Cytokines such as tumor necrosis factor alpha (TNF-alpha) are produced by the myocardium in heart disease and might be stimulated by reactive oxygen. In some cell types, cyclic adenosine monophosphate (AMP) inhibits TNF-alpha production. The authors tested the hypothesis that stimulation of cardiac beta-adrenergic receptors would inhibit cytokine gene transcription induced by reactive oxygen.
Rat hearts were perfused with buffer containing hypoxanthine. Reactive oxygen intermediates were generated by infusion of xanthine oxidase. Myocardial mRNA encoding 11 cytokines was determined. TNF-alpha, interleukin-6, and cyclic AMP were measured in the coronary effluent.
In control hearts, of the screened RNA, only mRNA encoding interleukin-1beta, -4, and -6 was detected. Stimulation with hypoxanthine-xanthine oxidase (HX-XO) induced detectable mRNA for TNF-alpha and interleukin-5 and increased mRNA band density for interleukin-1beta, -4, and -6. Simultaneous infusion of isoproterenol inhibited HX-XO-stimulated cytokine gene expression and caused release of cyclic AMP into the coronary effluent. In control hearts, TNF-alpha was not detected in the coronary effluent. After HX-XO administration, TNF-alpha was reliably detected at 60 min and interleukin-6 at 90 min. Simultaneous infusion of isoproterenol inhibited TNF-alpha and interleukin-6 release. Inclusion of propranolol in the perfusion buffer blocked the isoproterenol-induced inhibition of HX-XO-stimulated TNF-alpha release and release of cyclic AMP into the coronary effluent. In addition, elevating myocardial cyclic AMP with forskolin also blocked release of TNF-alpha stimulated by HX-XO. Finally, delaying infusion of isoproterenol until 30 min after HX-XO administration still suppressed release of TNF-alpha.
Reactive oxygen species activate cytokine gene transcription in the myocardium. The sympathetic nervous system, acting through beta-receptors to elevate myocardial cyclic AMP, regulates cardiac cytokine production by inhibition of transcription.
MYOCARDIAL production of inflammatory cytokines such as tumor necrosis factor α (TNF-α) and interleukin-6 appears to be triggered by events associated with myocardial ischemia. For instance, TNF-α is elevated in the plasma of patients with unstable angina and after myocardial infarction. The concentration is correlated with the extent of infarction, with complications such as rhythm disturbances, with the development of heart failure, and with creatinine kinase concentrations. 1–5Elevated interleukin-6 has been reported in patients with unstable angina and with uncomplicated myocardial infarction. 6–8Furthermore, both TNF-α and interleukin-6 are expressed in the myocardium and released by the heart after aortic declamping after cardiopulmonary bypass. 9–12The precise stimulus for production of these cytokines during ischemia and how myocardial production is regulated is currently not understood.
Myocardial infarction, unstable angina, and aortic cross-clamp–declamping are associated with episodes of ischemia–reperfusion during which reactive oxygen intermediates are generated. 13–15Oxidation of purines by xanthine oxidase (XO) during ischemia–reperfusion can produce superoxide, hydroxyl species, and hydrogen peroxide. 16In this regard, perfusion of isolated rat hearts with buffer containing hydrogen peroxide induced the expression of TNF-α in myocardial tissue. 17Furthermore, in rats, myocardial ischemia–reperfusion activated nuclear regulatory factor κB (NF-κB), an immediate response element regulating TNF-α gene transcription, 18and resuscitation from hemorrhagic shock caused both activation of NF-κB and expression of TNF-α in the heart. 19In addition, we found that treating coronary smooth muscle cells with the reactive oxygen intermediates generating system, hypoxanthine–XO (HX–XO), activated NF-κB and stimulated release of TNF-α. 20Thus, reactive oxygen intermediates may initiate myocardial expression of inflammatory cytokines through a mechanism dependent on the activation of NF-κB.
Studies of the regulation of TNF expression in macrophages and monocytes indicate that increased intracellular cyclic adenosine monophosphate (AMP) decreased bacterial lipopolysaccharide-stimulated TNF production. 21,22In minced ventricular tissue from rat hearts, in papillary muscle, and in isolated myocytes, agents that increase cyclic AMP, such as phosphodiesterase inhibitors, adenosine, and forskolin, have also been shown to inhibit TNF-α release. 23–26We previously reported that lipopolysaccharide stimulated release of TNF-α from human atria and rat heart, and that this release could be inhibited by treatment with isoproterenol. 27,28Therefore, it seems possible that TNF-α production by the heart might be regulated, at least in part, by sympathetic nervous stimulation of cardiac β-adrenergic receptors and the consequent increase in myocardial cyclic AMP. In the current study, we determined if reactive oxygen intermediates would stimulate expression of inflammatory cytokines in the myocardium and tested the hypothesis that stimulating cardiac β-adrenergic receptors with isoproterenol would inhibit this expression.
Materials and Methods
Rat recombinant TNF-α was purchased from Genzyme Diagnostics (Cambridge, MA). XO, hypoxanthine, and all other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). Murine fibroblast L929 cells were purchased from American Type Culture Collection (Rockville, MD). Cell culture plastic ware was obtained from Costar (Cambridge, MA).
Isolated Rat Heart
After obtaining approval from the Institutional Animal Care and Use Committee of Mercer University School of Medicine, male Sprague-Dawley rats (weight, 275–445 g) were anesthetized with pentobarbital. The hearts were removed and perfused Langendorf style at a constant pressure of 90 cm H2O with Krebs-Henseleit buffer at 37°C and gassed with 95% O2–5% CO2, pH = 7.4. The content of the buffer was 118 mm NaCl, 4.7 mm KCl, 1.2 mm NaH2PO4, 1.2 mm MgSO4, 25 mm NaHCO3, 10 mm glucose, and 2 mm CaCl2. All hearts were equilibrated on the perfusion apparatus for 30 min. For the generation of reactive oxygen species, 1 mm hypoxanthine was added to the buffer and, using a syringe pump, XO was infused to produce a final concentration in the perfusing buffer of 0.004 units/ml. In other experiments, isoproterenol (1 μm for 60 min) was either simultaneously infused with the XO or started 30 min after the initiation of the XO infusion (fig. 1). After equilibration, timed samples of coronary effluent were collected immediately before beginning infusion (time 0) and again at 30, 60, 90, 120, and 150 min for measurement of coronary flow and assay for cyclic AMP, TNF-α, and interleukin-6. In addition, to determine the role of cyclic AMP in cytokine release, the β-adrenergic receptor blocker propranolol (0.1 μm) was added to the perfusion buffer containing hypoxanthine, and then XO and isoproterenol were infused. Finally, to elevate myocardial cyclic AMP by a mechanism different from activation of the β-adrenergic receptor, forskolin (10 μm) was infused, and the effect on HX–XO-stimulated cyclic AMP and TNF-α release was determined.
Assay for Tumor Necrosis Factor-α
Biologically active TNF-α in the coronary effluent was measured by a cell cytotoxicity assay using murine fibroblast L929 cells as described previously. 29L929 cells were seeded into 96-well microtiter plates at an initial density of 5.0 × 104cells per well in 100 μl of Dulbecco modified Eagle medium plus 5% fetal bovine serum and antibiotics and incubated for 24 h at 37°C in 5% CO2. After incubation, the medium was removed, and 50 μl of Dulbecco modified Eagle medium with 5% fetal bovine serum containing 20 μg/ml actinomycin D (final concentration, 5 μg/ml) was added to all wells. In each plate, a standard curve was determined with serial dilutions of rat recombinant TNF-α. Test coronary effluent (150 μl) was added to the remaining wells in each 96-well plate. The plates were incubated for 18 h, and the medium was decanted. The remaining viable L929 cells were stained with 50 μl per well of 0.1% crystal violet in 20% ethanol, rinsed with phosphate-buffered saline, and air-dried. Methanol (100 μl) was then added to each well 5 min before reading to solubilize the dye. The optical density of each well was then read on an automated microplate reader Elx 800 (Bio-Tek Instrument, Inc., Winooski, VT) at 595 nm. TNF-α data are expressed as release in picograms per minute, determined as the concentration of TNF-α in the coronary effluent multiplied by the flow (picograms per minute times milliliter per minute). To determine if the cytotoxicity in the L929 assay was caused by TNF-α, a guinea pig antimouse TNF-α antiserum in a final dilution of 1:500 was added to selected samples of coronary effluent before L929 assay. All test agents, hypoxanthine, XO, isoproterenol, propranolol, and forskolin, were added directly into the L929 cell assay. All were without effect on the assay.
Slot Blot Analysis for Interleukin-6
Coronary effluent (800 μl) collected at 90 min was concentrated using a Microcon centrifugal filter tube with a nominal molecular weight cutoff of 3,000 (Millipore, Bedford, MA). Equal amounts of samples were slot blotted to supported nitrocellulose membranes by vacuum transfer. The membranes were also reversibly stained with Ponceau S to reconfirm equivalence of sample loading. After blocking with 5% nonfat milk, the membranes were incubated with a monoclonal antirat interleukin-6 antibody (R&D Systems Inc., Minneapolis, MN) and followed by horse radish peroxidase (HRP) conjugated secondary antibody (Amersham, Piscataway, NJ). The binding of antibody was detected by enhanced chemiluminescence.
Measurement of Cyclic Adenosine Monophosphate
For the measurement of cyclic AMP released from the heart, 1.8-ml aliquots of coronary effluent were collected into 200 μl 1N HCl at 0, 1, 2, 3, 4, 5, 10, 20, 30, 60, 90, 120, and 150 min. Cyclic AMP was determined by radioimmunoassay as described previously. 30Data are expressed as release in picomoles per minute, determined as the concentration of cyclic AMP in the coronary effluent multiplied by the flow (picomoles per minute times milliliters per minute).
RNase Protection Analysis
Five hearts in each of the groups (control, HX–XO, HX–XO plus isoproterenol, and isoproterenol alone) were perfused for 60 min after a 30-min equilibration period. The hearts were removed from the perfusion apparatus, and the atria were trimmed away. Ventricular tissue was cut into approximately 2-mm cubes and stored in 5 ml RNA/later®fluid (Ambion, Austin, TX) before extraction of RNA. Antisense RNA probes labeled with [α-32P] uridine triphosphate (UTP) were synthesized under the direction of T7 polymerase using DNA templates of rat cytokines, rCK1 multiprobe template set (Pharmingen, San Diego, CA), and purified by ammonium acetate–ethanol precipitation. The templates contained in the set included interleukin-1α and -1β, TNF-β, interleukin-3, -4, -5, -6, and -10, TNF-α, interleukin-2, INFγ, and the housekeeping genes, L32 and GAPDH . Total cellular RNA was prepared from ventricular tissue homogenates by acid guanidium isothyocynate-phenol-chloroform extraction as previously described. 31Determination of target mRNA was conducted using the Direct Protect®kit (Ambion) according to the manufacturer's instructions. Protected fragments of mRNA were separated by 5% polyacrylamide–8 m urea gel electrophoresis and exposed to x-ray film. Film was digitized, and integrated band density was analyzed using the NIH IMAGE program (written by Wayne Rasband, M.S., Computer Science, Research Service Branch, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/nih-image/index.html). Values for band density are expressed as the ratios of interleukin-1β, -4, and -6 band densities to that of the housekeeping gene, L32 .
Data were analyzed by two-way analysis of variance with repeated measures. Values are expressed as mean ± SE.
Figure 2shows results of an RNase protection assay that is typical of results obtained from assays performed on five hearts in each group. Control hearts were perfused for 60 min before RNA was purified from ventricles. Message encoding for interleukin-6 was detected in five of five control hearts, and for interleukin-1β and -4 in three of five control hearts. TNF-α and interleukin-5 message was not detected in any control heart (fig. 2). In hearts perfused for 60 min with HX–XO, mRNA encoding interleukin-1β, -4, -5, -6, and TNF-α was increased in all five hearts as represented by the increased intensity of each of the specific bands in figure 2. In hearts infused with isoproterenol and HX–XO simultaneously, the intensity of bands representing interleukin-1β, -4, and -6 message was reduced to less than control, and message for TNF-α and interleukin-5 was no longer detected. The administration of isoproterenol alone had no discernable effects on gene expression. Figure 3shows band densities of mRNA relative to control for interleukin-1β, -4, and -6 normalized to the L32 housekeeping gene. (Band densities for TNF-α and interleukin-5 mRNA were not included because this message could not be detected in the control state or after administration of HX–XO plus isoproterenol.) Messages for interleukin-1β, -4, and -6 were increased approximately threefold above control by HX–XO infusion. Simultaneous administration of isoproterenol with HX–XO prevented the increase of mRNA caused by HX–XO.
Cyclic AMP was not detected in the coronary effluent of control hearts or hearts perfused with HX–XO alone. However, as seen in figure 4, infusion of isoproterenol was associated with release of cyclic AMP. Concentrations of cyclic AMP peaked rapidly at 1 min (240 ± 65 pmol/min, n = 4 hearts), then declined but remained elevated for the duration of the experiment (33 ± 2 pmol/min at 150 min). The addition of 0.1 μm propranolol to the perfusion buffer significantly attenuated the isoproterenol-stimulated release of cyclic AMP at all time points (fig. 4).
In rat hearts perfused with buffer alone, no TNF-α was detectable in the coronary effluent for the entire duration of the experiment (150 min). It is possible that TNF-α was produced but was below the detection limit (1 pg) of the L929 cell cytotoxicity assay. Figure 5shows that in hearts treated with HX–XO, TNF was reliably detected in the coronary effluent at 60 min after the initiation of XO (728 ± 330 pg/min, n = 7 hearts). Even though the infusion of XO was stopped at 60 min, cardiac release of TNF-α continued to increase, with maximum release measured at 120 min (1,378 ± 251 pg/min). A simultaneous 60-min infusion of isoproterenol (1 μm) with XO significantly inhibited HX–XO-stimulated release of TNF-α at all time points (figure 5). For instance, at 120 min, isoproterenol infusion reduced release of TNF-α by 90% (135 ± 56 pg/min, n = 5 hearts) compared with HX–XO alone (P < 0.01). The cytotoxicity of the coronary effluent obtained from three hearts perfused with HX–XO on the L929 cells was completely blocked with an antibody to TNF-α, confirming cardiac release of biologically active cytokine.
The addition of propranolol to the perfusion buffer blocked the inhibition of HX–XO-stimulated release induced by isoproterenol (fig. 5). For instance, at 120 min with propranolol in the perfusion buffer and isoproterenol infusion, HX–XO-stimulated release of TNF-α was 1,156 ± 355 pg/min (n = 6 hearts), which was not significantly different from HX–XO alone (1,378 ± 251 pg/min) but significantly greater than HX–XO-stimulated release with infusion of isoproterenol in the absence of propranolol (135 ± 56 pg/min). In other experiments, forskolin induced release of cyclic AMP from hearts stimulated with HX–XO (fig. 6). In these experiments (n = 3 hearts) no TNF-α was detected in the coronary effluent.
In figure 7, slot blots of interleukin-6 released into the coronary effluent are shown and are typical of three sets of experiments each set conducted on the same day. In each of the three control hearts, interleukin-6 was detected and perhaps reduced when isoproterenol alone was given. Infusion of HX–XO was associated with interleukin-6 release in all three hearts, and release was reduced by simultaneous infusion of isoproterenol. Although band intensities varied among the individual hearts, the relation always held, i.e. , HX–XO always increased interleukin-6, and simultaneous infusion of isoproterenol always decreased it in three separate experiments.
Figure 8shows the effect of isoproterenol infusion begun 30 min after the initiation of XO perfusion on TNF-α released into the coronary effluent. In hearts perfused for 60 min with HX–XO alone, TNF-α was again reliably detected in the coronary effluent at 60 min (360 ± 79 pg/min, n = 8 hearts) and increased to a maximum of 1,501 ± 250 pg/min at 120 min. In hearts where isoproterenol (1 μm) was begun 30 min after the initiation of XO infusion and continued for 1 h, TNF-α in the coronary effluent at 60 min (421 ± 188 pg/min, n = 6 hearts) was no different from HX–XO alone. However, at 90, 120, and 150 min, delayed isoproterenol infusion significantly reduced the release of TNF-α. For instance, maximal release at 120 min was 638 ± 181 pg/min (P < 0.05 vs. HX–XO alone) or 45% of that seen with HX–XO alone.
In these experiments, we have shown that an enzyme–substrate system (HX–XO) that generates reactive oxygen intermediates and that is activated in episodes of myocardial ischemia–reperfusion, triggers myocardial gene expression for multiple cytokines, interleukin-1β, -4, -5, -6, and TNF-α. Infusion of the β-adrenergic agonist isoproterenol suppressed gene expression for each of these cytokines. Furthermore, we determined that TNF-α and interleukin-6 were both released into the coronary effluent by HX–XO and were inhibited by simultaneous infusion of isoproterenol and the consequent increase in cyclic AMP. Only interleukin-6 was detected in the coronary effluent of control hearts. These release data are consistent with the mRNA findings of expression of interleukin-6 in control hearts but not TNF-α, whereas both were expressed after administration of HX–XO, and both were inhibited by isoproterenol. Including propranolol in the perfusion buffer inhibited isoproterenol-stimulated release of cyclic AMP from the heart and blocked the isoproterenol-induced inhibition of HX–XO-stimulated TNF-α release. In addition, forskolin induced release of cyclic AMP and inhibited HX–XO-stimulated TNF-α release. These experiments with propranolol and forskolin indicate that the isoproterenol effect on cytokine gene expression was caused by activation of the cardiac β-adrenergic receptor and the consequent increase in myocardial cyclic AMP. Finally, when isoproterenol infusion was delayed 30 min after the start of HX–XO administration, TNF-α that was increasing in the coronary effluent was attenuated, suggesting a dynamic relation between elements regulating cytokine gene expression and β-adrenergic stimulation.
The cell type within the myocardium that is the source of cytokine expression was not identified in these experiments. In previous studies, the inflammatory cytokines interleukin-1 and -6 and TNF-α or their mRNAs have been identified in myocytes, endothelial cells, or vascular smooth muscle cells. 32–41To our knowledge, gene expression for interleukin-5 or the antiinflammatory cytokine, interleukin-4, has not been previously demonstrated in the myocardium. For the cytokines identified in this study for which NF-κB is a common regulator, namely, interleukin-1β, interleukin-6, and TNF-α, activation of this nuclear factor by reactive oxygen could initiate expression. On the other hand, expression of interleukin-4 and -5 by mast cells has been shown to be independent of NF-κB activation. 42The exact mechanism by which β-adrenergic stimulation inhibits myocardial cytokine gene transcription was not determined in the current study. Inhibition of activation of NF-κB by cyclic AMP can possibly be eliminated as a potential mechanism because activation was not affected by elevation of cyclic AMP in coronary smooth muscle cells, although TNF-α release by the cells was suppressed. 43Similar findings have been reported in endothelial cells where elevated cyclic AMP failed to block activation and translocation of NF-κB to the nucleus. 44A potential mechanism for the regulation of NF-κB–dependent genes by cyclic AMP was determined in human monocytes and endothelial cells. 45It is known that transcriptional activation of genes by NF-κB requires multiple cofactors. 46Activation of cyclic AMP–dependent protein kinase A induced phosphorylation of cyclic AMP response element binding protein (CREB). CREB then competed with NF-κB for a limited amount of an essential coactivator, cyclic AMP response element binding protein binding protein (CBP), to inhibit NF-κB–dependent gene transactivation. 45Perhaps such a mechanism is active in cells within the myocardium that produce these inflammatory cytokines.
In the current study we found that the acute administration of isoproterenol suppressed cytokine production by the heart. Just the opposite has been recently reported for chronic administration of isoproterenol to rats for 7 days. 47Chronic administration of catecholamines is known to be associated with free radical generation, myocardial ischemia, focal areas of myocardial necrosis, and failure. 48In these 7-day experiments, results similar to ours were found in normal myocardium in that mRNA encoding TNF-α was not detected while interleukin-6 was present in low levels. After chronic administration of isoproterenol, both mRNA and protein for TNF-α, interleukin-1β, and increased interleukin-6 message were detected in the myocardium. Chronic administration of isoproterenol at the doses used resulted in inflammatory cell infiltrates, but the cytokines were only detected in myocardial cells and blood vessels and were not found in serum. As stated by Murray et al. , 47their experiments could not distinguish between elevated cyclic AMP or increased heart rate, myocardial ischemia, free radical generation, and calcium overload as the cause of myocardial cytokine production. Our results suggest that free radicals will stimulate and elevated cyclic AMP will inhibit myocardial cytokine production, and this finding is supported by results in isolated myocytes where adenosine inhibited TNF-α expression through a cyclic AMP–dependent mechanism. 26It is possible that chronic administration of isoproterenol was associated with receptor desensitization and reduced myocardial cyclic AMP concentrations, thereby removing an inhibitory regulation as well as inducing reactive oxygen intermediates, which stimulated cytokine production. Collectively, these results indicate a role for the sympathetic nervous system in regulating myocardial inflammatory cytokine production. Moreover, they emphasize the complexity of this regulation in chronic disease states such as heart failure, which is accompanied by high circulating levels of catecholamines, reduced cardiac tissue levels, and reduced cardiac response to these adrenergic agonists.