This study was conducted with approval from the ethics board of the University of Bonn, Germany. Patients were included after written informed consent was given by them or their next-of-kin. The study included 16 patients who had been treated in three surgical intensive care units (ICUs) at the university hospital of Bonn after a diagnosis of severe sepsis according to the American College of Chest Physicians/Society of Critical Care Medicine consensus criteria 23. A blood sample was obtained from each patient on the day of inclusion. Ten patients treated in the ICU without sepsis or signs of infection served as controls. For both groups exclusion criteria were a lack of informed consent, age under 18 years, and preexisting haematological or immunological disease. The basic clinical characteristics of the critically ill patients and patients with severe sepsis (Table 1) have been published previously in part 24. As a second control group, 11 healthy subjects (six males, five females, median age = 34 years) were tested. All patients were Caucasian.

Blood samples were obtained within four hours after severe sepsis was first diagnosed to sample early stage severe sepsis. For flow cytometry, samples were anticoagulated with ethylenediaminetetraacetic acid (EDTA) and processed within 30 minutes. For RNA quantification, 2.5 ml whole blood was collected using blood RNA tubes (Paxgene system, PreAnalytiX, Qiagen, Hilden, Germany) and stored at -80°C until extraction. This method stabilised the RNA until further analysis.

Phosphatidylserine externalisation was measured by annexin V binding. Samples of 100 μl fresh whole blood were subjected to lysis with an ammonium chloride solution (Becton Dickinson, Heidelberg, Germany) for 15 minutes at room temperature to substantially lyse red blood cells without affecting the leucocyte populations. After washing, cells were phenotyped with phycoerythrin-labelled antibodies directed against CD4, CD8 and CD19 (Becton Dickinson, Heidelberg, Germany). They were then incubated with annexin-V-fluorescein-isothiocyanate (Becton Dickinson, Heidelberg, Germany) for 10 minutes at room temperature in Hepes buffer containing calcium (2.5 mM). Before flow cytometric analysis, cells were labelled with the DNA-dye 7-actinoaminomycin (7-AAD; Becton Dickinson, Heidelberg, Germany) for the detection of membrane leaks and subjected to data acquisition within 10 minutes. Cells that were phosphatidylserine positive but 7-AAD negative were selected by gating.

For caspase-3 activation and Bcl-2 expression measurement, samples of 100 μl whole blood were analysed. Erythrocytes were lysed with 1.5 ml PharmLyse, an ammonium chloride-based lysing reagent. (Pharmingen, Heidelberg, Germany) at room temperature. All subsequent steps were carried out at 4°C. Cells were washed with PBS (Sigma-Aldrich, Steinheim, Germany) and fixed with 750 μl PBS containing 4% paraformaldehyde (Sigma-Aldrich, Steinheim, Germany). After fixation, cells were permeabilised with 1ml PBS containing 0.5% saponin (Sigma-Aldrich, Steinheim, Germany) and 1% BSA (Sigma-Aldrich, Steinheim, Germany). Cells were washed twice with PBS (0.5% saponin/1% BSA). Active caspase-3 was detected by a phycoerythrin-labelled antibody directed against the active fragment (Becton Dickinson, Heidelberg, Germany). Bcl-2 was detected with a specific phycoerythrin-conjugated monoclonal antibody (Becton Dickinson, Heidelberg, Germany). Isotype control antibodies (Becton Dickinson, Heidelberg, Germany) were used to determine unspecific binding. After washing cells were phenotyped antibodies against CD4, CD8 and CD19 that were labelled with peridinin-chlorophyll-protein (PerCP; Becton Dickinson, Heidelberg, Germany). Cells were washed twice and resuspended in 250 μl PBS (0.5% saponin/1% BSA) for flow cytometric analysis.

Stained and fixed cells were measured by flow cytometry. Data were acquired in a flow cytometer (FACSCalibur, Becton Dickinson, Heidelberg, Germany) and analysed via CellQuest Pro software (CellQuest Pro, Version 4.0.2, Becton Dickinson, Heidelberg, Germany). Ten-thousand cells of interest were acquired. Populations of interest were selected by multiple gating using forward (cell size) and sideward (cell granularity) scatter and green or red regions on fluorescence channels. Antibody-binding was expressed as mean fluorescence intensity corrected for nonspecific binding using matched isotype control antibodies. Results of caspase-3 staining were given as percentages of positive cells. In case of Bcl-2 staining, daily calibration curves were generated using latex beads coupled to defined quantities of phycoerythrin (Quantibrite, Becton Dickinson, Heidelberg, Germany). Fluorescence intensities were calculated accordingly to compensate for potential deviations from linearity of the cytometer and to reduce day to day variability. Since the Bcl-2 antibody used, like most conjugated antibodies, is not guaranteed to be coupled to phycoerythrin in a one-to-one ratio, results were not given as molecules per cell but as linear fluorescence units.

Total RNA was extracted from whole blood using a PAXgene Blood RNA kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. cDNA was synthesised with a 1^st Strand cDNA Synthesis Kit for RT-PCR (Roche Diagnostics, Mannheim, Germany). The reaction mixture contained 8.2 μl of total RNA (equivalent to about 500 ng), 5 mM magnesium chloride, 2.1 mM deoxyribonucleotide triphosphate (dNTP), 3.2 μg of random primer p(dN)₆ (0.04 A₂₆₀ units/μl), 6.50 units of RNase inhibitor, 20 units of avian myeloblastosis virus reverse transcriptase, and one sample of reaction buffer to a total volume of 20 μl. The reaction was incubated as follows: 25°C for 10 minutes, 42°C for 60 minutes, 99°C for five minutes and 4°C for five minutes.

Real-time PCR with cDNA from blood samples was performed according to the manufactuer's manual in a total volume of 20 μl on a LightCycler instruments using the Light-Cycler FastStart DNA^Plus SYBR Green I (both from Roche Diagnostics, Mannheim, Germany). For each sample, reactions were performed in duplicates for target genes. Fluorescence was monitored at the end of the second segment of each cycle. PCR reactions were performed under the following conditions: the reactions started with an initial denaturation at 95°C for 10 minutes followed by 45 amplification cycles; Bcl-2: 95°C for 15 seconds, 65°C for five seconds, 72°C for five seconds followed by an additional heating to 85°C for melting curve detection.

The conditions for Bcl-xl and Bax differed only in annealing temperatures of 64°C and 70°C, respectively. Bid and Bak PCR reactions were performed at 95°C for 10 seconds, 60°C for 30 seconds and 72°C for one second. Bim PCR reaction was performed at 95°C for 15 seconds, 58°C for 30 seconds and 72°C for one second. Then, a melting curve was created for each reaction, and the product was cooled at 40°C for 30 seconds.

To calculate the amounts of transcripts relative to the housekeeping gene h-HPRT, housekeeping gene PCR was performed using the Light-Cycler-h-HPRT Housekeeping Gene Set (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions. For relative quantification analyses the expression target mRNA in each sample was calculated relative to the housekeeping gene using the LightCycler quantification software (Roche Diagnostics, Mannheim, Germany). An external calibrator was added in duplicates to each run to compensate for inter-run variability. PCR products were cloned into Jm109 plasmids (Promega, Mannheim, Germany) according to the manufacturer's instructions.

For specific genes the following primers were used: HPRT 5'-TGACCTTGATTTATTTTGCATACC, 5'-CGAGCAAGACGTTCAGTCCT (Operon, Cologne, Germany); Bim Refseq No. NM_006538 Band Size: 123 bp Reference Positions: 15–35 (SuperArray, Frederick, MD, USA), Bid Refseq No. NM_001196 Band Size: 176 bp Reference Positions: 331–351 (SuperArray, Frederick, MD, USA), Bak Refseq No. NM_001188, Band Size: 176 bp Reference Positions: 405–424 (SuperArray, Frederick, MD, USA), Bcl-2 5'-GCC AGC TGC ACC TGA CGC CCT TC, 5'-CCG CAT GCT GGG GCC GTA CAG TT (271 bp), Bcl-xl 5'-CAC AGT CAT GCC CGT CAG G, 5'-TGA ATG AAC TCT TCC GGG ATG (281 bp), Bax 5'-ACC CGG TGC CTC AGG ATG CGT, 5'-ACC CGG TGC CTC AGG ATG CGT (185 bp).

Dilution curves of PCR quantifications of cloned PCR products were created for initial calibration of the software. For quantification the crossing point method was employed. Concentrations were calculated as "normalised ratio" with Relative Quantification Software (Roche Diagnostics, Mannheim, Germany).

n o r m a l i s e d . r a t i o = c o n c . t arg ⁡ e t ( s a m p l e ) c o n c . r e f e r e n c e ( s a m p l e ) : c o n c . t arg ⁡ e t ( c a l i b r a t o r ) c o n c . r e f e r e n c e ( c a l i b r a t o r ) MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuqqRPxAKvMB6bYrY9gDLn3AGiuraeXatLxBI9gBaebbnrfifHhDYfgasaacPi6xNi=xI8qiVKIOFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@BB7E@

The mean expression level of each gene was set at the value 100 for healthy controls and relative changes in critically ill and sepsis patients were calculated accordingly.

To detect differences between clinical characteristics of the two patient groups a Mann-Whitney test was employed. To evaluate differences between the three groups, healthy controls, non-septic critically ill patients and patients experiencing severe sepsis, a non-Gaussian-distribution was assumed. Consequently, Kruskal-Wallis testing was performed. For post-hoc testing Dunn's multiple comparison test was employed when appropriate. Correlations were tested using the Spearman procedure and a two-tailed p value. Values were given as mean +/- standard error of the mean if not indicated. p < 0.05 was considered statistically significant.

Sixteen patients with severe sepsis (11 males and 5 females) and 10 critically ill non-septic patients (eight males and two females) were enrolled. The basic clinical characteristics of the critically ill patients and patients with severe sepsis are listed in Table 1 and have been published previously in part 24. Overall, 8 of 16 patients in the sepsis group died during the ICU stay. Patients with severe sepsis and non-critically ill patients did not differ in age but they did differ in Acute Physiology and Chronic Health Evaluation (APACHE) II score, simplified acute physiology score (SAPS) II and Sequential Organ Failure Assessment (SOFA) score.

Furthermore, patients with severe sepsis had higher interleukin (IL) 6 and higher procalcitonin serum concentrations. Briefly, underlying diseases of the sepsis group were necrotising fasciitis (n = 2), faecal peritonitis (n = 8) and pneumonia (n = 6). Nine of the ten critically ill non-septic patients were included in their post-operative period after trauma, abdominal or pharyngeal cancer, or aortic aneurysm rupture with a delayed recovery. These patients received prophylactic antibiotic treatment with no signs of infection in the perioperative phase. One patient had abacterial pancreatitis and did not receive antibiotics. Patients did not receive immunosuppressants or drotrecogin alfa (activated) before or during their treatment. Eight patients with severe sepsis received 3 mg/kg hydrocortisone before or at the time of sampling. No difference in white blood cells counts was found comparing critically ill patients with patients with sepsis. However, lymphocyte counts were decreased in severe sepsis as compared with critically ill patients and dropped below the local reference range of 1–4 G/l (Table 1).

Phosphatidylserine externalisation marks cells for phagocytosis as an early event of the apoptotic process. Cells were considered early apoptotic when phosphatidylserine was externalised on cells with a still intact membrane, as indicated by negative staining for 7-AAD. CD4⁺ and CD8⁺ T-cells and CD19⁺ B-cells exhibited significantly raised portions of phosphatidylserine-positive populations in severe sepsis, but not in critically ill patients (Figure 1a).

Caspase-3 is the central executioner caspase. Activation of caspase-3 leads to degradation of multiple intracellular substrates and to the typical morphological features of classical apoptosis. In the current study, activation of caspase-3 was measured by an antibody specific to the active fragment of cleaved caspase-3 (Figures 1b and 1c). In patients with severe sepsis, the subpopulation with active caspase-3 was elevated in CD4+ T-cells and CD8+ T-cells compared with critically ill patients or healthy controls. Also, B-cells from septic patients were found to contain significantly more activated caspase-3 than B-cells from critically ill patients or healthy controls.

Since the Bcl-2 family of proteins is known to regulate the mitochondrial integrity, we analysed the expression of Bcl-2 (Figure 2). The amount of Bcl-2 in CD4⁺ T-cells of critically ill non-septic patients did not differ significantly from healthy controls. However, in patients with sepsis, Bcl-2 protein levels dropped by about 25%. In CD8⁺ T-cells, no significant change between the three groups could be observed. The decrease in mitochondrial Bcl-2 was most pronounced in B-cells, where Bcl-2 dropped by 36% when compared with healthy controls.

When investigating the mRNA expression of mobile pro-apoptotic BH3-only proteins of the Bcl-2 family, massive induction was observed in severe sepsis (Figure 3a). When compared with healthy controls and critically ill patients, mRNA expression of Bim was upregulated. This corresponds to a 310.5-fold increase compared with critically ill patients and a 51.7-fold rise compared with healthy controls. While Bid was decreased in critically ill patients, it was markedly upregulated in severe sepsis.

Analysing pro-apoptotic members of the Bcl-2 family located in the mitochondrial membrane (Figure 3b), a lower expression of Bak was observed in critically ill patients, while a highly significant elevation of Bak mRNA was found in sepsis patients. Bax mRNA expression was not changed significantly in severe sepsis as compared with control patients.

The expression of pro-survival molecules was reduced in severe sepsis (Figure 3c). The mRNA expression of Bcl-2 was markedly downregulated in patients with severe sepsis as compared with healthy controls. Bcl-2 expression in critically ill patients was found at an intermediate level, which showed no statistical significance. The mRNA of the anti-apoptotic Bcl-xl decreased in severe sepsis as compared with critically ill patients.

When comparing survivors versus non-survivors no differences were observed for Bim (survivors 5548 ± 2510 versus non-survivors 4784 ± 2307, p > 0.5), Bid (538.3 ± 228.2 versus 291.8 ± 97.11, p > 0.5), and Bak (409.1 ± 124.6 versus 273.7 ± 79.17, p > 0.5). Corticoid treatment had no significant effect on gene expression of Bim (corticoid treatment 5493 ± 2417 versus no corticoid treatment 4839 ± 2406, p > 0.5), Bid (267.3 ± 102.2 versus 562.8 ± 283.1, p > 0.5) or Bak (343.9 ± 123.4 versus 338.8 ± 88.79, p = 1).

The degree of organ failure, as expressed by the SOFA score, correlated with pro-apoptotic Bim, Bid and Bax gene expression (Table 2). The severity of illness (SAPS II score) correlated with Bim and Bak gene expression (Table 2). Furthermore, mRNA expression of Bim, Bid and Bak demonstrated a positive correlation with procalcitonin, although Bim and Bid correlated with serum IL-6 (Table 2).