The use of moderate hypothermia during cardiac surgery is associated with repression of tumour necrosis factor-α via inhibition of activating protein-1: an experimental study

Introduction The use of moderate hypothermia during experimental cardiac surgery is associated with decreased expression of tumour necrosis factor (TNF)-α in myocardium and with myocardial protection. In order to identify the cellular mechanisms that lead to that repression, we investigated the effect of hypothermia during cardiac surgery on both main signalling pathways involved in systemic inflammation, namely the nuclear factor-κB (NF-κB) and activating protein-1 pathways. Method Twelve female pigs were randomly subjected to standardized cardiopulmonary bypass with moderate hypothermia or normothermia (temperature 28°C and 37°C, respectively; six pigs in each group). Myocardial probes were sampled from the right ventricle before, during and 6 hours after bypass. We detected mRNA encoding TNF-α by competitive RT-PCR and measured protein levels of TNF-α, inducible nitric oxide synthase and cyclo-oxygenase-2 by Western blotting. Finally, we assessed the activation of NF-κB and activating protein-1, as well as phosphorylation of p38 mitogen-activated protein kinase by electrophoretic mobility shift assay with super shift and/or Western blot. Results During and after cardiac surgery, animals subjected to hypothermia exhibited lower expression of TNF-α and cyclo-oxygenase-2 but not of inducible nitric oxide synthase. This was associated with lower activation of p38 mitogen-activated protein kinase and of its downstream effector activating protein-1 in hypothermic animals. In contrast, NF-κB activity was no different between groups. Conclusion These findings indicate that the repression of TNF-α associated with moderate hypothermia during cardiac surgery is associated with inhibition of the mitogen-activated protein kinase p38/activating protein-1 pathway and not with inhibition of NF-κB. The use of moderate hypothermia during cardiac surgery may mitigate the perioperative systemic inflammatory response and its complications.


Introduction
Myocardial damage is an important complication of cardiac surgery involving cardiopulmonary bypass (CPB) [1]. Synthesis of tumour necrosis factor (TNF)-α in the myocardium is thought to play a central role in its pathophysiology [2,3]. Indeed, there is a large body of evidence that, in experimental models, over-expression of TNF-α in the myocardium is related to adverse cardiac effects such as postinfarct remodelling and ventricular dilatation [4], transition from hypertrophic to dilated cardiomyopathy due to apoptosis [5] and impaired postischaemic functional recovery [6]. Additionally, local administration of soluble TNF-α receptor-1 gene reduced infarct size in a model of ischaemia/reperfusion injury [7]. In a study conducted in a neonatal model of ischaemia of the hypertrophied left ventricle, inhibition of the biological activity of TNF-α significantly improved postischaemic contractile function, myocar-AP = activating protein; COX = cyclo-oxygenase; CPB = cardiopulmonary bypass; IκB = NF-κB inhibitory protein; iNOS = inducible nitric oxide synthase; MAPK = mitogen-activated protein kinase; NF-κB = nuclear factor-κB; NO = nitric oxide; RT-PCR = reverse transcriptase polymerase chain reaction; TNF = tumour necrosis factor.
(page number not for citation purposes) dial energetics and intracellular calcium handing [8]. In humans there is a clear relationship between TNF-α expression in the myocardium and the severity of dilated cardiomyopathy [9,10].
The nuclear factor-κB (NF-κB) family of nuclear transcription factors is critical for the synthesis of TNF-α and for TNF-α induced secondary mediators of inflammation, such as inducible nitric oxide synthase (iNOS) and cyclo-oxygenase (COX)-2 [11]. Inflammatory stimuli lead to activation of NF-κB by inducing the phosphorylation of its inhibitory protein IκB, allowing its translocation into the nucleus [11][12][13]. Activating protein (AP)-1 is another major transcription factor for many inflammatory mediators, including TNF-α. It comprises a family of related transcription factors, consisting of heterodimers and homodimers of Jun, Fos and activating transcription factor [14]. AP-1 activity is regulated through interactions with extracellular and intracellular signals including p38 mitogen-activated protein kinase (MAPK), with phosphorylation of activating transcription factor-2 [14], which leads to expression of TNF-α [15].
Upon activation of NF-κB and AP-1 by inflammatory stimuli, expression of inflammatory genes such as that encoding TNFα and of proinflammatory enzymes such as iNOS and COX-2 takes place. In the myocardium, activation of NF-κB, p38 MAPK and AP-1 causes myocardial cell damage resulting from TNF-α production [16][17][18] and it contributes to perfusion maldistribution and to myocardial damage by nitric oxide and eicosanoids, caused by the activity of iNOS and COX-2, respectively [19].
Our previous experimental studies showed that moderate hypothermia during cardiac surgery involving CPB is related to repression of TNF-α, and that this is related to increased synthesis of interleukin-10 in myocardium [2,20]. In the present study we investigated the signalling pathways involved in this repression and found that the use of moderate hypothermia is associated with the inhibition of the p38-MAPK/AP-1 pathway but not with inhibition of the NF-κB pathway.

Animals
The study was approved by the supervising state agency for animal experiments. Twelve stress-resistant female pigs (deutsche Landrasse) weighing 40.3 ± 1.4 kg (mean ± standard deviation) were included. The animals were housed in the institute for animal experimentation located in our university hospital for at least 8 days before experiments were begun; this was to guarantee quiet care before scheduled cardiac surgery. After clinical veterinary examination was conducted, which confirmed that the animals were in good health, the pigs were randomly assigned to a temperature group during CPB (six pigs in each group): moderate hypothermia (28°C) and normothermia (37°C). Core temperature was monitored using an oesophageal probe (probe 1651; Datex-Ohmeda Division, Instrumentarium Corp., Helsinki, Finland).

Surgical procedure
General anaesthesia, and operative and CPB technique were as previously described [2]. Briefly, following sternotomy, cephotiam (50 mg/kg intravenously) and heparin were administered, and both caval veins, the aorta and the left atrium were cannulated and total CPB instituted for 120 minutes in all animals. This included 30 minutes perfusion during which animals subjected to moderate hypothermia during the operation were cooled down to 28°C; 60 minutes of perfusion during which the aorta was cross-clamped and right atriotomy performed; and 30 minutes of perfusion during which pigs undergoing hypothermic CPB were rewarmed to 37°C. In animals subjected to normothermia during the operation, perfusion was performed at 37°C for the 120 minutes. CPB was conducted with a flow index of 2.7 l/(minute. m 2 body surface area) in all animals, with a target mean systemic arterial pressure of 60 mmHg. Immediately after aortic cross-clamping, cardioplegia was achieved by a single injection of cold (4°C) cardioplegic solution (Bretschneider solution; 30 ml/kg) into the aortic root. Additional topical cooling of the myocardium was performed by application of 500 ml cold (4°C) saline solution. Myocardial temperature during CPB was monitored using a needle probe placed into the ventricular septum (Temperature Sensing Catheter; Medtronic Hemotec. Inc, Englewood, CO, USA). At the end of CPB, anticoagulation was reversed with protamine, mediastinal drains were placed and the chest was closed.

Postoperative care
The lungs of the pigs were mechanically ventilated until the end of the experiment. Postoperative monitoring included continuous registration of heart rate and rhythm, mean arterial blood pressure, left atrial pressure, oesophageal temperature and urine output, and measurement of arterial lactate levels and blood gases. Animals received dopamine and Ringer's lactate to optimize hemodynamics. Six hours after sternal closure the pigs were killed by phenobarbital overdose.

Tissue sampling
Samples for RT-PCR, Western blot and electrophoretic mobility shift assay were rapidly excised from the apex of the right ventricle before CPB, before aortic cross-clamping, before opening the aorta and immediately after death. Samples were snap frozen in liquid nitrogen and stored at -70°C.

Reverse transcriptase polymerase chain reaction
Total RNA was extracted using Rneasy Mini Kit (QIAGEN Inc., Hilden, Germany). RNA (3 µg) was reverse transcribed to cDNA using random hexamers. Using specific porcine primers for TNF-α and β-actin, cDNA products were coamplified by PCR as previously reported [2]. The PCR products were subjected to electrophoresis in 1.8% agarose gel, stained with ethidium bromide and photographed. The predicted lengths of amplification products for TNF-α and β-actin were 372 and 233 base pairs, respectively. Results are presented as ratio of band intensities of the mRNA of TNF-α to the corresponding β-actin mRNA (Quantity One ® Quantification software 4.1; Bio-Rad: BioRad Laboratories, Inc., Hercules, CA, USA).
For measurement of the concentrations of phospho-c-Jun and c-Jun, nuclear extracts (20 µg) were treated with SDS-PAGE sample buffer, followed by heating, and were then subjected to 10% gels. Western blots were performed with antibodies against monoclonal mouse anti-human phospho-c-Jun and polyclonal rabbit anti-human c-Jun (both Santa Cruz Biotechnology, Heidelberg, Germany) and polyclonal goat anti-human actin (DAKO). The bands were detected using a chemiluminescent system. Restaining with actin antibody ensured equal loading. Band intensities of phospho-c-Jun and c-Jun were normalized to that of actin (Quantity One ® Quantification software 4.1; Bio-Rad).

Electrophoretic mobility shift assay
Nuclear extracts were prepared as previously described [21]. Protein concentrations were determined using a Bio-Rad protein assay. The electrophoretic mobility shift assay was performed using a double-stranded 32 P-labelled mutated sisinducible element oligonucleotide from NF-κB (5'-AGT TGA GGG GAC TTT CC-3') and using a double-stranded 32 Plabelled consensus oligonucleotide from AP-1 (5'-CGC TTG ATG ACT CAG CCG GAA-3'; both MWG-Biotech AG, Ebersberg, Germany). For supershift assays, nuclear extracts were incubated with antibodies against NF-κB p50 and NF-κB p65 subunits and AP-1 c-Jun subunit (all polyclonal rabbit antihuman; Santa Cruz Biotechnology) for 30 minutes at room temperature before addition of the radiolabelled probe. The protein/DNA complexes were separated on a 6% polyacrylamide gel containing 7.5% glycerol in 0.25-fold tuberculin bacillary emulsion (20 mmol/l Tris, 20 mmol/l boric acid, 0.5 mmol/ l EDTA) at 210 V for 3 hours. Gels were dried and autoradiographed. Positive controls for AP-1 were performed on human HepG2 stimulated by TNF-α 10 ng/ml for 4 hours.

Statistical analysis
Results are expressed as mean ± standard deviation. Data were analyzed by analysis of variance with adjustment of P values by the t test. Differences between time points within groups were calculated using paired t tests using the Statistical Package for Social Sciences (SPSS; SPSS Software GmbH, Munich, Germany).

Oesophageal and myocardial temperatures
Oesophageal temperature during and after CPB was significantly lower in animals operated on under hypothermia than in the other animals. In contrast, myocardial temperature during CPB did not differ significantly between groups, with the exception of that measured before cross-clamping of the aorta (Table 1).

Haemodynamics and lactate levels
Heart rate, mean arterial and left and right atrial pressures, cathecolamine support and urine output were not significantly different between groups (data not shown). Lactate levels were lower at the end of CPB in animals operated on under moderate hypothermia than in the other animals (4.8 ± 0.3 mg/ ml versus 6.9 ± 0.3 mg/ml; P < 0.002).

Myocardial expression of tumour necrosis factor-α
There was no TNF-α expression before institution of CPB; however, there was expression 30 min after establishing CPB and before cross-clamping of the aorta. At that time, both TNFα gene expression and TNF-α concentrations tended to be Results (mean ± standard deviation) are shown as the ratio of protein levels of phospho-IκB-α, iNOS and COX-2 to actin. *P < 0.05 between groups. COX, cyclooxygenase-2; CPB, cardiopulmonary bypass; iNOS, inducible nitric oxide synthase; NF-κB, nuclear factor-κB; phospho-IκB-α, phosphorylation of inhibitor of NF-κB α.
lower in pigs operated on under moderate hypothermia than in the other animals (Figure 1a,b). Six hours after CPB, concentrations of TNF-α were lower in animals operated on under moderate hypothermia than in the others (P < 0.05; Figure  1a,b).

Activation of nuclear factor-κB
There was weak DNA binding activity of NF-κB before CPB that increased during and after CPB similarly in both animal groups ( Table 2). The supershift experiment showed that the NF-κB DNA binding complex containing p50 and p65 was already present before CPB (Figure 2a,b). Phosphorylation of IκB-α in the myocardium paralleled DNA binding activity of NF-κB before, during and after CPB in both animal groups ( Table   2).

Phosphorylation of p38 mitogen-activated protein kinase
There was weak phosphorylation of p38 MAPK before CPB in all animals. In those operated on under hypothermia, levels of phospho-p38 MAPK were lower during and after CPB than in animals operated on under normothermic conditions. In the latter animals, levels of phospho-p38 MAPK increased as soon as 30 minutes after CPB was established and reached a peak value just before removal of the aortic clamp (1 hour after ischaemia), and then decreased by 6 hours after CPB. Animals operated on under hypothermia exhibited lower levels of phospho-p38 MAPK 30 minutes after establishing CPB as compared with the other animals (P < 0.05; Figure 3a). Levels of phospho-c-Jun detected in the nuclear extract paralleled the activation of p38 MAPK in all animals during CPB, but continued to increase after CPB exclusively in those subjected to normothermia. In the other animals, levels of phospho-c-Jun were lower after removal of the aortic clamp and 6 hours after CPB (P < 0.05; Figure 3b).

Activation of activating protein-1
There was weak DNA binding activity of AP-1 before CPB in all animals. In animals operated on under hypothermia, AP-1 activity was lower during and after CPB than in those operated on under normothermia. In these latter animals, the DNA binding activity of AP-1 increased as soon as 30 minutes after CPB was established, remaining elevated for up to 6 hours after CPB (Figure 4a). Animals operated on under hypothermia exhibited lower DNA binding activity of AP-1 during and after CPB as compared with the others (P < 0.05 and P < 0.005, respectively; Figure 4a). The supershift experiment revealed that the AP-1 DNA binding complex containing c-Jun was present before and persisted during and after CPB (Figure 4b).
Synthesis of inducible nitric oxide synthase and cyclooxygenase-2 iNOS was not detected before but 30 minutes after establishing CPB in all animals. Its synthesis increased during and after CPB without any difference between groups (table 2). COX-2 levels increased during CPB in all animals. At 30 minutes after establishing CPB the COX-2 levels in animals operated on under moderate hypothermia were significantly lower than in the other animals (P < 0.005; Table 2).

Discussion
This study shows for the first time that the repression of TNFα associated with application of moderate hypothermia during CPB is associated with inhibition of the p38 MAPK/AP-1 pathway but not that of the NF-κB pathway.
In previous studies we reported that animals operated on under moderate hypothermic CPB have lower plasma levels and lower myocardial concentrations of TNF-α as well as less organ damage than do animals operated on under normothermic conditions [2,20]. In the present study we found that the anti-inflammatory effects of moderate hypothermia with repression of TNF-α and of its secondary mediator COX-2 are present as early as 30 minutes after initiation of CPB. This observation is supported by other investigators [22], who reported repression of TNF-α by hypothermia during ischaemia and early reperfusion periods in the liver.
The main aim of this work was to identify the upstream mechanisms that regulate the expression of TNF-α and COX-2, lev-els of which appear to be reduced by hypothermia. We therefore investigated both of the main signalling pathways involved in the expression of inflammatory mediators, namely the NF-κB [11] and AP-1 [23] pathways, and observed a differential effect of hypothermia on those transcription factors. Indeed, although in our series hypothermia led to reductions in phosphorylation of p38 MAPK and AP-1 activation, it did not affect activation of NF-κB in the myocardium. This confirms our previous work showing that hypothermia does not inhibit NF-κB activation in liver [21].
The inhibitory effect of hypothermia on p38 MAPK that we report here is supported by experimental work conducted in rat fibroblasts that showed inhibition of Raf [24], a mitogenstimulated protein kinase that is an important intermediate to p38 MAPK activation by cold stress [25]. Via inhibition of Raf, hypothermia suppresses the phosphorylation of p38 MAPK in vitro, thereby inhibiting phosphorylation of c-Jun and AP-1 activation [22].
Regulation of the activity of AP-1 occurs at two levels, depending on its concentrations and on the level of its phosphorylation [26]. In the present study, there was no difference in levels of c-Jun in nuclear extract between the groups, but animals operated on in hypothermia exhibited lower phosphorylation of c-Jun and lower DNA binding activity of AP-1 during and after CPB than did those operated on in normothermia. Therefore, we suggest that hypothermia during CPB inhibits phosphorylation of p38 MAPK, which in turn suppresses its nuclear targets, namely phosphorylation of c-Jun and activation of the transcription factor AP-1.
The clinical relevance of NF-κB and p38 MAPK/AP-1 activation in myocardium has been pointed out by several studies suggesting that both pathways are important actors in the failing heart [18,27]. NF-κB plays a central role in the myocardial inflammatory response, including TNF-α secretion by cardiomyocytes in response to systemic endotoxin [28]. However, NF-κB activation is also necessary for later preconditioning of the myocardium, and thus a potential role of this transcription factor for myocardial protection has been suggested [11]. The important role played by p38 MAPK/AP-1 in the myocardial inflammatory response is widely recognized [18,23]. Indeed, p38 MAPK activation is sufficient to induce inflammatory cytokine expression including TNF-α in cardiomyocytes in vitro and in vivo [18]. A recent study [18] showed that the inhibition of p38 MAPK activities blocks TNF-α secretion in transgenic hearts. The congruency of the effect of moderate hypothermia on phosphorylation of p38 MAPK and activity of AP-1, and on expression of TNF-α suggests that, in our model, moderate hypothermia during CPB represses the expression of TNF-α in the myocardium by selective inhibition of the p38 MAPK/AP-1 pathway.
In a similar in vivo model, we previously showed that hypothermia during CPB confers myocardial protection by inhibiting intramyocardial expression of TNF-α [2,20]. Cardiodepressive effects of TNF-α have been ascribed to immediate negative inotropic effects on cardiomyocytes mediated by sphyngosine, independent of NO, and to delayed NO-dependent negative inotropic effects [29]. High local concentrations of NO related to induction of iNOS are associated with myocardial cell damage and cell death [30]. In the present study, hypothermia during CPB did not inhibit synthesis of iNOS in myocardium despite repression of TNF-α. That expression of iNOS in cultured cardiac myocytes does not increase in response to TNF-α and that TNF-α does not influence iNOS mRNA expression and NO release in the isolated rat heart [31][32][33] supports our observation. In addition, endotoxin-induced TNF-α depresses myocardial contractility of isolated rat hearts independent of NO synthesis [34].
COX-2 is upregulated by various factors, including TNF-α, via the transcription factors NF-κB and AP-1 [11,35]. The latter is the major promotor element involved in COX-2 expression in cardiomyocytes [36]. Significant expression of COX-2 has been demonstrated in the myocardium of patients with congestive heart failure and in rat heart after treatment with endotoxin [37,38]. In our series expression of COX-2 paralleled the expression of TNF-α and activation of AP-1 during and after CPB, suggesting that COX-2 expression was mediated by TNF-α through AP-1. Myocardial depression in isolated rat hearts in response to staphylococcal α-toxin results from COX-2 derived thromboxane A 2 liberation, leading to coronary vasconstriction and perfusion mismatch [39,40]. In the present study, animals operated on under hypothermia did not differ from the other animals with respect to haemodynamics, but the fact that they had lower levels of lactate than did the pigs operated on under normothermia indicates that they had better tissue perfusion.

Conclusion
Our findings suggest that myocardial repression of TNF-α and COX-2 related to moderate hypothermia during CPB is due to inhibition of the p38 MAPK/AP-1 pathway but not that of NF-κB.