Introduction

Severe sepsis and septic shock are the leading cause of death in intensive care units [1]. Increased endothelial permeability is a hallmark of sepsis leading to plasma fluid loss, extravasation of plasma proteins and subsequent tissue edema formation [2, 3]. Endothelial barrier dysfunction and pro-inflammatory activation of endothelial cells contribute to microcirculatory disturbance resulting in multiorgan failure in sepsis [4]. Maintenance of hemodynamic stability and organ perfusion require fluid resuscitation in the form of crystalloid or colloid fluids at the expense of tissue edema formation that might promote organ dysfunction [5].

The endogenous peptide adrenomedullin (AM) has been shown to be elevated in human inflammatory disease, including sepsis and septic shock [6]. In experimental sepsis in mice overexpressing AM in their vasculature [7] and in AM-treated septic rats [8, 9] this peptide turned out to possess a high therapeutic potential. Besides anti-inflammatory effects, resulting in reduced liberation of pro-inflammatory cytokines in vivo [10], and antiapoptotic effects, as shown on vascular endothelial cells [11], AM enhances cardiac output and maintains cardiovascular stability in AM-treated rats and sheep, thereby contributing to improved animal survival [9, 12]. The mechanism by which AM improves circulatory function is not completely understood. The debate about direct myocardial AM effects is controversial because of inconsistent study results [13, 14, 15]. Furthermore, increased cardiac sympathetic nerve activity due to vasodilating AM effects or even independent of the baroreflex [16] is considered to promote hypercirculation.

In previous studies we demonstrated that AM reduces endothelial hyperpermeability in vitro and in isolated lung and intestine via cAMP-dependent relaxation of the endothelial microfilament system [17, 18]. Considering the high relevance of plasma fluid loss during sepsis, necessitating large volume resuscitation, stabilization of the endothelial barrier by AM might represent an eminent mechanism underlying its therapeutic potential.

We therefore hypothesized that therapeutic in vivo administration of AM in septic shock would improve endothelial barrier function and reduce plasma fluid loss. In addition, we intended to demonstrate that the therapeutic potential of AM, shown in models of endotoxemia [7] and cecal ligation and puncture [8, 9], could be extended to pathogens with completely different modes of action far beyond the signaling pathways of, for example, endotoxin. Pore-forming α-toxin is the major cytotoxin of Gram-positive S. aureus, a pathogen frequently causing sepsis [19]. It is known to provoke endothelial hyperpermeability [20], inflammatory mediator release [21], microcirculatory disturbance [22] and cardiac depression [23] in various models. Therefore we made use of α-toxin-exposed rats to produce vascular leakage and cardiocirculatory failure and to test the therapeutic potential of AM in this model.

Methods

Animal model

Male Sprague–Dawley rats (Harlan–Winkelmann GmbH, Borchen, Germany) weighing 271 ± 4 g were used in all experiments. The study was approved by the local animal care committee (Berlin, Germany; permit no. G0113/04).

After overnight fasting and water ad libitum, rats were anesthetized with a mixture of urethane and ketamine. All animals received mechanical ventilation with 50% oxygen adjusted to maintain arterial pCO2 between 33 and 43 mmHg and continuous infusion of 12 ml/kg per hour isotonic saline via the tail vein as volume supply. Body temperature was kept constant at 37 °C using a computer-controlled infrared lamp.

For continuous monitoring of mean arterial pressure (MAP) and heart rate, withdrawal of blood and injections, the right carotid artery and the right jugular vein were cannulated with polyethylene catheters. Oxygen delivery and uptake were calculated from blood gas analysis using standard equations [24]. Total blood sampling volume throughout the 6-h experimental period was 2.5 ml. Cardiac output was measured by means of the transpulmonary thermodilution technique as described previously [25] using a flexible 18-gauge thermomicroprobe (IT-18; Physitemp Instruments Inc., Clifton, NJ, USA) advanced into the thoracic aorta via the left femoral artery.

Experimental protocol

Rats were allocated to one of four experimental groups in predetermined continuous rotation: (a) administration of isotonic saline (control group, n = 10); (b) administration of 4.8 × 103 U/kg α-toxin (α-tox group, n = 15); (c) administration of 4.8 × 103 U/kg α-toxin and 24 μg/kg per hour AM (α-tox + AM group, n = 15); (d) administration of 24 μg/kg per hour AM (AM group, n = 6). α-Toxin was infused by a syringe pump via the central venous catheter over 45 min. AM was infused for 3 h at 24 μg/kg per hour via tail vein starting 1 h after the onset of α-toxin infusion. Equal fluid supply was granted throughout all experiments.

Chemically synthesized rat AM (amino acid 1–50) was obtained from Bachem Bioscience Inc. (Philadelphia, USA), and purified staphylococcal α-toxin was generously provided by Prof. Dr. S. Bhakdi (University of Mainz, Germany). Activity of α-toxin was 143 U/μg protein.

Determination of vascular permeability

At the end of all experiments (after 6 h in surviving animals or as soon as mean arterial blood pressure dropped below 40 mmHg in dying animals) 20 mg/kg Evans Blue Dye (EBD) was slowly infused (2 min) via the central venous catheter to assess vascular endothelial permeability [26]. Fifteen minutes after EBD-administration animals were killed and intravascular EBD was washed out of the organs. EBD extraction and determination of organ EBD content were performed as previously described [27]. (For more detail see electronic supplementary material, ESM.) The fluorescence properties of EBD in the near-infrared region (700 nm) was used to visualize disposal of EBD–albumin in organs by means of an Odyssey® infrared laser scanner (LI-COR, Lincoln, USA). In addition, intravascular fluid loss was assessed by depicting the course of hematocrit [28].

Statistics

All results are expressed as mean  ±  SEM. Normal distribution of variables was confirmed by Kolmogorov–Smirnov test. Differences between groups were analyzed by one-way analysis of variance followed by Scheffe's multiple comparison test. Comparisons within groups were performed with a paired-sample t-test. The survival rate was estimated by Kaplan–Meier method and compared by the log-rank test. P values < 0.05 were considered as significant. For statistical analysis the software SPSS 13.0 was used (SPSS Inc., USA).

Results

AM treatment reduces α-toxin induced mortality

All fully instrumented rats of the control group (10 of 10) and the AM group (6 of 6) survived the complete experimental period (Fig. 1). Challenge with 4.8 × 103 U/kg α-toxin for 45 min induced mortality of more than 50% of the animals (8 of 15) within 6 h. Therapeutic administration of 24 μg/kg per hour AM reduced mortality very significantly (7% vs. 53%), with 14 of 15 animals surviving the experimental period (Fig. 1).

Fig. 1
figure 1

Adrenomedullin (AM) treatment improves the survival rate of rats after central venous administration of staphylococcal α-toxin. A 3-h AM infusion was started 1 h after onset of α-toxin administration. Control (solid line, solvent only, n = 10); α-tox (dotted line, 4.8 × 103 U/kg α-toxin, n = 15); α-tox + AM (dashed and dotted line, 4.8 × 103 U/kg α-toxin + 24 μg/kg per hour AM; n = 15); AM (dashed line, 24 μg/kg per hour AM, n = 6). Data presented are mean  ±  SEM. #p < 0.05 vs. α-tox group

Full size image

AM abolishes hemodynamic impairment elicited by S. aureus α-toxin

While control animals were hemodynamically stable throughout the experimental period, the infusion of α-toxin provoked a transient pressure elevation and a subsequent continuous decrease of mean arterial pressure (MAP) from 116 ± 2 to 62 ± 4 mmHg within 6 h (Fig. 2a). The reduction in MAP could mainly be attributed to the α-toxin driven decline of cardiac index by nearly 50% (Fig. 2b). The decrease in cardiac index was due to an initial drop of heart rate by 20% and a progressive reduction of stroke volume index by 40% (Fig. S1, ESM). The consecutive drop of oxygen delivery was counterbalanced by a significant increase of oxygen extraction, while central venous oxygen saturation dropped significantly (Table S1, ESM). AM treatment significantly improved cardiac index (Fig. 2b) by a combined increase of heart rate and stroke volume (Fig. S1, ESM). This hemodynamic improvement resulted in a stabilization of MAP (Fig. 2a), oxygen delivery and central venous oxygen saturation (Table S1, ESM) compared to α-toxin animals without therapy. Effects of AM on hemodynamics per se consisted of a significant increase in cardiac index (Fig. 2b) and a concomitant reduction of MAP (Fig. 2a), indicating a decrease in systemic vascular resistance.

Fig. 2
figure 2

Adrenomedullin (AM) abolishes hemodynamic impairment of rats following central venous administration of staphylococcal α-toxin. a MAP; b cardiac index. A 3-h AM infusion was started 1 h after onset of α-toxin administration at T = 0. control (○, solvent only, n = 10); α-tox (Δ, 4.8 × 103 U/kg α-toxin, n = 15); α-tox + AM (▲, 4.8 × 103 U/kg α-toxin + 24 μg/kg per hour AM, n = 15); AM (●, 24 μg/kg per hour AM, n = 6). Data presented are mean  ±  SEM. *p < 0.05 vs. control group; #p < 0.05 vs. α-tox group

Full size image

AM reduces vascular hyperpermeability in vivo

Infusion of α-toxin resulted in a significantly increased EBD–albumin load of lung, liver, ileum, kidney and spleen (Fig. 3). Therapeutic administration of AM significantly reduced α-toxin induced EBD extravasation nearly to control levels. After AM alone, lung, liver and kidney had lower dye content than in control animals. The relationship between hyperpermeability and cardiac output was analyzed by depicting final cardiac index over an averaged value of organ EBD content (Fig. S2, ESM). For α-toxin-exposed animals a linear regression was performed and a significant negative correlation (r 2 = –0.90; p = 0.000004) was found, demonstrating an association of hemodynamic impairment with vascular leakage.

Fig. 3
figure 3

Adrenomedullin (AM) reduces extravasation of Evans blue dye–albumin complexes into lung, liver, ileum, kidney and spleen of anesthetized rats after central venous administration of staphylococcal α-toxin. Control (white bars, solvent only, n = 10); α-tox (black bars, 4.8 × 103 U/kg α-toxin, n = 15); α-tox + AM (grey bars, 4.8 × 103 U/kg α-toxin + 24 μg/kg per hour AM, n = 15); AM (pale grey bars, 24 μg/kg per hour AM, n = 6). Data presented are mean  ±  SEM. *p < 0.05 vs. control group; #p < 0.05 vs. α-tox group

Full size image

Scanning by an infrared laser displayed a slight and homogeneous disposal of EBD–albumin. α-Toxin exposure induced strong EBD–albumin disposal in lung, liver and ileum (Fig. 4). In the kidney, EBD–albumin was redistributed from the medulla towards the cortex (Fig. 4; Fig. S3, ESM). Therapeutic administration of AM diminished α-toxin effects on EBD–albumin extravasations in all investigated organs (Fig. 4) and normalized distribution within the kidney (Fig S3, ESM).

Fig. 4
figure 4

Increased Evans blue dye–albumin extravasation into lung, liver, ileum and kidney after exposure of rats to staphylococcal α-toxin was reduced by adrenomedullin (AM). After AM treatment the pattern of dye distribution in organs was similar to controls. Evans blue dye was visualized by an infrared laser scanner and scanned intensities were depicted in greyscale. Control (solvent only, n = 10); α-tox (4.8 × 103 U/kg α-toxin, n = 15); α-tox + AM (4.8 × 103 U/kg α-toxin + 24 μg/kg per hour AM, n = 15); AM (24 μg/kg per hour AM, n = 6)

Full size image

Withdrawal of blood samples and the continuous volume supply led to a decrease of hematocrit (52 ± 1% vs. 44 ± 1%) in controls (Fig. 5). Under equal fluid resuscitation and blood sampling, hematocrit was significantly higher in α-toxin-exposed animals than in controls (Fig. 5), indicating relative hemoconcentration due to plasma fluid loss into tissues. Subsequent AM treatment abolished this effect. Moreover, AM administration alone resulted in hemodilution, shown as a decrease of hematocrit (Fig. 5).

Fig. 5
figure 5

Adrenomedullin (AM) reversed the α-toxin induced hematocrit elevation above control values, indicating a reduction of plasma fluid loss in rats. A 3-h AM infusion was started 1 h after onset of α-toxin administration at T = 0. Control (○, solvent only, n = 10); α-tox (Δ, 4.8 × 103 U/kg α-toxin, n = 15); α-tox + AM (▲, 4.8 × 103 U/kg α-toxin + 24 μg/kg per hour AM; n = 15); AM (●, 24 μg/kg per hour AM, n = 6). Data presented are mean  ±  SEM. *p < 0.05 vs. control group; #p < 0.05 vs. α-tox group

Full size image

Discussion

In the presented study infusion of staphylococcal α-toxin induced massive endothelial hyperpermeability resulting in 6-h mortality of over 50%. Cardiac output and MAP dropped due to decreased heart rate and stroke volume as a consequence of intravascular fluid loss. Therapeutic administration of AM significantly improved survival after α-toxin exposure, and stabilized the endothelial barrier as well as circulatory function.

Effect of Staphylococcus aureus α-toxin on fully instrumented and monitored rats

Clinically, S. aureus is an important sepsis-causing pathogen, and isolates expressing the α-toxin-encoding gene are frequently found in staphylococcal septicemia [29, 30, 31, 32]. The presence of the α-toxin-encoding hla gene in S. aureus was considered as a lethal factor in murine sepsis models [33, 34].

Continuous infusion of α-toxin was performed in order to mimic α-toxin-related effects of circulating staphylococci or systemically distributed α-toxin from local sites of infection. Since no data are yet available concerning α-toxin concentration in tissues or blood, we empirically chose a dose of 4.8 × 103 U/kg in order to achieve a 6-h mortality of approximately 50%. After α-toxin, deterioration of cardiac index was followed by a progressive decline in MAP and in oxygen delivery.

α-Toxin provoked significantly enhanced extravasation of EBD-labeled albumin in vivo in all investigated organs, such as lung, liver, ileum and kidney, demonstrating loss of endothelial barrier function. In accordance, Blomqvist et al. noted strong binding of 3H-labeled α-toxin in lung, liver and kidney after venous α-toxin injection in mice [35]. Redistribution of EBD–albumin from renal medulla to cortex resembled findings in acute renal failure [36]. Considering that 3H-labeled α-toxin injected in mice bound strongly in the area of proximal renal tubuli [35] the toxin might promote renal dysfunction by a direct impact on renal vasculature.

Impairment of cardiac function by high doses of α-toxin has been observed in ex vivo isolated rat hearts [23]. In contrast, Blomqvist et al. noted no binding of radiolabeled α-toxin in the heart of mice [35]. Even though we cannot exclude direct cardiac effects in our model, we suggest from our data that intravascular volume loss following general vascular hyperpermeability indicated by a relative hemoconcentration has contributed greatly to the α-toxin-induced deterioration of cardiac index. Overall, since α-toxin expression has to be considered as a lethal factor in murine sepsis [33, 34] these data suggest that S. aureus α-toxin-related endothelial hyperpermeability as described in models in vitro [18, 37] and in ex vivo [20] may contribute to cardiocirculatory breakdown and septic shock in systemic S. aureus infections in vivo.

Therapeutic intervention with AM

Beneficial effects of AM have been demonstrated in endotoxic mice and rats undergoing cecal ligation und puncture [7, 8]. We intended to prove that the therapeutic benefit of AM can be transferred to septic shock induced by factors of Gram-positive bacteria and made use of S. aureus α-toxin as a stimulus [38]. In septic shock, plasma levels of AM in humans rise up to 40-fold (from 33 ng/l in healthy volunteers to 1.2 μg/l) [39]. The protective effects in septic animals were observed at continuous infusion of 6 μg/kg per hour or a bolus of 12 μg/kg, which exceeded the physiological plasma levels by far. In the face of the rapid and strong responses to α-toxin [18, 20, 23] we decided to test a therapeutic intervention with 24 μg/kg per hour AM started 1 h after onset of S. aureus α-toxin infusion.

AM significantly improved the survival of α-toxin-exposed rats. Although AM is known to have vasodilating effects via increased NO liberation by the endothelium and direct effects on vascular musculature mediated by increased intracellular cAMP [40, 41], our and several other studies showed improved survival in animals with sepsis and septic shock and an overall stabilization of hemodynamics [7, 9]. Treatment of α-toxin-exposed rats with AM induced a significant increase in cardiac index due to combined augmentation of heart rate and stroke volume. Since linear regression analysis demonstrated an association of hyperpermeability with circulatory failure in our model, AM-related reduction of endothelial hyperpermeability might contribute to the stabilization of cardiocirculation. Enhanced global hemodynamics as well as an AM-related improvement of the microcirculation [22] may have contributed to a normalization of renal hemodynamics, which was indicated by a restoration of the normal dye distribution within kidney [36]. Furthermore, AM reduced α-toxin-related endothelial hyperpermeability in all investigated organs. An overall reduction of vascular permeability was underlined by the AM-related reduction of α-toxin-induced relative hemoconcentration. Interestingly, AM by itself reduced endothelial permeability, resulting in an decreased extravasation of EBD–albumin compared to controls. Hemodilution from blood sampling and fluid administration appeared in all groups but was most pronounced in the AM group. We speculate that AM inhibited fluid extravasations into tissue that occurred in controls under the conditions of anesthesia, mechanical ventilation and surgical trauma.

In previous studies we demonstrated that AM stabilized endothelial barrier function via a cAMP-dependent pathway in human endothelial cells [17] and in isolated organs [17, 18]. Taken in account that mice with disrupted AM genes have displayed an extreme hydrops fetalis and cardiovascular anomalies, AM may act as a central endothelial barrier stabilizing endogenous peptide in health as well as in sepsis [42]. Septic animals receiving AM had a significantly higher mean central venous oxygen saturation, a lower lactate concentration, a lower base deficit, and a higher pH than the animals assigned to α-toxin only. These findings are in accordance with the findings of the study by Rivers et al. [43], where treatment of severe sepsis and septic shock by goal-directed therapy with intravenous fluids and vasoactive drugs provided significant benefits and improved outcome. However, it should be noted that our study was performed in an animal model with limited duration of observation (6 h) and results may differ from human sepsis. A further limitation is the lack of variables indicating systemic inflammation and oxidative stress.

Conclusions

In this study we demonstrated that therapeutic administration of AM reduces endothelial permeability in vivo. Stabilization of the endothelial barrier occurred with or without prior inflammatory stimulation with staphylococcal α-toxin and was accompanied with an enhancement of hemodynamic parameters. Considering the high importance of an adequate volume supply in clinical sepsis, we conclude that a reduction of plasma fluid loss by AM would significantly contribute to the maintenance of hemodynamic stability and prevent the transition from the hyperdynamic to the fatal hypodynamic septic state. Overall, we believe that further analysis of AM-related stabilization of endothelial barrier function may pave the way to an intervention to stabilize endothelial barrier integrity directly in sepsis and septic shock.