Myocardial function is depressed in sepsis and is an important prognosticator in the human condition. Using echocardiography in a long-term fluid-resuscitated Wistar rat model of faecal peritonitis we investigated whether depressed myocardial function could be detected at an early stage of sepsis and, if so, whether the degree of depression could predict eventual outcome. At 6 h post-insult, a stroke volume <0.17 ml prognosticated 3-day mortality with positive and negative predictive values of 93 and 80%, respectively. Subsequent fluid loading studies demonstrated intrinsic myocardial depression with poor-prognosis animals tolerating less fluid than either good-prognosis or sham-operated animals. Cardiac gene expression analysis at 6 h detected 527 transcripts significantly up- or down-regulated by the septic process, including genes related to inflammatory and cell cycle pathways. Predicted mortality was associated with significant differences in transcripts of genes expressing proteins related to the TLR2/MyD88 (Toll-like receptor 2/myeloid differentiation factor 88) and JAK/STAT (Janus kinase/signal transducer and activator of transcription) inflammatory pathways, β-adrenergic signalling and intracellular calcium cycling. Our findings highlight the presence of myocardial depression in early sepsis and its prognostic significance. Transcriptomic analysis in heart tissue identified changes in signalling pathways that correlated with clinical dysfunction. These pathways merit further study to both better understand and potentially modify the disease process.
- animal model
- faecal peritonitis
- gene transcript
- heart failure
• Myocardial function is depressed in sepsis and is an important prognosticator in patients.
• Using a long-term fluid-resuscitated rat model of faecal peritonitis, we demonstrate that significant differences in stroke volume and HR measured 6 h post-insult could predict 3-day mortality with positive and negative predictive values of 93 and 80% respectively. In separate studies, cardiac gene expression analysis at 6 h detected 527 transcripts significantly up- or down-regulated by the septic process, including genes related to inflammatory and cell cycle pathways. The degree of change in these signalling pathways correlated with clinical dysfunction.
• These findings suggest a crucial role for early cardiovascular performance in determining subsequent outcome, with clear therapeutic implications.
Sepsis, the systemic inflammatory response syndrome to infection, is the leading cause of death in the critically ill. This occurs primarily through the development of multiple organ failure , in which cardiac dysfunction is a well-recognized manifestation [2,3]. Identified mechanisms for myocardial depression include alterations in adrenergic signalling, intracellular calcium cycling, impaired electromechanical coupling and mitochondrial dysfunction . The hierarchy of these mechanisms, however, remains uncertain and individual variations in severity inadequately explained.
A clinically relevant sepsis model achieving a 50% mortality rate raises the important but hitherto unaddressed question as to why some animals survive while the remainder die despite similar, if not, identical characteristics in terms of genotype, gender, age, weight and environment, and the septic insult received. Though these factors differ considerably in humans with sepsis, resulting in variable degrees and combinations of multiple organ dysfunction, a wide selection of inflammatory , hormonal [6,7], circulatory , bioenergetic  and organ dysfunction biomarkers [7,10] can prognosticate as early as the first few days of intensive care admission. Notably, survivors and non-survivors are often clinically indistinguishable at this early time point. This implies that mortality may be determined to a large extent by a pattern of molecular genetic and functional responses that differentiate survivors from non-survivors at an early stage.
Although the degree of myocardial depression has been reported in several clinical studies to be an indicator of poor outcome [11,12], others suggest this may have a protective role . This disparity may relate to timing of the study relative to the duration of critical illness. However, the role and prognostic significance of early myocardial dysfunction has received relatively scant attention. We hypothesized that early cardiac performance could predict eventual outcome, and performed transcriptomic analysis to probe putative pathways that could reconcile the physiological and molecular mechanisms of sepsis-induced cardiac dysfunction.
MATERIALS AND METHODS
All experiments were performed according to local ethics committee (University College London) and Home Office (U.K.) guidelines under the 1986 Scientific Procedures Act. Under 2% isoflurane anaesthesia, male Wistar rats (Charles River) had PVC tubes (inner diameter, 0.58 mm; outer diameter, 0.96 mm) inserted into the right jugular vein and left carotid artery, and tunnelled subcutaneously to emerge at the nape of the neck. These lines were subsequently mounted on to a swivel-tether system allowing the rat, on recovery from anaesthesia, to have unimpeded movement in its cage and free access to food and water. After 24 h, sepsis was induced by i.p. (intraperitoneal) injection of faecal slurry (1.8 ml of a 40% preparation obtained from the bowel content of a rat from the same batch). Fluid resuscitation, consisting of a 1:1 solution of 6% hetastarch (Elohaes®; Fresenius-Kabi) and 5% glucose, was commenced 2 h later through the central venous line. Glucose supplementation prevented hypoglycaemia in the animals with sepsis due to decreased appetite and intake. The infusion rate was maintained at 10 ml/kg of body weight per h between 2 h and 24 h, halving thereafter at 24 h intervals. Animals received additional 25 ml/kg of body weight boluses of 6% hetastarch at 6 h and 10 ml/kg of body weight at 24 h to optimize fluid resuscitation (results from pilot studies). Sham animals of similar age, gender and weight underwent identical instrumentation and fluid administration, but did not receive the i.p. injection of faecal slurry. At 6, 24, 48 and 72 h post-sepsis induction, and in sham-operated controls, clinical scoring was performed. This included an assessment of appearance, alertness and movement, as previously described .
Echocardiography studies were carried out using a 14 MHz linear-array transducer connected to a digital ultrasound system (Vivid 7; GE Healthcare). Images were acquired in left parasternal long and short axis views at the papillary muscle level. Radius (r) and length (l) of the left ventricle were measured at end-diastole by M-mode and two-dimensional mode respectively. LV (left ventricular) end-diastolic volumes were calculated using a prolate spheroid formula (4/3·π·r2·l). Stroke volume was measured in the ascending aorta (radius 1.3 mm) from the velocity–time integral using pulsed-wave Doppler. The vessel diameter at this level remains stable despite intravascular volume variations .
Study 1: long-term outcome study
A total of 20 rats with sepsis and eight sham-operated rats (body weight, 334±4.5 g) were monitored for 72 h, and time of death was recorded. At baseline, 3, 6, 24, 48 and 72 h, spontaneously breathing animals were removed from their cage, anaesthetized with isoflurane and placed in a supine position on a warming mat for echocardiography. The concentration of isoflurane was kept between 1.0 and 1.5%, so that the animals tolerated the study but could react to a leg pinch with a withdrawal response.
Study 2: fluid loading study
In 14 rats with sepsis and six sham rats (body weight, 305±6.0 g) i.v. (intravenous) fluid resuscitation was started 2 h after the insult with echocardiography performed at 6 h, as described above. Following echocardiography, animals received a 2.5 ml bolus (approximately 8 ml/kg of body weight) of 6% hetastarch intravenously over 2 min, and this was repeated at approximately 5 min intervals. Echocardiography was performed immediately after each fluid bolus. Bolus fluid administration was repeated until death occurred.
Study 3: histology
Two sham animals and 11 animals with sepsis were prepared as above. At 6 h post-insult, echocardiography was performed to distinguish predicted sepsis survivors from non-survivors. Animals were killed at 24 h. Cardiac samples for histology and EM (electron microscopy) were taken from the LV free wall at the papillary muscle level and were analysed by blinded investigators.
Study 4: network-based gene expression analysis
The model was prepared as described above. At 6 h post-insult, clinical severity was recorded  and echocardiography performed to distinguish predicted survivors from non-survivors. Freshly culled (naïve) and sham-operated animals served as controls. Hearts (four per group) were promptly removed and stored in liquid nitrogen. Heart muscle samples (30–50 mg each) were transferred into 900 μl of lysis solution (RNeasy Mini Kit; Qiagen) including mercaptoethanol, homogenized in a tissue lyser at 30 Hz for 4 min at room temperature (Qiagen), then centrifuged and processed further, according to manufacturer's instructions. RNA integrity and purity were analysed (Agilent 2100; Agilent), and 300 ng of total RNA was reverse-transcribed and amplified (Illumina RNA amplification Kit IL1791; Ambion). All cRNA samples were quantified with a NanoDrop spectrophotometer before proceeding to sample hybridization. A portion (0.75μg of cRNA) of each sample was hybridized to Sentrix Rat Ref-Seq-12 Beadchips (BD-27-302; Illumina) with >24000 gene-specific oligonucleotide probes per array, targeting all known genes and splice variants. The hybridized arrays were stained with Streptavidin-Cy3 (FluoroLinkTM Cy3™; Amersham Biosciences), diluted to 1 mg/ml or 1 μg/μl with RNase-free water, washed, dried and scanned immediately on the Illumina Bead Station. Data were normalized using a variance stabilizing transformation . For the heatmap, the expression values of selected genes were Z-score standardized (S.D.=1, mean=0). Data were analysed using Ingenuity Pathway Analysis (Ingenuity Systems) for functional analyses of the dataset to identify significant biological functions and diseases connected with the expression profile. The microarray experiment description file according to the MIAME checklist is available at ftp://rudiger2012:Phee5eTh@ftp.sirslab.de/
To confirm changes in mRNA transcript levels, ATP2A3 (ubiquitous calcium transporting ATPase), PRKAG2 (protein kinase AMP-activated protein kinase γ2 non-catalytic subunit), PRKACA (protein kinase cAMP-dependent catalytic α) and PPP1CB (protein phosphatase 1 catalytic subunit β-isoform) were quantified by real-time PCR. Results were computed with specialized software (qBase plus; Biogazelle) and given as the ratio between the gene of interest and the housekeeping gene HMBS (hydroxymethylbilane synthase).
Results are expressed as means±S.E.M. or percentages. Continuous variables were compared with ANOVA. If appropriate, post-hoc testing was performed using Bonferroni corrections. χ2 testing compared the categorical variables. The quality of stroke volume as a diagnostic test was described by the area under the ROC (receiver operator characteristic) curve. Relationships between gene array and PCR values are described by the square of the sample correlation coefficient. All testing was two-tailed; P values <0.05 were considered significant.
Study 1: long-term outcome study
Animals with sepsis developed clinical manifestations of illness including hunched appearance, piloerection, bloating and lethargy, reaching a nadir between 16–48 h. A spectrum of clinical severity was observed, even between animals from the same litter, which received the same batch of faecal slurry. Animals with sepsis showed early signs of sepsis at 6 h after i.p. slurry injection; their clinical severity score of 2.6±0.4 contrasted with 0±0 for sham animals (P<0.001). However, the clinical severity was similar between animals predicted to survive or not. Survival rates at 6, 24, 48 and 72 h were 95, 55, 30 and 25% respectively. At post-mortem, purulent peritonitis was present with ascites, adhesions and enlarged, fluid-filled bowel loops. Survivors showed signs of recovery by 72 h with increased interest in their surroundings, improved appetite and appearance. Sham animals became oedematous yet continued to eat, drink and maintain an interest in their environment. One sham animal developed a local wound infection at the tether anchorage site. Two other sham animals developed right ventricular dilatation, with one dying between 48 and 72 h. All available results are included in the analyses. Clinical and haemodynamic parameters are displayed in Table 1 and Figure 1. Fluid resuscitation, starting at 2 h and optimized by a fluid bolus at 6 h, restored preload deficiency in the animals with sepsis. As a result of the fluid loading, stroke volume and, consequently, cardiac output increased in all groups.
Stroke volume measured at 6 h was a good discriminator of outcome, with an area under the ROC curve of 0.83 [95% CI (confidence interval), 0.57–1.1; P=0.033]. A total of 13 out of the 14 animals with a stroke volume <0.17 ml died (mortality 93%) compared with one of five animals with a stroke volume ≥0.17 ml (mortality 20%, P=0.006). Hence a 6 h stroke volume <0.17 ml could prognosticate 3-day mortality with high positive and negative predictive values of 93 and 80%, respectively. Consequently, this cut-off value was used in subsequent experiments to distinguish predicted survivors from non-survivors. Unlike body temperature and clinical severity, HR (heart rate) measured at 6 h could also discriminate outcomes [ROC 0.93 (95% CI, 0.81–1.0); P=0.005], with similar positive (93%) and negative (80%) predictive values. Animals with sepsis with a HR ≥430/min had a mortality of 93%, whereas those with a HR below this cut-off had a 3-day mortality of only 20% (P=0.006). Supplementary Table S1 (at http://www.clinsci.org/cs/124/cs1240391add.htm) shows the reproducibility of the echocardiography parameters.
Study 2: fluid loading study
Figure 2(A) shows changes in stroke volume with repeated fluid administration. Predicted non-survivors were fluid-responsive but failed to reach maximal stroke volumes attained by predicted survivors or sham-operated animals. Total volume infused before death was 31±7.3 ml in predicted non-survivors, 49±6.4 ml in predicted survivors and 61±5.6 ml in sham animals (P<0.05). Right ventricular dilatation with increasing septal shift and impaired LV filling, suggestive of right heart failure, was seen as a pre-terminal event (Figure 2B).
Study 3: histology
Only three out of 11 animals with severe sepsis survived until 24 h in this substudy. Tissues from these three animals were compared with tissues from two sham controls. H&E (haematoxylin and eosin)-stained cardiac tissues showed no signs of necrosis and minimal immune cell infiltration. Oil Red-O staining failed to demonstrate intracellular lipid accumulation. EM of hearts from animals with sepsis revealed glycogen accumulations along clusters of mitochondria and between myofibrils. Although the contractile apparatus appeared normal, many mitochondria showed morphological derangements (Figure 3).
Study 4: network-based gene expression analysis
Gene expression profiling of heart tissue from naïve, sham, and predicted sepsis survivor and sepsis non-survivor subgroups revealed 527 significantly altered gene transcript activities. Figure 4 shows the differences in mean gene expression values between the groups, displayed as four clusters of co-regulated transcripts. Cluster 2 represents transcripts of decreased abundance related to anaesthesia and instrumentation. These differed significantly in amount between naïve and sham rats; sepsis did not further alter these transcript levels so this can be interpreted as an ‘injury’ cluster. Here, Ingenuity Pathway Analysis revealed a pronounced effect on transcript levels reflecting tissue proliferation status (cell cycle and replication), and a decrease in cAMP signalling transcripts. Cardiac actin and mitochondrial ribosomal protein transcripts were also decreased. In addition, down-regulation of ARDB2 (adrenergic receptor β2) and PRKCA (protein kinase Cα) may have contributed to the cardiac dysfunction observed in the animals with sepsis by providing an initial basic trigger for subsequent sepsis-specific signalling in the adrenergic pathway. Cluster 4 represents transcripts elevated in sham-operated compared with naïve animals, but not further affected by sepsis. Hence, this can be interpreted as a ‘danger’ cluster.
Clusters 1 and 3 represent the changes in transcripts specific to sepsis. Supplementary Table S2 (at http://www.clinsci.org/cs/124/cs1240391add.htm) lists all of the increased (65 genes) and decreased (55 genes) transcripts in predicted non-survivors compared with survivors. To study sepsis-specific events in more detail, subsequent analyses focused on these two clusters. Compared with the sham group, the main signalling pathways affected by sepsis were the acute-phase response plus TLR (Toll-like receptor), IL (interleukin)-6 and IL-10 signalling. As depicted for the TLR and IL-6 signalling pathways (Figure 4C), central players of innate immunity [e.g. TLR2, MyD88 (myeloid differentiation factor 88), LBP (lipopolysaccharide-binding protein) and the JAK/STAT (Janus kinase/signal transducer and activator of transcription) pathway] responded with an increase in transcript abundance. Ingenuity Pathway Analysis of the sepsis-regulated transcripts revealed profound effects on gene expression activities in animals with sepsis, including the functional categories of inflammatory disease, organ injury and abnormalities, skeletal and muscular disorders, and cardiovascular disease. Transcript levels of STAT3, REL, c-Myc, BCL10 and p21, potent regulators of proliferation and apoptosis at the transcription level, were increased in response to sepsis. Key mediators of inflammation such as PLA2G4A (phospholipase A2 group IVA), PTGER4 (prostaglandin E receptor 4) and PTGS2 [prostaglandin-endoperoxidase synthase 2: COX2 (cyclo-oxygenase 2)] were likewise elevated.
Pathways associated with predicted poor outcome included transcripts encoding for amino-sugar metabolism [PDE10A (phosphodiesterase 10A), UAP1 (UDP-N-acetylglucosamine pyrophosphorylase 1) and CYB5R2 (cytochrome b5 reductase 2)] and p53-dependent cell-cycle arrest [CDKN1A (cyclin-dependent kinase inhibitor 1A), GADD45G (growth-arrest and DNA-damage-inducible protein 45G) and STAG1 (stromal antigen 1)]. Consistent with the progressive worsening in myocardial function demonstrated physiologically, significant differences in transcripts within the β-adrenergic signalling and calcium-cycling pathways were also linked with an unfavourable outcome (Figure 5). These included increases in PDE10A, PP1 (protein phosphatase 1) and ITPKC (inositol 1,4,5-trisphosphate 3-kinase C) and decreased SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase). In addition, the PKA (protein kinase A) transcript was decreased in animals with sepsis compared with sham animals. The correlations between gene array and PCR results were good for ATP2A3 and PRKAG2, moderate for PRKACA and absent for PPP1CB (see Supplementary Figure S1 and Supplementary Table S3 at http://www.clinsci.org/cs/124/cs1240391add.htm).
In the present study, we describe a clinically relevant fluid-resuscitated long-term (3-day) rodent model of sepsis where outcome could be determined with high accuracy from haemodynamic measurements taken early in the illness. This demonstration of an early difference in phenotype in animals that proceeded to either recover fully or die occurred notwithstanding similarities in genotype, age, gender and upbringing of these animals, and receipt of a similar insult. They reflect patient data where, despite the marked heterogeneity of patient populations, baseline cardiovascular variables such as HR [8,17], stroke volume  and cardiac output [18,19] could differentiate between survivors and non-survivors of septic shock. We used echocardiography-derived measurements of stroke volume for the present study, which predicted mortality with a high positive and negative predictive value. Though our focus was on ventricular performance, we also found a similar predictive ability could be made for HR. This may offer a viable option for future studies if the capability to measure stroke volume is not available. Our findings suggest that an increased systemic inflammatory response will compromise myocardial function through circulating myocardial depressant factors  including NO (nitric oxide) , and through altered gene expression of proteins involved in cardiac contractile and relaxation pathways (as demonstrated at 6 h in the present study) and, subsequently (at 12–48 h), in bioenergetic pathways .
The lower values of cardiac output in the predicted non-survivor group did not simply reflect more profound hypovolaemia related to a greater degree of capillary leak, as fluid loading failed to generate either the maximal stroke volumes attained by the sham or sepsis survivor groups, nor did it reduce HR. Notably, all animals with sepsis at 6 h were normotensive (Table 1) and had received 40 ml/kg of body weight (equivalent to 3 litres in humans) between 2 and 6 h. Intrinsic myocardial depression was, however, confirmed by a progressive decrease in tolerance to large volume fluid loading in the animals with sepsis; this occurred to a much greater extent in predicted non-survivors.
Hollenberg et al.  reported myocardial depression in mice with sepsis, despite the co-existence of a high cardiac output. These findings reflect those made in clinical echocardiographic studies, some of which could also prognosticate as early as the first day of ICU (intensive care unit) admission [12,24,25]. As seen in this study, the failing thinner-walled right ventricle copes less well with excessive fluid loading; this will likely be compounded in sepsis by co-existing (acute) lung injury, increasing pulmonary artery pressures and right ventricular afterload. A careful balance must therefore be reached between an avoidance of excess fluid that will compromise myocardial function and worsen interstitial oedema , and sufficient fluid administration to rescue the macrocirculation, improve organ perfusion and enhance survival . As our data demonstrate, the sicker the animals with sepsis become, the less fluid loading they are able to tolerate.
Histology demonstrated negligible cardiac tissue necrosis, supporting previous findings in animals with sepsis [15,27,28] and in non-survivors of human septic shock [29,30]. EM did display evidence of mitochondrial swelling, suggesting organelle damage, confirming previous work by ourselves and others in animals [21,22,28,31–33] and patients  with sepsis. The finding of glycogen accumulation in cardiomyocytes from animals with sepsis, as reported previously , has also been identified in hibernating myocardium of pigs following ischaemia . In summary, structural tissue damage was minor, although changes were apparent at the subcellular level. These observations support our hypothesis that myocardial dysfunction is a functional rather than a structural derangement [4,37].
This descriptive approach to myocardial depression in sepsis can be systematically studied in the present model by applying genome-wide analysis of alterations in the myocardial transcriptome. To this end, we employed a structured network knowledge-based approach that can provide insights into regulation of cell function and interaction, as demonstrated in healthy volunteers treated with endotoxin . Although we cannot yet describe precise mechanisms through which these changes occur, sepsis clearly elicits significantly different responses compared with sham operation alone, and that some of these are differentially expressed and associated with poor outcome.
The genome-wide profiling revealed early up-regulation of TLR2/MyD88 and JAK/STAT3-dependent signalling with sepsis. This supports findings of improved cardiac function, with normal sarcomere shortening and peak change in intracellular Ca2+, in TLR2−/− mice undergoing caecal ligation and puncture compared with their wild-type equivalents . Although the JAK/STAT3 pathway is generally considered to be cardio-protective , it too can contribute to cardiac dysfunction during ischaemia . A study of rats undergoing cecal ligation and puncture reported attenuated organ failure and improved survival rates following JAK/STAT3 pathway inhibition, although the heart was not specifically examined .
Importantly, in relation to the physiological changes recorded in this model, myocardial transcriptomics revealed alterations in transcript abundance linked to β-adrenergic signalling and Ca2+ flux that depended on disease severity and prognosis (Figure 5). Although mindful of the limitations in interpreting transcriptomic data, it is conceivable that such transcript differences could impact on myocardial function. In our model, the PKA transcript level was significantly decreased in the animals with sepsis, whereas PDE10 transcript abundance was elevated. PDE increases breakdown of cAMP, further reducing PKA activity . Increased activity of PP1 dephosphorylates phospholamban, reducing Ca2+ uptake into the SR (sarcoplasmic reticulum) . Down-regulation of SERCA has a similar effect . Up-regulation of PP1 and down-regulation of PKA also reduce Ca2+ entry into the cell by inhibiting L-type calcium channels . A diminution in L-type calcium current was reported recently in ventricular myocytes isolated from pigs with sepsis . High cytosolic and low SR Ca2+ levels lead to failure of diastolic relaxation and impaired systolic contraction, respectively. Both these aspects are also recognized in septic cardiomyopathy [4,48].
We focused our transcriptomic analyses on heart tissue as our prognostic tool was based on an assessment of myocardial dysfunction. Analyses of liver tissue taken concurrently show significant differences in other signalling pathways . It is thus reasonable to speculate that each organ has its own transcriptomic response to infection. A better understanding of each organ's response, and interactions between organs, may provide useful clues to developing novel therapies.
In summary, our findings emphasize both the presence and prognostic significance of cardiac dysfunction with intrinsic myocardial depression during early sepsis, at a time when overall clinical severity was not marked. The poor-prognosis animals had reduced tolerance to i.v. fluid loading, highlighting the importance of adequate but not excessive fluid administration in clinical management. Our study implies that outcome is already determined at an early stage in sepsis. Greater or lesser degrees of gene up- or down-regulation, with the likely addition of post-transcriptional modifications that were not measured in this study, will distinguish eventual survivors and non-survivors. If confirmed, this has major consequences for patient management as a proportion are clearly not benefiting from current approaches; new paradigms would need to be introduced to improve outcomes in such patients.
This work was supported by the Swiss National Science Foundation (Basel, Switzerland), the Stiefel Zangger Foundation (Zurich, Switzerland) and the Siegenthaler Foundation (Zurich, Switzerland) (grants to A.R.), and the U.K. Medical Research Council (grants to A.D., J.C. and V.T.). The echocardiography equipment was funded by the British Heart Foundation. Transcriptomic analyses were funded by the Krokus Foundation (Basel, Switzerland) (to A.R.), and the Federal Ministry for Education and Research (within the ‘Center for Sepsis Control and Care’) [grant number 01 EO 1002, Project D1.2 (to M.B.)]. This work was undertaken at UCL Hospitals/UCL, which received support from the National Institute of Health Research Biomedical Research Centre funding scheme.
Alain Rudiger conceived and designed the study, performed the experimental work, analysed and interpreted the data, and drafted and approved the paper. Alex Dyson, Karen Felsmann and Jane Carré performed the experimental work, and drafted and approved the paper. Valerie Taylor performed the experimental work. Sian Hughes performed the experimental work and interpreted the data. Innes Clatworthy, Jana Lemm and Ralf Claus performed the experimental work and interpreted the data. Alessandro Protti and Denis Pellerin assisted with experimental design and interpreted the data. Michael Bauer interpreted the data, and drafted and approved the paper. Mervyn Singer conceived and designed the study, interpreted the data, supervised the study, and drafted and approved the paper.
↵1 A member of the scientific advisory board of SIRS-Lab GmbH, Jena, Germany, where the microarray experiments were performed.
Abbreviations: ARDB2, adrenergic receptor β2; ATP2A3, ubiquitous calcium-transporting ATPase; CDKN1A, cyclin-dependent kinase inhibitor 1A; CI, confidence interval; COX2, cyclo-oxygenase 2; EM, electron microscopy; H&E, haematoxylin and eosin; HMBS, hydroxymethylbilane synthase; HR, heart rate; IL, interleukin; i.p., intraperitoneal; i.v., intravenous; JAK, Janus kinase; LBP, lipopolysaccharide-binding protein; LV, left ventricular; MyD88, myeloid differentiation factor 88; PDE10A, phosphodiesterase 10A; PKA, protein kinase A; PLA2G4A, phospholipase A2 group IVA; PP1, protein phosphatase 1; PPP1CB, protein phosphatase 1 catalytic subunit β-isoform; PRKACA, protein kinase cAMP-dependent catalytic α; PRKAG2, protein kinase AMP-activated γ2; PTGS2, prostaglandin-endoperoxidase synthase 2, ROC, receiver operator characteristic; SERCA, sarcoplasmic/endoplasmic reticulum Ca2+-ATPase; SR, sarcoplasmic reticulum; STAT, signal transducer and activator of transcription; TLR, Toll-like receptor
- © The Authors Journal compilation © 2013 Biochemical Society