Levels of circulating angiopoietin-2 (Ang-2) increase in sepsis, raising the possibility that Ang-2 acts as a modulator in the sepsis cascade. To investigate this, experimental sepsis was induced in male C57BL6 mice by a multidrug-resistant isolate of Pseudomonas aeruginosa; survival was determined along with neutrophil tissue infiltration and release of proinflammatory cytokines. Survival was significantly increased either by pretreatment with recombinant Ang-2 2 h before or treatment with recombinant Ang-2 30 min after bacterial challenge. Likewise, Ang-2 pretreatment protected against sepsis-related death elicited by Escherichia coli; however, Ang-2 failed to provide protection in lipopolysaccharide (LPS)-challenged mice. The survival advantage of Ang-2 in response to P. aeruginosa challenge was lost in tumor necrosis factor (TNF)-deficient mice or neutropenic mice. Infiltration of the liver by neutrophils was elevated in the Ang-2 group compared with saline-treated animals. Serum TNF-α levels were reduced by Ang-2, whereas those of interleukin (IL)-6 and IL-10 remained unchanged. This was accompanied by lower release of TNF-α by stimulated splenocytes. When applied to U937 cells in vitro, heat-killed P. aeruginosa induced the secretion of IL-6 and TNF-α; low levels of exogenous TNF-α synergized with P. aeruginosa. This synergistic effect was abolished after the addition of Ang-2. These results put in evidence a striking protective role of Ang-2 in experimental sepsis evoked by a multidrug-resistant isolate of P. aeruginosa attributed to modulation of TNF-α production and changes in neutrophil migration. The protective role of Ang-2 is shown when whole microorganisms are used and not LPS, suggesting complex interactions with the host immune response.
Severe sepsis and septic shock are the third-leading causes of mortality. They affect more than 1.5 million people annually in North America and another 1.5 million people annually in Europe, with mortality ranging between 35 and 50% (Annane et al., 2003). The great mortality of severe sepsis/shock has intensified research targeting the underlying key mechanisms of pathogenesis.
The predominant theories of pathogenesis support that sepsis develops when well conserved structures of the bacterial cells known as pathogen-associated molecular patterns (PAMPs) stimulate cells of the innate immune system like circulating monocytes and tissue macrophages for the biosynthesis and release of proinflammatory cytokines such as tumor necrosis factor-α (TNF-α), interleukin (IL)-6, and interferon-γ (IFN-γ) and anti-inflammatory cytokines such as IL-10. Both proinflammatory and anti-inflammatory cytokines orchestrate septic phenomena (Giamarellos-Bourboulis and Raftogiannis, 2012).
However, over the last decade many factors/mediators released by the vascular endothelium have been recognized as of paramount importance in the pathogenesis of sepsis. Angiopoietins (Angs) were discovered in 1996 as growth factors that regulate new blood vessel formation, but it was soon recognized that they also modulate inflammatory processes (Tsigkos et al., 2003). Both Ang-1 and Ang-2 share a common receptor tyrosine kinase, namely Tie2. However, they have also been shown to bind to integrins and activate additional/alternative signaling pathways. Ang-1 is produced by mesenchymal cells, binds to the extracellular matrix, and exists in multimers (Augustin et al., 2009). Binding of Ang-1 to Tie2 leads to receptor dimerization, phosphorylation, and activation. In contrast, responses to Ang-2 have been more variable; Ang-2 has been shown to exert agonist or antagonist behavior, leading many researchers to classify Ang-2 as a partial agonist or context-dependent antagonist (Eklund and Olsen, 2006; Huang et al., 2010). Whether Ang-2 triggers Tie2 phosphorylation and signaling depends on the concentration, vascular bed, and source of Ang-2 used. Functionally, Ang-1 promotes vessel stability, attracts pericytes to growing vessels, inhibits endothelial apoptosis, and reduces vascular permeability (Tsigkos et al., 2003; Augustin et al., 2009). Ang-2 is produced mainly by endothelial cells. It is stored in the Wadel-Palade bodies of endothelial cells and released upon stimulation into the circulation (Fiedler et al., 2006). Ang-2 has been reported to sensitize the endothelium to inflammatory stimuli and increase vascular permeability (Roviezzo et al., 2005; Fiedler et al., 2006).
Evidence suggests Ang-2 is implicated in the pathogenesis of sepsis. Circulating concentrations of Ang-2 are reported to be considerably increased in patients with sepsis, and their levels are related with disease severity (Giuliano et al., 2007; Orfanos et al., 2007; Kranidioti et al., 2009). It is unclear, however, whether this is a circumstantial association or there is a functional consequence of Ang-2 increased levels related to the course of the disease. The aim of the present study was to evaluate the pharmacological effect of Ang-2 administration in a model of lethal sepsis in mice challenged with a multidrug-resistant (MDR) isolate of Pseudomonas aeruginosa. This model has direct implications in translational medicine, because MDR Gram-negative isolates are worldwide and emerging as major pathogens of nosocomial sepsis in intensive-care units (Micek et al., 2011).
Materials and Methods
The studies were carried out in 7- to 9-week-old C57BL/6 mice (body weight 20–25 g; Hellenic Pasteur Institute, Athens, Greece) in the animal facilities of Attikon University Hospital (Athens, Greece). After acclimatization, mice were kept in cages with a constant rotation rate of 70 air changes per hour to ensure sterility in an isolated room with a controlled temperature of 24°C and a 12-h day-night cycle. Mice were fed standard chow (AnaLab, Athens, Greece) and allowed water ad libitum. The study was approved by the Ethics Committee of Attikon University General Hospital and the Veterinary Directorate of the Prefecture of Athens, Greece (license number K/3727/2007).
Ang-2 Pretreatment in Sepsis.
Each experimental series was performed on multiple days; no more than three mice per group were studied each day.
Experimental sepsis was induced by the intravenous challenge of 1 × 106 cfu/mice of the live isolate P. aeruginosa 2. This is a hospital-acquired bloodstream isolate from a male patient with severe sepsis; it was multidrug-resistant to piperacillin/tazobactam, ceftazidime, imipenem, ciprofloxacin, and amikacin as defined according to criteria of the Clinical Laboratory Standards Institute (Wayne, PA). Single colonies were suspended in Mueller-Hinton broth (Oxoid Basingstoke, Hampshire, UK) and incubated overnight at 37°C in a shaking water bath. The inoculum was adjusted to 5 × 107 cfu/ml by using 0.5 of the McFarland climax; 0.2 ml was used for animal challenge.
Animals were randomly assigned to three groups. Group A (sham-treated) mice were administered 0.15 ml of saline (0.9% NaCl) intraperitoneally, and 2 h later they were intravenously challenged with Mueller-Hinton broth. Group B (saline controls) mice were administered 0.15 ml of saline intraperitoneally, and 2 h later they were intravenously challenged with the test isolate. Group C (Ang-2 pretreated) mice were administered 25 μg/kg of human recombinant Ang-2 (R&D Systems Inc., Minneapolis, MN) intraperitoneally diluted into 0.15 ml of saline; 2 h later they were intravenously challenged with the test isolate. Although it was lower, the dose of Ang-2 was selected in analogy with former publications (Fiedler et al., 2006). All injections were performed under light ether anesthesia.
Survival was recorded every 12 h for a total of 7 days. At specific times after bacterial challenge, animals were sacrificed with an intramuscular injection of 25 mg/kg of ketamine. After a midline abdominal incision under sterile conditions, the intestines were displaced to the left. The abdominal cavity was drained with 2 ml of saline. The fluid was then aspirated and used for the measurement of absolute cell counts. After this, the lower vena cava was punctured under aseptic conditions, and 0.3 ml of whole blood was sampled and collected in pyrogen-free tubes or heparin-coated tubes. Segments of 0.5 g of liver, spleen, and lung were cut by disposable sterile blades and collected into separate sterile containers.
Additional survival experiments were performed by using mice deficient for the tumor necrosis factor gene [TNF(−/−); Alexander Fleming Institute, Vari, Athens, Greece], generated as described previously (Pasparakis et al., 1996). TNF(−/−) mice and wild-type control animals were maintained in a random C57BL/6 × 129/Sv (H-2b) genetic background. Both groups originated from C57BL/6 × 129/Sv F2 littermates, homozygous for each mutation, and subsequently kept as individual lines. TNF-deficient mice were backcrossed to C57BL/6 mice for at least six generations [referred to as B6.TNF(−/−)]. Sepsis survival experiments were also performed by using mice rendered neutropenic by two rounds of intraperitoneal injection of cyclophosphamide (Baxter, Athens, Greece): 150 mg/kg 4 days before the bacterial challenge and 100 mg/kg 2 days before the bacterial challenge. Mice were randomly assigned to each of the aforementioned three treatment groups (sham, saline control, and Ang-2). Additional sham-operated mice were sacrificed as described above either before (n = 10), 1 day after (n = 10), or 2 days after (n = 10) bacterial challenge. Neutrophil counts were measured in EDTA-whole blood sampled from the lower vena cava by a counter (Beckman Coulter, Fullerton, CA). In all three cases, neutrophil counts were below 500 neutrophils/mm3, confirming successful induction of neutropenia.
Survival experiments were also performed in C57B6 mice challenged with live P. aeruginosa 2 and pretreated with the intraperitoneal administration of 25 μg/kg of human recombinant Ang-1 (R&D Systems Inc.) in 0.15 ml of saline. Additional mice pretreated with saline alone (control group) were also used in these experiments.
To define the effect of Ang-2 administered as treatment postinduction of sepsis, in an additional series of experiments mice were challenged with P. aeruginosa as described above; 30 min later mice were intravenously administered either 0.15 ml of saline or 25 μg/kg of recombinant human Ang-2.
Finally, survival experiments were conducted with C57BL/6 mice challenged with either 1 × 106 cfu/mice of live Escherichia coli 15941 or 30 mg/kg of lipopolysaccharide (LPS) from E. coli O55:B5 (Sigma, St. Louis, MO). The live isolate was a bloodstream isolate from a female patient with urosepsis; it was susceptible to antimicrobials. Bacterial inoculum was prepared for challenge as described for P. aeruginosa 2.
Quantitative Tissue-Derived Bacterial Cultures, Histopathology, Peritoneal Cell Counting, and Neutrophil Apoptosis.
After sacrifice, tissue segments were weighed and homogenized. One aliquot of 0.1 ml was diluted four consecutive times 1:10 into sterile saline, and 0.1 ml of each dilution was plated onto McConkey agar and incubated at 35°C for a total of 3 days. The number of viable colonies was estimated after multiplying with the appropriate factor of dilution. Identification of colonies was performed by the API20E system (BioMérieux, Paris, France). The number of viable cells was expressed as its log10 value in cfu/g. The lower detection limit was 10 cfu/g.
Liver segments were fixed with formalin and embedded in paraffin, and tissue sections were stained with hematoxylin and eosin. A semiquantitative scoring system was used where infiltration by neutrophils was scored as 0 (absent), 1 (sparse), 2 (moderate), and 3 (intense) corresponding to 0, 1 to 3, 4 to 6, and >6 cells/high-power field, respectively. Scoring of each tissue sample represented the mean score of five different high microscopic power fields. Scoring was done by two expert pathologists, unaware of the treatment group of each animal.
Counting of cells in the peritoneal fluid was performed in an automated cell counter (Beckman Coulter) reporting for both the total number of cells in the peritoneal fluid and subpopulations. To define the rate of apoptosis of circulating neutrophils, red blood cells were lysed after incubation of whole blood with ammonium chloride (1.0 mM) for 10 min. After three consecutive washings, white blood cells were incubated for 15 min in the dark with the protein Annexin-V conjugated with the fluorochrome fluorescein isothiocyanate (emission 525 nm; Immunotech, Marseille, France) and propidium iodide (emission 575 nm; Immunotech). Cells were analyzed with the FC500 counter (Beckman Coulter). Cells staining positive for Annexin-V and negative for propidium iodide after forward scatter/side scatter gating for neutrophils were considered apoptotic.
Determination of Vascular Permeability.
The effect of Ang-2 pretreatment on vascular permeability was assessed by Evans Blue leakage into tissues. More precisely, the abdominal aorta was punctured under aseptic conditions, and 3 ml of Evans Blue (AlterChem Co, Athens, Greece) was injected into the abdominal aorta within 5 min. Ten minutes after the end of the infusion of Evans Blue, repeated intravenous infusions of 0.9% sodium chloride were performed followed by animal sacrifice; the liver and the right lung of the mice were removed, homogenized, and centrifuged. The optical density of the supernatant was visualized at 450 nm.
Cytokine Production by Splenocytes.
In some experiments, the spleen was removed from animals euthanized at various time points postbacterial challenge and transferred into 1 ml of RPMI 1640 (Biochrom, Berlin, Germany). It was then gently squeezed and passed through a sterile filter (250 mm; 12 × 13 cm; AlterChem Co). Isolated splenocytes were collected and counted in a Neubauer chamber, excluding dead cells by trypan blue staining. Splenocytes were incubated into wells of a 24-well plate at a density of 5 × 106/ml in 1 ml of RPMI 1640 (final volume) supplemented with 2 mM glutamine (Biochrom), 10% fetal bovine serum (Biochrom), 100 U/ml of penicillin G, and 0.1 mg/ml of streptomycin (Sigma) at 37°C in 5% CO2 in the absence or presence of stimuli. Incubation lasted for 24 h after stimulation with 10 ng/ml of LPS of E. coli O55:B5 and for 5 days after stimulation with 5 × 105 cfu/ml of heat-killed Candida albicans. At the end of incubation, the plates were centrifuged, and the supernatants were collected and stored at −70°C until they were assayed. Supernatants from LPS-exposed splenocytes were used to determine TNF-α and IL-6; supernatants from splenocytes exposed to C. albicans were used to determine IL-10, IL-17, and IFN-γ.
Cytokine Production by U937 Cells.
Cells of the U937 human monocytic cell line were exposed to 1 × 106 cfu/ml of the heat-killed isolate P. aeruginosa 2, and the supernatant was used to determine TNF-α and IL-6 released by the cells. The isolate was heat-killed after exposure of one log-phase inoculum of 1 × 108 cfu/ml at 75°C for 5 h in a shaking water bath. Lack of growth of the heat-killed bacterium was confirmed after subcultures of serial dilutions onto MacConkey agar. Confluent U937 cells were thoroughly washed with Hanks' solution (Merck, Darmstadt, Germany), resuspended in RMPI 1640 supplemented with 2 mM glutamine and 10% fetal bovine serum, and distributed in a 24-well plate at a density of 1 × 106 cells per well in 1-ml final volume. Incubation with the bacteria was done in the absence or presence of 250 pg/ml of recombinant human TNF-α (Sigma) with/without 100 pg/ml of recombinant human Ang-2 (R&D Systems Inc.). The plate was incubated at 37°C in a 5% CO2 atmosphere. After 24 h the plate was centrifuged for 8 min at 1400g in room temperature, and the supernatant was collected for determination of cytokine concentrations.
Blood collected into pyrogen-free tubes was centrifuged; serum was kept refrigerated at −70°C until being assayed for TNF-α, IL-6, IL-10, and soluble triggering receptor expressed on myeloid cells-1 (sTREM-1). Cytokine concentrations in serum or in splenocyte supernatants were determined by enzyme-linked immunosorbent assay (R&D Systems Inc.) specific for the mice antigens. The lowest limit of detection for all cytokines was 8 pg/ml; for sTREM-1 it was 31.25 pg/ml.
Concentrations of TNF-α and of IL-6 in supernatants of U937 were determined by enzyme-linked immunosorbent assay (R&D Systems Inc.) specific for the human antigens. The lowest limit of detection for both cytokines was 5 pg/ml.
Survival was calculated after Kaplan-Meier analysis; comparisons between groups were done by the log-rank test. Quantitative variables were expressed as means ± S.E. Comparisons between groups were done by the Kruskall-Wallis test. Any p value < 0.05 after adjusting for multiple comparisons was considered significant.
None of the sham-treated animals (i.e., no bacterial load) died. Cumulative mortality of saline-pretreated mice was 77.3% (n = 22), whereas the mortality of the Ang-2-pretreated animals was only 10.7% (n = 28; p < 0.0001 between groups; Fig. 1A). It is noteworthy that, from a therapeutic approach standpoint, a statistically significant survival benefit of Ang-2 was also observed when the protein was administered 30 min after challenge with P. aeruginosa; cumulative mortality in the saline- and Ang-2-treated groups was 80 and 60%, respectively (Fig. 1B). As such, Ang-2 was effective when administered either as pretreatment or treatment. Because the effect of Ang-2 pretreatment was clearly more significant and thus better amenable to further experimental analysis, we chose to conduct the rest of the experimentation using the pretreatment mode.
To find out whether the protection by Ang-2 in this model was specific to this protein or could be mimicked by other related protein family members, a second series of similar experiments was undertaken, where Ang-1 was used instead of Ang-2 in the pretreatment phase. As shown in Fig. 2A, even by increasing the number of animals per group to 20, no statistically significant survival benefit was demonstrated with Ang-1.
We next asked whether the protective role of Ang-2 against sepsis elicited by P. aeruginosa was specific to this organism and indeed whether Ang-2 could be a beneficial pretreatment in sepsis elicited by other types of challenge. For this reason, we performed two additional experimental series, using one live isolate of E. coli and bacterial LPS as challenges. Results showed that pretreatment with Ang-2 offered survival benefit against challenge by live E. coli but not against LPS (Fig. 2, B and C).
From the pattern of cumulative survival rates (Fig. 1), it can be safely determined that the first 24 h after bacterial infection constitutes the critical time window, during which the survival benefit of Ang-2 pretreatment is already manifested and established, i.e., the pivotal pharmacological activity of Ang-2 occurs at a very early time point. For this reason, subsequent investigations focused on changes that are detectable within a few hours from P. aeruginosa injection. To identify the early inflammatory mediators against which the action of Ang-2 is expressed, serum concentrations of TNF-α, IL-6, IL-10, and sTREM-1 in the three modes of treatment (sham, saline control, and Ang-2) were determined at 0.5 to 6 h after P. aeruginosa inoculation (Fig. 3). sTREM-1 is the soluble counterpart of the TREM-1 receptor found on the cell membrane of neutrophils and monocytes. TREM-1 is shed from the cellular membrane so that circulating sTREM-1 is an indirect index of the activation of neutrophils (Dimopoulou et al., 2012). All cytokine concentrations of sham-treated mice (no bacterial load) were below the respective lower limit of detection. In pathogen-challenged mice, the only statistically important difference was observed between the saline and Ang-2 groups related to the ability of Ang-2 to significantly reduce serum concentration of TNF-α and sTREM-1 1 h after bacterial challenge.
In vivo pretreatment with Ang-2 impaired the ability of splenocytes from P. aeruginosa-exposed mice to respond to LPS in vitro by increased TNF-α (Fig. 4). In contrast, production of IL-6, IL-10, IL-17, and IFN-γ was not significantly affected by prior in vivo treatment with Ang-2 (Fig. 4). The findings from the analyses of circulating cytokines and the release of cytokines by leukocytes (splenocytes) of Ang-2-pretreated mice imply that the action of Ang-2 is mediated, at least in part, through down-regulation of TNF-α.
Given that neutrophils are important players in the cascade leading to full-blown sepsis, we analyzed the presence of neutrophils in a major tissue affected by bacterial sepsis, the liver. Histological analysis of liver sections revealed that the kinetics of liver neutrophil infiltration were affected by Ang-2 treatment: Ang-2 significantly reduced by half the neutrophil score at 2 h postbacterial challenge, whereas an Ang-2-dependent increase in liver neutrophil counts was observed at 8 h after bacterial inoculation (Fig. 5). In the same groups of animals and during the same time frame, the neutrophil count in the peritoneal fluid was elevated by Ang-2 pretreatment (statistically significant at 8 h), probably implying that, as a result of Ang-2 pretreatment, increased peritoneal neutrophil numbers gave rise to subsequent migration to other tissues, such as the liver. This is unlikely to result from increased extravasation of neutrophils in response to elevated permeability of the vasculature elicited by Ang-2 pretreatment, because leakage of Evans Blue in tissues was actually lower in Ang-2-pretreated mice (Fig. 5). To provide an alternative explanation for this observation, we assessed the rate of apoptosis of circulating neutrophils at a preceding time point, 6 h after bacterial challenge, and found that Ang-2 pretreatment increased this rate approximately 5-fold (Fig. 5). This may partly explain why, despite the effect of Ang-2 on the accumulation of neutrophils, there was no change in the bacterial load of two major target organs in sepsis, lung and liver within the first 6 h (Fig. 6). However, when bacterial growth was measured in the liver and lung of Ang-2-pretreated mice that survived for 168 h postbacterial challenge the mean ± S.E. bacteria in the liver and lung were 3.11 ± 0.50 and 3.01 ± 0.51 log10 (cfu/g), respectively, significantly lower than the respective growth at 6 h (p < 0.0001), showing that the pathogen load was substantially contained by then. Not enough numbers of saline-treated mice survived until day 7 to allow a similar analysis.
Taken together, the above results indicate that the survival benefit that results from Ang-2 pretreatment in lethal P. aeruginosa-induced sepsis could be mediated through 1) modulation of TNF-α production and 2) changes in neutrophil organ migration. To test these hypotheses, P. aeruginosa-induced sepsis survival experiments were, respectively, performed in 1) TNF(−/−) mice of similar genetic background (back-crossed to C57B6) and 2) neutropenic mice (<500/mm3). In both of these modified physiological backgrounds, the protective effect of Ang-2 was lost (Fig. 7), indicating that TNF-α and neutrophils play a role in the Ang-2 effect, and both are disease-modulating factors that are targeted by Ang-2. To test whether the inability of Ang-2 to protect TNF-deficient mice was caused by inadequate dosing, we conducted an additional series of experiments in TNF-deficient mice by using a double amount of Ang-2 (50 μg/kg) but could not detect a statistically significant survival benefit (p = 0.168; data not shown).
The importance of TNF-α release in the effect of Ang-2 was further investigated in vitro by stimulating human U937 monocytes with the MDR heat-killed P. aeruginosa isolate 2. When a low priming-dose of TNF-α was added during the bacterial exposure (250 pg/ml), the concentration of TNF-α and of IL-6 in the supernatant increased by almost 10-fold (note that the endogenous, released TNF-α concentration in the supernatant is >10-fold higher than that of the exogenously added factor; Fig. 8). This effect was largely attenuated by the concomitant presence of Ang-2 in the well and was visible only when TNF-α was present; in contrast, Ang-2 had no effect on cytokine production in response to P. aeruginosa alone (Fig. 8). It seems that Ang2 can effectively contain the TNF-α-induced production of IL-6 as well as the positive-feedback loop of TNF-α auto-induction and therefore can interfere with relatively early steps of the proinflammatory cascade.
Accumulating evidence from several clinical studies performed over the last few years clearly indicates that circulating Ang-2 is greatly elevated in the event of sepsis (Giuliano et al., 2007; Orfanos et al., 2007; Kümpers et al., 2008, 2009). Serum Ang-2 increases in parallel with disease severity, and circulating levels are positively linked with the expansion of interstitial edema and the development of adult respiratory distress syndrome (van der Heijden et al., 2009). This has been explained by the increase of vascular permeability caused by Ang-2, leading to increased pulmonary leakage (Roviezzo et al., 2005; Parikh et al., 2006). The present study investigated the effect of pharmacological administration of Ang-2 and challenges the concept that increased levels of this growth factor are detrimental. Our results indicate that pretreatment of mice with Ang-2 protected them from lethal sepsis by MDR P. aeruginosa. These results are challenging the traditional concept that Ang-2 is a harmful mediator of disruption of the vascular integrity. On the contrary, they indicate an anti-inflammatory effect of Ang-2.
Existing data suggest that Ang-2 is released from the vascular endothelium. Ang-2 is stored in Wadel-Palade bodies and secreted upon response to bacterial stimuli. LPS is the most widely recognized bacterial PAMP that stimulates the production of Ang-2 from the vascular endothelium. As a consequence, increased tissue production has been documented in experimental mice and human volunteers (Kümpers et al., 2008; Mofarrahi et al., 2008).
However, evidence from our group suggests that endothelial cells are not the only site of production of Ang-2 and in the event of sepsis circulating monocytes are able to secrete Ang-2 (Kranidioti et al., 2009). More precisely, it was described that a circulating factor existed in patients with septic shock; this factor could stimulate in vitro Ang-2 production by healthy monocytes. Secretion of Ang-2 by monocytes was modulated by LPS so that stimulation of Toll-like receptor-4 by LPS inhibited release of Ang-2, whereas inhibition of Toll-like receptor-4 stimulated release of Ang-2. Furthermore, addition of one TNF-α blocking antibody hampered the ability of monocytes to produce Ang-2. These results corroborate the present findings and suggest that Ang-2 may be an endogenous anti-inflammatory mediator; excess circulating levels in critically ill patients may simply indicate the need of the host to counterbalance for hypersecretion of proinflammatory mediators. However, persistent overproduction of Ang-2 may have a deleterious effect by disrupting the function of vascular endothelium.
Our present data show that a single treatment with recombinant Ang-2 offers a striking survival benefit, whether the polypeptide is administered 2 h before bacterial challenge or 30 min after bacterial challenge. As is the case with most, if not all, examined molecules in sepsis models, the benefit is more pronounced in the pretreatment experimental setting. It should be noted that Ang-2 is a short-lived molecule. This fact has two implications: first, a more “aggressive” treatment mode with multiple administrations of Ang-2 or use of a more stable analog should be tested in the future to see whether a more pronounced survival effect is obtained; and second, it is likely that after bacterial challenge there is an excess of pathogen- or immune-derived proteases, resulting in an accelerated degradation of Ang-2 delivered after the pathogen, which could at least in part explain the diminished effectiveness of Ang-2 in this mode of treatment. The related cytokine Ang-1 failed to elicit a similar protective effect. It has been reported (Witzenbichler et al., 2005) that Ang-1 administered by an adenovirus protected mice from LPS-induced endotoxic shock, improving cardiopulmonary function and survival. More recently, it was shown that the administration of Ang-1 reduced lethality in mice subjected to polymicrobial sepsis after cecal ligation and puncture (David et al., 2011). In the present study, although Ang-1 somewhat improved survival of mice inoculated with P. aeruginosa, its beneficial effect failed to reach statistical significance. The discrepancy between the results observed herein and those in the literature could be explained by the different septic challenges used or the different routes of administration of the growth factor used.
In the studied model of lethal sepsis, Ang-2 required the following to prevent death: 1) challenge by a bacterial PAMP other than LPS, 2) a host capable of TNF-α production, and 3) the presence of neutrophils. Our results pointed to TNF-α as the main mediator modulated by Ang-2. TNF-α was the only cytokine, among all tested, to be decreased in serum and down-regulated after stimulation of isolated splenocytes. As a consequence, the protective effect of Ang-2 was lost in TNF(−/−) mice. The ability of Ang-2 to reduce TNF-α production was further corroborated by in vitro studies showing that release of TNF-α and of IL-6 by U937 monocytes was down-regulated by Ang-2. Ang-2 can therefore mitigate proinflammatory cytokine release perhaps by interfering with a positive feedback loop initiated by TNF-α.
TNF-α is a central molecule in the pathogenesis of Gram-negative sepsis. As such, inhibition of TNF-α has provided prolonged survival in a variety of animal models of sepsis induced after challenge with LPS or Gram-negative bacteria such as E. coli (Tracey et al., 1987; Giamarellos-Bourboulis et al., 2003). The biological significance of TNF-α as the acute proinflammatory cytokine leading to early death has been shown in experiments using TNF-deficient mice. In these experiments, mice deficient for the TNF gene were more tolerant to fecal peritonitis or lethal endotoxemia compared with wild-type mice (Dharmana et al., 2002; Secher et al., 2009). Given the crucial role of TNF-α in Gram-negative sepsis, it can be explained, at least in part, why down-regulation of TNF-α production by Ang-2 confers significant survival benefit.
Another serum inflammatory index that was down-regulated by Ang-2 was sTREM-1. This finding is compatible with overall lower leukocyte activation in the presence of Ang-2 or altered leukocyte distribution between peritoneum, blood, and tissues (Routsi et al., 2005).
From our data on neutropenic mice, it is clear that part of the effect of Ang-2 is mediated through multiple effects on circulating neutrophils, including distribution, activation, and viability. Histology results suggest that by 8 h after pathogen inoculation Ang-2 had primed the migration of neutrophils from the peritoneal cavity to the liver. Although neutrophils act to contain infection in tissues through the process of phagocytosis, bacterial load in the liver and lung was not fully suppressed, perhaps because Ang-2 in parallel induced significant neutrophil apoptosis. The increased presence of neutrophils in the liver in response to Ang-2 treatment at 8 h after bacterial challenge is unlikely to be caused by overall increased vascular permeability by Ang-2. Evans Blue tissue content, a classic index of vascular permeability, was actually decreased in animals treated with Ang-2, suggesting an alternative explanation is needed, such as increased neutrophil mobility/migration (Sturn et al., 2005; Brkovic et al., 2007). Furthermore, it is obvious that the effect of Ang-2 on permeability is context-dependent. Ang-2 has been described as a molecule that increases vascular permeability, whereas in the present context, in the presence of bacterial challenge, it seems to decrease leakage of macromolecules into tissue. This unexpected effect of Ang-2 is paralleled by its contextual effect on angiogenesis, where Ang-2 can both inhibit angiogenesis via inhibition of Tie2 signaling or act as an angiogenesis promoter in a Tie2-independent manner in situations where Tie2 expression on endothelial cells is diminished (Felcht et al., 2012).
Neutrophils play a pivotal role in the innate immune response because they manage to contain the offending microorganism through the process of phagocytosis. The overall orchestration of the immune response after a severe infection is directed toward activation of neutrophils for migration at the infection site (Giamarellos-Bourboulis and Raftogiannis, 2012). However, neutrophil migration is impaired in individuals with sepsis (Reddy and Standiford, 2010). Neutrophils bear Tie2 receptors on their cell membranes and are responsive to both Ang-1 and Ang-2; both are reported to modulate migration of neutrophils (Sturn et al., 2005; Brkovic et al., 2007), which may partly explain the findings of the present study. In septic mice pretreated with Ang-2, the presence of neutrophils in the liver persisted for 8 h postbacterial challenge to levels similar to those found within the first 2 h after bacterial inoculation; in contrast in mice pretreated with saline, liver infiltration by neutrophils decreased at 8 h. This correlated timewise to the increase of neutrophils in the peritoneum and showed that Ang-2 treatment can result in the maintenance of stable, elevated counts of neutrophils in the liver. Failure of neutrophil recruitment in organs is described as a driver to earlier death (Alves-Filho et al., 2010). In agreement with these previous reports, neutropenic animals displayed increased death rate in response to challenge with P. aeruginosa and by the same token lost the ability to benefit from Ang-2 treatment. However, inappropriate accumulation of neutrophils in remote organs can cause tissue damage and compromise survival (Souto et al., 2011). It is obvious that a right balance is required and the sustained presence of neutrophils in the liver after pretreatment with Ang-2, achieved in the present study, seems to have contributed to survival benefit.
The inability of Ang-2 to protect against E. coli LPS lethality contrasts with the observed protective effect against the whole organism. This, although surprising, has two possible explanations: 1) Ang-2 protects against a PAMP or a combination of PAMPs other than LPS, and 2) immune responses differ between substantially purified LPS and the whole microorganism (Christiansen et al., 2012).
The presented results provide an entirely new concept for the role of Ang-2 in sepsis; Ang-2 participates in the anti-inflammatory response of the host. Its action is mediated on the monocyte level through selective modulation of the release of TNF-α and on the neutrophil level through the priming of tissue migration and infiltration, accompanied by an increase of blood neutrophil apoptosis. These results increase our understanding of the complexity of the pathogenesis of sepsis and argue that Ang-2 orchestrates a multilayered effect on immune cell distribution and activation. Moreover, the protective effect of Ang-2 when administered after bacterial challenge creates promising novel perspectives for immunointervention.
Participated in research design: Giamarellos-Bourboulis and Papapetropoulos.
Conducted experiments: Tzepi, Carrer, Tsaganos, Claus, Vaki, Pelekanou, Kotsaki, Tziortzioti, Topouzis, and Bauer.
Performed data analysis: Giamarellos-Bourboulis.
Wrote or contributed to the writing of the manuscript: Tzepi, Giamarellos-Bourboulis, Bauer, and Papapetropoulos.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- pathogen-associated molecular pattern
- colony-forming units
- normal saline
- Pseudomonas aeruginosa
- triggering receptor expressed on myeloid cells-1
- soluble TREM-1
- tumor necrosis factor.
- Received April 6, 2012.
- Accepted July 31, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics