Lung Vascular Injury after Administration of Viable Hemolysin-forming Escherichia coli in Isolated Rabbit Lungs 1 - 3
WERNER SEEGER, RUDOLF OBERNITZ, MICHAEL THOMAS, DIETER WALMRATH, N. SUTTORN, IAN B. HOLLAND, FRIEDRICH GRIMMINGER, BETTINA EBERSPACHER, FERDINAND HUGO, and SUCHARIT BHAKDI
Introduction
Thirty to fifty percent of Escherichia coli isolates causing extraintestinal infections in humans elaborate a proteinaceous cytolysin that damages target cell membranes through the generation of circumscribed, transmembrane pores (1-3). Structurally related toxins are also produced by other Enterobacteriaceae including Morganella, Proteus, and Pasteurella ssp. The relevance of E. coli hemolysin as a determinant of bacterial pathogenicity has been established in animal models with the use of genetically engineered, isogenic bacterial strains (4, 5). An analogous role as a virulence factor in human infections has been inferred from the high association of hemolysin production with disease, including pyelonephritis and septicemia (6-8). Bacterial sepsis is the most consistent factor associated with the development of acute respiratory failure in adults (ARDS), with gram-negative rods including E. coli currently representing the predominant infectious agents (9, 10). Moreover, acute lung injury of different etiology is often complicated by nosocomial pneumonia caused by Enterobacteriaceae (11). Against this background it was of interest that direct intravascular application of very low doses of E. coli hemolysin induced key events of acute respiratory failure in blood-free perfused rabbit lungs (12). In particular, development of acute thromboxane-mediated pulmonary hypertension as well as protracted vascular leakage was noted. In the present study, these pathophysiologic events were dosedependently reproduced by infusion of live hemolysin-forming E. coli. In contrast, an E. coli strain that secretes an inactive form of hemolysin was without effect. These findings support a possible role of pore-forming bacterial exotoxins in the pathogenesis of acute lung injury
SUMMARY Escherichia coli hemolysin, a transmembrane pore-forming exotoxin, is considered an important virulence factor. In the present study, the possible significance of hemolysin production was Investigated In a model of septic lung failure through infusion of viable bacteria in Isolated rabbit lungs; 104 to 107 E.coli/ml perfusate caused a dose- and tlme-dependent appearance of hemolysin, accompanied by release of potassium, thromboxane A 2 , and PGl 2 into the perfusate. Concomitantly, marked pulmonary hypertension developed. Inhibitor studies suggested that the pressor response was predominantly medlate~ by pulmonary thromboxane generation. Administration of hemolysin-forming E. coli additionally caused a protracted, dose-dependent Increase In the lung capillary filtration coefficient, followed by severe edema formation. The permeability increase was independent of lung prostanoid generation. An E. coli strain that releases an Inactive form of hemolysin completely failed to provoke the described biophysical and biochemical responses. Preappllcation of 2 x 108 human granulocytes was without effect in the present experimental model. Weconclude that the hemolysin produced by low numbers of E. coli organisms can provoke thromboxanemediated pulmonary hypertension and severe vascular leakage. E. coli hemolysin and, possibly, other related cytolyslns may thus contribute directly to the pathogenesis of acute respiratory failure AM REV RESPIR DIS 1991; 143:797-805 under conditions of sepsis or pneumonia.
under conditions of sepsis and severe pneumonia. Methods Reagents BM 13.177 was a gift from Boehringer Mannheim GmbH (Mannheim, Germany). D,LLysin-mono-acetylsalicylate/glycin (9:1; ASA) was obtained from Bayer AG (Leverkusen, Germany). Bovine albumin (98lt7o purity, reduced in free fatty acids to < 5 ug/g), rabbit anti-6-keto-PGF ta , and rabbit anti-Ixls, were obtained from Paesel AG (Frankfurt, Germany). Tritium-labeled lXB 2 and 6-ketoPGF HI were from New England Nuclear (Dreieich, Germany), and E. coli endotoxin (serotype 0111:B4) was purchased from Sigman Chemical (Munich, Germany). All other biochemicals were obtained from Merck AG (Darmstadt, Germany) in p.a. quality. Isolated Lung Model The model has been previously described (12-15). Briefly, rabbits of either sex (body weight, 2.2 to 2.6 kg) were deeply anesthetized and anticoagulated with 1,000U of heparin per kg body weight. The lungs were excised while being perfused with KrebsHenseleit-albumin buffer (KHAB) through cannulas in the pulmonary artery and the left
atrium. The buffer contained 132.8mM NaCI, 4.3 mM KCI, 1.1 mM KH 2P04 , 24.1 mM NaHC03 , 2.4 mM CaCI 2 , and 1.3 mM MgP0 4 , as well as 240 mg glucose and 1 g albumin per 100 ml. The lungs were placed in a temperature-equilibrated housing chamber at 37° C, freely suspended from a force transducer. They were ventilated with 4lt7o CO 2 , 17lt7o O 2 , and 79lt7o N 2 , and the pH of the perfusion fluid ranged between 7.35 and 7.45. After extensive rinsing of the vascular bed, the lungs were recirculatingly perfused with a pulsatile flow of 100 ml/min. The alternate use of two separate perfusion circuits,
(Received in original form February 20, 1990 and in revised form September 7, 1990) 1 From the Department of Internal Medicine, Division of Clinical Pathophysiology and Experimental Medicine, the Institute of Medical Microbiology, Justus-Liebig University, Giessen, Germany, and the Department of Microbiology and Genetics, University of Leicester, Leicester, United Kingdom. 2 Supported by the Deutsche Forschungsgemeinschaft (SFB 249, Teilprojekte A6 and A7). 3 Correspondence and requests for reprints should be addressed to Werner Seeger, Department of Internal Medicine, Justus-Liebig University Giessen, Klinikstrasse 36, D-63 Giessen, Germany.
797
798
SEEGER, OBERNITZ, THOMAS, WALMRATH, SUTTORN, HOLLAND, GRIMMINGER, EBERSPACHER, HUGO, AND BHAKDI
each containing 200 ml, allowed exchange of perfusion fluid. Perfusion pressure, ventilation pressure, and the weight of the isolated organ were registered continuously. The left atrial pressure was set 2 mm Hg under baseline conditions (zero referenced at the hilus) to guarantee zone III conditions at endexpiration throughout the lung. The capillary filtration coefficient (Kfc) and the total vascular compliance were determined gravimetrically from the slope of weight gain, induced by a 7.5-mm Hg step elevation of the venous pressure for 8 min. The application of this method to the present model and the use of zero time extrapolation of the slope of weight gain for the calculation of Kfc have been described (14).In lungs, which already revealed a constant rate of weight gain before onset of the hydrostatic challenge, this slope of weight increase was subtracted from the hydrostatic challenge-induced rate of weight gain for calculation of Kfc. In addition, the hydrostatic challenge-induced net lung weight gain (6. W) was calculated for each period of venous pressure elevation. Lungs selected for the study were those that (1) had a homogenous white appearance without signs of hemostasis or edema formation, (2) had pulmonary artery and ventilation pressures in the normal range, and (3) wereisogravimetricduring a steady-state period of 40 min. Absence of circulating cells in the lung effluent was ascertained in each experiment.
Preparation of Human Granulocytes and Rabbit Plasma Heparinized human donor blood was centrifuged in a discontinuous Percoll gradient (16, 17) to yield a PMN fraction of approximately 970/0 purity. The granulocytes were kept in RPMI 1640 medium with 200/0 calf serum for 90 to 120min. Immediately before use, the cellswerewashed twice and suspended in Krebs-Henseleit-albumin buffer. Rabbit blood was obtained from anesthetized animals by femoral artery catheterization after anticoagulation with heparin. Blood was centrifugated for 10min at 3,000 X g, and plasma was stored at - 70 0 C until experimental use. Preparation of E. Coli Hemolysin producing E. coli strain LE200i (Hly") and an E. coli strain (MC41OO; Hly") carrying the hemolysin HlyA + gene but lacking the Hlye gene necessary for activation of the toxin were used. As described previously (18) both strains secrete a 107-kD protein, but in the case of MC4100in a completely inactive form. Both strains were grown to the late logarithmic phase of bacterial growth, were washed three times in KHAB (3 min at 690 x g), and quantified by absorbance measurement at 578 nm. A calibration curve established for each E. coli strain was used. On the basis of this approximate quantification, aliquots of the E. coli suspension were mixed with the lung buffer medium. directly after the final washing procedure to achieve the required bacterial concentrations. Aliquots were plated on agar, and colony-forming units
(CFU) were counted for definitive quantification in each experiment. Data evaluation was undertaken only when the approximate quantification differed from quantification by CFU less than 30%.
sis products lXB2 and 6-keto-POF t a as described (19). The hemolysin concentration was determined by an ELISA that uses a monoclonal antihemolysin antibody to capture the antigen and a second polyclonal antibody for development, as described (20). Lactate dehydrogenase (LDH) and potassium in the perfusion fluid were measured according to standard techniques.
Biochemical Measurements TxA2 and POI 2 in the recirculating buffer fluid were assayed by RIA as their stable hydroly-
lE.
KHAB ASA (SOOp"')
I
20j /5
coli - Hly+{/06/m lJ
PAP (mmHg)
/o....r--t-,I' 5
o PVP
/5~
-
(mmHg)
1"""-_"'"
/0
~ -
i
Lf
.toW
20
(g)
I
/5 /0
5Jl.
o - - . - - - - "'zfL--r#e.,.,-----,--"7'J'---r-----. , ;/1 I -15
-25
60 (min)
0 30
Fig. 1. Lung permeability increase induced by hemolysin-forming E. coli in an isolated rabbit lung. Perfusion was performed with Krebs-Henseleit-albumin buffer (KHAB) in the presence of 500 11M acetylsalicylic acid (ASA) in order to block any major E. coli-Hly+evoked rise in pulmonary artery pressure (PAP). Three sequential sudden venous pressure(PVP)elevations for determination of Kfc and vascular compliance were performed. Time was set zero, when the E. coli-Hly+ were mixed with the buffer medium. The markedly increased fluid filtration in the subsequent hydrostatic challenges indicates severe, progressive rise in lung vascular permeability. (Note the interruptions in time scale: li.W = lung weight gain.)
70
jKHAB PAP (mmHg) 60
.toW (g)
l
50
30
20 /0
o
30
E.COIi - Hly+ (106/ m l) PAP
/ j" --",./~ ...
... ------' i
i
15
30
i
o
Fig. 2. Pulmonary artery pressure (PAP) rise and lung weight gain (Ii.W) induced by hemolysin-forming E. coli in an isolated rabbit lung. Perfusion was performed with Krebs-Henseleit-albumin buffer (KHAB), in the absence of PMN or rabbit plasma. E. coli-Hly+were admixed to the recirculating buffer medium to achieve the stated final concentration at the time indicated by arrow.
o
i
60 (min)
PAP (mmHg)
60
Fig. 3. Time- and dose-dependency of
E. coli-Hly+-evoked pulmonary artery 50
30 20
10
o I
o
I
I
i
i
30
60
90
120
I
150
180 (min)
pressure (PAP) rise. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. As controls, 106/ml and 107/ml E. coli MC4100 (Hly-), which secretes an inactive toxin, were used (n = 3 each).The E. coli-Hly+induced pressor responses differ significantly from the Hly--controls at all bacterial concentrations used (p < 0.001; two-way analysis of variance).
799
HEMOLYSIN-FORMING ESCHERICHIA COLI AND WNG VASCULAR LEAKAGE
Fig. 4. Time- and dose-dependency of
E. coli-Hly+-evoked thromboxane (TxB2) liberation into the recirculating medium. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero, when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. All TxB2 data were corrected for baseline values determined in the recirculating medium directly before addition of the E. coli. The shaded area indicates the detection limit of the assay. As control, 108/ml E. coli MC4100 (Hly-) were used as in figure 3. The E. coli-Hly+-induced thromboxane release differs significantly from the Hly- controls at all bacterial concentrations used (p < 0.001;two-way analysis of variance).
500 ~OO
300 200
----~----~----
100
106/ ml Ecoli-Hly---0
o I
Experimental Protocol After a steady-stateperiod of 40 min, the recirculating perfusate was exchanged by fresh buffer medium, which in some experiments was supplemented with 15070 vol/vol rabbit plasma; 10 min later, 2 x 108 PMN in 2 ml KHAB or the same volume of cell-free buffer were slowly injected into the pulmonary artery, followed by another lO-minperiod. Random cell counts in the recirculating medium
i i i
I
o
30
60
90
120
i
150
I
180 (min)
3 and 10 min after PMN application documented a nearly quantitative sticking of these cells in the lung vasculature « 3070 circulating PMN). Microscopic examinations revealed that the vast majority of the PMN were sequestered in the pulmonary capillaries. Next, live E. coli bacteria were admixed to the recirculating buffer volume to give final concentrations of 104 , 105 , 106 , or 107 CFU/ml, and time was set zero. Then l-ml
TABLE 1
perfusate samples for determination of prostanoids, potassium, LDH, and hemolysin were taken at 5, 10, 20, and 30 min and subsequently every 30 min. Perfusion was terminated after 180 min or when the E. coliinduced increase in total lung weight exceeded 15 g. Control experiments, without application of E. coli, included perfusions with buffer with or without rabbit plasma and PMN. In additional control studies, 106/ml and 107 Iml nonhemolytic E. coli (E. coli-Hly-) were applied. Inhibitor studies were performed with addition of 500 IJMASA or 50 IJMBM 13.177 to the recirculating buffer medium before application of bacteria. Experiments addressing the influence of E. coli (in the presence or absence of plasma or PMN) on lung vascular permeability wereperformed in the presence of 500 IJM ASA in order to block any significant pressor response to the bacteria. Venous pressure challenges for gravimetric determination of Kfc were undertaken after completion of the steady-state period as well as every 30 min after E. coli application (time schedule given in figure 1). Controls corresponded to those performed in the absence of sequential venous pressure challenges. In additional control studies in KHAB perfused lungs, E. coli endotoxin was admixed to the buffer medium to give a final concentration of 100 nglml, and recirculation was continued for 180 min. Kfc was determined according to the time schedule given in figure 1.
E. COL/-Hly+-INDUCED PRESSOR RESPONSE·
Results
Time after Application of E. coli-Hly+ (min)
104/ml E. coli E. coli E. coli E. coli E. coli E. coli 1Q5/ml E. coli E. coli E. coli E. coli E. coli E. coli 1Q8/ml E. coli E. coli E. coli E. coli E. coli E. coli 107/ml E. coli E. coli E. coli E. coli E. coli E. coli
+ PMN
+ +
PL PMN + PL + ASA t + BM 13.177
+
PMN + PL + PMN + PL + ASA + BM 13.177
+ +
PMN PL + PMN + PL + ASA + 8M 13.177
+
PMN + PL + PMN + PL + ASA + 8M 13.177
o
5
10
20
30
60
90
120
180
11.00 9.25 8.25 8.75 8.00 9.00
11.00 11.25 8.00 8.00 8.00 8.00
10.50 11.25 8.25 8.00 8.25 8.50
10.50 11.75 8.25 9.00 8.75 8.50
13.50 13.25 8.25 10.75 9.00 9.00
13.00 25.00 8.00 12.00 8.75 9.00
16.50 40.00 12.00 11.30 9.00 10.00
30.00 49.00 12.00 11.30 10.00 11.00
60.00
14.00 8.30 , 8.00 10.67 11.00 8.50
15.00 7.83 16.50 9.83 11.00 7.50
16.50 8.30 19.00 11.17 11.25 7.50
17.50 9.00 25.00 16.33 11.50 7.50
26.50 12.30 26.00 26.33 11.25 7.50
34.50 27.00 28.00 31.00 11.25 8.00
40.00 30.00 22.00 37.30 12.00 9.00
8.50 14.00 8.00 11.50 8.75 9.50
10.50 18.50 18.00 10.50 7.75
20.50 33.50 32.00 13.00 8.25 8.50
30.00 55.00 47.70 21.50 8.75 9.00
47.50 59.00 63.30 38.50 10.75 10.50
58.00
55.00 62.67 75.00
3.00 9.50 8.00 9.50 7.75 10.00
Definitionof abbreViations: PMN
=
8.50 40.00
40.00
45.00
24.50
27.33 50.00 20.00 8.25 11.00
40.00
25.00 10.25 7.75 10.00
70.00 48.00 8.75 11.50
=
12.25 22.00
52.00 12.00 22.50
80.00 13.25 12.00
polymorphonuclear leukocytes; PL 15% vollvol rabbit plasma; ASA = acetylsalicylic acid . • The table gives the pulmonary artery pressure (mm Hg) directly before E. coli-Hly" application (zero min) as well as at the indicated times after administration of the bacteria. Mean values of 2 to 3 separate experiments, performed at the different concentrations of E. coli and the different perfusion media, are given. t For inhibition Qf cyclooxygenase or antagonism of thromboxane, ASA (500 11M) as well as 8M 13.1n (50 11M) were used.
In accordance with previous results, pulmonary artery pressure (Ppa) ranged between 8 and 14mm Hg in all experiments in the initial baseline period (12, 14,21). The application of hemolysin-forming E. coli in KHAB-perfused lungs caused a time- and dose-dependent increasein Ppa (figures 2and 3). At 104 bacteria/ml, a progressive pressure rise occurred after an initial lag period of 60 min, whereas 107 E. coli/ml caused a rapid Ppa increase to values approximating 40 mm Hg within 5 min. The pressor responses were paralleled by a marked release of ThB2 into the recirculating buffer medium (figure 4); this surpassed 2,500 pg/ml at the highest E. coli concentration. There was a significant correlation between the height in Ppa and the ThB2 level, determined at different times after application of the various E. coli concentrations (r = 0.86, n = 62, p < 0.001). In the presence of the cyclooxygenaseinhibitor ASA or the thromboxane receptor-antagonist BM 13.177, the pressor responses were blocked by > 85070 at all bacterial concentrations (table 1). Similar to the thromboxane liberation, a time- and dose-dependent release of 6-keto-PGF 1u
800
SEEGER, OBERNITZ. THOMAS, WALMRATH, SUTTORN, HOLLAND. GRIMMINGER, EBERSPACHER, HUGO. AND BHAKDI
6-keto-PGF1 t>: (pg/mfJ
Fig. 5. Time- and dose-dependency of
E. coli-Hly+-evoked PGI 2 (6-keto-PGF 1a)
",'" ~
liberation into the recirculating medium. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero, when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. AII6-keto-PGF1a data were corrected for baseline values determined in the recirculating medium directly before addition of the E. coli. The shaded area indicates the detection limit of the assay. As control, 106/ml of the E. coli (Hly-) were used as in figure 3. The PGI 2 release evoked by 106 and 107 E coli-Hly+/ml differs significantly from the Hly· controls (p < 0.001; two-way analysis of variance).
", '1r:5
2000
1500
1000 I061ml E.coli-Hly· ---
IO'lml
500
o
I
o
I
30
60
I
I
I
90
120
150
I
180 (min)
(figure 5), potassium (table 2), and hemolysin (figure 6) into the recirculating medium was noted. In contrast, liberation of LDH was modest and exhibited no clear dose dependency with respect to the different E. coli concentrations used (table 3). The E. coli strain releasing inactive hemolysin did not induce any significant pressor response within the observation period of 180min (figure 3). Correspondingly, no substantial release of TxB2 , potassium, or LDH was noted (figure 4,
tables 2 and 3). A moderate, progressive liberation of POI 2 occurred to a similar extent as observed in control lungs (figure 5 and table 4). The administration of 15070 vol/vol rabbit plasma or 2 X 108 PMN or both caused only transient rises in Ppa, which did not surpass 3 mm Hg, and the perfusion pressure remained constant over a total observation period of 180 min (n = 3 of each control). No significant amounts of ThB2 « 100 pg/ml), potassium « 0.2 mM), or LDH « 25 U/L) were
TABLE 2
E. COLl-Hly'-INDUCED POTASSIUM RELEASE* Time after Application of E. coli-Hly+ (min) 10
20
30
60
90
120
0.07 0.07 0.05 0.05
0.05 0.13 0.02 0.05
0.07 0.13 0.03 0.05
0.07 0.14 0.03 0.10
0.27 0.20 0.11 0.15
0.31 0.24 0.13 0.17
0.40 0.33 0.21 0.24
+ PL
0.08 0.06 0.06 0.05
0.08 0.13 0.06 0.09
0.08 0.17 0.07 0.15
0.12 0.32 0.08 0.23
0.47 0.43 0.15 0.37
0.69 0.65 0.17 0.57
0.11 0.16 0.10 0.13
0.24 0.16 0.17 0.38
0.33 0.40 0.30 0:64
0.67
+ PL
0.07 0.08 0.07 0.07
+ PL
0.16 0.07 0.10 0.05
0.40 0.13 0.14 0.10
0.80 0.13 0.34 0.69
0.80 0.42 0.55 0.69
0
0
0
0
5
104/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 10S/mi E. coli E. coli + PMN E. coli + PL E. coli + PMN 106/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 10 7/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 106/ml E. coli-Hly-t
+ PL
<
WERNER SEEGER, RUDOLF OBERNITZ, MICHAEL THOMAS, DIETER WALMRATH, N. SUTTORN, IAN B. HOLLAND, FRIEDRICH GRIMMINGER, BETTINA EBERSPACHER, FERDINAND HUGO, and SUCHARIT BHAKDI
Introduction
Thirty to fifty percent of Escherichia coli isolates causing extraintestinal infections in humans elaborate a proteinaceous cytolysin that damages target cell membranes through the generation of circumscribed, transmembrane pores (1-3). Structurally related toxins are also produced by other Enterobacteriaceae including Morganella, Proteus, and Pasteurella ssp. The relevance of E. coli hemolysin as a determinant of bacterial pathogenicity has been established in animal models with the use of genetically engineered, isogenic bacterial strains (4, 5). An analogous role as a virulence factor in human infections has been inferred from the high association of hemolysin production with disease, including pyelonephritis and septicemia (6-8). Bacterial sepsis is the most consistent factor associated with the development of acute respiratory failure in adults (ARDS), with gram-negative rods including E. coli currently representing the predominant infectious agents (9, 10). Moreover, acute lung injury of different etiology is often complicated by nosocomial pneumonia caused by Enterobacteriaceae (11). Against this background it was of interest that direct intravascular application of very low doses of E. coli hemolysin induced key events of acute respiratory failure in blood-free perfused rabbit lungs (12). In particular, development of acute thromboxane-mediated pulmonary hypertension as well as protracted vascular leakage was noted. In the present study, these pathophysiologic events were dosedependently reproduced by infusion of live hemolysin-forming E. coli. In contrast, an E. coli strain that secretes an inactive form of hemolysin was without effect. These findings support a possible role of pore-forming bacterial exotoxins in the pathogenesis of acute lung injury
SUMMARY Escherichia coli hemolysin, a transmembrane pore-forming exotoxin, is considered an important virulence factor. In the present study, the possible significance of hemolysin production was Investigated In a model of septic lung failure through infusion of viable bacteria in Isolated rabbit lungs; 104 to 107 E.coli/ml perfusate caused a dose- and tlme-dependent appearance of hemolysin, accompanied by release of potassium, thromboxane A 2 , and PGl 2 into the perfusate. Concomitantly, marked pulmonary hypertension developed. Inhibitor studies suggested that the pressor response was predominantly medlate~ by pulmonary thromboxane generation. Administration of hemolysin-forming E. coli additionally caused a protracted, dose-dependent Increase In the lung capillary filtration coefficient, followed by severe edema formation. The permeability increase was independent of lung prostanoid generation. An E. coli strain that releases an Inactive form of hemolysin completely failed to provoke the described biophysical and biochemical responses. Preappllcation of 2 x 108 human granulocytes was without effect in the present experimental model. Weconclude that the hemolysin produced by low numbers of E. coli organisms can provoke thromboxanemediated pulmonary hypertension and severe vascular leakage. E. coli hemolysin and, possibly, other related cytolyslns may thus contribute directly to the pathogenesis of acute respiratory failure AM REV RESPIR DIS 1991; 143:797-805 under conditions of sepsis or pneumonia.
under conditions of sepsis and severe pneumonia. Methods Reagents BM 13.177 was a gift from Boehringer Mannheim GmbH (Mannheim, Germany). D,LLysin-mono-acetylsalicylate/glycin (9:1; ASA) was obtained from Bayer AG (Leverkusen, Germany). Bovine albumin (98lt7o purity, reduced in free fatty acids to < 5 ug/g), rabbit anti-6-keto-PGF ta , and rabbit anti-Ixls, were obtained from Paesel AG (Frankfurt, Germany). Tritium-labeled lXB 2 and 6-ketoPGF HI were from New England Nuclear (Dreieich, Germany), and E. coli endotoxin (serotype 0111:B4) was purchased from Sigman Chemical (Munich, Germany). All other biochemicals were obtained from Merck AG (Darmstadt, Germany) in p.a. quality. Isolated Lung Model The model has been previously described (12-15). Briefly, rabbits of either sex (body weight, 2.2 to 2.6 kg) were deeply anesthetized and anticoagulated with 1,000U of heparin per kg body weight. The lungs were excised while being perfused with KrebsHenseleit-albumin buffer (KHAB) through cannulas in the pulmonary artery and the left
atrium. The buffer contained 132.8mM NaCI, 4.3 mM KCI, 1.1 mM KH 2P04 , 24.1 mM NaHC03 , 2.4 mM CaCI 2 , and 1.3 mM MgP0 4 , as well as 240 mg glucose and 1 g albumin per 100 ml. The lungs were placed in a temperature-equilibrated housing chamber at 37° C, freely suspended from a force transducer. They were ventilated with 4lt7o CO 2 , 17lt7o O 2 , and 79lt7o N 2 , and the pH of the perfusion fluid ranged between 7.35 and 7.45. After extensive rinsing of the vascular bed, the lungs were recirculatingly perfused with a pulsatile flow of 100 ml/min. The alternate use of two separate perfusion circuits,
(Received in original form February 20, 1990 and in revised form September 7, 1990) 1 From the Department of Internal Medicine, Division of Clinical Pathophysiology and Experimental Medicine, the Institute of Medical Microbiology, Justus-Liebig University, Giessen, Germany, and the Department of Microbiology and Genetics, University of Leicester, Leicester, United Kingdom. 2 Supported by the Deutsche Forschungsgemeinschaft (SFB 249, Teilprojekte A6 and A7). 3 Correspondence and requests for reprints should be addressed to Werner Seeger, Department of Internal Medicine, Justus-Liebig University Giessen, Klinikstrasse 36, D-63 Giessen, Germany.
797
798
SEEGER, OBERNITZ, THOMAS, WALMRATH, SUTTORN, HOLLAND, GRIMMINGER, EBERSPACHER, HUGO, AND BHAKDI
each containing 200 ml, allowed exchange of perfusion fluid. Perfusion pressure, ventilation pressure, and the weight of the isolated organ were registered continuously. The left atrial pressure was set 2 mm Hg under baseline conditions (zero referenced at the hilus) to guarantee zone III conditions at endexpiration throughout the lung. The capillary filtration coefficient (Kfc) and the total vascular compliance were determined gravimetrically from the slope of weight gain, induced by a 7.5-mm Hg step elevation of the venous pressure for 8 min. The application of this method to the present model and the use of zero time extrapolation of the slope of weight gain for the calculation of Kfc have been described (14).In lungs, which already revealed a constant rate of weight gain before onset of the hydrostatic challenge, this slope of weight increase was subtracted from the hydrostatic challenge-induced rate of weight gain for calculation of Kfc. In addition, the hydrostatic challenge-induced net lung weight gain (6. W) was calculated for each period of venous pressure elevation. Lungs selected for the study were those that (1) had a homogenous white appearance without signs of hemostasis or edema formation, (2) had pulmonary artery and ventilation pressures in the normal range, and (3) wereisogravimetricduring a steady-state period of 40 min. Absence of circulating cells in the lung effluent was ascertained in each experiment.
Preparation of Human Granulocytes and Rabbit Plasma Heparinized human donor blood was centrifuged in a discontinuous Percoll gradient (16, 17) to yield a PMN fraction of approximately 970/0 purity. The granulocytes were kept in RPMI 1640 medium with 200/0 calf serum for 90 to 120min. Immediately before use, the cellswerewashed twice and suspended in Krebs-Henseleit-albumin buffer. Rabbit blood was obtained from anesthetized animals by femoral artery catheterization after anticoagulation with heparin. Blood was centrifugated for 10min at 3,000 X g, and plasma was stored at - 70 0 C until experimental use. Preparation of E. Coli Hemolysin producing E. coli strain LE200i (Hly") and an E. coli strain (MC41OO; Hly") carrying the hemolysin HlyA + gene but lacking the Hlye gene necessary for activation of the toxin were used. As described previously (18) both strains secrete a 107-kD protein, but in the case of MC4100in a completely inactive form. Both strains were grown to the late logarithmic phase of bacterial growth, were washed three times in KHAB (3 min at 690 x g), and quantified by absorbance measurement at 578 nm. A calibration curve established for each E. coli strain was used. On the basis of this approximate quantification, aliquots of the E. coli suspension were mixed with the lung buffer medium. directly after the final washing procedure to achieve the required bacterial concentrations. Aliquots were plated on agar, and colony-forming units
(CFU) were counted for definitive quantification in each experiment. Data evaluation was undertaken only when the approximate quantification differed from quantification by CFU less than 30%.
sis products lXB2 and 6-keto-POF t a as described (19). The hemolysin concentration was determined by an ELISA that uses a monoclonal antihemolysin antibody to capture the antigen and a second polyclonal antibody for development, as described (20). Lactate dehydrogenase (LDH) and potassium in the perfusion fluid were measured according to standard techniques.
Biochemical Measurements TxA2 and POI 2 in the recirculating buffer fluid were assayed by RIA as their stable hydroly-
lE.
KHAB ASA (SOOp"')
I
20j /5
coli - Hly+{/06/m lJ
PAP (mmHg)
/o....r--t-,I' 5
o PVP
/5~
-
(mmHg)
1"""-_"'"
/0
~ -
i
Lf
.toW
20
(g)
I
/5 /0
5Jl.
o - - . - - - - "'zfL--r#e.,.,-----,--"7'J'---r-----. , ;/1 I -15
-25
60 (min)
0 30
Fig. 1. Lung permeability increase induced by hemolysin-forming E. coli in an isolated rabbit lung. Perfusion was performed with Krebs-Henseleit-albumin buffer (KHAB) in the presence of 500 11M acetylsalicylic acid (ASA) in order to block any major E. coli-Hly+evoked rise in pulmonary artery pressure (PAP). Three sequential sudden venous pressure(PVP)elevations for determination of Kfc and vascular compliance were performed. Time was set zero, when the E. coli-Hly+ were mixed with the buffer medium. The markedly increased fluid filtration in the subsequent hydrostatic challenges indicates severe, progressive rise in lung vascular permeability. (Note the interruptions in time scale: li.W = lung weight gain.)
70
jKHAB PAP (mmHg) 60
.toW (g)
l
50
30
20 /0
o
30
E.COIi - Hly+ (106/ m l) PAP
/ j" --",./~ ...
... ------' i
i
15
30
i
o
Fig. 2. Pulmonary artery pressure (PAP) rise and lung weight gain (Ii.W) induced by hemolysin-forming E. coli in an isolated rabbit lung. Perfusion was performed with Krebs-Henseleit-albumin buffer (KHAB), in the absence of PMN or rabbit plasma. E. coli-Hly+were admixed to the recirculating buffer medium to achieve the stated final concentration at the time indicated by arrow.
o
i
60 (min)
PAP (mmHg)
60
Fig. 3. Time- and dose-dependency of
E. coli-Hly+-evoked pulmonary artery 50
30 20
10
o I
o
I
I
i
i
30
60
90
120
I
150
180 (min)
pressure (PAP) rise. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. As controls, 106/ml and 107/ml E. coli MC4100 (Hly-), which secretes an inactive toxin, were used (n = 3 each).The E. coli-Hly+induced pressor responses differ significantly from the Hly--controls at all bacterial concentrations used (p < 0.001; two-way analysis of variance).
799
HEMOLYSIN-FORMING ESCHERICHIA COLI AND WNG VASCULAR LEAKAGE
Fig. 4. Time- and dose-dependency of
E. coli-Hly+-evoked thromboxane (TxB2) liberation into the recirculating medium. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero, when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. All TxB2 data were corrected for baseline values determined in the recirculating medium directly before addition of the E. coli. The shaded area indicates the detection limit of the assay. As control, 108/ml E. coli MC4100 (Hly-) were used as in figure 3. The E. coli-Hly+-induced thromboxane release differs significantly from the Hly- controls at all bacterial concentrations used (p < 0.001;two-way analysis of variance).
500 ~OO
300 200
----~----~----
100
106/ ml Ecoli-Hly---0
o I
Experimental Protocol After a steady-stateperiod of 40 min, the recirculating perfusate was exchanged by fresh buffer medium, which in some experiments was supplemented with 15070 vol/vol rabbit plasma; 10 min later, 2 x 108 PMN in 2 ml KHAB or the same volume of cell-free buffer were slowly injected into the pulmonary artery, followed by another lO-minperiod. Random cell counts in the recirculating medium
i i i
I
o
30
60
90
120
i
150
I
180 (min)
3 and 10 min after PMN application documented a nearly quantitative sticking of these cells in the lung vasculature « 3070 circulating PMN). Microscopic examinations revealed that the vast majority of the PMN were sequestered in the pulmonary capillaries. Next, live E. coli bacteria were admixed to the recirculating buffer volume to give final concentrations of 104 , 105 , 106 , or 107 CFU/ml, and time was set zero. Then l-ml
TABLE 1
perfusate samples for determination of prostanoids, potassium, LDH, and hemolysin were taken at 5, 10, 20, and 30 min and subsequently every 30 min. Perfusion was terminated after 180 min or when the E. coliinduced increase in total lung weight exceeded 15 g. Control experiments, without application of E. coli, included perfusions with buffer with or without rabbit plasma and PMN. In additional control studies, 106/ml and 107 Iml nonhemolytic E. coli (E. coli-Hly-) were applied. Inhibitor studies were performed with addition of 500 IJMASA or 50 IJMBM 13.177 to the recirculating buffer medium before application of bacteria. Experiments addressing the influence of E. coli (in the presence or absence of plasma or PMN) on lung vascular permeability wereperformed in the presence of 500 IJM ASA in order to block any significant pressor response to the bacteria. Venous pressure challenges for gravimetric determination of Kfc were undertaken after completion of the steady-state period as well as every 30 min after E. coli application (time schedule given in figure 1). Controls corresponded to those performed in the absence of sequential venous pressure challenges. In additional control studies in KHAB perfused lungs, E. coli endotoxin was admixed to the buffer medium to give a final concentration of 100 nglml, and recirculation was continued for 180 min. Kfc was determined according to the time schedule given in figure 1.
E. COL/-Hly+-INDUCED PRESSOR RESPONSE·
Results
Time after Application of E. coli-Hly+ (min)
104/ml E. coli E. coli E. coli E. coli E. coli E. coli 1Q5/ml E. coli E. coli E. coli E. coli E. coli E. coli 1Q8/ml E. coli E. coli E. coli E. coli E. coli E. coli 107/ml E. coli E. coli E. coli E. coli E. coli E. coli
+ PMN
+ +
PL PMN + PL + ASA t + BM 13.177
+
PMN + PL + PMN + PL + ASA + BM 13.177
+ +
PMN PL + PMN + PL + ASA + 8M 13.177
+
PMN + PL + PMN + PL + ASA + 8M 13.177
o
5
10
20
30
60
90
120
180
11.00 9.25 8.25 8.75 8.00 9.00
11.00 11.25 8.00 8.00 8.00 8.00
10.50 11.25 8.25 8.00 8.25 8.50
10.50 11.75 8.25 9.00 8.75 8.50
13.50 13.25 8.25 10.75 9.00 9.00
13.00 25.00 8.00 12.00 8.75 9.00
16.50 40.00 12.00 11.30 9.00 10.00
30.00 49.00 12.00 11.30 10.00 11.00
60.00
14.00 8.30 , 8.00 10.67 11.00 8.50
15.00 7.83 16.50 9.83 11.00 7.50
16.50 8.30 19.00 11.17 11.25 7.50
17.50 9.00 25.00 16.33 11.50 7.50
26.50 12.30 26.00 26.33 11.25 7.50
34.50 27.00 28.00 31.00 11.25 8.00
40.00 30.00 22.00 37.30 12.00 9.00
8.50 14.00 8.00 11.50 8.75 9.50
10.50 18.50 18.00 10.50 7.75
20.50 33.50 32.00 13.00 8.25 8.50
30.00 55.00 47.70 21.50 8.75 9.00
47.50 59.00 63.30 38.50 10.75 10.50
58.00
55.00 62.67 75.00
3.00 9.50 8.00 9.50 7.75 10.00
Definitionof abbreViations: PMN
=
8.50 40.00
40.00
45.00
24.50
27.33 50.00 20.00 8.25 11.00
40.00
25.00 10.25 7.75 10.00
70.00 48.00 8.75 11.50
=
12.25 22.00
52.00 12.00 22.50
80.00 13.25 12.00
polymorphonuclear leukocytes; PL 15% vollvol rabbit plasma; ASA = acetylsalicylic acid . • The table gives the pulmonary artery pressure (mm Hg) directly before E. coli-Hly" application (zero min) as well as at the indicated times after administration of the bacteria. Mean values of 2 to 3 separate experiments, performed at the different concentrations of E. coli and the different perfusion media, are given. t For inhibition Qf cyclooxygenase or antagonism of thromboxane, ASA (500 11M) as well as 8M 13.1n (50 11M) were used.
In accordance with previous results, pulmonary artery pressure (Ppa) ranged between 8 and 14mm Hg in all experiments in the initial baseline period (12, 14,21). The application of hemolysin-forming E. coli in KHAB-perfused lungs caused a time- and dose-dependent increasein Ppa (figures 2and 3). At 104 bacteria/ml, a progressive pressure rise occurred after an initial lag period of 60 min, whereas 107 E. coli/ml caused a rapid Ppa increase to values approximating 40 mm Hg within 5 min. The pressor responses were paralleled by a marked release of ThB2 into the recirculating buffer medium (figure 4); this surpassed 2,500 pg/ml at the highest E. coli concentration. There was a significant correlation between the height in Ppa and the ThB2 level, determined at different times after application of the various E. coli concentrations (r = 0.86, n = 62, p < 0.001). In the presence of the cyclooxygenaseinhibitor ASA or the thromboxane receptor-antagonist BM 13.177, the pressor responses were blocked by > 85070 at all bacterial concentrations (table 1). Similar to the thromboxane liberation, a time- and dose-dependent release of 6-keto-PGF 1u
800
SEEGER, OBERNITZ. THOMAS, WALMRATH, SUTTORN, HOLLAND. GRIMMINGER, EBERSPACHER, HUGO. AND BHAKDI
6-keto-PGF1 t>: (pg/mfJ
Fig. 5. Time- and dose-dependency of
E. coli-Hly+-evoked PGI 2 (6-keto-PGF 1a)
",'" ~
liberation into the recirculating medium. The experiments were performed in KHAB-perfused lungs, in the absence of human granulocytes and rabbit plasma. Time was set zero, when the hemolysin-forming E. coli were administered to the buffer medium at the given final concentrations. Mean values of 2 to 3 experiments at each concentration are given. AII6-keto-PGF1a data were corrected for baseline values determined in the recirculating medium directly before addition of the E. coli. The shaded area indicates the detection limit of the assay. As control, 106/ml of the E. coli (Hly-) were used as in figure 3. The PGI 2 release evoked by 106 and 107 E coli-Hly+/ml differs significantly from the Hly· controls (p < 0.001; two-way analysis of variance).
", '1r:5
2000
1500
1000 I061ml E.coli-Hly· ---
IO'lml
500
o
I
o
I
30
60
I
I
I
90
120
150
I
180 (min)
(figure 5), potassium (table 2), and hemolysin (figure 6) into the recirculating medium was noted. In contrast, liberation of LDH was modest and exhibited no clear dose dependency with respect to the different E. coli concentrations used (table 3). The E. coli strain releasing inactive hemolysin did not induce any significant pressor response within the observation period of 180min (figure 3). Correspondingly, no substantial release of TxB2 , potassium, or LDH was noted (figure 4,
tables 2 and 3). A moderate, progressive liberation of POI 2 occurred to a similar extent as observed in control lungs (figure 5 and table 4). The administration of 15070 vol/vol rabbit plasma or 2 X 108 PMN or both caused only transient rises in Ppa, which did not surpass 3 mm Hg, and the perfusion pressure remained constant over a total observation period of 180 min (n = 3 of each control). No significant amounts of ThB2 « 100 pg/ml), potassium « 0.2 mM), or LDH « 25 U/L) were
TABLE 2
E. COLl-Hly'-INDUCED POTASSIUM RELEASE* Time after Application of E. coli-Hly+ (min) 10
20
30
60
90
120
0.07 0.07 0.05 0.05
0.05 0.13 0.02 0.05
0.07 0.13 0.03 0.05
0.07 0.14 0.03 0.10
0.27 0.20 0.11 0.15
0.31 0.24 0.13 0.17
0.40 0.33 0.21 0.24
+ PL
0.08 0.06 0.06 0.05
0.08 0.13 0.06 0.09
0.08 0.17 0.07 0.15
0.12 0.32 0.08 0.23
0.47 0.43 0.15 0.37
0.69 0.65 0.17 0.57
0.11 0.16 0.10 0.13
0.24 0.16 0.17 0.38
0.33 0.40 0.30 0:64
0.67
+ PL
0.07 0.08 0.07 0.07
+ PL
0.16 0.07 0.10 0.05
0.40 0.13 0.14 0.10
0.80 0.13 0.34 0.69
0.80 0.42 0.55 0.69
0
0
0
0
5
104/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 10S/mi E. coli E. coli + PMN E. coli + PL E. coli + PMN 106/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 10 7/ml E. coli E. coli + PMN E. coli + PL E. coli + PMN 106/ml E. coli-Hly-t
+ PL
<
