Apelin and APJ orchestrate complex tissue-specific control of cardiomyocyte hypertrophy and contractility in the hypertrophy-heart failure transition

Abstract

The G protein-coupled receptor APJ is a promising therapeutic target for heart failure. Constitutive deletion of APJ in the mouse is protective against the hypertrophy-heart failure transition via elimination of ligand-independent, β-arrestin-dependent stretch transduction. However, the cellular origin of this stretch transduction and the details of its interaction with apelin signaling remain unknown. We generated mice with conditional elimination of APJ in the endothelium (APJ^endo−/−) and myocardium (APJ^myo−/−). No baseline difference was observed in left ventricular function in APJ^endo−/−, APJ^myo−/−, or control (APJ^endo+/+, APJ^myo+/+) mice. After exposure to transaortic constriction, APJ^endo−/− mice displayed decreased left ventricular systolic function and increased wall thickness, whereas APJ^myo−/− mice were protected. At the cellular level, carbon fiber stretch of freshly isolated single cardiomyocytes demonstrated decreased contractile responses to stretch in APJ^−/− cardiomyocytes compared with APJ^+/+ cardiomyocytes. Ca²⁺ transients did not change with stretch in either APJ^−/− or APJ^+/+ cardiomyocytes. Application of apelin to APJ^+/+ cardiomyocytes resulted in decreased Ca²⁺ transients. Furthermore, hearts of mice treated with apelin exhibited decreased phosphorylation in cardiac troponin I NH₂-terminal residues (Ser²² and Ser²³) consistent with increased Ca²⁺ sensitivity. These data establish that APJ stretch transduction is mediated specifically by myocardial APJ, that APJ is necessary for stretch-induced increases in contractility, and that apelin opposes APJ’s stretch-mediated hypertrophy signaling by lowering Ca²⁺ transients while maintaining contractility through myofilament Ca²⁺ sensitization. These findings underscore apelin’s unique potential as a therapeutic agent that can simultaneously support cardiac function and protect against the hypertrophy-heart failure transition.

NEW & NOTEWORTHY These data address fundamental gaps in our understanding of apelin-APJ signaling in heart failure by localizing APJ’s ligand-independent stretch sensing to the myocardium, identifying a novel mechanism of apelin-APJ inotropy via myofilament Ca²⁺ sensitization, and identifying potential mitigating effects of apelin in APJ stretch-induced hypertrophic signaling.

INTRODUCTION

Heart failure is a common cause of morbidity and mortality, accounting for more than $30 billion of health care spending per year in the United States alone. Therapies for heart failure currently target compensatory mechanisms that become maladaptive over time. Although mortality is reduced, the burden of disease remains significant. In addition, the utility of these therapies in the acute heart failure setting is lacking. The search for molecular targets that improve both acute and chronic disease continues.

Apelin was identified (26, 46) as one of the endogenous ligands for APJ, a G protein-coupled receptor cloned over a decade ago (31). Since its discovery, apelin has been noted to have many actions in the cardiovascular system, ranging from inotropy to vasodilatory properties, both directly in a nitric oxide- and endothelium-dependent manner (1, 4, 28, 30, 47) and indirectly through control of vasopressin release in the central nervous system (16, 57, 59). Apelin was first noted to play a role in cardiovascular disease through a “reverse heart failure” model in patients implanted with ventricular assist devices who later went on to transplantation (8). Apelin-null mice display depression in cardiac systolic function at rest and significant detriments in exercise capacity (8, 10, 24). Furthermore, administration of apelin can rescue multiple heart failure phenotypes (1, 14). Apelin acts both to inhibit hypertrophy and to support contraction in the myocardium (1). While the physiology of these findings is well established, the underlying cellular basis of these effects remains incompletely understood.

Consistent with findings in zebrafish (32, 40, 56), homozygous elimination of APJ in mice leads to frequent embryonic mortality and significant cardiac developmental abnormalities (56). Surviving APJ-null mice show no difference in baseline blood pressure, although APJ interacts directly with ANG II type 1 receptors (AT₁Rs) to antagonize its effects on blood pressure (2, 7, 10). Additionally, surviving APJ-null mice show modest decrements in baseline cardiac contractility as well as severe exercise limitations, similar to apelin-null mice (7, 24). However, elimination of APJ confers protection against hypertrophy and heart failure in response to transaortic constriction (TAC) (39). This counterintuitive finding may be due to a second function of APJ as a stretch sensor. When cells expressing APJ are stretched, it leads to increased β-arrestin signaling and a consequent hypertrophic response. This signaling pathway is antagonized by apelin-APJ binding (39).

One explanation for the differences observed in embryonic lethality, developmental effects, and differential myocardial responses to APJ versus apelin elimination may be the presence of multiple stimulators of APJ signaling, including multiple ligands and stretch transduction. For example, ELABELA/Toddler, described nearly a decade after apelin, binds APJ and is critical for vasculogenesis and normal development in zebrafish, independent of apelin (9, 25, 33). Additionally, stretch-induced, prohypertrophic APJ signaling through β-arrestins is induced by stretch, potentially leading to a paradoxical protective effect of APJ elimination in the myocardium (39), even as apelin itself is antihypertrophic via a non-β-arrestin-based APJ signaling mechanism. Although a downstream pathway for this surprising finding has been proposed (39), it remains to be determined whether this stretch function of APJ is localized in the myocardium or endothelium. Furthermore, the downstream mechanism of increased contractility due to apelin-APJ binding remains to be elucidated. Here, we used inducible tissue-specific elimination of APJ as well as in vitro cellular physiology to address these questions. We hypothesized that myocardial APJ is solely responsible for the observed stretch-induced prohypertrophic signaling, whereas endothelial APJ is not itself prohypertrophic. We also elucidate one mechanism by which downstream apelin-APJ signaling leads to increased Ca²⁺ sensitivity in the sarcomere without increasing Ca²⁺ transients, explaining the complex effects of apelin in supporting inotropy and countersignaling hypertrophy.

METHODS

Generation of tissue-specific APJ elimination.

All procedures involving animal use, housing, and surgeries were approved by the Stanford Animal Administrative Panel on Laboratory Animal Care. Animal care and interventions were provided in accordance with the Laboratory Animal Welfare Act.

All mouse strains were maintained in C57BL/6 background. The APJ^−/− (7), floxed APJ allele, APJ^F (7) (Fig. 1A), α-MHC-MerCreMer [purchased from the Jackson Laboratory (Bar Harbor, ME), termed α-MHC-Cre], and Tie2Cre (20, 45) strains were as previously described. For tissue-specific deletions, α-MHC-MerCreMer; APJ^F/F mice and Tie2Cre; APJ^F/F mice were generated to eliminate APJ in cardiomyocytes (termed APJ^myo−/−) and in endothelial cells (termed APJ^endo−/−), respectively. Robust description and confirmation of the α-myosin heavy chain (αMHC) and Tie2 promoters as cardiomyocyte- and endothelium-specific drivers of Cre expression, respectively, were first published in 2001 [myocardial α-MHC-MerCreMer (44) and endothelial Tie2Cre (20)]. These strains have been used successfully for the cardiomyocyte- and endothelium-specific knockout of many genes, as we demonstrate here as well (Fig. 1, B and C). Of note, given previously observed transient cardiomyopathy caused by tamoxifen induction (23), great care was taken to avoid confounding: we used tamoxifen-treated, α-MHC-Cre⁺ animals as APJ^myo+/+ controls, treated with low-dose tamoxifen over 12 days, and compared cardiac function measurements after ample recovery time (TAC was performed at least 4 wk after tamoxifen treatment).

Fig. 1.

Tissue-specific elimination of APJ. A: floxed APJ construct for APJ elimination. B: RNA expression in whole heart (APJ^endo+/+: 1 ± 0.6 vs. APJ^endo−/−: 0.003 ± 0.007, n = 8 and 6, P = 0.002) and in situ hybridization to APJ transcripts depicting expression of APJ in cardiomyocytes of APJ^endo−/− and APJ^endo+/+ hearts. Red stain indicated by black arrows indicates RNA expression; hematoxylin stained cardiomyocyte nuclei blue. C: RNA expression in whole heart (APJ^myo+/+: 1.0 ± 0.3 vs. APJ^myo−/−: 0.66 ± 0.2, n = 7 and 4, P = 0.05) and in situ hybridization to APJ transcripts depicting expression of APJ in cardiomyocytes of APJ^myo−/− and APJ^myo+/+ hearts. Red stain indicated by black arrows indicates RNA expression; hematoxylin stained nuclei blue.

Primers used for genotyping were as follows: APJ flox forward: 5′-CTGTCCAAAGCGTCCTCATT-3′ and APJ floxed reverse: 5′CCCATTGTTATGTGGTTTCC-3′, α-MHC-Cre forward: 5′-AGGTGGACCTGATCATGGAG-3′ and α-MHC-Cre reverse: 5′ATACCGGAGATCATGCAAGC-3′, and Cre reverse: 5′-TCCATGAGTGAACGAACCTG-3′ and Cre forward: 5′-TCGATGCAACGAGTGATGAG-3′.

Additional experiments were performed using C57BL/6 mice (Jackson Laboratory) with and without constitutive APJ knockout [APJ^+/+, APJ^−/− (18)]. These were used to generate primary cardiomyocytes for carbon fiber experiments, as described below (39).

Animal surgery and phenotyping.

For tamoxifen induction of the cardiomyocyte-specific CreER lines, male mice were intraperitoneally injected with tamoxifen at 4 wk of age (25 mg/kg wt daily for 12 days).

Mice underwent TAC as previously described (42) at 12–16 wk of age. Tissue-specific APJ knockout and control animals were divided into two groups: TAC and sham controls. Briefly, mice were anesthetized using an isoflurane inhalation chamber, intubated, and ventilated. After surgical exposure of the thoracic aorta, a 6.0 silk suture was placed between the innominate and left carotid arteries to induce a constriction diameter of ∼0.4 mm. In sham control mice, an identical procedure was conducted except that the aorta was constricted.

In vivo left ventricular (LV) systolic function was evaluated by echocardiography in the short-axis view, as previously described (42). Measurements occurred at 1 day before surgery (baseline), 7 days, and 14 days after surgery and then every 14 days before euthanasia and tissue collection. Male APJ^endo−/− and APJ^endo+/+ mice were followed until 10 wk after TAC, as a significant proportion of control mice had not progressed to heart failure until that time. APJ^myo−/− and APJ^myo+/+ mice were followed until 4 wk after TAC, as APJ^myo+/+ mice had started to display heart failure in that group at that time.

Upon euthanasia, heart weight, body weight, and tibia length were measured by standard methods. Hearts were paraffin fixed, sectioned, and mounted on slides. Trichrome staining as well as wheat germ agglutinin (WGA; rhodamine labeled) staining (1:200 in PBS, Vector Laboratories, Burlingame, CA) was performed. Fibrosis quantification and cell size measurements were performed using ImageJ after image capture at ×20.

Cardiomyocyte isolation.

Adult LV cardiac myocytes were isolated from 8- to 12-wk-old male C57BL6 [APJ^+/+ and APJ^−/− (18)] using collagenase type II (Worthington Biochemical, Lakewood, NJ). Experiments were performed immediately after isolation, with myocytes resuspended in a HEPES-buffered solution containing (in mM) 1 CaCl₂, 137 NaCl, 5.4 KCl, 15 dextrose, 1.3 MgSO₄, 1.2 NaH₂PO₄, and 20 HEPES (pH 7.4).

Cardiomyocyte sretch by the two-carbon fiber technique.

Cardiomyocytes were isolated, prepared, and mounted on the inverted microscope. A pair of micro carbon fibers each attached to miniature hydraulic manipulators (SM-28, Narishige, Tokyo, Japan) and computer controlled with a piezoelectric translator (P-621.1CL, Physik Instrument, Karlsruhe/Palmbach, Germany) mounted on a custom-made railing system (IonOptix, Milton, MA) were attached to single isolated cardiomyocytes robustly twitching in response to field stimulation (IonOptix, Milton, MA). Cells were axially stretched by the piezoelectric translator movement of the carbon fibers using custom software (MATLAB).

Single cardiomyocyte contraction and Ca²⁺ dynamics.

Cardiomyocytes were loaded with 0.2 µM of fluo-5f-acetoxymethyl ester (Molecular Probes, Eugene, OR) for 30 min and then allowed to incubate in dye-free HEPES-buffered saline with or without pyroglutamylated ([Pyr1])apelin-13 (10 nM) for an additional 30 min to allow for deesterification of the Ca²⁺ dye. Cardiomyocytes were electrically field stimulated at 1 Hz, and spatially averaged Ca²⁺ transients and sarcomere length were measured using a standard FITC cube (Chroma) using the HyperSwitch system (IonOptix). Fractional sarcomeric shortening (FSS) and fluorescence transients were normalized to ΔF/F units by taking the average proportional increase in fluorescence across 10 beats per condition per cell. The dye was assumed to be at near equilibrium with the Ca²⁺ transient. For Ca²⁺ transient and unloading sarcomeric shortening experiments, data were analyzed in a blinded fashion. [Pyr1]apelin-13 was used in all experiments and was obtained from American Peptide (Sunnyvale, CA).

Apelin infusion.

Apelin was dissolved in distilled, autoclaved, degassed water, frozen at −20°C at 100 µM, and aliquoted on the morning of use. Minipumps contained either apelin or saline. The pumps infused 2 ml over 7 days, after which mice were euthanized according to Animal Administrative Panel on Laboratory Animal Care guidelines.

Immunoblot analysis and in situ hybridization.

LV tissue was homogenized in 400 μl RIPA buffer, and the resultant protein concentration was measured by BCA assay (Thermo Scientific, Waltham, MA). Western blot analysis was performed using the following primary antibodies: anti-cardiac troponin I (cTnI; phospho-Thr¹⁴³, ab58546, Abcam, Cambridge, MA), anti-cTnI (phospho-Ser²² + Ser²³, ab58545, Abcam) at a dilution of 1:1,000, and anti-GAPDH (sc-25778, Santa Cruz Biotechnology, Santa Cruz, CA). Secondary antibody was horseradish peroxidase-conjugated anti-rabbit IgG antibody (1:5,000, no. 18-8816-33, Rockland Antibodies, Limerick, PA). For confirmation of tissue specificity of APJ elimination, in situ hybridization was performed on formalin-fixed, paraffin-embedded sections mounted on glass slides using RNAscope technology, with the probe specific to mouse Aplnr (ACDbio no. 436171 and RNAscope 2.5 HD Assay-RED).

Statistics.

Continuous variables were compared using Student’s t-test for two sample comparisons and ANOVA for multiple groups. Echocardiography data were analyzed using a paired Student’s t-test comparing fractional shortening (FS) within each animal before and after TAC.

RESULTS

Endothelium- and cardiomyocyte-specific APJ-null mice display normal cardiac size and function at baseline but exhibit differential responses to pressure overload.

At baseline, neither APJ^endo−/− nor APJ^myo−/− lines showed a difference in LV size or systolic function from APJ^endo+/+ and APJ^myo+/+ lines, respectively. APJ^myo−/− mice displayed a mean FS of 37 ± 2.8% (means ± SD, n = 8), and APJ^myo+/+ mice (α-MHC-Cre⁺, also tamoxifen treated) displayed a mean FS of 39 ± 5.7% (n = 8, P = 0.59). APJ^endo−/− mice displayed a mean FS of 39 ± 3.9% (n = 15), and APJ^endo+/+ mice displayed a mean FS of 40 ± 3.3% (n = 11, P = 0.41). There was also no difference in LV internal dimension at diastole, interventricular septal thickness at diastole, or posterior wall thickness at diastole at baseline (Table 1).

Table 1.

Echocardiographic characteristics

	APJendo−/− TAC	APJendo−/− Sham	APJendo+/+ TAC	APJendo+/+Sham
n	8	7	6	6
Baseline
FS, %	39.1 ± 6.2	37.9 ± 3.7	39.0 ± 6.2	41.0 ± 4.2
LVIDd, mm	3.5 ± 0.3	3.5 ± 0.5	3.7 ± 0.3	3.4 ± 0.4
IVSd, mm	1.0 ± 0.1	1.0 ± 0.2	1.2 ± 0.2	1.1 ± 0.1
PWd, mm	1.0 ± 0.1	1.0 ± 0.2	1.2 ± 0.3	1.1 ± 0.2
Heart rate, beats/min	532 ± 58	559 ± 76	531 ± 38	492 ± 66
Week 4
FS, %	31.0 ± 12.5	42.6 ± 6.7	34.5 ± 10.1	37.8 ± 6.6
LVIDd, mm	3.9 ± 0.6	3.4 ± 0.4	3.7 ± 0.3	3.7 ± 0.9
IVSd, mm	1.0 ± 0.2	0.83 ± 0.2	1.1 ± 0.1*	0.9 ± 0.05*
PWd, mm	1.1 ± 0.2	1.0 ± 0.4	1.1 ± 0.2	0.7 ± 0.2
Heart rate, beats/min	538 ± 28	534 ± 38	535 ± 88	522 ± 54
Heart weight/body weight (10 wk)	7.3 ± 2.3	4 ± 0.45	5.8 ± 1.2	4.6 ± 0.45
Heart weight/tibia length (10 wk)†	11.8 ± 3.2*	6.3 ± 0.8*	9.8 ± 1.6*	7.5 ± 1.0*
	APJmyo−/− TAC	APJmyo−/− Sham	APJmyo+/+ TAC	APJmyo+/+ Sham
n	3	4	5	3
Baseline
FS, %	36.1 ± 4.0	38.0 ± 3.3	37.6 ± 5.9	36.9 ± 3.1
LVIDd, mm	3.5 ± 0.1	3.6 ± 0.1	3.5 ± 0.5	3.4 ± 0.2
IVSd, mm	0.9 ± 0.1	0.65 ± 0.05	0.9 ± 0.2	0.73 ± 0.03
PWd, mm	0.9 ± 0.1	0.87 ± 0.09	1.1 ± 0.1	0.8 ± 0.1
Heart rate, beats/min	413 ± 15	529 ± 38	435 ± 70	508 ± 12
Week 4
FS, %	40.1 ± 10.6	38 ± 5.6	29.7 ± 8.2	37 ± 1.9
LVIDd, mm	3.5 ± 0.2	3.3 ± 0.3	3.4 ± 0.4	3.5 ± 0.2
IVSd, mm	0.9 ± 0.2	0.7 ± 0.08	1.1 ± 0.3	0.7 ± 0.06
PWd, mm	1.1 ± 0.2	0.8 ± 0.09	1.3 ± 0.2*	0.9 ± 0.09*
Heart rate, beats/min	444 ± 35	498 ± 70	421 ± 69	466 ± 35
Heart weight/body weight†	6.7 ± 0.7	5.5 ± 0.7	9.2 ± 1.8	5.3±0.4
Heart weight/tibia length†	9.0 ± 0.9	7.3 ± 0.8	12.2 ± 1.3*	8.3±0.4*

Values are expressed as means ± SD. TAC, transaortic constriction; FS, fractional shortening; LVIDd, left ventricular internal dimension at diastole; IVSd, interventricular septal thickness at diastole; PWd, posterior wall thickness at diastole.

^*P < 0.01 by post hoc Studentʼs t-test;

^†P < 0.05 by ANOVA.

Four weeks after TAC, APJ^myo+/+ mice treated with TAC had significantly reduced FS (37.6 ± 5.9% before vs. 29.7 ± 8.2% after TAC, means ± SD, n = 3, P = 0.01; Fig. 2A), whereas the systolic function of APJ^myo−/− mice was preserved (36.1 ± 4.0% before vs. 40.9 ± 10.6% after TAC, n = 4, P = 0.6; Fig. 2A). Posterior wall thickness was also greater in hearts of APJ^myo+/+ mice that underwen TAC compared with sham-treated APJ^myo+/+ mice (P = 0.004), whereas no difference in wall thickness was detected in APJ^myo−/− mice between TAC and sham-treated groups (Table 1). At 4 wk after TAC, the heart weight-to-tibia length ratio was significantly greater in APJ^myo+/+ TAC mice than in any other group (APJ^myo+/+ TAC: 12.2 ± 1.3 vs. APJ^myo+/+ sham: 8.3 ± 0.4, APJ^myo−/− TAC: 9.0 ± 0.9, and APJ^myo−/− sham: 7.3 ± 0.8, P < 0.01 by ANOVA; Table 1). This indicates that APJ^myo−/− mice were protected from the hypertrophy and LV dysfunction caused by TAC in APJ^myo+/+ mice.

Fig. 2.

Endothelium- and cardiomyocyte-specific APJ elimination differentially affect the cardiac response to pressure overload. A: 4 wk after transaortic constriction (TAC), APJ^endo−/− mice displayed a significant decrease in fractional shortening (FS; n = 8, P = 0.02), where APJ^endo+/+ mice did not (n = 6, P = 0.6). APJ^myo−/− mice did not display a decrease in FS at this time point (n = 3, P = 0.6), whereas APJ^myo+/+ mice did display decreased FS (n = 5, P = 0.01). Error bars show means ± SE. B: APJ^endo−/− mice displayed a significant increase in cardiomyocyte size 10 wk after TAC compared with sham-treated control mice (P < 0.01), and this difference was not greater than APJ^endo+/+ mice treated with TAC. Four weeks after TAC, APJ^myo−/− mice displayed a significant decrease in cardiomyocyte size compared with APJ^myo+/+ TAC-treated mice (P < 0.05). C: APJ^endo−/− mice displayed a trend toward increased interstitial fibrosis compared with APJ^endo+/+ mice 10 wk after TAC, whereas no significant change was seen in fibrosis in APJ^myo−/− or APJ^myo+/+ mice.

Conversely, APJ^endo−/− mice that underwent TAC showed reduced FS after 4 wk (39.1 ± 6.2% before vs. 31.0 ± 12.4% after TAC, n = 8, P = 0.03; Fig. 2A), whereas APJ^endo+/+ mice that underwent TAC did not have significantly reduced FS at this time point (39.0 ± 3.4% before vs. 34.5 ± 10% after TAC, n = 6, P = 0.6; Fig. 2A). A small number of APJ^endo+/+ TAC animals had failed at the 4-wk time point compared with most APJ^myo+/+ TAC animals. For this reason, APJ^endo animals were followed until 10 wk after TAC, at which time five of six APJ^endo+/+ TAC mice and five of eight APJ^endo−/− TAC mice displayed a FS of <30% (data not shown). There was an increase in interventricular septal thickness at diastole in APJ^endo+/+ TAC animals compared with sham animals at 4 wk after TAC (P = 0.02), although no difference was observed in the APJ^endo−/− groups (Table 1). At 10 wk after TAC, the heart weight-to-tibia length ratio was significantly greater in APJ^endo+/+ TAC mice versus APJ^endo+/+ sham mice and in APJ^endo−/− TAC mice versus APJ^endo−/− sham mice but not across genotypes (APJ^endo+/+ TAC: 9.8 ± 1.6 vs. APJ^endo+/+ sham: 7.5 ± 1.0, P < 0.01; APJ^endo−/− TAC: 11.8 ± 3.2 vs. APJ^endo−/− sham: 6.3 ± 0.8, P = 0.01; Table 1). This indicates that APJ^endo−/− show a more rapid worsening of FS and wall thickening in response to TAC but that end-stage measures of cardiac size and function did not differ between APJ^endo−/− and APJ^endo+/+ mice exposed to TAC.

APJ^myo−/− mice were clearly protected from cellular hypertrophy 4 wk after TAC, as measured by WGA cell membrane stain. In response to TAC, the increase in cardiomyocyte size in APJ^myo−/− compared with sham-treated animals was significantly smaller compared with APJ^myo+/+ animals [APJ^myo−/− (n = 4): 1.32 ± 0.03-fold increase over sham (n = 5, 1 ± 0.05-fold, P = 0.006) vs. APJ^myo+/+ (n = 5): 1.94 ± 06-fold increase over sham (n = 3, 1 ± 0.2-fold, P = 0.008), and APJ^myo−/− vs. APJ^myo+/+ after TAC; P = 0.01; Fig. 2B]. This indicates that APJ expression in the myocardium is necessary for TAC-induced cardiomyocyte hypertrophy. Neither APJ^myo−/− mice nor APJ^myo+/+ mice displayed significant fibrosis on trichrome staining 4 wk after TAC. APJ^myo−/− mice after TAC displayed only 2 ± 0.4% total fibrosis area (n = 4), similar to sham APJ^myo−/− mice that displayed 2.2 ± 0.2% fibrosis (n = 5). APJ^myo+/+ mice after TAC displayed 1.1 ± 0.2% fibrosis (n = 5), whereas sham APJ^myo+/+ control mice showed 0.8 ± 0.02% fibrosis (n = 3). Fibrosis in APJ^myo−/− and APJ^myo+/+ mice after TAC was significantly different, but the biological meaning of this is uncertain, given overall fibrosis is low in both groups (P = 0.03; Fig. 2C).

Both APJ^endo−/− and APJ^endo+/+ mice displayed significantly increased cardiomyocyte size compared with sham-treated control mice [APJ^endo−/− (n = 6): 1.98 ± 0.18-fold increase over sham (n = 6, 1 ± 0.1-fold, P < 0.01) vs. APJ^endo+/+ (n = 8): 1.95 ± 0.16-fold increase over sham (n = 4, 1 ± 0.06-fold, P < 0.01; APJ^endo−/− vs. APJ^endo+/+ after TAC: P = 0.9; Fig. 2B]. Ten weeks after TAC, both APJ^endo−/− and APJ^endo+/+ mice displayed increased fibrosis compared with sham mice [APJ^endo−/− (n = 8): 5.6 ± 1.9% total fibrosis area vs. 0.3 ± 0.09% fibrosis in sham (n = 8), P = 0.03; APJ^endo+/+ (n = 6): 2.2 ± 1.1% total fibrosis vs. 0.4 ± 0.08% fibrosis in sham (n = 6), P = 0.04; Fig. 2C], with a higher mean fibrosis area in APJ^endo−/− mice compared with APJ^endo+/+ mice after TAC, although this did not reach statistical significance (P = 0.22; Fig. 2B). Therefore, in accordance with the cardiac function data above, while myocardial expression of APJ drives hypertrophy, conditional tissue-specific elimination of APJ from the endothelium does not protect from pressure overload hypertrophy and cardiac dysfunction.

APJ is necessary for stretch-induced augmentation of contraction in single cardiomyocytes.

Having established that the previously observed stretch-sensing role of APJ is specific to APJ expression in the myocardium and further that endothelial loss of APJ is associated with a poor ventricular response to pressure overload, we went on to examine potential mechanisms downstream of apelin-APJ signaling in single isolated cardiomyocytes (1). The G protein-independent stretch transduction of APJ has been previously attributed to β-arrestin-dependent signaling in APJ^−/− mice (39). This lack of hypertrophic signaling is thought to prevent the hypertrophy-heart failure transition in APJ^−/− mice in response to pressure overload. We chose to further investigate the mechanism of increased contractility observed downstream of apelin-APJ signaling.

We first asked whether APJ is necessary for increased contractility due to stretch at the cellular level as reported in APJ^+/+ single cardiomyocytes stretched with carbon fibers (41). To measure contractility, we used a metric of FSS, which is defined as the change in sarcomere length with contraction divided by its baseline length. We then sequentially stretched cardiomyocytes and determined the slope of increase in FSS per percent increase from initial sarcomere length [referred to here as stretch-augmented FSS (SAFSS)]. We found that APJ^−/− cardiomyocytes augmented their contractility significantly less with stretch than APJ^+/+ cardiomyocytes [APJ^−/− SAFSS = 0.37 ± 0.04 and APJ^+/+ = 0.84 ± 0.05, P = 0.04 (mean slope ± SE; n = 3 mice/group); Fig. 3A].

Fig. 3.

Apelin lowers Ca²⁺ transients in isolated cardiomyocytes and simultaneously increases myofilament sensitivity, whereas length-dependent activation by APJ is independent of Ca²⁺ transients. A: representative sarcomere length tracings and slopes of fractional sarcomere shortening (FSS) per increase in diastolic sarcomere length (stretch) [APJ^−/− FSS/%: 0.37 ± 0.04/% sarcomere length (SL) and APJ^+/+: 0.84 ± 0.05/% SL, P = 0.04 (mean slopes ± SE, n = 3 mice/group]. B: change in Ca²⁺-induced Ca²⁺ release in APJ^−/− compared with APJ^+/+ cardiomyocytes [APJ^−/− ΔF/F: 1.2 ± 0.43 and APJ^+/+ ΔF/F: 0.54 ± 0.2, P = 0.2 (means ± SE, n = 3 mice/group)]. No difference in ΔF/F was found within genotype between baseline and stretch. C: apelin significantly decreased peak Ca²⁺ transients in APJ^+/+ cardiomyocytes (49% reduction in ΔF/F in apelin-treated animals, apelin: n = 39 cells from four animals and saline: n = 23 cardiomyocytes from four animals in the control group, P = 3 × 10⁻⁴). D: immunoblot analysis of cardiac troponin I (cTnI) phosphorylation in apelin- and saline-treated mice using anti-phospho (Ser²² and Ser²³) antibody against the PKA-mediated site demonstrated decreased phosphorylation in the presence of apelin (P < 0.01). No difference was found in PKC-phosphorylated residue Thr¹⁴³ (n = 6 per group). E: immunoblot showing fluorescence detection of antibodies to phospho-cTnI (Ser²² and Ser²³), phospho-cTnI (Thr¹⁴³), and GAPDH in cardiac tissue of APJ^myo+/+ and APJ^myo−/− mice before and after transaortic constriction (TAC). Fluorescence quantification of phospho-cTnI (Ser²² and Ser²³, open bars) and phospho-cTnI (Thr¹⁴³, solid bars) normalized to GAPDH fluorescence is shown (bars show means ± SD of arbitrary units). *P < 0.01.

We went on to investigate whether this difference in contractile response to stretch is associated with changes in the peak Ca²⁺ transient in APJ^−/− and APJ^+/+ cardiomyocytes. We measured the Ca²⁺ transient as the change in fluorescence with contraction as a proportion of baseline fluorescence of calcium dyes (ΔF/F). We found a decrease in the Ca²⁺ transient in APJ^−/− cardiomyocytes compared with APJ^+/+ cardiomyocytes [APJ^−/− ΔF/F: 1.2 ± 0.43, APJ^+/+ ΔF/F: 0.54 ± 0.2, P = 0.2 (means ± SE, n = 3 mice/group); Fig. 3B]. No difference was discovered in ΔF/F within genotype between baseline and stretch (Fig. 3B).

Apelin-APJ binding decreases the Ca²⁺ transient in spontaneously beating cardiomyocytes and alters TnI phosphorylation of Ser²² and Ser²³, residues known to control myofilament Ca²⁺ sensitivity.

Given this lack of change in Ca²⁺ transients with stretch, we hypothesized that the documented inotropic effect of apelin-APJ signaling might also be independent of changes in cytoplasmic Ca²⁺ concentration (7, 14). Further supporting such a hypothesis, APJ is necessary for apelin’s effect on cardiomyocyte contractility (35) and it does not activate G_s (3, 55) but stimulates extracellular acidification (46, 51). To investigate this, we assessed Ca²⁺ transients in freely contracting APJ^+/+ cardiomyocytes with and without apelin. Surprisingly, apelin significantly decreased peak Ca²⁺ transients in APJ^+/+ cardiomyocytes (49% reduction in ΔF/F in apelin-treated animals, apelin: n = 39 cells from four animals, saline: n = 23 cardiomyocytes from four animals in the control group, P = 3 × 10⁻⁴; Fig. 3C).

Since apelin decreases cardiomyocyte Ca²⁺ transients, its observed inotropic effect remains to be explained. One way in which force generation can be augmented without increasing prohypertrophic Ca²⁺ transients is myofilament sensitization. During β-adrenergic stimulation, PKA phosphorylates cTnI at its NH₂ terminus (human Ser²³ and Ser²⁴; mouse Ser²² and Ser²³), leading to decreased myofilament Ca²⁺ sensitivity and increased lusitropy (37). Furthermore, cTnI phosphorylation at these and other residues is increased in hypertensive heart failure, perhaps representing a maladaptive mechanism of disease (13, 58). Apelin-APJ signaling leads to decreased PKA activation via the inhibition of adenylyl cyclase and cAMP production, and we have recently reported increased diastolic relaxation and an increase in FSS in single cardiomyocytes exposed to apelin (35). Thus, we hypothesized that apelin-APJ binding leads to decreased PKA-dependent phosphorylation of cTnI at Ser²² and Ser²³, increasing myofilament Ca²⁺ sensitivity and thus also contractile capability in the absence of increased Ca²⁺ transient.

To test this hypothesis, we treated APJ^+/+ C57BL/6 mice with apelin infusion for 7 days and examined cardiac tissue by phospho-Western blot analysis. We found that, unlike Thr¹⁴³, a residue phosphorylated by PKC (a protein kinase canonically activated by apelin-APJ signaling), treatment with apelin reduced phosphorylation at PKA phosphorylated at Ser²² and Ser²³ compared with saline treatment [n = 6 mice/group, apelin-treated density at PKA residues (cTnI-Ser²²+ Ser²³): 0.76 ± 0.03, saline- treated density: 1.27 ± 0.06 (means ± SE, arbitrary units of density normalized to GAPDH expression, P = 0.01); Fig. 3D]. APJ^myo−/− mice did not display increased cTnI phosphorylation at Ser²² and Ser²³ at baseline or after TAC (Fig. 3E), indicating that endogenous levels of apelin do not overcome maladaptive phosphorylation at this residue in heart failure.

DISCUSSION

Here, we present data demonstrating that 1) cardiomyocyte-specific elimination of APJ is protective against pressure-overload heart failure, 2) this protection extends to abrogation of cellular hypertrophy and fibrosis, 3) APJ is necessary for the stretch-dependent force increase in single cardiomyocytes, 4) the presence of apelin reduces Ca²⁺ transients regardless of preload, and 5) apelin may preserve inotropy through myofilament Ca²⁺ sensitization via reduced cTnI Ser²³ and Ser²⁴ phosphorylation.

These findings add to our understanding of the fundamental mechanisms of cardiac hypertrophy and failure. They confirm that afterload-induced stretch signaling is mediated by cardiac, not endothelial, APJ. They further underscore apelin’s therapeutic potential as an agent capable of antagonizing this stretch-APJ-β-arrestin-based hypertrophy signaling. Prior data have shown that constitutive deletion of APJ during development was frequently lethal and leads to marked developmental cardiac abnormalities (7). Given the importance of APJ for cardiac development, a question arose over the extent to which the function of these remaining mice reflected a developmental abnormality or adult loss of APJ. Later data showed a protective effect of APJ elimination, but the cellular basis of that effect remained speculative at that time (39). Here, we show that elimination of myocardial APJ in the adult is also protective, which underlines the likelihood of a developmentally driven explanation for cardiac decrements in the original APJ^−/− mice. It is also particularly important in light of recent findings that endothelial progenitor populations expressing APJ can compensate for one another during heart development (43).

Here, we provide mechanistic insights clarifying these findings, in particular, with respect to contractile response to stretch and associated changes in Ca²⁺ handling in the presence or absence of APJ. We demonstrate that loss of APJ significantly reduces stretch-augmented FS and show that no increase in Ca²⁺ transients is observed with stretch in either these cells or APJ^+/+ cardiomyocytes. These findings complement our recent work (35), which demonstrates that the increase in Frank-Starling gain observed with apelin treatment is mediated by a lusitropic effect on carbon fiber-stretched cardiomyocytes. Here, we show that APJ^−/− cells have a significantly smaller increase in FSS in response to stretch, suggesting that that increases in Frank-Starling gain are mediated by APJ’s stretch-sensing function. Our data showing decreased phosphorylation of cTnI at Ser²² and Ser²³ in response to apelin treatment are consistent with increased Ca²⁺ sensitivity and therefore reduced lusitropy, indicating that apelin’s lusitropic effect is mediated via a different mechanism and potentially relies on a combinatorial effect of myofilament residue phosphorylation to increase contractility and lusitropy simultaneously (36, 58).

Given the preserved ejection fraction of APJ^myo−/− mice after TAC, the loss of contractile response to stretch associated with APJ elimination in cardiomyocytes requires some consideration. First, it must be noted that single cell contractility assessments test the response of single myocytes outside of the tissue context of the hypertrophy-heart failure transition; they are not exposed to chronic pressure-induced hypertrophy, fibrosis, or cell death. It is possible, for example, that although a single cell exhibits less contractile response to stretch without APJ, this does not translate to an organ level measure of ejection fraction at one preload (as we observed in APJ^−/− mice). Further, it is possible that, because of decreased contractility, APJ^−/− cardiomyocytes are actually protected in the long term from the high basal cytoplasmic Ca²⁺ accumulation associated with the hypertrophy-heart failure transition. Finally, the small loss in contractility may not outweigh the detrimental effect of β-arrestin signaling downstream of APJ-stretch transduction, leading in sum to protection against the hypertrophy-heart failure transition.

In light of these findings, it is particularly interesting that in the absence of stretch, apelin decreases the peak Ca²⁺ transient (Fig. 3C). While stretch-dependent APJ signaling likely activates the β-arrestin and ERK1/2 pathways to induce prohypertrophic transcriptional program in cardiomyocytes (39), this is a Ca²⁺-independent mechanism that, to our knowledge, would not be expected to directly activate calcineurin signaling. As phasic Ca²⁺ concentration changes are well known to be associated with eccentric cardiomyocyte hypertrophy via calcineurin activation (11, 51), this represents a contributory explanation for the antihypertrophic effect of apelin-APJ binding previously described (39). Potentially, this effect of apelin-APJ binding could bypass β-arrestin-ERK1/2-mediated concentric-hypertrophic signaling to ameliorate the hypertrophy associated with hypertensive remodeling (11). In addition, apelin-APJ binding leads to decreased PKA activation via G_i. PKA-dependent phosphorylation of ryanodine receptor 2 leads to increased Ca²⁺ transients (29). Thus, we expected and confirmed here that apelin should reduce Ca²⁺ transients due to its deactivation of cAMP/PKA signaling. One prior research group showed increased Ca²⁺ transients with apelin application to isolated rat cardiomyocytes (48). The discrepancy may be explained by the use of a different apelin isoform (apelin-16), which is not the predominantly active form in human hearts [as is [Pyr1]apelin-13, used here (28)] or may be due to different pacing frequency (0.5 Hz in the referenced report compared with 1 Hz here). We also used low-affinity fluorescent Ca²⁺ dyes to avoid stacking of Ca²⁺ transients with repeated stimulation.

Although our finding of apelin-induced decrease in Ca²⁺ transients is consistent with the known antihypertrophic effects of apelin-APJ binding (1, 39, 54) and PKA control of ryanodine receptor 2 phosphorylation (29), apelin’s observed inotropic effect remains to be explained (1, 8, 14). Although some data have implicated PKC-ε and increased ERK1/2 phosphorylation in apelin-mediated increased contractility (34), our understanding of the underlying mechanism at the level of Ca²⁺ handling and the sarcomere remains limited. Here, we examined myofilament phosphorylation as a potential intermediary of apelin’s inotropic effect, showing reduced Ser²² and Ser²³ phosphorylation of cTnI, consistent with increased Ca²⁺ sensitivity that could explain increased contractility despite lower Ca²⁺ transients.

Certainly, the control of myofilament Ca²⁺ sensitivity by phosphorylation is complex. Multiple phosphorylation sites on TnI control contractility and Ca²⁺ sensitivity. A significant body of in vitro and in silico work exists to suggest that TnI phosphorylation is dynamic between disease states and that unique residues confer different and sometimes opposing effects on contractility (27, 38, 50, 58). For Ser²² and Ser²³ specifically, phosphorylation at this site has consistently been found to decrease myofilament Ca²⁺ sensitivity (12, 13, 37), suggesting that apelin, in fact, increases myofilament Ca²⁺ sensitivity by reducing phosphorylation at this site. That the immediate inotropic effect of apelin is not mediated by increased Ca²⁺ transients but is due to differential myofilament Ca²⁺ sensitization is novel. It is also consistent with the known extracellular acidification and intracellular alkalinization resulting from apelin binding that leads to a leftward shift in the Ca²⁺-contractility curve (15, 22). This would make apelin one of only a handful potential inotropic therapeutics to work through nonadrenergic signaling by directly affecting myofilament Ca²⁺ sensitivity, the other major therapeutics being pimobendan, levosimendan, and omecamtiv mecarbil. Apelin is unique in this class of potential therapeutics, as, on top of its inotropic, antihypertrophic, and vasodilatory effects, it has been implicated in protection against vascular remodeling and diabetes as well as myocardial infarction and ischemic stroke (5, 6, 17, 49, 52, 53, 59).

Overall, these data address fundamental gaps in our understanding of apelin-APJ signaling in heart failure and cardiac hypertrophy. We have developed tissue-specific elimination of APJ to localize the ligand-independent stretch-sensing function of this receptor to the myocyte. Consistent with this, we show loss of the APJ receptor decreases the contractile response to stretch. We offer evidence for a novel mechanism of apelin-APJ-induced contractility via cTnI phosphorylation-mediated Ca²⁺ sensitivity in the face of reduced phasic Ca²⁺ release, helping to explain mitigating effects of apelin in APJ stretch-based signaling and providing a rationale for the antiarrhythmic effect. Taken together, these results confirm the fundamental role of APJ as a stretch sensor in both acute and chronic increased afterload, localize its stretch transduction to the myocardium, and further support the rare therapeutic combination afforded by apelin-APJ signaling: increased cardiac output due to improved contractility and decreased afterload and simultaneous protection against the deleterious consequences of chronically raised phasic Ca²⁺ in the hypertrophy-heart failure transition.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grants F32-HL-134233-02 (to V. N. Parikh) and R01-HL-105993-03 (to E. A. Ashley).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

E. A. Ashley is a Founder of Personalis and DeepCell, Inc., and an advisor for SequenceBio and Genome Medical.

AUTHOR CONTRIBUTIONS

J.L., C.S., C.W., D.B., T.Q., and E.A.A. conceived and designed research; V.N.P., J.L., C.S., C.W., M.Z., D.N.C., Z.G., and Y.H. performed experiments; V.N.P., J.L., C.S., C.W., M.Z., K.S., and E.A.A. analyzed data; V.N.P., J.L., C.S., C.W., A.C.Y.C., P.S.T., P.R.-L., T.Q., and E.A.A. interpreted results of experiments; V.N.P. and J.L. prepared figures; V.N.P. drafted manuscript; V.N.P., A.C.Y.C., K.S., D.B., P.R.-L., T.Q., and E.A.A. edited and revised manuscript; V.N.P., J.L., C.S., P.R.-L., T.Q., and E.A.A. approved final version of manuscript.