The heart is a resistin target tissue and can function as an autocrine organ. We sought to investigate whether cyclic mechanical stretch could induce resistin expression in cardiomyocytes and to test whether there is a link between the stretch-induced TNF-α and resistin. Neonatal Wistar rat cardiomyocytes grown on a flexible membrane base were stretched by vacuum to 20% of maximum elongation at 60 cycles/min. Cyclic stretch significantly increased resistin protein and mRNA expression after 2–18 h of stretch. Addition of PD-98059, TNF-α antibody, TNF-α receptor antibody, and ERK MAP kinase small interfering RNA 30 min before stretch inhibited the induction of resistin protein. Cyclic stretch increased, whereas PD-98059 abolished, the phosphorylated ERK protein. Gel-shift assay showed a significant increase in DNA-protein binding activity of NF-κB after stretch, and PD-98059 abolished the DNA-protein binding activity induced by cyclic stretch. DNA binding complexes induced by cyclic stretch could be supershifted by p65 monoclonal antibody. Cyclic stretch increased resistin promoter activity, whereas PD-98059 and p65 antibody decreased resistin promoter activity. Cyclic stretch significantly increased TNF-α secretion from myocytes. Recombinant resistin protein and conditioned medium from stretched cardiomyocytes reduced glucose uptake in cardiomyocytes, and recombinant small interfering RNA of resistin or TNF-α antibody reversed glucose uptake. In conclusion, cyclic mechanical stretch enhances resistin expression in cultured rat neonatal cardiomyocytes. The stretch-induced resistin is mediated by TNF-α, at least in part, through ERK MAP kinase and NF-κB pathways. Glucose uptake in cardiomyocytes was reduced by resistin upregulation.
- cyclic stretch
- glucose uptake
cardiomyocytes have been identified as a principal target of the proinflammatory actions of TNF-α (14). TNF-α can be induced in stretched myocytes and in hemodynamic-overloaded myocardium (9, 13, 18, 22). TNF-α is recognized as a significant contributor to myocardial dysfunction (11). In neonatal cardiomyocytes, TNF-α activates NF-κB (15). TNF-α can also modulate resistin expression in adipocytes and peripheral blood mononuclear cells (4, 11, 15). The link between TNF-α and resistin in cardiomyocytes has not been reported.
More recently, the heart was shown to be a resistin target tissue (12). In cardiomyocytes, mouse and human resistins directly impair glucose transport (12). Many studies demonstrated that isolated cardiomyocytes are insulin responsive and share many characteristics of adipocytes and skeletal muscle in terms of insulin stimulation of glucose transport (1, 3, 10, 19). Murine resistin is expressed not only in adipose tissue but, also, in the gastrointestinal tract, adrenal gland, skeletal muscle, brain, and pituitary gland (20, 21). There have been no reports, however, on resistin expression in cardiomyocytes. Since the heart is a resistin target tissue and can function as an autocrine organ, we hypothesize that cardiomyocytes express resistin gene. In diseased heart, glucose transport in the myocardium may be impaired. Left ventricular end-diastolic pressure is elevated in most of the diseased heart. The elevated end-diastolic pressure will stretch the myocardium. Whether mechanical stretch can induce resistin expression in cardiomyocytes has not been reported. Therefore, we sought to investigate whether cyclic mechanical stretch could induce resistin expression in cardiomyocytes and test whether there is a link between the stretch-induced TNF-α and resistin. Furthermore, we also tried to seek possible molecular mechanisms and signal pathways mediating resistin expression in cardiomyocytes by cyclic mechanical stretch.
MATERIALS AND METHODS
Primary cardiomyocyte culture.
Cardiomyocytes were obtained from 2- to 3-day-old Wistar rats by trypsinization, as previously described (27). Cultured myocytes thus obtained were >95% pure as revealed by observation of contractile characteristics with a light microscope and stained with anti-desmin antibody (Dako Cytomation, Glostrup, Denmark). Cardiomyocytes were seeded on a flexible membrane base of six culture wells at a density of 1.6 × 106 cells/well in Ham's F-10 containing 10% horse serum and 10% fetal calf serum. After 2 days in culture, cells were transferred to serum-free medium (Ham's F-12-DMEM, 1:1) and maintained for another 2 days. The enriched myocytes were then subjected to cyclic stretch. The study conforms with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, Revised 1996). The study was reviewed and approved by the Institutional Animal Care and Use Committee of Shin Kong Wu Ho-Su Memorial Hospital.
In vitro cyclic stretch on cultured cardiomyocytes.
Cardiomyocytes cultured on the flexible membrane base were subjected to cyclic stretch produced by a Flexcell FX-2000 strain unit with computer-controlled application of sinusoidal negative pressure at a frequency of 1 Hz (60 cycles/min) for 2–24 h. The roles of JNK, p38 MAP kinase, or ERK kinase in stretch-induced resistin expression were determined by pretreatment of the myocytes with 20 μM SP-600125, 3 μM SB-203580, or 50 μM PD-98059 (all from Calbiochem, San Diego, CA) for 30 min before cyclic stretch. SP-600125 is a potent, cell-permeable, selective, and reversible inhibitor of JNK. SB-203580 is a highly specific, cell-permeable inhibitor of p38 kinase. PD-98059 is a specific and potent inhibitor of ERK kinase.
Total RNA from rat abdominal fat, cardiomyocytes, skeletal muscle, liver, and kidney was extracted using the single-step acid guanidinium thiocyanate-phenol-chloroform extraction method (8). Real-time RT-PCR was performed as described previously (8). The rat resistin primers were 5′-ACTTCAGCTCCCTACTG-3′ and 5′-GTCTATGCTTCCGCACT-3′.
Western blot analysis.
Western blot was performed as previously described (29). Rabbit anti-resistin rat polyclonal antibody was obtained from Chemicon (Temecula, CA), anti-rat TNF-α and anti-rat TNF-α receptor antibodies from R & D Systems (Minneapolis, MN), and polyclonal anti-ERK and monoclonal anti-phosphorylated ERK kinase antibodies from Cell Signaling (Beverly, MA).
Neonatal cardiomyocytes were transfected with 800 ng of ERK-annealed small interfering RNA (siRNA; Dharmacon, Lafayette, CO) or resistin siRNA oligonucleotide (Invitrogen, Carlsbad, CA). ERK and resistin siRNAs are target specific 20- to 25-nt siRNAs designed to knock down gene expression. siRNA sequences were 5′-GACCGGAUGUUAACCUUUAUU (sense) and 5′-PUAAAGGUUAACAUCCGGUCUU (antisense) for ERK and ACACAUUGUAUCCUCACGGACGUCCC (sense) and GGACGUCCGUGAGGATACAAUGUGU (antisense) for resistin. As a negative control, a nontargeting (control) siRNA (Dharmacon) was used. For transfection of neonatal cardiomyocytes with siRNA oligonucleotides, Effectene transfection reagent was used according to the manufacturer's instructions (Qiagen, Valencia, CA). After incubation at 37°C for 24 h, cells were stretched for 18 h and analyzed by Western blot.
Nuclear protein concentrations from cultured myocytes were determined by Bio-Rad protein assay. Consensus and control oligonucleotides (Research Biolabs, Singapore) were labeled by polynucleotide kinase incorporation of [γ-32P]dATP. The consensus oligonucleotide sequence of NF-κB was 5′-AGTTGAGGGGACTTTCCCAGGC-3′. The NF-κB mutant oligonucleotide sequence was 5′-AGTTGAGGCGACTTTCCCAGG-3′. EMSA was performed as previously described (8). In each case, mutant or cold oligonucleotide was used as control to compete with labeled sequences.
Chromatin immunoprecipitation assay.
Chromatin immunoprecipitation (CHIP) assays were carried out with the CHIP assay kit (Upstate Biotech, Temecula, CA) according to the manufacturer's instructions. One-third of the cell lysate from a stretched cardiomyocyte was immunoprecipitated by anti-NF-κB p65 monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and then analyzed by PCR of the resistin promoter, and the remaining two-thirds of the cell lysate was added with anti-acetylated histone H3 antibody and analyzed by PCR of the resistin promoter. The primers for the resistin promoter were 5′-GAAGGAGCTGTGGGAC-3′ and 5′-GCAGTAGGGAGCTGAAG-3′. The primers for GAPDH were 5′-CATCACCATCTTCCAGGAGC-3′ and 5′-GGATGATGTTCTGGGCTGCC-3′. The PCR products of the resistin promoter and GAPDH (210 and 359 bp, respectively) were separated by agarose gel electrophoresis.
Promoter activity assay.
A −741- to +22-bp rat resistin promoter construct was generated as follows. Rat genomic DNA was amplified with forward (ACGCGTCTCAGCGGTAGAGCTCTTG) and reverse (AGATCTGGAGAAATGAAAGGTTCTTCATC) primer. The amplified product was digested with Mlu I and Bgl II restriction enzymes and ligated into pGL3-basic luciferase plasmid vector (Promega, Madison, WI) digested with the same enzymes. The resistin promoter contains NF-κB conserved sites (GGGACTT) at −285 to −279 bp. For the mutant, the NF-κB binding sites were mutated using the mutagenesis kit (Stratagene, La Jolla, CA). Site-specific mutations were confirmed by DNA sequencing. Plasmids were transfected into cardiomyocytes using a low-pressure accelerated gene gun (Bioware Technologies, Taipei, Taiwan) essentially according to the manufacturer's protocol. Two micrograms of test plasmid and 0.02 μg of control plasmid (pGL4-Renilla luciferases) were cotransfected with the gene gun in each well and then replaced by normal culture medium. After 6 h of cyclic stretching, cell extracts were prepared using the Dual-Luciferase Reporter Assay System (Promega), and dual-luciferase activity was measured using a luminometer (Turner Designs).
Measurement of TNF-α concentration.
Conditioned medium from stretched myocytes and control (unstretched) cells was collected for TNF-α measurement. The level of TNF-α was measured by a quantitative sandwich enzyme immunoassay technique (R & D Systems). The lower limit of detection of rat TNF-α was 5 pg/ml. Intra- and interobserver coefficients of variance were <10%.
Glucose uptake in cardiomyocytes.
Cardiomyocytes were seeded on ViewPlate (Packard Instrument, Meriden, CT) for 60 min at a density of 5 × 103 cells/well in serum-free medium with transferrin (5 μg/ml) and insulin (5 μg/ml) and incubated overnight. Recombinant mouse resistin (20 μg/ml; R & D Systems), resistin siRNA, TNF-α antibody, or conditioned medium was added to the plate. Glucose uptake was studied by addition of 0.1 mmol/l glucose and 500 nCi/ml d-[3-3H]glucose (Perkin Elmer, Boston, MA) for 2–8 h. Cells were washed twice with PBS. Nonspecific uptake was studied in the presence of 10 μM cytochalasin B and subtracted from the measured value. MicroScint-20 (50 μl) was added, and the plate was read with TopCount (Packard Instrument, Meriden, CT).
Rat model of aorta-caval shunt.
On the day of surgery, 290- to 320-g Wistar rats were anesthetized with pentobarbital sodium (80 mg/kg), and aorta-caval shunt was induced as described previously (28). The animals were killed 17 days after aorta-caval shunt, and blood was obtained from the right ventricle for measurement of circulating resistin levels.
Values are means ± SE. Statistical significance was performed with Student's t-test or ANOVA (GraphPad Software, San Diego, CA) where appropriate. Dunnett's test was used to compare multiple groups with a single control group. The Tukey-Kramer comparison test was used for pairwise comparisons between multiple groups after ANOVA. P < 0.05 was considered to denote statistical significance.
Cardiomyocytes express resistin gene.
RT-PCR was performed to investigate whether cardiomyocytes express resistin gene. Resistin mRNA was expressed most intensely in fat tissue (see supplemental Fig. 1 in the online version of this article at the American Journal of Physiology-Heart and Circulatory Physiology website). Cardiomyocytes and skeletal muscle expressed resistin mRNA, whereas liver and kidney did not.
Cyclic stretch enhances resistin protein and mRNA expression in cardiomyocytes.
The levels of resistin protein shown by Western blot analysis began to increase as early as 2 h after stretch at 20% of maximum elongation, reached a maximum of 3.7-fold (P < 0.01) over the control by 6 h, and remained elevated up to 18 h (Fig. 1) . Resistin protein returned to the baseline level after 24 h of stretch. Stretch-induced resistin protein expression was load dependent (Fig. 1). Stretch at 10% of maximum elongation increased resistin protein expression from 18 to 24 h.
Real time-PCR showed a significant increase in resistin messages from 2 to 18 h of stretch at 20% of maximum elongation (Fig. 1C). Resistin mRNA, similar to protein expression, returned to the baseline level after 24 h of stretch. Addition of the angiotensin type 1 receptor antagonist losartan (100 nM) 30 min before stretch did not significantly attenuate stretch-induced resistin mRNA expression (data not shown).
Stretch-induced resistin protein expression in myocytes is mediated by TNF-α and ERK kinase.
Western blot demonstrated a significant reduction in the stretch-induced increase of resistin protein after addition of TNF-α antibody (5 μg/ml), TNF-α receptor antibody (5 μg/ml), or PD-98059 30 min before stretch (Fig. 2, A and B). The stretch-induced increase in resistin protein was not affected by SP-600125. The stretch-induced increases in resistin protein were also completely blocked after addition of U-0126 (25 μmol/l), a specific and potent inhibitor of ERK kinase, 30 min before stretch (data not shown). SB-203580 partially decreased stretch-induced resistin protein expression. ERK siRNA also completely blocked the resistin expression induced by cyclic stretch (P < 0.001). ERK siRNA knocked down the ERK protein expression. The control siRNA did not affect the resistin expression induced by cyclic stretch. The inhibitor used in the study did not affect basal resistin gene expression (data not shown). These findings imply that the ERK pathway and TNF-α mediated the induction of resistin protein by cyclic stretch in myocytes. Exogenous addition of other proinflammatory cytokines, such as interleukin-6 (10 ng/ml) and angiotensin II (10 nM), did not induce resistin protein expression (Fig. 2, C and D) in cultured cardiomyocytes. This finding confirms the specificity of TNF-α in stretch-induced resistin expression. Addition of losartan before stretch also did not significantly attenuate the stretch-induced resistin protein expression.
Phosphorylated ERK protein was induced by cyclic stretch to 20% of maximum elongation (Fig. 3). The stretch-induced increase in phosphorylated ERK protein occurred slightly earlier than the stretch-increase in resistin protein. The phosphorylated ERK was abolished by PD-98059 and ERK siRNA. Addition of TNF-α antibody (5 μg/ml) or TNF-α receptor antibody (5 μg/ml) attenuated the phosphorylation of ERK protein induced by cyclic stretch. Addition of losartan before stretch did not abolish the stretch-induced phosphorylated ERK.
Cyclic stretch increases NF-κB binding activity.
Cyclic stretch of myocytes for 2–24 h significantly increased the DNA-protein binding activity of NF-κB (Fig. 4A). An excess of unlabeled NF-κB oligonucleotide competed with the probe for binding NF-κB protein, whereas an oligonucleotide containing a 2-bp substitution in the NF-κB binding site did not compete for binding. Addition of PD-98059 30 min before stretch abolished the DNA-protein binding activity induced by cyclic stretch. DNA binding complexes induced by cyclic stretch could be supershifted by a specific p65 antibody (a specific antibody for NF-κB), indicating the presence of this protein in these complexes. After immunoprecipitation with p65 antibody, CHIP assay showed a resistin promoter band (Fig. 4B). This implies that NF-κB binds to resistin promoter and confirms the specificity of DNA-protein binding activity of NF-κB by gel-shift assay.
Cyclic stretch increases resistin promoter activity.
The rat resistin promoter construct contains signal transducer and activator of transcription (Stat-3), activator protein-1, NF-κB, and hypoxia-inducible factor-1α binding sites. Cyclic stretch for 6 h significantly increased the resistin promoter activity by 2.3-fold compared with control without stretch (Fig. 5). When the NF-κB binding sites were mutated, the increased promoter activity induced by cyclic stretch was abolished. Addition of PD-98059 and NF-κB p65 antibody 30 min before stretch abolished the increased resistin promoter. This finding indicates that cyclic stretch regulates resistin in cardiomyocytes at the transcriptional level and that NF-κB binding sites in the resistin promoter are essential for transcriptional regulation.
Cyclic stretch stimulates secretion of TNF-α from myocytes.
The increase in TNF-α secretion from myocytes induced by cyclic stretch began 2 h after stretch and continued for 18 h (see supplemental Fig. 2 in the online version of this article). The mean concentration of TNF-α rose from 36.5 ± 1.7 pg/ml before stretch to 98.9 ± 5.1 pg/ml after 2 h of stretch (P < 0.01). The increased resistin expression levels in cultured myocytes upon stretch were associated with TNF-α secretion.
Recombinant resistin reduces glucose uptake.
Recombinant mouse resistin (20 μg/ml) and conditioned medium from stretched cardiomyocytes significantly reduced glucose uptake over 2–8 h of incubation compared with control untreated cardiomyocytes (Fig. 6). The dose of recombinant mouse resistin was based on the study by Graveleau et al. (12). Addition of resistin siRNA or TNF-α antibody before recombinant resistin reversed the glucose uptake to baseline levels. Resistin siRNA also reversed the glucose-lowering effect of conditioned medium. After an overnight incubation, the insulin in the medium measured by enzyme immunoassay (CRYSTAL CHEM, Downers Grove, IL) was still measurable (700 pg/ml). Use of serum-free medium without insulin and resistin resulted in glucose uptake by cardiomyocytes of 150 ± 10 counts/min, whereas addition of resistin reduced glucose uptake to 110 ± 8 counts/min (P < 0.05, n = 3).
In vivo aorta-caval shunt increases resistin protein expression.
Resistin protein and mRNA expression significantly increased 1 and 3 days after induction of aorta-caval shunt and tended to decrease 5 and 7 days after shunt (Fig. 7). The left ventricular end-diastolic dimension increased from 6.1 ± 0.3 to 6.5 ± 0.4 mm after 3 days of aorta-caval shunt. Aorta-caval shunt resulted in a pulmonary-to-systemic flow ratio of 1.7. The circulating resistin levels increased from 1 to 5 days after aorta-caval shunt and returned to baseline 7 days after shunt (see supplemental Fig. 3 in the online version of this article). The circulating TNF-α also significantly increased (P < 0.05, n = 3) from 1 day (140 ± 7 pg/ml) to 5 days (122 ± 9 pg/ml) after shunt compared with the sham group (80 ± 5 pg/ml). Although the circulating TNF-α level remained elevated 7 days after shunt (115 ± 5 pg/ml), the difference did not reach statistical significance compared with the sham group.
In the present study, we demonstrated several significant findings: 1) cardiomyocytes express resistin gene; 2) cyclic stretch upregulates resistin expression in cardiomyocytes; 3) TNF-α acts as an autocrine factor to mediate the increased resistin expression induced by cyclic stretch; 4) ERK kinase and NF-κB transcription factor are involved in the signaling pathway of resistin induction; 5) resistin impairs glucose uptake in cardiomyocytes; and 6) the in vivo aorta-caval shunt acutely increases resistin protein expression. The aorta-caval shunt is the animal model for cardiac volume overload. Resistin in cardiomyocytes was upregulated in a time- and a load-dependent manner by cyclic stretch. Our data clearly indicate that hemodynamic forces play a crucial role in the modulation of resistin expression in cardiomyocytes. Our data also demonstrated that a functional consequence of resistin upregulation by stretch was reduction of glucose uptake.
The induction of resistin protein by cyclic stretch was largely mediated by the ERK kinase pathway, because the specific and potent inhibitors of an upstream ERK kinase, PD-98059 and U-0126, inhibited the induction of resistin protein. This signaling pathway of ERK was further confirmed by the finding that ERK siRNA inhibited the induction of resistin protein by cyclic stretch. In the present study, stretched myocytes secreted TNF-α, and TNF-α monoclonal antibody and TNF-α receptor antibody blocked the increases of resistin protein induced by cyclic stretch. These results provide the first evidence for TNF-α mediation of cyclic stretch-induced expression of resistin in cardiomyocytes. These results further confirm the autocrine or paracrine production of cardiomyocytes in response to cyclic stretch.
Previous studies showed that obesity and atherosclerosis are increasingly viewed as inflammatory states. Biomarkers that integrate metabolic and inflammatory signals are attractive candidates for defining risk of atherosclerotic cardiovascular disease (30). Rodent resistin is derived almost exclusively from fat tissue, and adipose expression and serum levels are elevated in models of obesity and insulin resistance (16, 25, 23). Hyperresistinemia impairs glucose tolerance and induces hepatic insulin resistance in rodents (2), whereas resistin-deficient mice are protected from obesity-associated insulin resistance (26). The present study using an isolated cardiomyocyte culture system demonstrated that glucose transport was impaired by resistin. The glucose uptake in cardiomyocytes was reduced by resistin upregulation. Resistin has been demonstrated to impair insulin-mediated GLUT4 translocation in cardiomyocytes (12). Thus, impairment of glucose transport may explain the potential mechanism of resistin induction of insulin resistance. In hemodynamic overload, especially volume overload, stretched myocytes may impair glucose uptake and contractile function. The present study suggests that resistin is a metabolic link between mechanical stress and hypertrophic heart. Therefore, the transient increase in resistin gene expression after cyclic stretch or acute volume overload may be important in patients with hemodynamic overload. Recently, using an isolated perfused rat heart model, Rothwell et al. (26) demonstrated that resistin impaired cardiac recovery after ischemia-reperfusion injury. Their study showed no significant effect of resistin on myocardial glucose uptake. The different findings may be explained by the difference in glucose metabolism between whole myocardium and cultured cardiomyocytes. Therefore, the role of resistin in dilated ventricle with contractile dysfunction needs further investigation.
TNF-α is recognized as a significant contributor to myocardial dysfunction (13). Cardiomyocytes have been identified as a principal target of the proinflammatory actions of TNF-α and cause a series of pathological changes in cardiomyocytes. In neonatal cardiomyocytes, TNF-α activates NF-κB (9). Resistin was also shown to have potent proinflammatory properties (4). Resistin promotes endothelial cell activation (17) and causes endothelial dysfunction of porcine coronary arteries (5). Recently, resistin was found to have a potential role in atherosclerosis, because resistin increases monocyte chemoattractant protein-1 and soluble vascular cell adhesion molecule-1 expression in vascular endothelial cells (7) and promotes vascular smooth muscle cell proliferation (6). The link between TNF-α and resistin in cardiomyocytes may indicate that resistin plays a role as a downstream protein of TNF-α to contribute to cardiomyocyte dysfunction. The present study confirms that heart is a resistin target tissue as well as a resistin autocrine organ.
The role of TNF-α on the effect of resistin is controversial. Fasshauer et al. (11) reported that TNF-α is a negative regulator of resistin gene expression in adipocytes, whereas Kaser et al. (15) and Bokarewa et al. (4) reported that TNF-α is a positive regulator of resistin gene expression in peripheral blood mononuclear cells. In the present study, TNF-α was found to be a positive regulator of resistin gene expression in stretched cardiomyocytes. Taken together, the effect of TNF-α on the regulation of resistin gene expression is different in different cell types.
NF-κB is a proinflammatory master switch that controls the production of several inflammatory markers and mediators. Cae et al. (6) demonstrated that hepatic activation of NF-κB caused local and systemic insulin resistance. In the present study, we demonstrated that cyclic stretch-stimulated NF-κB-DNA binding activity required at least phosphorylation of ERK, since ERK inhibitor abolished the NF-κB binding activity. The p65 monoclonal antibody, a specific antibody for NF-κB, shifted the NF-κB-DNA binding complex, indicating the specificity of the cyclic stretch-induced NF-κB-DNA binding activity. In the present study, we used CHIP assay to confirm that the resistin gene upstream region contains an NF-κB site. We further demonstrated that cyclic stretch increased resistin promoter activity and that the binding site of NF-κB in the resistin promoter is essential for the transcriptional regulation. Taken together, our results indicate that cyclic stretch may increase NF-κB transcriptional activity in cardiomyocytes.
In summary, our study is the first report of cyclic mechanical stretch enhancement of resistin expression in cultured rat neonatal cardiomyocytes. The stretch-induced resistin is mediated by TNF-α, at least in part, through the ERK kinase and NF-κB pathway. Glucose uptake in cardiomyocytes was reduced by resistin upregulation.
This study was sponsored in part by a grant from the New Century Health Care Promotion Foundation (Taipei, Taiwan).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society