INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

PREVIOUS STUDIES from this laboratory (1, 2, 21, 25, 32) have shown that a period of ischemia followed by reperfusion (I/R) in isolated perfused rat lungs increases microvascular permeability. This model of lung endothelial injury requires O₂ radicals, the xanthine oxidase system, leukocyte rolling factors, endothelial cell and leukocyte adherence factors, and activation of the myosin light-chain kinase system (1, 2, 21, 25, 32). There is increasing evidence indicating that tumor necrosis factor- (TNF-) is required to produce the pathophysiology occurring in lungs subjected to different inflammatory models and endotoxin shock (3, 10). TNF- is a 17-kDa proinflammatory cytokine secreted by a number of cells, including macrophages, mast cells, and epithelial cells, in response to a variety of stimuli, especially endotoxin lipopolysaccharide (LPS) (4, 30). Other studies have suggested that TNF- alters the selectivity of the endothelial barrier and produces experimental pulmonary edema (14, 23). TNF- has also been detected in the bronchoalveolar lavage fluid obtained from patients with acute respiratory distress syndrome (36). Studies by Hocking et al. (14) have also shown that TNF- increases pulmonary vascular resistance that results in alveolar edema as a result of the release of thromboxane A₂ in response to activation of polymorphonuclear leukocytes (PMN). TNF- also increases the neutrophil population in the lung caused by inducing endothelium-derived PMN chemotactic and adherent factors (5, 36).

Serrano et al. (34) have shown that TNF- enhances intercellular adhesion molecule-1 (ICAM-1), E-selectin, and vascular cell adhesion molecule-1 (VCAM-1) expression in human aortic endothelial cells monolayers, and Ferro et al. (11) have shown that a 4-h incubation of TNF- reduces pulmonary arterial endothelial monolayer selectivity by an nitric oxide-dependent mechanism. However, the studies that evaluate cytokine involvement in producing lung damage usually require 3-4 h before the endothelial barrier damage is present in measurable amounts. Even when TNF- is administered into animals or placed into bathing solutions surrounding endothelial cell monolayers, 3-4 h are required before monolayer damage and the cell upregulation of adhesion and rolling factors occurs (34, 36). The I/R models that have been developed for isolated rat lungs in our and other laboratories (1, 2, 9, 12, 17, 21, 25, 32) use much shorter periods of ischemia (45 min to 1 h) followed by 1-2 h of reperfusion. We had previously thought that cytokines would not play a significant role in this relatively short-termed I/R model (1, 9, 12, 17, 32); however, Eppinger et al. (10) have recently demonstrated an increase in mRNA for TNF- after 30 min of reperfusion in rat lungs subjected to 1.5 h of ischemia in situ. From these data, we hypothesized that TNF- would be involved even in the relatively short-termed I/R models of lung microvascular injury. To test this hypothesis, perfusate TNF- levels were measured and an anti-TNF- antibody was studied in both an I/R model and an air-ventilated ischemia model (/R), used by Fisher's group (9, 12, 17).

The present study clearly shows that TNF- is released after either 45 min of ischemia with lungs extended at end-expiratory pressure and then reperfused and ventilated or when the lungs are ventilated with an O₂ gas mixture during ischemia for 45 min and then reperfused. Perfusate TNF- increased in both models, and the damage associated with I/R was totally eliminated by pretreating the animals with a specific anti-TNF- antibody. However, TNF- introduced into the perfusate produced no endothelial damage over the same time frame as used in I/R, but exogenous TNF- significantly increased the amount of damage associated with I/R endothelial injury.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Isolated perfused lung. Male CD rats (250-350 g body wt, Charles River Laboratories) were anesthetized with pentobarbital sodium (50 mg/kg body wt ip). A tracheostomy was performed, and lungs were ventilated with 21% O₂-5% CO₂ (Harvard rodent ventilator, model 683) at a rate of 50 breaths/min, a tidal volume of 15 ml/kg body wt, and a positive end-expiratory pressure of 2 cmH₂O. After median sternotomy, heparin (300 IU) was injected into the right ventricle, and cannulas were placed into the pulmonary artery and left ventricle. The heart, lungs, and mediastinal structures were removed en bloc and suspended from a force-displacement transducer (Grass, model FT03) into a humidified chamber to monitor weight changes. The lungs were perfused at a constant flow of 0.03 ml/g body wt with Earle's balanced salt solution [(in mg/l) 200 CaCl₂, 400 KCl, 97.7 MgSO₄, 6,800 NaCl, 140 NaH₂PO₄ · H₂O, 1,000 glucose, 10 phenol red] containing 0.21% NaHCO₃ and 4% bovine serum albumin. The first 75 ml of perfusate, which contained large amounts of residual blood cells and plasma, were discarded. An additional 50 ml of perfusate were then used for recirculation. Pulmonary arterial (Ppa) and pulmonary venous pressures (Ppv) were continuously monitored with pressure transducers (Gould-Statham, model P23 ID) and recorded on a polygraph recorder (Grass, model 7E).

Measurement of the pulmonary capillary pressure (Ppc). The Ppc was estimated by using the double-occlusion method. When arterial and venous catheters are simultaneously occluded, Ppa and Ppv will equilibrate to the same pressure. This equilibration pressure is equal to Ppc (27, 39).

Capillary permeability [the filtration coefficient (K_fc)]. The K_fc was used as a measurement of capillary permeability, by using methods previously described (8, 27). Briefly, after an isogravimetric state was achieved, Ppv was rapidly elevated by 6-8 cmH₂O for 15 min. The increase in lung weight was recorded, and the rate of weight change (W/t) during the 6- to 15-min interval was analyzed with a linear regression of the log₁₀-transformed rates of weight changes per minute. The initial rate of weight gain was then calculated by extrapolating W/t to time 0. K_fc was calculated by dividing W/t at time 0 by the change measured in Ppc that occurred after venous pressure elevation and was normalized by using the baseline wet lung weight and expressed as milliliters per minute per centimeters H₂O per 100 grams of lung tissue.

	SPECIFIC PROTOCOLS

Control lungs (n = 5). All lungs were allowed to equilibrate and attain isogravimetric conditions for 25 min. After equilibration, baseline hemodynamic profiles (Ppa, Ppc, and Ppv) were measured, and the permeability parameter K_fc was calculated. Then, the lungs were ventilated and perfused for 45 min, which corresponds to the time used for the ischemic periods in the models described below. Next, an additional 90 min of ventilation and perfusion were applied to mimic the total experimental time frame used for both the I/R and /R models, as described below, and K_fc and hemodynamic profiles were determined at 135 min.

I/R lungs (n = 5). All lungs were allowed to equilibrate and attain isogravimetric conditions for 25 min. After equilibration, baseline hemodynamic profiles (Ppa, Ppc, and Ppv) were measured, and the permeability parameter K_fc was calculated. Then the lungs were not ventilated but held at end-expiratory pressure and not perfused for 45 min (ischemia), followed by the resumption of both ventilation and perfusion (reperfusion) until the end of the experiment. Hemodynamic profiles and K_fc values were obtained after 30 and 90 min of reperfusion (20, 24, 31). In addition, perfusate samples were taken after each K_fc measurement and frozen to determine perfusate TNF- levels. TNF- measurements on perfusate samples were made by Dr. J. Fuseler at Louisiana State University Medical Center using the L929 cell cytotoxicity assay as described below (13, 30).

/R lungs (n = 5). The same experimental protocol was used in this ischemia model, except lungs were ventilated with 21% O₂-5% CO₂-74% N₂ gas mixture during the ischemic period. Hemodynamic profiles (Ppa, Ppc, and Ppv) and the permeability parameter K_fc were evaluated at the same time intervals as in the I/R model, and perfusate samples were also taken after each K_fc measurement to measure perfusate TNF- levels.

TNF- control lungs (n = 5). To examine the effects of TNF- on isolated rat lungs not exposed to ischemia, the following experimental protocol was used. After an equilibration period of 25 min, a baseline hemodynamic profile (Ppa, Ppc, and Ppv) was measured and the permeability parameter K_fc was calculated. Then, recombinant rat TNF- (50, 000 U) was introduced into the venous reservoir. Lungs were continuously ventilated and perfused for 45 min and for an additional 90-min period, which corresponds to the time frame of the I/R and /R protocols. Hemodynamic profiles and K_fc values were obtained after 90 min of reperfusion.

TNF- and I/R lungs (n = 5). To determine the effects of TNF- in isolated rat lungs subjected to I/R, the following experimental protocol was used. After a 25-min equilibration period, a baseline hemodynamic profile (Ppa, Ppc, and Ppv) and the permeability parameter K_fc were measured. Recombinant rat TNF- (50,000 U) was then introduced into the venous reservoir and allowed to circulate for 30 min. Then, the lungs were not ventilated or perfused for 45 min (ischemia), followed by the resumption of both ventilation and perfusion (reperfusion) until the end of the experiment. Hemodynamic profiles and K_fc values were measured after 90 min of reperfusion.

Anti-TNF- antibody control (normal goat IgG) (n = 4). To assess any nonspecific effects of the goat IgG in the isolated rat lungs, the standard I/R protocol was followed, except that animals were pretreated with 10 mg ip of a normal goat IgG 2 h before the lung was isolated as described above.

Anti-TNF- antibody in I/R and /R lungs (n = 8). To assess the effects of an anti-TNF- antibody in I/R and /R models, animals were pretreated with goat anti-TNF- IgG 2 h before the beginning of the standard I/R or /R protocols as described above.

TNF- measurements. TNF- levels in the perfusate samples were measured by using the L929 cell cytotoxicity assay. Confluent monolayers of L929 cells, which are grown in 96-well plates, were exposed to serial dilution of the sample and cultured in the presence of actinomycin-D for 18 h, after which cell viability was determined. Cell viability was determined with mitochondrial indicator dye 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide, 5 mg/ml of which were added to each well and incubated for 2 h at 37°C. The cells were lysed in buffer containing 10% SDS and 40% N,N-dimethylformamide in distilled water at pH = 4.95, and color was allowed to develop for 2 h at 37°C. Mean absorbance from triplicate samples was converted to units per milliliter by fitting experimental titrations to standard curves generated from TNF- of known activity.

Materials. Earle's balanced salt and bovine serum albumin were purchased from Sigma Chemical (St. Louis, MO). Rat recombinant TNF- was purchased from Endogen (Woburn, MA). Polyclonal anti-TNF- antibodies were prepared at Louisiana State University Medical Center (New Orleans, LA). They were produced as previously described (3). Briefly, anti-TNF- antibodies were produced in goats by using the RIBI adjuvant system containing 0.5 mg each of monophosphoryl lipid A, trehalose dimycolate, and cell wall skeleton in 0.2% Tween 80 (RIBI Immunochem Research, Hamilton, MT). The serum IgG fraction was obtained by polyethylene glycol 4000 precipitation and column chromatography with diethylaminoethyl Bio-Gel A (Bio-Rad, Richmond, CA). The neutralizing capacity of the anti-TNF- IgG fraction was determined by mixing equal volumes of murine recombinant TNF- (400 U/ml) or a dilution of rat serum containing TNF- (400 U/ml) with serial dilutions of the anti-TNF- IgG. This mixture was incubated for 1 h at 37°C and then tested for residual TNF- activity. Under these conditions, the antibody was determined to contain 6.5 and 9.0 × 10⁵ 50% neutralizing U/mg of IgG protein against murine recombinant TNF- and TNF--containing rat serum, respectively. Normal goat IgG prepared in the same way had no detectable TNF--neutralizing activity. The binding properties of anti-TNF- IgG and normal goat IgG were also tested in microtiter plates precoated with LPS or murine recombinant interleukin-1, interferon-, or TNF-. Neither IgG preparation demonstrated effective binding to LPS, interleukin-1, or interferon-. Only the anti-TNF- IgG bound TNF-.

Statistics. All results are expressed as means ± SE. Comparisons were made by using analysis of variance with Newman-Keuls as the post hoc test. Significance was determined when P < 0.05 was obtained (43).

	RESULTS

Top Abstract Introduction Methods Results Discussion References

I/R models of lung injury. Figure 1 shows the microvascular permeability as measured by K_fc in I/R (open bars) and /R (solid bars) models of lung endothelial injury. A large and statistically significant increase in microvascular permeability occurred in both groups. A threefold increase in K_fc occurred after 30 min of reperfusion (P < 0.01), which increased to a fivefold level after 90 min (P < 0.01) in the I/R group. A larger increase in K_fc occurred in the /R group, in which lungs were ventilated with 21% O₂-5% CO₂ gas mixture during the ischemic period as K_fc increased fivefold after 30 min of reperfusion and tenfold after an additional 60 min of reperfusion compared with controls. There were no changes in microvascular permeability in time-matched control lungs (gray bars) not subjected to either I/R or /R.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Effect of nonventilated ischemia followed by reperfusion (I/R) and ventilation (open bars) and of ventilation with 21% O2-5% CO2-74% N2 mixture during ischemia followed by reperfusion and continued ventilation (/R; solid bars) on endothelial damage, as measured by filtration coefficient (Kfc). Note that after 30 min of reperfusion permeability was increased significantly in both types of ischemic injury (* P < 0.05) and damage in both ischemic models was exacerbated after an additional 60 min compared with 30 min of reperfusion (# P < 0.05). Also, /R model produced more endothelial damage than I/R at 30 and 90 min after reperfusion (@ P < 0.05). Gray histograms represent time-matched control lungs not subjected to I/R or /R.

Perfusate TNF- levels in I/R and /R models. Figure 2 shows the total TNF- measured in the perfusates for both the I/R and /R ischemic models. TNF- levels in the perfusate of I/R lungs increased significantly after 30 min of reperfusion compared with baseline [from 44.66 ± 3.71 to 68.66 ± 8.08 (SE) U/ml, respectively, P < 0.05], and no further increase occurred after additional 60 min of reperfusion. In the /R group, TNF- levels increased significantly after 30 min of reperfusion compared with baseline values (from 67.33 ± 2.90 to 94.66 ± 14.04 U/ml, respectively, P = 0.002), and the levels increased to even higher levels after 90 min of reperfusion (averaging 490.00 ± 261.28 U/ml).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Measurement of perfusate tumor necrosis factor- (TNF) levels in I/R (open bars) and /R (solid bars). Note that TNF- significantly increased after 30 min of reperfusion in both groups compared with controls (* P < 0.05), and TNF- levels were higher in the /R group compared with I/R group (@ P < 0.05). TNF- levels were greater in /R than in I/R ischemic model after both 30 and 90 min (@ P < 0.05); however, in the I/R group, amount of TNF- after 60 min of reperfusion did not increase above 30-min values.

Effects of TNF- in isolated rat lungs. Figure 3 shows that TNF- added to the perfusate of the lungs not exposed to I/R failed to alter the microvascular permeability as measured by the K_fc (TNF- control) over 135 min compared with no TNF- lungs. However, a slight but statistically greater increase (P < 0.05) in microvascular permeability occurred in I/R lungs that were pretreated with TNF- and subjected to 90 min of reperfusion (TNF-+I/R) compared with I/R alone.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Effects of exogenous administration of TNF- after baseline measurements on producing lung injury. Histograms show Kfc after 135 min for control in presence or absence of TNF-. Note that TNF- in lungs not exposed to I/R (designated as TNF- control) showed no damage over 135-min period and was no different than in lungs only perfused for 135 min. However, in I/R lungs, presence of exogenous TNF- significantly increased endothelial damage during I/R, as shown by TNF-+I/R. * Significant difference (P < 0.05) between I/R and TNF-+I/R compared with controls. # Significant difference (P < 0.05) between TNF-+I/R and I/R alone.

Effects of anti-TNF- antibody in I/R and /R models. Figure 4 shows that pretreatment of the animals with a goat anti-TNF- antibody significantly attenuated the increase in microvascular permeability (K_fc) present after 90 min of reperfusion associated with both the I/R (AbTNF-+I/R, where Ab is antibody) and /R (AbTNF-+/R) models of lung injury. Nonimmune goat IgG produced no inhibitory effect on I/R injury (Ab-contr+I/R). Because of the small amount of this antibody available for the study, nonimmune IgG was evaluated on the I/R model.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Effects of using a specific antibody (Ab) to TNF- on endothelial damage (Kfc) associated with I/R (AbTNF-+I/R) and /R (AbTNF-+ /R), compared with untreated lungs (I/R) and /R. Note that antibody totally blocked damage associated with both forms of ischemia. When an inactive antibody was used (Ab-contr+I/R), lung endothelium was not protected against I/R. * Significant difference (P < 0.05) compared with controls; @ significant difference between /R and I/R, respectively.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Although TNF- had not previously been measured in our models of I/R and /R in lungs, several publications indicate that it is released over relatively long periods of ischemia and results in alterations of vascular reresistance, endothelial damage, and the formation of pulmonary edema. Chang (6) has shown that pretreatment of rats with TNF- before lung isolation increased myeloperoxidase level and enhanced platelet-activating factor-induced vasoconstriction. This TNF- potentiation of platelet-activating factor-induced vasoconstriction is neutrophil independent but thromboxane dependent. Studies by Serrano et al. (34) show increases in ICAM-1, VCAM-1, and E-selectin in aortic endothelial cell cultures after 4 h of exposure to TNF-. Reignier et al. (28) have shown an increase in endothelial barrier permeability in blood-perfused rat lungs subjected to 1 h of ischemia followed by 1.5 h of reperfusion. This damage was prevented with 3'-sulfated Lewis, which blocks both E and L selectins. Knowles et al. (22) evaluated the effects of hypoxemia and reoxygenation plus TNF- on PMN phagocytosis and bacterical activity. During hypoxemia, PMN phagocytosis activity increased in the presence of TNF-, IL-8, and IL-1. Johnson and Ferro (19) have shown that guinea pig lungs perfused for 4 h after isolation and then challenged with TNF- produced more pulmonary edema through the formation of peroxynitrites, and nitrous oxide accelerated the process. Also, several studies have shown that visceral I/R injury causes TNF-- and IL-1-dependent lung injury in models incorporating occlusion of supraceleac area, bowel ischemia, hepatic ischemia, hindlimb ischemia, and mesenteric artery occlusion (7, 31, 33, 35, 40, 42).

Horgan et al. (15) challenged guinea pigs with LPS over a 2-h period and found that TNF- increased in lungs' effluent and more leukocytes were present in the lungs. When PMNs were activated with phorbol myristate and placed into the lung, an even greater accumulation of PMNs occurred in the lung, resulting in endothelial damage that was blocked by a TNF- antibody. Also, Bagby et al. (3) have shown that deaths due to LPS, but not due to peritonitis, were greatly decreased when an anti-TNF- antibody was used. Longer time periods of TNF- exposure have also been shown to produce damage in guinea pig and sheep lungs [5 h (16), 18 h (19), respectively] and in isolated endothelial monolayers (12 h) (23, 29).

Only two studies have been published on the role of TNF- in lungs subjected to some type of ischemia followed by reperfusion (10, 26). Palace et al. (26) produced ischemia in the left lung of a rabbit over a 24-h period. On reperfusion, TNF- increased to its maximal value after 45-60 min, and myeloperoxidase activity also increased, indicating that leukocyte sequestration also occurred in these lungs. Recently, Eppinger et al. (10) subjected in situ rat lungs to 1.5 h of ischemia followed by 4 h of reperfusion, and the transcapillary permeability of albumin (calculated as albumin clearance) was used as an estimate of endothelial damage. In these studies, the permeability increased after 30 min of reperfusion, decreased significantly at 1 h, and then continued to increase in an almost linear fashion for the next 3 h. In studies presented in the present paper, K_fc increased after 30 min of reperfusion and was further increased after an additional hour of reperfusion in both I/R and /R ischemia models. It should be emphasized that albumin clearance, which was used as an index of microvascular permeability in the study by Eppinger et al. (10), and the K_fc used in the present study measure two different membrane parameters: macromolecule permeability and solvent permeability of the pulmonary vascular system, respectively. Albumin clearance is a function of both transcapillary fluid flux and endothelial permeability, whereas vascular pressures do not affect K_fc measurements, which are, in fact, increased by known amounts to measure this membrane parameter.

It is interesting that the permeability index in the study by Eppinger et al. (10) actually decreased to lower levels after 1-h reperfusion compared with 30 min. This is an important finding, since it indicates that some endothelial protective effect may be present in the in situ lung I/R model compared with isolated lungs. However, the clearance measured in that study may be complicated by albumin flux increasing if pulmonary microvascular pressures increased in their model on reperfusion. Also, a larger accumulation of edema fluid at 1 h may decrease the albumin clearance calculation, since albumin flux is divided by wet lung weight to obtain the albumin transcapillary clearance. In studies conducted in our laboratory (18) in lungs, albumin clearance increases as transcapillary filtration increases. First, albumin clearance is nonlinear and then becomes almost linear at increasing filtration rates. If clearance decreases, it is required that the residual radioactivity accumulating over the first 30 min of reperfusion be cleared from the tissues either by lymphatics, by leaking of tissue fluid into the pleural cavity, or by albumin diffusing back into the plasma, which necessitates an uphill albumin transport. It is well known that elevated microvascular pressures increase transvascular fluid flux, which decreases as Starling forces readjust with the expansion of tissue spaces. Thus a rapid transvascular flux of albumin occurs into the tissues early at reperfusion when microvascular pressures are highest, followed by a decrease in transcapillary fluid flux with time, as Starling forces readjust. Radioactive albumin moves out of the plasma into the tissues of the total body at ~3%/h, so the gradient producing transvascular albumin movement into lung tissues has not changed significantly over the next 30 min. It is quite likely that fluid flux was higher at 1 h in the study by Eppinger et al. (10) because the Starling forces could no longer adjust, and the lung weight increase was much greater than the albumin flux, resulting in a decreased albumin clearance compared with the 30-min albumin clearance. Otherwise, one must somehow explain how the tissue albumin was removed. However, these arguments are only speculative, and future studies are necessary to evaluate this most interesting finding of an apparent decrease in vascular permeability occurring during reperfusion of in situ rat lungs.

We have shown that pulmonary vascular pressures increase in isolated rabbit (1) and dog (2) lungs reperfused with whole blood after a period of ischemia; however, in the saline-albumin perfused rat lung models used in the present study, the vascular changes occurring at reperfusion are very small. If microvascular pressure increased for in situ lungs, then protein clearance could increase and possibly wane as the vascular pressures decrease back toward normal levels. Another possibility for the differences between our studies and those of Eppinger et al. (10) is that reperfusion in the intact lung may cause the release of a ₂-adrenergic compound that would reverse the damage, as observed in our I/R studies (32). The difference could also be caused by isolating the lung in our study in our I/R models and perfusing it with a saline-albumin solution. Blood-perfused and saline-perfused lungs produce different amounts of nitric oxide that could affect both vascular permeability and vascular pressures (41). Obviously, TNF- is involved in both types of damage, yet we do not see reductions in endothelial damage in the ischemic models used in this study, once damage has occurred, without some type of intervention. Finally, it may be possible that 1.5 h of ischemia causes the release of different inflammatory mediators or alters their concentrations compared with 45 min of ischemia.

However, there are also similarities between the studies by Eppinger et al. (10) and our studies, e.g., the permeability at 30 min was significantly decreased after the use of an anti-TNF- antibody in their study, and this protective effect of a TNF- antibody on the endothelial damage occurred in both types of ischemic injury used in the present study after 90 min of reperfusion.

In our studies, TNF- in the perfusate increased after 30 min of reperfusion in both the I/R and /R 45-min ischemic models. After an additional 60 min of reperfusion, TNF- levels did not increase to higher levels in the I/R model but doubled in the /R model. These levels of TNF- correlated with the endothelial damage occurring in both models, and an antibody to TNF- totally blocked the damaging effect of both I/R and /R, whereas an inactive antibody produced no protection. However, when TNF- was added into the perfusate of control lungs not subjected to either I/R or /R conditions, no damage occurred over 3 h of perfusion. In contrast, when exogenous TNF- was introduced into the perfusate of lungs subsequently subjected to I/R injury, an even greater endothelial damage occurred, indicating that additional TNF- exposure during lung ischemia produced a potentiating effect.

The present study also indicates that air-ventilated ischemia is more injurious than nonventilated ischemia. This finding is consistent with studies of Eckenhoff et al. (9), which also showed that increased O₂ content in the ventilated gas mixture during ischemia produced more O₂ radical production in rat lungs. Also, in the present study, TNF- augmented microvascular injury associated with I/R, and even greater levels of TNF- were found in /R group after 90 min of reperfusion, compared with I/R lungs. These findings indicate that TNF- is also involved in the production of endothelial injury in these two different models of lung ischemic damage.

In summary, TNF- is released in both I/R and /R inflammatory models of endothelial damage. Thus TNF- release is required to produce the damage, most likely by upregulating ICAM-1 and P-selectin on endothelial cells and CD-18 on neutrophils (24), since both are known requirements in the sequence of events that must be present to produce the endothelial damage associated with the I/R and /R models used in the present study (25). However, other mechanisms may also be involved in the production of the I/R endothelial lung damage, in addition to TNF- in a time-dependent fashion, as indicated by the study of Eppinger et al. (10).