricular tissues were mechanically disaggregated with a bore pipette. Ventricular cells were filtered through a nylon mesh, and stored at 4 °C in low Cl⁻, high K⁺ Kraft–Bruhe (KB) solution (in mM: l-glutamic-acid 50, KCl 40, taurine 20, KH₂PO₄ 20, MgCl₂ 3, glucose 10, EGTA 1, HEPES 10, pH 7.4 with 1 M KOH) until the electrophysiological study. Only Ca²⁺-tolerant cells with clear cross striations and without spontaneous contractions or significant granulation were selected for experimental studies.

2.2.2 Electrophysiological studies

All current recordings were obtained in the whole-cell, voltage-clamp configuration of the patch clamp technique by using 1.60 OD borosilicate glass electrodes (Garner Glass Company). Cell capacitance was calculated by integrating the area under an uncompensated capacity transient elicited by a 25 mV hyperpolarizing test pulse (25 ms) from a holding potential of 0 mV. Series resistance was within the range of 2 to 11 MΩ. Most of the data presented in these studies were obtained with electrodes having a resistance of 0.5–3 MΩ. After formation of a high resistance seal between the recording electrode and the myocyte membrane, electrode capacitance was fully compensated electronically before breaking the membrane patch.

I_Ca (inward calcium current) through the L-type voltage-dependent calcium channel (LVDCC) was elicited by depolarizing voltage steps (380 ms) from −50 to +60 mV in 10 mV increments from a holding potential of −60 mV. The recorded currents were filtered at 2 kHz through a four-pole low-bass Bessel filter and digitized at 5 kHz. Experiments were controlled using pClamp 5.6 software (Axon Instruments) and analyzed using Clampfit 6.0.3. Ca²⁺ currents were recorded using an external solution containing, in mM: CaCl₂ 2, tetraethyl-ammonium chloride (TEA-Cl) 135, 4-aminopyridine (4-AP) 5, glucose 10, HEPES 10, pH 7.3. The pipette solution contained (mM): cesium aspartate 100, CsCl 20, MgCl₂ 1, Mg-ATP 2, Na₂-GTP 0.5, EGTA 5, HEPES 5, pH 7.3 with 1 N CsOH.

I_to was elicited by depolarizing voltage steps (680 ms) from −40 to +80 mV in 10 mV increments from a holding potential of −40 mV at a frequency of 0.5 Hz. In these studies, the transient outward current was separated from the other currents by the following procedures. The holding potential was held at −40 mV to inactivate the sodium current. I_Ca was largely blocked by 0.3 mM CdCl₂ included in the recording solution. The external solution for recording I_to was (in mM): NaCl 138, KCl 4, CaCl₂ 2, MgCl₂ 1, glucose 10, HEPES 10, NaH₂PO₄ 0.33, pH 7.4. The patch clamp electrodes contained (in mM): potassium-glutamate 120, KCl 10, MgCl₂ 2, HEPES 10, EGTA 5 and Mg-ATP 2, pH 7.2 with KOH. EGTA was included in the pipette solution to minimize Ca²⁺-induced inactivation. Ventricular myocytes exhibit two components of transient outward current: a Ca²⁺-independent 4-AP-sensitive current and a Ca²⁺-dependent current. The Ca²⁺-dependent component was suppressed by the inclusion of 5 mM EGTA in the intracellular solution.

Steady-state inactivation parameters of I_to,peak were obtained with a double-pulse protocol using 250 ms (or 1000 ms) prepulses to voltages from −100 to +40 mV from a holding potential of −80 mV. Peak currents at the test pulse to +40 mV were normalized to the maximal transient outward current elicited by the test pulse and fitted individually for each cell by a single Boltzmann distribution. The time course of decay for I_to in both control and transgenic myocytes was determined directly using a least-squares fit of the inactivating section of the current trace (1600 ms) with a double exponential function by using the Chebyshev algorithm of CLAMPFIT. The time course of inactivation of the transient outward current (which has been shown to be independent of voltage) was best fit by an equation in the following form:  

I_K1 was evoked by a series of depolarizing voltage steps (480 ms) from −140 to −50 mV in 20 mV increments from a holding potential −40 mV.

For action potential (AP) recordings, cardiomyocytes were perfused with normal Tyrode's solution composed of (in mM): NaCl 138, KCl 4, CaCl₂ 2, MgCl₂ 1, glucose 10, HEPES 10, NaH₂PO₄ 0.33, pH 7.4 with NaOH. The ‘physiological’ pipette solution consisted of (in mM): KCl 140, MgCl₂ 1, Mg-ATP 4, NaCl 5, HEPES 10, EGTA 2, pH 7.4 with KOH. Action potentials were recorded in the current-clamp mode by injecting suprathreshold current pulses through the patch-clamp electrode. Action potentials were recorded at 1 Hz. At least 15 steady-state action potentials were recorded at this cycle length. The action potential duration (50 and 90% repolarization: APD₅₀ and APD₉₀) was measured from the traces.

All the recordings were done after 3 min of membrane rupture at room temperature (22–24 °C).

2.3 Isolated retrograde perfused heart preparation

Mice were anesthetized intraperitoneally with 100 mg/kg sodium Nembutal and protected with 1.5 U heparin to prevent intracoronary microthrombi. The heart was rapidly excised, the aorta was canulated with a 20 gauge needle and perfused in a retrograde fashion at a constant hydrostatic pressure of 55 mmHg with a modified Krebs–Henseleit solution (KHS) containing (in mM): 118 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, 25 NaHCO₃, 0.5 EDTA 11 glucose. Buffer was equilibrated with 95% O₂ and 5% CO₂, giving a pH of 7.4. Hearts were bathed in the perfusate in a water-jacketed bath maintained at 37.4 °C.

A home-made fluid-filled catheter (PE-50) was inserted into the left atrium through the pulmonary vein, guided through the mitral valve opening, advanced into the left ventricle and forced through the ventricular apex. The proximal end of the catheter remained in the left ventricle. The distal end of the catheter was connected via fluid-filled polyethylene tubing to a further Baxter pressure transducer. Aortic pressure in the aortic outflow catheter was measured through a sidearm. A micrometer clamp was attached to the outflow catheter so that the Starling resistance (aortic pressure) could be monitored. Aortic flow was measured with a graduated biuret in milliliters per minute. Coronary flow was allowed to drip from the pulmonary artery and measured by an electronic balance to determine coronary flow rates. Following instrumentation, hearts were allowed to equilibrate for 30 min.

Heart rate, left ventricular pressure (LVP) and the mean coronary perfusion pressure were continuously monitored using a Biobench (‘National Instruments’) DAQ (Data Acquisition) system and digitized at 1 kHz. The pressure curve was used to calculate the rate of pressure development (+dP/dt) and decline (−dP/dt), time to peak pressure (TPP) and time to half of relaxation (TR_1/2). TPP and TR_1/2 were normalized with respect to peak LVP since they are dependent upon the extent of pressure development. Lowering the extracellular calcium concentration from 2.0 to 0.75 mM tested the role of calcium influx. Experiments at 0.75 mM did not require special protocols or additional modifications for obtaining a functioning heart.

2.4 Drug infusion

Identical dose–response curves to isoproterenol were obtained at 0.75 and 2.0 mM [Ca²⁺]_o. Isoproterenol was added to the KHS entering the heart using a microperfusion pump (multispeed infusion pump, model 600, Harvard Apparatus). Hearts were exposed to each dose level for 60 s for maximal response. Effect of increased [Ca²⁺]_o (up to 4.0 mM) was tested after complete isoproterenol wash-out when contractile values and heart rates were restored to control level. Forskolin was tested in separate sets of experiments. Drugs were obtained from Sigma Chemicals (St. Louis, MO, USA).

2.5 Immunoblot analysis

A polyclonal antibody recognizing AC5 and AC6 protein (Santa Cruz Biosciences) was used in Western blots conducted on cardiac homogenates. Left ventricular samples were homogenized in 1 ml cold Tris buffer (10 mM Tris HCl, 1 mM EDTA and 1% SDS, protease inhibitors, pH 7.4) followed by centrifugation at 16,000 g for 20 min at 4 °C. The supernatant was subjected to Western blotting with AC5/6 antibody. To determine the protein level expression of Kv1.5 as well as that of the other K⁺ channels, Kv2.1, Kv1.4, Kv4.2, equal amounts of solubilized (50 μg) membrane proteins from 6–7- and 13–14-week-old mouse ventricle were isolated using a previously described method [16]. Proteins (200 μg for AC5/6 and 50 μg for K⁺-channel) were separated on 4–15% SDS–polyacrylamide gels (PAGE) and transferred to a nitrocellulose membrane (Hybond ECL, Amersham Pharmacia Biotech.). The membranes were incubated with anti-AC5/6 antibody diluted 1:100 and anti-Kv1.4 (ALOMONE), anti-Kv4.2 (CHEMICON International, Inc.), Anti-Kv1.5 (ALOMONE) and Anti-Kv2.1 (UPSTATE Biotech.) primary antibodies (1:250, 1:200, 1:100 and 1:400 final dilutions, respectively, in TBS-T with 5% nonfat dry milk). Primary antibodies were detected with anti-mouse IgG, horseradish peroxidase diluted 1:1500 (Amersham Life Science) for AC5/6 and with anti-rabbit IgG, peroxidase-linked species-specific whole antibody from donkey, diluted 1:5000 for K⁺-channel antibodies. The antibodies were visualized by enhanced chemiluminescence (ECL Western blot detection reagent, Amersham Pharmacia Biotech.).

To ensure equivalent loading of proteins, the nitrocellulose membrane was probed with the primary antibody for mouse anti-GAPDH (glyceraldehyde-3-phosphatase dehydrogenase) (CHEMICON International, Inc., 1:200).

The degree of labeling was quantitated by a computer program (AlphaImager™) and all measurements were normalized to protein levels of GAPDH and the value was expressed in relative ratio.

2.6 Statistical analysis

Data are expressed as mean±S.E.M. Between group statistical comparisons were made using the unpaired Student's t-test. The statistical significance of the difference between control and serial hemodynamic changes within groups was determined by one-way ANOVA and multiple comparisons with the Turkey procedure. The differences in the dose response curves were analyzed by two-way ANOVA. P<0.05 was considered significant.

3 Results

3.1 Cardiac function under baseline conditions

To investigate the effects of calcineurin overexpression on cardiac function a group of aged-matched Tg and Ntg mice 4–7 weeks of age were subjected to physiological and pharmacological analyses. The functional consequences of calcineurin overexpression were determined in the isolated retrograde perfused heart preparation and compared to the Ntgs, under identical afterload conditions (55 mmHg mean aortic pressure). The baseline functional parameters for hearts from Tg and Ntg animals are summarized in Table 1. In the Tg heart +dP/dt, as a measure of contractility, was increased from 3300±259 to 5488±515 mmHg/s (P<0.001). Diastolic pressure and −dP/dt (lusitropic or relaxation index) were increased, probably because these parameters are dependent upon the extent of peak intraventricular pressure (IVP). In the hyperdynamic hearts without diastolic decompensation the relaxation parameters were also enhanced. The Tg hearts showed a markedly reduced heart rate compared with age-matched Ntgs (223±32 vs. 381±11 b/min, respectively, P<0.001). The spontaneous frequency of contraction affects contractility and relaxation due to the well-known phenomenon of frequency potentiation and restitution. Because there was a substantial difference of intrinsic heart rate in Tg versus Ntg mice, a number of hearts from each group were tested at a wide range of frequencies. When the contractile parameters were normalized between groups for the differences in heart rate, a significant difference in the left ventricular maximal rate of pressure development (Fig. 1) and other parameters (Table 2) were also detected, indicating that the intrinsic differences in contractility in the Tg hearts can be observed over a wide range of stimulation frequencies.

Fig. 1

Relation between stimulation frequency and LV +dP/dt in Tg and Ntg isolated retrograde perfused hearts. The HR-LV dP/dt relation was biphasic with an initial positive slope (ascending limb) and a subsequent negative slope (descending limb); *P<0.05 versus 4 beats/s.

Table 1

Baseline contractile parameters of the Ntg and Tg mouse heart

Hemodynamic parameter	Nontransgenic (Ntg) (n=6)	CN-transgenic (Tg) (n=4)
Intraventricular pressure (IV) systolic (mmHg)	89.2±1.2	112.9±15.4*
Intraventricular pressure (IV) diastolic (mmHg)	−7.8±2.9	−19.6±5.4*
Contraction +dP/dt (mmHg/s)	3300±253	5488±515*
Relaxation −dP/dt (mmHg/s)	2662±242	4575±671*
Heart rate (beats/min)	381±11	223±32**
TPP (ms/mmHg)	0.35±0.3	0.24±0.01*
TR1/2 (ms/mmHg)	0.77±0.06	0.50±0.03*
Body weight (g)	18.72±0.74	17.58±0.88
Heart weight (mg)	92±20	190.7±20.4*
Heart/body weight ratio (×1000)	4.9±2.0	11.3±4.8

+dP/dt (mmHg/s), maximal rate of left ventricular pressure development; −dP/dt (mmHg/s), maximal rate of left ventricular pressure decline; TPP (ms/mmHg), time to peak pressure normalized to peak pressure; TR_1/2 (ms/mmHg), time to half relaxation normalized to half relaxation pressure. The indicated values are mean±S.E.M. *P<0.05, **P<0.01.

Table 2

Frequency-dependent changes in contractile parameters in Tg and Ntg isolated retrogradely perfused mouse heart

Stimulation frequency (Hz)	LVSP (mmHg)		LVDP (mmHg)		−dP/dT (mmHg/s)
	Ntg (n=4)	CN-Tg (n=4)	Ntg (n=4)	CN-Tg (n=4)	Ntg (n=4)	CN-Tg (n=4)
4	85.8±10.5	106.8±11.6	−6.9±5.0	−11.2±0.9	−2540±158	−4553±330
5	94.2±10.2	110.3±10.2	−7.4±4.7	−17.2±1.9	−2697±136	−4694±265
6	92.2±2.6	112.7±8.4	−8.3±4.9	−22.1±0.6*	−2860±311	−5837±468
7	90.8±0.9	110.1±5.9	−8.3±5.2	−24.6±0.3*	−2963±409	−6451±369*
8	88.3±1.6	106.8±6.2	−8.3±5.0	−25.5±0.8*	−2744±378	−6720±283**
9	84.9±3.3	101.3±8.6	−8.6±5.9	−25.4±0.8*	−2806±422	−6820±310**
10	82.1±3.8	98.7±9.7	−7.9±5.2	−24.9±1.3*	−2812±433	−6651±404**
11	79.3±4.7	89.5±9.8	−7.4±5.8	−25.0±1.5*	−2794±430	−6520±296**
12	75.8±4.4	81.4±10.7*	−4.9±5.5	−20.7±0.7*	−2641±429	−5926±245**
13	70.6±6.1	75.1±10.3*	0.8±4.4	−14.3±1.2	−2530±416	−5390±190**
14	62.6±3.7	71.1±8.6*	5.4±2.9*	−13.8±1.2	−2262±312	−4739±243

LVSP, left ventricular systolic pressure (mmHg); LVDP, left ventricular diastolic pressure (mmHg); −dP/dt, first derivative of pressure decline (mmHg); *,**P<0.05 and 0.01, respectively, compared to values at 4 Hz.

The increase in contractility in the Tg hearts at basal heart rate did not attenuate the positive force–frequency relationship. It is known that increased stimulation frequency increases calcium entry into the cell per unit time, so that the ascending limb of the force–frequency relation can be attributed to the increasing amount of systolic calcium available to the contractile proteins. Increased calcium re-sequestration into the sarcoplasmic reticulum (SR) results in increased SR Ca²⁺ release during the subsequent contraction cycle. The flattened contraction–frequency relationship in nontransgenic hearts was enhanced compared to calcineurin overexpressors, especially with respect to composite parameters such as +dP/dt and −dP/dt, which reflect both the absolute magnitude as well as the time course of the force development. The pronounced ascending limb of the biphasic force–frequency relationship in the Tg hearts may reflect an increased SR capacity to resequestrate Ca²⁺. Frequency-induced potentiation of force of contraction provides evidence that the Ca²⁺-dependent contractile reserve is preserved in the Tg hearts.

To further investigate whether calcineurin overexpression can affect Ca²⁺-dependent contractile responses, we increased [Ca²⁺]_o from 2.0 mM up to 4.0 mM (Fig. 2). The positive inotropic effect observed after the increase in [Ca²⁺]_o confirmed that calcineurin overexpression did not lead to defects in the mechanism of calcium myofibril activation or in the ability of contractile proteins to bind calcium when the latter is delivered to the sarcomere.

Fig. 2

Effect of increased calcium concentration (4.0 mM) on +dP/dt in isolated retrograde perfused heart. The percentage of increase in +dP/dt of left ventricular pressure from baseline is depicted. Data represent mean±S.E.M. n=4 for increased [Ca²⁺]_o, NS, Ntg vs. Tg.

3.2 cAMP-dependent contractile responses

To study the effects of cAMP-dependent inotropic interventions, the hearts were exposed to increasing concentrations of the β-AR agonist isoproterenol and we assessed peak positive and negative left ventricular dP/dt under basal conditions (37 °C, no pacing, 2.0 mM Ca²⁺ in the perfusate). Changes in the indices of contractility (+dP/dt) and relaxation (−dP/dt) are displayed as changes from baseline in Fig. 3A and B.

Fig. 3

Concentration–response curves for the effect of isoproterenol in the isolated retrograde perfused heart at 2.0 mM [Ca²⁺]_o in the perfusion solution. n=4 for Ntg, n=4 for Tg, +dP/dt (A) and −dP/dt (B) were used as indices of contractility and relaxation and are displayed as relative changes from baseline. (C) Concentration–response curves for the effect of isoproterenol on heart rate. Values represent mean±S.E.M. *P<0.05, **P<0.001 Ntg vs. Tg. C represents the unstimulated control level.

The positive inotropic and lusitropic effect of β-AR stimulation was markedly less in the Tg hearts compared to the Ntgs, indicating that the β-adrenergic signaling pathway and subsequent inotropic response is significantly blunted in this model of cardiac hypertrophy. The Tg hearts displayed the usual positive chronotropic response to adrenergic stimulation (Fig. 3C), but the increase in the heart rate was not accompanied by a substantial increase in +dP/dt and −dP/dt.

Forskolin is known to be a direct activator of multiple isoforms of AC. As shown in Fig. 5, the positive inotropic effects of forskolin (10⁻⁶ M) were also reduced in the Tg hearts compared to the Ntgs. The data clearly indicate that the β-adrenergic signaling pathway is affected downstream of the β-ARs and the G proteins at the cAMP production level.

Fig. 5

Effect of forskolin (10⁻⁶ M) on maximal rate of pressure development (+dP/dt) at 2.0 and 0.75 mM [Ca²⁺]_o in the perfusion solution. The percentage of increase in peak rate of left ventricular pressure rise from baseline is depicted. Data represent mean±S.E.M. *P<0.05 Ntg vs. Tg. n=4 for Ntg and Tg, respectively.

It is known that, in general, the mouse like the rat heart has a higher intracellular [Na⁺] and [Ca²⁺] than other species and therefore functions in a type of ‘superphysiological’ calcium mode that can aggravate calcium overload and therefore hypertrophic remodeling. Lowering the calcium concentration in the perfusate decreases intracellular calcium and allows the mouse heart to respond to a variety of drugs [17]. Under conditions of [Ca²⁺]_o of 0.75 mM there was no statistically significant difference in contractility and relaxation between the Tg and Ntg hearts (Fig. 4). Furthermore, as expected, the maximal rate of pressure development was decreased in the Tg hearts from 5488±515 at 2.0 mM [Ca²⁺]_o to 2118±260 mmHg/s at 0.75 mM [Ca²⁺]_o, whereas in the Ntgs the drop in contractility was less pronounced from 3307±253 to 2467±159 mmHg/s. This observation suggests that the enhanced contractile performance as a result of calcineurin-induced hypertrophic remodeling depends on Ca²⁺ mobilized from the extracellular space and Ca²⁺ entry is a critical step in maintaining enhanced Ca²⁺ handling cycle. Certainly, we cannot exclude that the enhanced contractility of the Tg mouse heart is also due to mobilization of intracellular calcium stores as well. As stated above, lowering the calcium concentration in the extracellular perfusate from 2.0 mM Ca²⁺ to 0.75 mM allows for an almost complete restoration of the β-AR positive inotropic and lusitropic response in calcineurin-overexpressed transgenic model of cardiac hypertrophy (Fig. 6), in which a blunted response is detected at the normally used [Ca²⁺]_o. In contrast, lowering [Ca²⁺]_o to 0.75 mM did not influence the depressed heart rate, which remained low at 222±27 beats/min, and did not differ from that recorded at 2.0 mM [Ca²⁺]_o. The inotropic effect of forskolin observed at low calcium concentrations was similar between the Tg and Ntg (Fig. 5).

Fig. 6

Concentration-response curves for the effect of isoproterenol in the isolated retrograde perfused heart at 0.75 mM [Ca²⁺]_o in the perfusion solution. n=5 for Ntg, n=4 for Tg. +dP/dt and −dP/dt were used as indices of contractility and relaxation and are displayed. Values represent mean±S.E.M., NS Ntg vs. Tg. C represents the unstimulated control level.

Fig. 4

Maximal rate of pressure development (+dP/dt) and decline (−dP/dt) (A) and heart rate (B) at 0.75 mM [Ca²⁺]_o in the perfusion solution in Tg and Ntg isolated retrograde perfused hearts. n=4 for Tg, n=4 for Ntg. Data represent mean±S.E.M. **P<0.001 Ntg vs. Tg.

Our data have demonstrated that calcineurin overexpression leads to a phenotype with enhanced contractility and relaxation with a negative feedback effect of calcium, mobilized from the extracellular pool, on β-adrenoreceptor-mediated physiological responses. Calcium-inhibited AC5/6 is a cardinal signaling molecule to augment intracellular calcium cycling, leading to positive inotropism. On the other hand, calcineurin-activated nuclear import of NFAT is adversely regulated by several kinases, including PKA. The AC/cAMP/PKA pathway is supposed to be negatively regulated by hypertrophic stimuli initiating NFAT activation and translocation to the nuclei [6]. However, quantitative immunoblot assay of AC5/6 protein expression revealed that hypertrophic remodeling in calcineurin overexpressors was associated with elevated protein expression of AC5/6 (Fig. 7). This might be one of the mechanisms responsible for the increased baseline contractility.

Fig. 7

Western blot analysis of AC5/6 protein expression in Tg and Ntg hearts (200 μg protein/lane). Shown are representative results of Western blots with cardiac homogenates. Glyceraldehyde-3-phosphatase dehydrogenase (GAPDH) levels are shown as a loading control. Histogram shows average levels of AC5/6 which were determined by scanning individual immunoblots (n=3), *P<0.05.

3.3 Cellular electrophysiology

We examined the electrophysiological (EP) properties of cardiomyocytes from 6- to 7-week-old Tg and Ntg littermates. Fig. 8 displays action potentials recorded from normal and two hypertrophied ventricular myocytes and Table 3 summarizes the data on action potential characteristics. Control APs were typically short (spiked) and showed no evidence of a ‘plateau’, which is typical of the neonatal and the adult mouse heart. The average APD₅₀ and APD₉₀ values were significantly prolonged in the Tg compared with the Ntg mice. Despite changes in the APD, the resting potential did not differ between the two groups. The duration and the shape of the AP are the result of a precise balance of inward and outward currents that are active during the plateau phase. The role of K⁺ currents responsible for repolarization in the mouse ventricle was examined using the whole-cell patch-clamp technique. Barry et al. [18] have demonstrated the importance of I_to in controlling the waveform of the mouse ventricular action potential, but electrophysiological studies in ventricular myocytes isolated from Kv2.1 N216FLAG transgenic mice revealed that I_K,slow also has a prominent role [19].

Fig. 8

Differences in the action potential shape and duration in Ntg and Tg myocytes. Representative action potentials were recorded in single ventricular myocytes isolated from Ntg (A) and Tg mice (B, C) at room temperature at a stimulation frequency of 1.0 Hz.

Table 3

Action potential characteristics recorded from Ntg and Tg cardiac myocytes

	N	APD50 (ms)	APD90 (ms)	RP (−mV)
Ntg	14	9.4±0.3	19.6±1.2	78.9±1.1
CN-Tg	18	14.1±1.2*	41.0±3.9*	77.9±1.1

*

P<0.5.

Although the main focus of the study was to compare ventricular repolarization between Ntg and Tg mice, we also examined whether there was a difference in I_Ca. Our observation is in agreement with Yatani et al. [20], that is Tg myocytes showed enhanced I_Ca density and inactivation kinetics compared with Ntg (Fig. 9A and B).

Fig. 9

Comparison of I_Ca and IRK₁ currents in Ntg and Tg mouse ventricular myocytes. Representative I_Ca recordings obtained with 380 ms pulses in a Ntg (A) and Tg (B) cell. There were no significant changes in the inward rectifier K⁺ currents in Ntg and Tg animals. Families of representative current traces elicited from a holding potential of −40 mV using voltage steps of 500 ms from −140 to +100 mV in 20 mV increments from Ntg (C) as compared with Tg (D) myocytes at 7–8 weeks old. Averaged peak current density–voltage relationship plotted for cells from Ntg and Tg animals (E).

The I_K1 is important in maintaining the resting membrane potential and in shaping the contour of the action potential.

Representative I_K1 traces from Ntg (Fig. 9C) and Tg (Fig. 9D) myocyte are shown. I_K1 density–voltage relationships for Ntg and Tg mice are shown in Fig. 9E. I_K1 densities were not significantly different between Ntg and Tg cells over a broad range of membrane potentials, particularly the membrane potential range encountered during phase III repolarization of the action potential, from −40 to −80 mV. At −80 mV, I_K1 density is −3.6±0.4 pA/pF (n=32) in Ntg and −3.4±0.3 pA/pF (n=22) in Tg myocytes.

Representative current density traces of I_to recorded from myocytes of the Ntg and Tg mouse cardiomyocytes are shown in Fig. 10B–D (from 13-week-old Tg mice), respectively. The current–voltage relationships of the transient outward peak current (I_to,peak) and the current remaining at the end of the 680 ms pulse (I_sus, but certainly reflects the sum of I_ss (steady state) and I_K,slow.) for the same cells are summarized in Fig. 10A (upper and middle panels). The difference between the peak outward current and the I_sus (I_sus, include part of the I_to and I_K,slow) is defined as the transient outward potassium current (I_to) (Fig. 10A, bottom panel). Both I_to,peak and I_sus were normalized by the corresponding cell capacitance (C_m). The membrane capacitance was significantly larger in the Tg myocytes (269.1±12.8 pF, n=22) compared to the Ntg cells (187.9±11.3 pF, n=23). In both Ntg and hypertrophied cells, the threshold for the activation of I_to was −20 mV. Contrary to our expectation, the mean peak I_to density in the Tg cardiomyocytes was significantly larger than in the Ntg cells at all depolarization steps. At +40 mV, I_to,peak density increased from 21.5±1.4 pA/pF (n=25) in the Ntg to 28.0±1.9 pA/pF (n=16) (P<0.05) in the Tg group, and in 13–14-week-old Tg cardiomyocytes decreased to 17.9±3.0 pA/pF (n=4). Similar to I_to,peak density, the current density at the end of the 680 ms pulse step, I_sus, was significantly larger in the Tg than in the Ntg myocytes at V_m=30–80 mV. Although the I_to [I_to,peak−I_sus] in the Ntg myocytes does not differ significantly from the Tg myocytes (at +40 mV: Ntg 6.7±0.7 pA/pF vs. Tg 9.9±0.9 pA/pF; NS) there was a tendency toward an increased current density compared to the Ntg. We have evaluated the effects of mice with cardiac heart failure (age 13 weeks) on I_to to a lesser extent, but the results showed that I_to was decreased significantly (7.8±1.3, n=4) (Fig. 10D) compared to Ntg mice.

Fig. 10

Voltage-dependent activation of transient outward potassium current (I_to) in mouse ventricular myocytes. Plots showing the voltage dependence of the current density of I_to,peak (A, upper panel), I_sus (A, middle panel) and I_{to peak}−I_sus (A, lower panel). Families of I_to from Ntg (B) and Tg (C and D) myocytes were recorded during a series of voltage clamps from a holding potential of −40 mV to selected test potentials ranging from −40 to +80 mV.

Steady-state inactivation curves were determined in Ntg and Tg myocytes to compare the availability of I_to channels for activation and to determine whether the observed changes in current density are associated with altered kinetics. Steady-state inactivation parameters of I_to,peak were obtained with the double-pulse protocol with 250 ms prepulses to different conditioning potentials (from −100 to +40 mV) from a holding potential of −80 mV, and the interval between pulses was 5 s. Peak currents at the test pulse to +40 mV (500 ms) were normalized to the maximal transient outward current elicited by the test pulse. Under these conditions the residual steady-state current was approximately 50% of the peak current level, as illustrated in Fig. 11A (bottom panel, original current tracings from a Tg myocyte). In Fig. 11B, the normalized magnitude of I_to,peak was plotted against conditioning potentials. The data were fitted to the Boltzmann distribution. The mean values of V_0.5 (half inactivation potential) and k (slope constant) for Ntg and Tg cells were determined in the presence of 0.3 mM CdCl₂. Cadmium is known to cause a shift in the voltage dependence of both activation and inactivation of I_to toward more positive potentials, but the rate of inactivation is unaffected [21]. There was no significant difference in V_0.5 (Ntg, −23.9±1.9 mV, n=10 versus CN-Tg, −18.6±3.3 mV, n=13, NS) and k between the Ntg (7.8±0.5 mV) and the Tg (6.6±0.4 mV) cells, indicating that hypertrophy did not alter the fraction of channels available for activation at a given membrane potential. Interestingly, in the Tg myocytes the remaining steady-state current after the depolarizing protocol was slightly smaller compared with that of Ntg (the difference was still present between Tg and Ntg cells after using a 1 s prepulse in the steady-state inactivation protocol). This provides further evidence that more than one channel contributes to I_to in the gating process, and I_sus is probably one of them.

Fig. 11

Steady-state inactivation of transient outward potassium current (I_to). Voltage dependence of inactivation was assessed using a double-pulse protocol. From a holding potential of −80 mV, cells were depolarized to various conditioning potentials for 250 ms followed by a 500 ms long test pulse to +40 mV. Representative traces are from a Tg cardiomyocyte (A). Graph showing the voltage dependence of steady-state inactivation for Ntg (n=9) and hypertrophied Tg (n=13) myocytes. The data are normalized to the largest peak I_to current measured using a prepulse of −100 mV (I_t/I_max). The curves were fitted to a Boltzmann distribution (see ‘Methods’) (B). The total peak outward current includes a steady-state component that may not represent I_to.

We next compared the time course of inactivation of I_to in Ntg and hypertrophied Tg myocytes. The τ_fast and τ_slow are summarized in Table 4. At all test potentials, there were no significant differences in either τ_fast or τ_slow between the Ntg and Tg myocytes. The rapidly inactivating current is referred to as I_to and the slowly inactivating current as I_K,slow[22].

Table 4

Inactivation properties of transient outward current (I_to) in myocytes from Ntg and Tg mice

mV	Ntg		CN-Tg
	τfast (ms) (n=12)	τslow (ms) (n=13)	τfast (ms) (n=12)	τslow (ms) (n=13)
10	74.24±4.9	1275.27±298.2	77.56±7.6	1203.15±140.8
20	64.98±4.7	1162.78±360.5	74.02±8.6	895.50±83.7
30	60.44±4.0	785.39±202.0	66.23±6.6	738.41±91.0
40	58.67±2.7	659.47±142.5	65.87±5.0	603.03±30.2
50	50.98±2.6	511.05±88.5	62.33±4.7	553.01±38.7
60	60.66±2.7	619.38±164.8	61.25±4.5	561.56±34.7
70	61.84±3.0	456.33±97.8	61.08±4.3	501.68±54.5
80	64.28±3.0	470.68±117.2	61.09±4.7	512.35±41.6

Inactivation decay courses of outward currents recorded as described in Methods during 1600 ms depolarizing voltage steps to test potential +40 mV. The time courses were determined in individual Ntg and Tg myocytes using double-exponential fits to the decay phases of the currents. Inactivation time constants for the fast (τ_fast) and slow (τ_slow) components of inactivation are presented.

To determine if there were changes in the expression of voltage-gated K⁺ channel α-subunit proteins, Western blot analysis was performed on membrane proteins prepared from ventricle tissue. The same age groups were also used for electrophysiological studies. Typical Western blots with anti-Kv1.4, anti-Kv1.5, anti-Kv2.1 and anti-Kv4.2 K⁺ channel α-subunit-specific antibodies are presented in Fig. 12A. The density of the specific α-subunit bands was determined relative to the GAPDH bands, as described in the ‘Methods’ section. Quantitative analysis of Western blots revealed that Kv1.5 (underlying I_Kur) and Kv2.1 (encoding for I_K,slow) expression levels were reduced significantly by 68% (n=6) and 78% (n=8), respectively, in Tg ventricles (Fig. 12B) compared to those of Ntgs. Kv4.2 (encoding for I_to,f) and Kv1.4 (underlying I_to,s) expression levels were not significantly different from that of Ntg. These results confirm the electrophysiological data in this age group, which also showed no difference in their corresponding K⁺ currents. In our EP studies we did not separate the 4-aminopyridine-sensitive ultrarapid delayed rectifier K⁺ current (I_Kur) from the total potassium current and did not determine I_K,slow. The finding that Kv4.2 expression is reduced in 13-week-old Tg mice by 45% in parallel with the reduction in transient outward current (I_to) supports the suggestions that Kv4.2 (and Kv4.3) most likely underlies the major component of I_to.

Fig. 12

Comparison of different K⁺ channel expression in Ntg and Tg mouse ventricles. Representative Western blot analysis of Kv1.4, Kv1.5, Kv2.1 and Kv4.2 on proteins (50 μg/lane) isolated from two Ntg and two Tg hearts from each group. Arrows, position of K⁺ channel isoforms (A). Statistical evaluation of data obtained by densitometry of the immunoblots. Relative protein levels of K⁺ channel subunits normalized to GAPDH (n=4–8 in all groups), *P<0.05 (B).

4 Discussion

Left ventricular hypertrophy is an important compensatory mechanism in response to neurohormonal stimuli, volume or pressure overload and intrinsic factors that increase the basal concentration of Ca²⁺ and activate calcium-sensing signaling molecules in the heart, such as calcineurin. We observed increased systolic and diastolic indices of cardiac performance and an enhanced force–frequency relationship in the Tg hearts with calcineurin overexpression. These results demonstrate that calcineurin can be considered as one of the molecular determinants of hyperdynamic performance leading to cardiac hypertrophic remodeling. Calcineurin triggers a signal that is sufficient to activate a physiological adaptive response that may compensate and protect the heart against functional insufficiency under conditions of elevated afterload or disturbed contractility secondary to abnormal myofibrillar/cytoskeletal architecture. Reduced spontaneous frequency of contraction can be considered as one possible adaptation leading to a decreased energy demand in the highly hypercontractile myocardium by increasing the period of diastole. On the other hand, the reduced heart rate we observed may be a direct consequence of a shifted balance of phosphorylation/dephosphorylation systems providing continuous modulation of excitability and firing properties [23,24]. Reactive functional compensation and structural remodeling in the calcineurin-overexpressed hearts appears to reflect an important compensatory mechanism such that, without this adaptation, the heart could not sustain the pressure overload and might decompensate to failure more rapidly. As reported by Meguro et al. [25] the incidence of decompensation leading to heart failure increased with the inhibitory action of cyclosporine on the development of LVH (left ventricular hypertrophy). As recently shown, concentric hypertrophy and decompensated heart failure share a common signaling pathway, but calcineurin activation is predominant in stable hypertrophy versus heart failure [26]. In the calcineurin-overexpressed heart the structural remodeling seems not to be chiefly responsible for increased systolic performance since it was not present under conditions of decreased calcium influx. The intrinsic myocardial functional remodeling appears to be more important in maintaining hypercontractile performance.

Prolongation of the AP is well documented by many to be a prominent feature of the heart in hypertrophy or cardiac failure in a variety of species, involving altered expression of depolarizing and repolarizing currents [27]. In contrast to other models of cardiac hypertrophy that show a significant downregulation of I_to and I_K1[28–30], in the calcineurin-overexpressed hypertrophic model, we observed a significant increase in I_to,peak and I_sus but no significant change in I_to. Although there is evidence that I_to,fast is a key contributor in shaping the early phase of the cardiac ventricular action potential, I_to,s and the delayed rectifier currents (I_Kur and I_K,slow) also play a major role in repolarization of the mouse myocytes. According to model predictions, reduction in both I_to,fast and I_K1 current densities measured in terminal heart failure have only modest effects on the APD and the AP prolongation is primarily due to altered expression of intracellular Ca²⁺-handling proteins [31]. Our interpretation of the Western blot results is that the observed reduction in Kv1.5 and Kv2.1 α-subunits may contribute to the prolongation of the action potential, with the latter phase of repolarization being particularly affected. In membranes prepared from 13-week-old Tg mice the Kv4.2 α-subunit protein expression also decreased and did account for the decreased I_to current density in heart failure in this transgenic model. Taken together, our observations suggest that the attenuation of the Kv1.5 and Kv2.1 α-subunits and the increased I_Ca[20] may explain the marked action potential prolongation in transgenic cardiomyocytes. Previous electrophysiological recordings from cardiomyocytes isolated from transgenic mice expressing a mutant Kv2.1 α-subunit (Kv2.1 N216), which functions as a dominant negative, revealed that I_K,slow is selectively attenuated and led to marked increases in action potential duration [19]. The results of the present study on I_to and I_Ca[20] in 6–8-week-old Tg mice provide important but opposing roles in the prolongation of the APD. The observed changes in I_to,peak and I_sus should in fact contribute to a shortening of the action potential duration. Therefore, it remains conceivable that the increased I_to,peak may provide a ‘compensatory’ response to prevent excessive lengthening of the action potential duration and Ca²⁺ influx through LVDCC during excitation [32]. Overall, several reviews show that electrophysiological studies have failed to identify a uniform ionic mechanism that can be associated with different hypertrophic models [33,34].

According to our finding, the I_sus current was increased and τ_fast of the inactivation decay was slowed in the Tg myocytes compared to that of Ntg (although this fell short of statistical significance due to inter-cell variability). Interestingly, phorbol 12-myristate 13-acetate (PMA)-induced hypertrophy of cultured rat ventricular myocytes also resulted in an increase in I_sus[35]. As a major regulator of Ca²⁺ homeostasis and gene expression, CaMKII (Ca²⁺/calmodulin-dependent protein kinase-II) is essential for heart function and together with calcineurin are well-characterized downstream targets of calmodulin regulation. In a recent study, Passier et al. [36] showed that CaMKIV-overexpressing mice developed cardiac hypertrophy. Tessier et al. [37] suggested that CaMKII modulates the inactivation of voltage-dependent K⁺ channels. The most likely explanation for the elevated I_sus in the calcineurin-overexpressed Tg myocytes is that increasing [Ca²⁺]_i causes an activation of calmodulin which in turn activates CaMKII. Interestingly, in chronic human atrial fibrillation the Kv1.5 expression was reduced [38] and δ-CaMKII protein expression was significantly enhanced [37]. It has been reported that Kv1.5 (the main molecular basis for I_sus in human atrial myocytes) has consensus sites for CaMKII [39], raising the possibility of direct phosphorylation of these channels. Therefore, it remains possible, according to our present study, that CaMKII indeed can be responsible for the observed changes in I_sus in Tg cardiomyocytes, but additional investigation is warranted.

A new electrophysiological profile in myocardium with chronically activated calcineurin can contribute to the enhanced calcium influx, leading to a hypercontractile phenotype. In addition, the augmented calcium influx through LVDCC acts as a trigger for SR Ca²⁺ release and there is evidence of an interaction between calcineurin and Ryr2 increasing SR calcium loading [40,41].

Direct mechanisms, whereby the calcineurin-directed dephosphorylation would augment calcium release or affect subcellular pools of reactive calcium in the cardiomyocyte, are unknown. In excitable cells, calcineurin plays a role in the negative feedback regulation of Ca²⁺ entry through the LVDCC [42], weakening intracellular Ca²⁺ signaling [43]. One of the possible pathway to increase contractility is AC/cAMP/PKA-dependent phosphorylation of calcium-handling proteins. On the other hand, phosphorylation of PKA responsive sites on the NH₂-terminal region of NFAT may cause export of NFAT from the nuclei to the cytoplasm and therefore inhibition of NFAT-mediated transcription [44]. Although the signaling pathway through AC/cAMP/PKA can negatively modulate the activity of transcriptional factors, directing the complex hypertrophic reprogramming, its activation appears to be necessary to enhance contractility and express the full hypertrophic phenotype. Our data suggest that, in the myocardium, calcineurin initiated transcriptional response leads to an enhanced AC5/6 protein expression which may be indicative of different patterns of genetic reprogramming of cAMP signaling in cardiac hypertrophic remodeling versus nonexcitable cells. It is of interest that hyperdynamic remodeling initiated by calcineurin did not appear to affect Ca²⁺-dependent contractile reserve, but might have predisposed the hearts to a general decline in responsiveness to receptor systems which operate at the level of PKC and βARK1 [45,46].

Upregulation of Ca²⁺-dependent PKC isoform(s) intrinsic to the myocardium with chronically activated calcineurin [47] can cause desensitization of the β-AR through activation of βARK1. Thus, PKC-mediated functional uncoupling may affect upstream signaling and may be responsible to some extent for the observed blunting or uncoupling of the β-AR signaling pathway. Direct stimulation of AC activity with forskolin also produced a reduced inotropic response, suggesting that cAMP production is significantly affected in the transgenic model under conditions of physiological extracellular calcium concentration and enhanced contractility. It is clear that by simply lowering the calcium in the perfusing solution, a restoration of the contractile response to β-AR stimulation occurs. The prevalence of AC5 and AC6 in the cardiomyocytes hints at a crucial role for the susceptibility of cAMP-dependent signaling to calcium inhibition [48]. Upregulation of the cellular signaling pathway through AC5/6 found in the transgenic heart may represent a part of maladaptive changes, leading to enhanced baseline contractile performance, and in addition serving as a mechanism limiting cAMP production through stimulated β-adrenoreceptors.

Our data indicate that Ca²⁺–AC interaction independent from calcineurin is the most dynamic target affected in the hypertrophied heart with β-AR hyporesponsiveness. Secondarily, increased calcium influx and upregulated AC5/6 may represent a transcriptional mechanism by which this regulatory cascade acquires calcium to inhibit cAMP production. The calmodulin/calcineurin/AC9 pathway can be ruled out as a possible potential target for increasing cardiac responsiveness to adrenergic stimulation, and targeting Ca²⁺-dependent AC isoforms may represent an additional approach to manipulate the effectiveness of the β-adrenergic pathway in cardiac hypertrophy and failure.

Acknowledgements

The authors appreciate the assistance of Dr. Sheryl E. Koch in the preparation of the manuscript. This work was supported by NIH PO1 HL22619 (A.S.) and by NIH training grant T32 HL07382 (A.S., N.N.P., M.R.) and by NIH grant HL60562 (J.D.M.).