Endogenous regulation of cardiovascular function by apelin-APJ

David N. Charo, Michael Ho, Giovanni Fajardo, Masataka Kawana, Ramendra K. Kundu, Ahmad Y. Sheikh, Thomas P. Finsterbach, Nicholas J. Leeper, Kavita V. Ernst, Mary M. Chen, Yen Dong Ho, Hyung J. Chun, Daniel Bernstein, Euan A. Ashley, Thomas Quertermous


Studies have shown significant cardiovascular effects of exogenous apelin administration, including the potent activation of cardiac contraction. However, the role of the endogenous apelin-APJ pathway is less clear. To study the loss of endogenous apelin-APJ signaling, we generated mice lacking either the ligand (apelin) or the receptor (APJ). Apelin-deficient mice were viable, fertile, and showed normal development. In contrast, APJ-deficient mice were not born in the expected Mendelian ratio, and many showed cardiovascular developmental defects. Under basal conditions, both apelin and APJ null mice that survived to adulthood manifested modest decrements in contractile function. However, with exercise stress both mutant lines demonstrated consistent and striking decreases in exercise capacity. To explain these findings, we explored the role of autocrine signaling in vitro using field stimulation of isolated left ventricular cardiomyocytes lacking either apelin or APJ. Both groups manifested less sarcomeric shortening and impaired velocity of contraction and relaxation with no difference in calcium transient. Taken together, these results demonstrate that endogenous apelin-APJ signaling plays a modest role in maintaining basal cardiac function in adult mice with a more substantive role during conditions of stress. In addition, an autocrine pathway seems to exist in myocardial cells, the ablation of which reduces cellular contraction without change in calcium transient. Finally, differences in the developmental phenotype between apelin and APJ null mice suggest the possibility of undiscovered APJ ligands or ligand-independent effects of APJ.

  • pressure-volume hemodynamics
  • cardiomyocyte

the apelin peptide and its G protein-coupled receptor (GPCR) APJ constitute a signaling pathway implicated in numerous cardiac and vascular functions. The APJ receptor was first cloned by homology to other known GPCRs, and the human APJ amino acid sequence is 31% identical to that of the angiotensin II type-1A receptor (26). Apelin is the only known ligand for APJ and is synthesized as a pre-pro-peptide consisting of 77 amino acids with shorter biologically active forms including the 36, 17, 16, 13, and 12 amino acids encoded in the COOH-terminal region (5, 35, 36). Although each of these isoforms have biological activity, the predominant isoform in the cardiovascular system appears to be [Pyr1]apelin-13 (21). Both apelin and APJ are predominantly expressed in the cardiovascular system. Studies have shown that apelin expression is primarily linked to the endothelium, and APJ expression to the myocardium, suggesting a paracrine mode of signaling (22, 32). However, additional studies have demonstrated that rat atrium and cultured cardiomyocytes can express apelin as well, thus postulating a role for autocrine signaling within the myocardium (30). Radioligand binding assays have shown that human heart tissue binds apelin with high affinity and with kinetics that suggest a single binding site (17).

The apelin-APJ pathway regulates systemic perfusion and has effects both on the heart and the vasculature. Exogenous apelin is a potent activator of cardiac contractility when added to isolated rat cardiomyocytes or intact heart preparations, as well as when administered in vivo to rats or mice (3, 12, 34). In the vasculature, apelin is a vasodilator in both the arterial and venous circulations, and these effects appear to be mediated at least in part by nitric oxide (7, 14, 19, 36). In heart failure models, apelin has a positive inotropic effect on failing myocardium (4, 15). We have previously reported that chronic infusion of apelin to mice increases cardiac output without causing hypertrophy (3). These numerous studies have well characterized the effects of exogenous apelin, but the role of endogenous apelin-APJ signaling is less clear. Plasma apelin levels increase in early heart failure but are diminished during functional decompensation, suggesting that endogenous apelin may be secreted in an attempt to increase contractility (6). Consistent with this hypothesis, we found that APJ is one of the most differentially expressed genes in the human failing heart (6). Aside from these direct effects on the heart and circulation, the apelin-APJ pathway has actions that suggest a beneficial role in heart failure. There is evidence for apelin antagonism of vasopressin release, and we have recently found that apelin signaling antagonizes the renin-angiotensin system in murine models of atherosclerosis (8, 9).

To understand the role of endogenous apelin-APJ signaling in cardiovascular physiology, we created mice lacking either the ligand (apelin−/−) or the receptor (APJ−/−) of this pathway. Here we report that both apelin−/− and APJ−/− mice have decreased basal cardiac contractility at the whole animal level with a more dramatic decrement at the cellular level, suggesting a direct role for endogenous apelin-APJ signaling in basal contractile function with the potential for autocrine signaling in vitro. Additionally, the differences in phenotype between the apelin−/− and APJ−/− mice suggest the possibility of an undiscovered ligand or ligand-independent effects of APJ.


Studies were approved by the Stanford University Administrative Panel on Laboratory Animal Care and conform to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication No. 85-23, Revised 1996).

Generation and genotyping of the apelin null mouse.

Both 5′ and 3′ homology regions isolated from a murine bacteria artificial chromosome clone were assembled in the pKO Scrambler NTKV-1901 vector (Stratagene) with the lacZ gene from the pPD46.21 vector. The neomycin phosphotransferase (neo) gene allowed for positive selection, and the herpes simplex virus-thymidine kinase gene positioned outside the region of homology allowed negative selection. TL-1 129SVJ embryonic stem cells were transfected, selected, and injected into C57Bl/6 blastocysts, and two chimeric male mice were obtained and bred to C57Bl/6 females (Jackson Laboratories, Bar Harbor, ME) to obtain heterozygous pups. Male heterozygous mice were bred to C57Bl/6 females five times before homozygous animals were generated. Thus all experiments were carried out on mixed background C57Bl/6-129SVJ animals and their littermate controls.

Generation and genotyping of the APJ null mouse.

To direct homologous recombination in the APJ locus, a single targeting construct was made containing C57Bl/6 APJ homology regions, three LoxP sites, and the neomycin and diphtheria toxin genes. The construct was transfected into B6-3 embryonic stem (ES) cells, targeted ES cell clones were transfected with pBSCre, and resulting clones found to have deleted both the neo and APJ genes were injected into C57Bl/6J-Tyrc-2J blastocysts to establish the targeted allele fully in the C57Bl/6 background. All breeding has been performed with mice on the C57Bl/6 background, and experiments were conducted on these offspring.


Two dimensional and M-mode echocardiography were carried out using the GE Vivid ultrasound platform with the small animal probe (13 MHz). Male and female mice aged 15 to 16 wk (apelin+/+, n = 13 and apelin−/−, n = 13; and APJ+/+, n = 12 and APJ−/−, n = 13) were anesthetized using 1% isoflurane in oxygen (1.5 l/min). After the anterior chest was shaved, mice were warmed using a heating pad. Short-axis views of the left ventricle (LV) were captured at the level of the papillary muscles in two-dimensional and M modes. Analysis was carried out offline, and fractional shortening was calculated as [LVEDD − LVESD]/LVEDD, where LVEDD is left ventricular end-diastolic diameter (measured at the tip of the mitral valve leaflets) and LVESD is left ventricular end-systolic diameter.

Magnetic resonance imaging.

Male and female mice aged 14–18 wk (apelin+/+, n = 8; and apelin−/−, n = 7) were scanned as previously described (3). Briefly, the mice were anesthetized with 2% isoflurane with an oxygen flow rate of 1 l/min. ECG leads were inserted subcutaneously, and respiration was monitored using a pneumatic pillow placed on the abdomen. The mouse body temperature was monitored by a rectal probe and kept at 37°C by heated air. Magnetic resonance images were acquired on a 4.7 T Oxford magnet controlled by a Varian Inova console (Varian, Palo Alto, CA). Image acquisition was gated to respiration and to the ECG R wave (SA Instruments, Stony Brook, NY). Coronal and sagittal scout images were followed by six contiguous 1-mm-thick, short-axis slices orthogonal to the interventricular septum. Nine cine frames were taken at each slice level and were acquired through more than one cardiac cycle, guaranteeing acquisition of systole and diastole. Measurements of end systole and end diastole were derived from these short-axis views of the LV at the level of the papillary muscles using ImageJ software (NIH, Bethesda, MD). Ejection fraction was calculated as [LVEDA − LVESA]/LVEDA, where LVEDA is left ventricular end-diastolic area and LVESA is left ventricular end-systolic area.

Cardiovascular hemodynamics.

We performed simultaneous measurements of pressure and volume using a specialized conductance catheter (Millar Instruments, Houston, TX). Male and female mice aged 10–14 wk (apelin+/+, n = 13 and apelin−/−, n = 12; and APJ+/+, n = 14 and APJ−/−, n = 12) were anesthetized with 1% to 2% isoflurane in oxygen (1.5 l/min). They were then intubated and ventilated at a tidal volume of 200 μl at 100 breaths/min (Harvard Apparatus, Holliston, MA). The left internal jugular vein was cannulated with polyethylene-10 tubing and a 10% albumin solution infused at 5 μl/min. Subsequently, the right carotid artery was cannulated with the pressure-volume catheter and advanced retrograde across the aortic valve into the LV along the long axis. Invasive blood pressures were acquired while the catheter was in the aorta before crossing the valve. The temperatures of the mice were constantly monitored by a rectal probe, and they were maintained at 37°C by a self-regulating heating pad (Fine Science Tools, Foster City, CA). The abdomen was then opened with a small transverse incision, and the inferior vena cava was visualized just superior to the liver and inferior to the diaphragm. After baseline loops were obtained, preload was reduced by manual occlusion of the inferior vena cava for 5 s and averaged for 3 occlusions/animal.

Graded treadmill running.

Male and female mice aged 10–14 wk (apelin+/+, n = 17 and apelin−/−, n = 12; and APJ+/+, n = 18 and APJ−/−, n = 14) were placed into an a Simplex II metabolic rodent treadmill (Columbus Instruments, Columbus, OH), which allows measurement of oxygen consumption (V̇o2) and carbon dioxide production (V̇co2) using a closed-chamber volumetric method of gas analysis. The respiratory exchange ratio (RER) was calculated as the ratio of V̇co2 to V̇o2. Calibration of the O2 and CO2 gas sensors was performed before each experiment using a standard gas mixture (0.5% CO2-20.5% O2-79% N2) and an offset gas (100% N2). Mice were subjected to a graded exercise protocol similar to those previously described in our laboratory (10). Briefly, mice were placed into the exercise chamber and allowed to equilibrate for 6 min. Treadmill activity was initiated at 7.5 m/min at a 4° incline and was increased 2.5 m/min and 2° of incline every 3 min until the mice showed signs of exhaustion. Mice were then allowed to recover in the chamber for 15 min. This exercise protocol was performed only once on each mouse so that conditioning was not a confounder.

Isolation of left ventricular myocytes.

Adult left ventricular myocytes were isolated from 12- to 14-wk-old male and female mice (apelin+/+, n = 40 cells from 5 animals and apelin−/−, n = 40 cells from 4 animals; and APJ+/+, n = 39 cells from 4 animals and APJ−/−, n = 30 cells from 3 animals) based on previously published protocols with modifications (25, 38). Briefly, mice were injected with heparin (100 IU) and anesthetized with pentobarbital sodium (100 mg/kg ip). The heart was excised and retrogradely perfused at 37°C with a calcium free solution containing (in mM) 120 NaCl, 14.7 KCl, 0.6 KH2PO4, 0.6 Na2HPO4, 1.2 MgSO4-7H2O, 4.6 NaHCO3, 10 Na-HEPES, 30 taurine, 10 2,3-butanedione monoxime, and 5.5 glucose (pH 7.0) for 4 min followed by an enzymatic digestion with collagenase type 2 (Worthington Biochemical, Lakewood, NJ) 394 U/mg. The digestion was initially performed in calcium free solution for 2 min, and CaCl2 was then added for a final concentration of 50 μM and perfused for 6 additional minutes. The heart was then removed, and the LV was first cut into small pieces and then further digested by gently pipetting with plastic transfer pipettes for 3–5 min. Stop buffer [calcium free solution + 12.5 mM CaCl2 + 10% bovine calf serum (Hyclone, Logan, UT)] was added, and the cell suspension was collected in a 15-ml tube and centrifuged at 30 g for 3 min. Myocytes were resuspended in stop buffer containing 100 mM CaCl2, allowed to rest for 2 min, and then centrifuged at 30 g for 3 min using a Sorvall RT6000B benchtop centrifuge. These steps were repeated using stop buffer with increasing CaCl2 concentrations until 1 mM was achieved. Experiments were performed with freshly isolated 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).

Myocyte shortening and relaxation.

Cell contraction properties of myocytes were evaluated with a video-based sarcomere spacing acquisition system (SarcLen; IonOptix, Milton, MA). Rod-shaped myocytes with clear striation patterns that were quiescent when unstimulated were chosen. Cells were placed in a culture chamber stimulation system (Cell MicroControls, Norfolk, VA), mounted on an inverted microscope (Nikon TE2000U; Nikon, Melville, NY), and electrically stimulated with suprathreshold voltage at 0.5 Hz and superfused with a HEPES-buffered solution at 25°C as previously described (11, 27). Changes in sarcomere length were recorded, and further analysis was performed using IonWizard software (IonOptix).

Calcium transient measurements.

A separate set of myocytes was loaded with 0.5 μM fura 2-acetoxymethyl ester (Molecular Probes, Eugene, OR) for 15 min. Cells were then washed and rested for an additional 40 min to allow the deesterification of the fura 2-ester. Myocytes were stimulated at 0.5 Hz and superfused with a HEPES-buffered solution at 25°C. Cells were excited at 340 and 380 nm, continuously alternated, at rates as high as 250 pairs/s using a HyperSwitch system (IonOptix). Background-corrected fura 2 ratios were collected at 510 nm. This ratio is independent of cell geometry and excitation light intensity and reflects the intracellular calcium concentration (24, 33). Experiments employed 12- to 14-wk-old male and female mice (apelin+/+, n = 16 cells from 2 animals and apelin−/−, n = 16 cells from 2 animals; and APJ+/+, n = 27 cells from 3 animals and APJ−/−, n = 19 cells from 2 animals).


Troponin I expression and Ser 23/24 phosphorylation was measured by immunoblotting. Whole tissue lysates were obtained by homogenization of snap-frozen left ventricular myocardium in lysis buffer. Samples were heated at 110°C on a heat block for 15 min, cooled at room temperature, and centrifuged for 10 min at 800 g. Supernatant was collected, and total protein concentration was quantified using the Molecular Probes Quant-IT kit (Invitrogen, Carlsbad, CA). Samples were electrophoresed on Nupage 4–12% Bis-Tris gel (Invitrogen), and proteins were transferred to a polyvinylidene difluoride membrane (Hybond-P; Amersham Pharmacia Biotech, Buckinghamshire, UK). Blocking was accomplished with 5% nonfat milk in Tris-buffered saline-Tween. Immunoblotting was performed using Troponin I and Phospho-Troponin I (Cardiac, Ser23/24) antibodies (Cell Signaling Technology, Danvers, MA) at 1:1,000 dilutions. The secondary antibody was goat anti-rabbit IgG at 1:5,000 dilution. Protein was visualized using the enhanced chemiluminescence method (ECL; Amersham). Western blots were quantitatively analyzed using ImageJ and were expressed as a ratio of phosphorylated Troponin I divided by Troponin I.

Analytical methods.

Pairwise comparisons were made using Student's t-test. One-way ANOVA was used in the event of multiple groups. Post hoc testing was according to the method of Fisher. Exact P values are stated in most cases. All data are presented in the text as means ± SE.


Apelin null mice show normal physiological development.

Apelin null mice were generated using a construct that replaced the start codon and signal sequence of the apelin gene with a bacterial lacZ reporter gene (Fig. 1A). Knockout mice were viable, fertile, and demonstrated a Mendelian pattern of inheritance with respect to the mutant allele. Germline transmission of the apelin targeted allele was evaluated by Southern blots with the wild-type allele migrating at 12 kb and the targeted allele migrating at 17 kb (Fig. 1B). Northern blots and RT-PCR demonstrated no evidence of apelin RNA in the mutant mice (data not shown). Postmortem inspection of major organs of 12-wk-old adult mice revealed no gross or histological abnormality. Organ weights were similar, and histology revealed no gross microscopic differences (data not shown). Cell counting of cardiac nuclei within predefined microscopy fields revealed no evidence of cardiac cellular hypertrophy (data not shown).

Fig. 1.

Targeting of the apelin and APJ loci. A: a construct encoding a nuclear targeted lacZ expression cassette was integrated by homologous recombination into the apelin locus, resulting in deletion of the first exon including the native ATG and leader peptide. Transfections were performed in 129SVJ embryonic stem cells, and targeted cells were microinjected into C57Bl/6 blastocysts. B: germline transmission of the apelin targeted allele in male mice. Male agouti offspring of chimeric founder mice were evaluated by Southern blot with an apelin genomic probe. The wild-type allele (+) migrates at 12 kb, and the targeted allele (−) migrates at 17 kb. There is only one allele for apelin in these male mice since apelin is on the X chromosome. C: a construct was created with loxP recombination sites flanking the APJ gene, and a loxP flanked neomycin resistance (neo) gene 3′ to the APJ gene. D: germline transmission of the APJ targeted allele was evaluated by PCR. The wild-type band (+) is 100 bp, and the targeted band (−) is 225 bp.

Deletion of APJ reveals a role in murine development.

To allow both complete and conditional deletion of the APJ locus in mice, we developed a targeting construct containing loxP sites flanking the APJ gene and the 3′ neomycin resistance gene (Fig. 1C). Homologous recombination was performed in C57Bl/6 embryonic stem cells, and expression of the Cre recombinase in successfully targeted cells led to the identification of clones that had deleted both the APJ and neo genes, or only the neo gene. Microinjection of these clones into C57Bl/6 blastocysts allowed development of one mouse line per recombination type, and the line with deletion of the APJ gene was employed in these studies.

Heterozygous APJ+/− animals were viable and fertile; however, the litter size for these animals was small, and the majority of pups failed to survive, due at least in part to poor mothering behavior. This phenotypic trait was even more pronounced in the homozygous knockout mothers. This problem was ameliorated by fostering the pups onto an outbred CD1 strain mother and was not further investigated. Even with survival of all pups resulting from a heterozygote × heterozygote cross, the litter sizes remained small. With 175 offspring studied, there were 106 heterozygotes, 52 wild-type, and 17 null mutants, reflecting a statistically significant lower number of homozygous animals (Chi square, P = 1.8 × 10−5). Preliminary studies of early stage embryos suggest multiple cardiac developmental defects and evidence of overall developmental failure (Anderson et al., manuscript in preparation).

Evaluation of APJ mRNA in various murine tissues did not identify a transcript in the APJ−/− animals, whereas a prominent APJ mRNA was identified in heart, lung, aorta, and kidney of wild-type animals (data not shown). Germline transmission of the targeted allele was evaluated by PCR with the wild-type band migrating at 100 bp and the targeted allele migrating at 225 bp (Fig. 1D). Primer sequences are as follows: APJ-F (CCAAAGGCAGGCATGAAGTAGTGG), APJ-ck-F (GCTGCAAAGGGCTGGAAGGAG), and APJ-R (GTGGTTTCCAGAGCAGGAGGCTTAG).

Imaging studies reveal differences in load-dependent measures of contractility.

Changes in fractional shortening and ejection fraction are readily available and widely accepted measures of contractility in mammalian cardiovascular physiology. Echocardiography using a 13-Mhz probe revealed significantly reduced fractional shortening in the APJ null mice compared with their littermate controls (P = 0.0002; Fig. 2A). This trend was also apparent in the apelin null mice (P = 0.09; Fig. 2B). Although the mean ejection fraction measured by magnetic resonance imaging was in the same direction in the apelin null mice, this was not a significant change (P = 0.23; Fig. 2C).

Fig. 2.

Imaging studies reveal modest differences in load-dependent measures of contractility. A: APJ null mice have significantly decreased fractional shortening by echocardiography (APJ+/+, 44.9 ± 2.7% and APJ−/−, 31.3 ± 1.6%; P = 0.0002, n = 12/group). B: apelin null mice reveal a nonsignificant trend toward decreased contractility using fractional shortening by echocardiography (apelin+/+, 40.4 ± 1.1% and apelin−/−, 37.9 ± 0.8%; P = 0.09, n = 13/group) and ejection fraction (C) by magnetic resonance imaging (apelin+/+, 87.9 ± 1.5%, n = 8 and apelin−/−, 83.9 ± 2.9%; P = 0.2, n = 7/group). *P < 0.05.

Apelin null mice have decreased load-independent contractility.

Since apelin affects vascular resistance as well as cardiac contractility, it is important to measure intact heart function in a way that is not as affected by changes in vascular loading conditions. Pressure-volume relationships are considered the gold standard for measuring load-independent contractility. Therefore, we performed pressure-volume loops and analyzed various regression parameters of systolic contractility including the end-systolic pressure volume relationship (ESPVR), preload recruitable stroke work (PRSW), and maximal elastance (Emax). We found that apelin null mice had significantly lower ESPVR (P = 0.0002; Fig. 3A) as well as a trend toward lower PRSW (P = 0.13; Fig. 3C) and Emax (P = 0.11; Fig. 3E). The APJ null mice had similar trends toward decreased contractility parameters, although none reached statistical significance (ESPVR, P = 0.15 and PRSW, P = 0.19; Fig. 3, B and D). Both apelin and APJ null mice had similar heart rates and blood pressures compared with littermate controls under conditions of anesthesia and ventilation (data not shown). Additionally, measures of load-dependent contractility such as the rate of rise and fall of left ventricular pressure were not different between groups (data not shown).

Fig. 3.

Apelin null mice have decreased intrinsic contractility and APJ null mice trend toward decreased contractility as measured by load-independent parameters. There was a significantly lower end-systolic pressure volume relationship (ESPVR) in the apelin null mice [A; apelin+/+, 11.1 ± 0.73 and apelin−/−, 6.9 ± 0.56 mmHg/relative volume units (RVU); P = 0.0002] and a trend toward decreased ESPVR in the APJ null mice (B; APJ+/+, 10.2 ± 0.59 and APJ−/−, 8.9 ± 0.51 mmHg/RVU; P = 0.15). Similar trends were seen with preload recruitable stroke work (PRSW) in the apelin null mice (C; apelin+/+, 44.1 ± 3.2 and apelin−/−, 36.5 ± 3.7 mmHg × RVU; P = 0.13) and APJ null mice (D; APJ+/+, 39.0 ± 3.7 and APJ−/−, 32.0 ± 3.4 mmHg × RVU; P = 0.19) as well as with maximal elastance (Emax) in the apelin null mice (E; apelin+/+, 25.2 ± 1.9 and apelin−/−, 20.9 ± 1.7 mmHg/RVU; P = 0.11) and APJ null mice (F; APJ+/+, 25.6 ± 1.6 and APJ−/−, 22.9 ± 2.7 mmHg × RVU; P = 0.38). Representative pressure-volume loops of apelin null (G) and wild-type (H) mice are shown. *P < 0.05.

Both apelin and APJ null mice have reduced exercise capacity.

To investigate the role of endogenous apelin-APJ signaling on functional capacity, both apelin and APJ null mice underwent graded treadmill exercise. This is a clinically relevant method of providing a severe, yet physiological, stress to the cardiovascular system. When compared with their littermate controls, both apelin and APJ-deficient mice manifested decreased exercise capacity as measured by treadmill time in minutes (apelin null, P = 0.03 and APJ null, P = 0.009; Fig. 4, A and B). Since motivation can be a significant confounder in exercise testing, maximum exercise was defined metabolically when mice achieved a RER greater than 1.0. At this point, CO2 production begins to exceed O2 consumption. Additionally, exercise conditioning was avoided by only running each animal once. We further found that the apelin null mice had decreased maximum oxygen consumption (V̇o2max; P = 0.03; Fig. 4C) and trended toward lower resting oxygen consumption (P = 0.09; Fig. 4E). This is a measure of metabolic activity and is defined by the Fick equation as equal to the cardiac output multiplied by the tissue oxygen extraction. With the assumption of no differences in oxygen extraction, this would imply that the apelin null mice have a decreased ability to augment cardiac output in response to an exercise challenge.

Fig. 4.

Apelin and APJ null mice have reduced exercise capacity and apelin null mice have lower maximal oxygen consumption (V̇o2). Graded treadmill running reveals that apelin (A) and APJ (B) null mice have impaired exercise capacity (apelin+/+, 25.65 ± 0.77 and apelin−/−, 23.31 ± 0.59 min, P = 0.034; and APJ+/+, 23.64 ± 0.74 and APJ−/−, 20.30 ± 0.97 min, P = 0.009). C: apelin null mice have lower maximal oxygen consumption (apelin+/+, 6269.2 ± 116 and apelin−/−, 5885.0 ± 120 ml·kg−1·h−1; P = 0.033). D: APJ null mice have similar maximal oxygen consumption [APJ+/+, 6,534 ± 162 and APJ−/−, 6,740 ± 329 ml·kg−1·h−1; P = not significant (ns)]. E: apelin null mice have a trend toward lower resting oxygen consumption (apelin+/+, 4,240 ± 124 and apelin−/−, 3,944 ± 107 ml·kg−1·h−1; P = 0.09). F: mean oxygen consumption over time in apelin null (n = 12) and wild-type (n = 17) mice during treadmill exercise. *P < 0.05.

In vitro cardiomyocyte studies reveal decreased sarcomeric function with no change in intracellular calcium.

Freshly isolated cardiomyocytes from apelin and APJ null mice demonstrated clear deficits in their sarcomeric function (Fig. 5). Sarcomeric shortening was less (apelin null, P = 0.02; and APJ null, P = 0.002; Fig. 5, A and B) with no difference in starting sarcomere length [apelin and APJ null, P = not significant (NS)]. There were also differences in the rate of contraction and relaxation, with myocytes from apelin and APJ null mice exhibiting a lower maximum velocity of contraction (apelin null, P = 0.01 and APJ null, P = 0.003; Fig. 5, C and D) and lower maximum velocity of relaxation (apelin null, P = 0.04 and APJ null, P = 0.001; Fig. 5, E and F).

Fig. 5.

Apelin and APJ null cardiomyocytes show decreased contractile function. Sarcomeric shortening was significantly decreased in the apelin null myocytes (A; apelin+/+, 4.52 ± 0.26% and apelin−/−, 3.69 ± 0.23%; P = 0.02) and APJ null myocytes (B; APJ+/+, 6.98 ± 0.40% and APJ−/−, 5.07 ± 0.41%; P = 0.002). The speed of contraction was also significantly reduced in the apelin null myocytes (C; apelin+/+, −1.10 ± 0.06 and apelin−/−, −0.89 ± 0.05 μm/s; P = 0.011) and the APJ null myocytes (D; APJ+/+, −2.41 ± 0.15 and APJ−/−, −1.77 ± 0.14 μm/s; P = 0.003). The velocity of relaxation was lower in the apelin null myocytes (E; apelin+/+, 0.65 ± 0.06 and apelin−/−, 0.51 ± 0.04 μm/s; P = 0.04) and the APJ null myocytes (F; APJ+/+, 1.63 ± 0.12 and APJ−/−, 1.07 ± 0.11 μm/s; P = 0.001). G: representative raw loop of sarcomeric function demonstrating decreased contractility of cardiomyocytes from the apelin null mice. *P < 0.05.

To begin to determine a mechanism for these changes, we assessed calcium fluorescence in a separate group of apelin and APJ null cardiomyocytes and their littermate controls. There was no difference in calcium transient in apelin and APJ null myocytes compared with controls (P = NS for both groups; Fig. 6, A and B). Additionally, troponin I phosphorylation at serine 23–24 was assessed by immunoblotting of left ventricular lysates from the apelin null mice and control mice and was found to be similar between groups (P = NS; Fig. 6C).

Fig. 6.

Apelin and APJ null cardiomyocytes show similar intracellular calcium kinetics and troponin I phosphorylation. There were similar calcium transients in the apelin null myocytes (A; apelin+/+, 31.0 ± 2.6% and apelin−/−, 34.0 ± 2.5%; P = NS) and APJ null myocytes (B; APJ+/+, 38.1 ± 2.4% and APJ−/−, 37.6 ± 2.6%; P = NS). C: phosphorylation levels of troponin I at serine residues 23–24 were similar at basal conditions between the apelin null and wild-type mice as expressed by the ratio of phos-troponin I (phos-trop I) to troponin I (trop I; apelin+/+, 0.69 ± 0.03 and apelin−/−, 0.64 ± 0.03; P = NS).

Baseline values for the wild-type controls were sometimes different between the two experimental groups of isolated cardiomyocytes. The genetic background of mice is well known to be an important determinant of phenotype, and the apelin−/− mice were on a mixed C57Bl/6-129SVJ background, whereas the APJ−/− mice were on a pure C57Bl/6 background. Wild-type controls were always chosen from littermates, thereby eliminating variation between the control and experimental groups due to strain-specific effects; however, it is likely that the differences between the wild-type controls reflect strain-dependent effects on baseline cardiovascular function.


Studies employing isolated cardiomyocytes, rat heart preparations, and in vivo normal and diseased rodent models have consistently shown that administration of exogenous apelin has a positive effect on cardiac contractility (3, 4, 12, 34). To better understand the role of the endogenous apelin-APJ pathway in physiological conditions, we created and studied mice deficient in either the ligand (apelin−/−) or the receptor (APJ−/−). The major finding in this paper is that physiological profiling including noninvasive imaging, in vivo pressure-volume hemodynamics, exercise stress testing, and in vitro cardiomyocyte sarcomeric shortening reveals that the endogenous apelin-APJ pathway contributes to cardiac function.

In vivo measures of cardiac function are consistent with this pathway serving a modest physiological role under basal conditions, with accentuation of function of the pathway under conditions of stress. Both load-dependent and load-independent in vivo measures revealed variability between the two knockout lines, with the APJ−/− mice having decreased fractional shortening by echocardiography and the apelin−/− mice having a decreased ESPVR by invasive hemodynamics. Overall, these results appear consistent with measurement of a small effect. Thus our conclusion is that the two lines of mice have similar modest impairments in basal cardiac function. Interestingly, a reduction in the intercept of the ESPVR revealed that at normal preload there is some difference in left ventricular performance, but the absolute magnitude of change was greatest at higher end-diastolic volumes, such as would be the case during conditions of stress. In keeping with this observation, we found a consistent decrement in exercise capacity in the two lines of targeted mice. These data suggest that mice without a functional apelin-APJ pathway have a decreased ability to augment cardiac output in response to a cardiovascular stress such as exercise.

The most striking finding in our data came from the in vitro cardiomyocyte studies. Using isolated single left ventricular myocytes from apelin−/− and APJ−/− mice, we observed impaired sarcomeric function with no change in baseline sarcomeric length. We also noted changes in the rate of sarcomere contraction and relaxation, suggesting overall that the apelin-APJ pathway may increase myocyte contractility in an autocrine fashion. To determine whether the loss of endogenous apelin-APJ signaling altered calcium transients, we measured intracellular calcium concentrations in field-stimulated left ventricular myocytes from apelin−/− and APJ−/− mice and their wild-type littermate controls. Despite a clear difference in contractility, there was no significant difference in intracellular calcium baseline level or contraction excursion, suggesting that loss of apelin or APJ affects myofilament sensitivity or cross-bridge cycling kinetics. Since the phosphorylation state of troponin I (the inhibitory unit of the troponin complex) is differentially affected by PKA and PKC signaling and directly affects cross-bridge kinetics, we examined troponin I phosphorylation at the serine 23/24 residue. Phosphorylation at this residue reduces myofilament calcium sensitivity, and we hypothesized an increase in phosphorylation state in the knockout animal. In fact, we found no differences between the two genotypes, suggesting a troponin I-independent mechanism of calcium sensitization. These observations are consistent with recent data obtained with isolated rat cardiomyocytes showing that exogenous apelin does not increase calcium transients but activates the sarcolemmal Na+/H+ exchanger and increases intracellular pH, suggesting increased myofilament sensitivity to calcium as a mechanism for inotropy (12).

We interpret these data in isolated cardiomyocytes to suggest an autocrine apelin-APJ pathway may be activated in the myocardium under conditions of cardiovascular stress. Autocrine myocardial signaling is consistent with prior studies showing that myocardial cells in vitro express both apelin and APJ (3, 30, 34). Furthermore, this pathway appears to be upregulated under hypoxic conditions, likely through HIF-1α, since we previously demonstrated increased apelin and APJ expression in the hearts of hypoxic mice (32). Similar conditions may be present for cardiomyocytes during the isolation process since they are subjected to hypoxia, low temperature, and variable pH. Additionally, myocardial cells in the developing heart can express both apelin and APJ, suggesting that this is an embryonic pathway that can be reactivated during conditions of cardiovascular stress (3). This hypothesis deserves further study through investigation of apelin and APJ expression in both animal disease models and human diseased tissues.

It is important to note that the cardiovascular phenotype seen in this study differed between the apelin−/− and APJ−/− mice. Currently, apelin is considered to be the only ligand for APJ, and APJ is considered to be the only receptor for apelin. One would expect, therefore, that the phenotype generated by functional deletion of apelin and APJ genes should be very similar if not identical. The most striking phenotypic difference is the embryonic developmental defect in the APJ, but not in the apelin, null animals. This is not due to background strain differences, since we have recently finished ten generations of breeding the apelin null allele on the C57Bl/6 background, and offspring of heterozygote crosses arise in the expected Mendelian ratios (data not shown). Similarly, developmental studies in zebrafish have suggested that the loss of apelin does not exactly phenocopy the loss of APJ (agtrl1b) in this model organism, with the phenotype of the receptor loss being more consistent and more dramatic (28, 31, 37). Observed differences between the null lines could reflect the existence of other unknown ligands and/or receptors, or various ligand-independent functions of APJ such as heterodimerization with other GPCRs. Receptor heterodimerization is a known mechanism for modulating the function of this class of receptors and the bradykinin receptor and the angiotensin II type 1 receptor (AT1R) interact in such a manner (1, 2, 23). Furthermore, our group recently published data showing cross talk between the apelin-APJ and angiotensin II pathways, specifically that APJ and AT1R physically associate in the cell (8).

The physiological profile of this pathway reported here varies in some basic respects from studies by other groups with apelin and APJ knockout mice. Kuba et al. (18) found no evidence of cardiac dysfunction in young apelin knockout mice by echocardiographic assessment and concluded that there was no hemodynamic compromise in animals less than 6 mo of age. Although these data are consistent with our load-dependent measures provided by echocardiography and magnetic resonance imaging in mice ∼3 mo of age, we did find evidence of cardiovascular compromise by exercise testing, and this linked to impaired cardiac contractility by in vivo and isolated cell experiments. This likely reflects a difference in sensitivity of methodological approaches, and the additional methods employed in the current study may have identified a modest basal contractile phenotype not apparent in the work by Kuba et al. (18). Regarding the comparison of our APJ null phenotype with other knockout lines reported in the literature, there are unfortunately very little cardiovascular phenotypic data available on the other two lines that have been described (13, 14, 16). Ishida et al. (14) did look at blood pressure, and our data confirms their finding of no difference in resting blood pressure. Importantly, all three APJ null lines appear to show significant loss of homozygous null animals during embryogenesis. Ishida et al. commented that there was not a problem with development of their animals but reported the results of their breeding, which does seem to indicate non-Mendelian distribution of genotypes among the offspring (14). Our statistical evaluation of their data indicates a significantly lower than expected number of homozygous null APJ mice (Chi Square, P = 0.03), suggesting that there is significant embryonic lethality in their model. Roberts et al. (29) found APJ-deficient mice to be born in a non-Mendelian ratio that is strikingly consistent with our own findings, with heterozygous matings producing 6% knockout, 28% wild-type, and 66% heterozygous offspring for the APJ mutation (29). Therefore, it appears likely that each of the APJ-deficient mouse models shows a significant loss of a portion of homozygous null embryos.

These data establish an endogenous role for the apelin-APJ pathway in basal cardiac function with more dramatic effects under conditions of stress. The potential for exploiting this pathway for medical benefit is significant. Balanced arterial and venous vasodilatation and load-independent inotropy via calcium sensitization, as well as angiotensin II and vasopressin antagonism, all separately denote areas of pharmacotherapeutic development in decompensated heart failure (20). To date, no single agent has been described, which targets all of these processes, thus making the apelin-APJ pathway a very promising target in the treatment of heart failure.


This work was supported in part by National Institutes of Health RO1 Grants HL-077676 (to T. Quertermous), NIH KO8 HL-083914-01 (to E. A. Ashley), NRSA T32 HL-07708 (to K. Ernst), and NRSA F32 HL-097615-01 (to D. N. Charo) and a fellowship from the Stanley J. Sarnoff Foundation (to D. N. Charo). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the article.


No conflicts of interest are declared by the authors.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
View Abstract