INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nitric oxide (NO), which is a key mediator of many regulatory processes, has profound effects on vascular tone and barrier function. In healthy organisms NO is produced by the so-called constitutive isoform of the nitric oxide synthase (NOS) that is also present in vascular endothelial cells. As a highly soluble gas NO can diffuse into vascular smooth muscle cells where it then activates soluble guanylate cyclase to produce cyclic guanosine monophosphate (cGMP), a potent second messenger of vasodilation. In addition, NO is involved in the maintenance of microvascular barrier function (1).

Evidence suggests that the constitutive NOS is absent during systemic inflammation (5). There is, however, a large amount of data supporting the hypothesis that NO may play a major role in the pronounced vasodilation that may lead to shock and mortality during sepsis. When compared with the physiologic rates of NO synthesis, manifold quantities of NO are present in rodents during states of systemic inflammation (6). This overproduction of NO is due to the synthesis of an inducible isoform of NOS in endothelial cells, vascular smooth muscle cells, and monocytes/macrophages. Several of the well-known stimuli of the inflammatory response have been identified as potent inducers of the production of NO (7). Elevated plasma levels of nitrites and nitrates, which may be the stable end products of the short-lived NO in patients with inflammatory processes (8, 9), are regarded as further evidence for a role of NO during systemic inflammation. NOS inhibitors have been administered to a small number of patients (10) and animals (13, 14) under conditions of systemic inflammation, where this treatment reversed the pronounced systemic vasodilation and decrease in blood pressure.

NOS inhibition, however, may elicit side effects on the pulmonary vasculature. The role NO plays in the pulmonary microvasculature is not yet well defined. On the one hand, some data suggest that reducing the overproduction of NO in the lung may have some beneficial effects: peroxynitrite, which is a strong oxidant, can be formed when NO reacts with superoxide (15). Peroxynitrite has been stained in human pulmonary tissue after lung injury (16). And in rodent models, the administration of NOS inhibitors prevented lung alveolar injury from smoke inhalation (3) and the endotoxin-induced acute pulmonary extravasation of Evans blue (17). On the other hand, NOS inhibition may result in injury to the pulmonary microvasculature. First of all, damage to the capillaries may merely be due to the hemodynamic effects of NOS blockade. Increased pulmonary artery pressure and pulmonary vascular tone after NOS inhibition have been demonstrated in different experimental models (18, 19). High hydrostatic pressures in the pulmonary microvasculature have clearly been shown to be able to cause stress failure of the microvascular barrier in several experimental models (20). Besides that, NOS inhibition may be associated with microvascular injury, because NO per se may be protective to pulmonary endothelial cells (4).

We performed two studies to investigate the pulmonary microvascular effects of NOS inhibition in sheep. The aim of the first experiment was to investigate lung microvascular changes after the administration of the NOS inhibitor N-nitro L-arginine methylester (L-NAME) in a large animal model of systemic inflammation with systemic vasodilation. We hypothesized that continuous administration of endotoxin would lead to a phasic response with early transient lung microvascular injury and late pulmonary vasodilation. NOS inhibition in this model would lead to pulmonary vasoconstriction and increased pulmonary microvascular permeability to proteins. In the second study the effect of the nonselective NOS inhibitor L-NAME (21) was investigated in healthy sheep because impaired microvascular barrier function after NOS inhibition had particularly been found in healthy animals where NOS inhibitors may have acted on the constitutive enzyme (2, 22). Again, we hypothesized that pulmonary endothelial permeability to protein would increase after NOS inhibition.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Adult ewes (n = 30, 43.8 ± 2.5 kg) were instrumented for two chronic studies under halothane anesthesia. The cutdown technique was used to insert polyvinyl catheters into the right femoral arteries and veins. The arterial catheter was further advanced into the thoracic aorta for continuous monitoring of mean arterial pressure (). A Swan-Ganz thermodilution catheter (Model 93A 131 7F; American Edwards Laboratories, Santa Ana, CA) was percutaneously inserted via the right jugular vein and advanced into the pulmonary artery to measure mean pulmonary arterial pressure (), pulmonary capillary wedge pressure (Ppcw), effective pulmonary capillary pressure (Pmv), central venous pressure (PCV) and to measure body core temperature.

The animals were then thoracotomized on the right side in the sixth intercostal space to cannulate the efferent vessel of the caudal mediastinal lymph node with silastic medical grade tubing (0.025 inches inner diameter [I.D.], 0.047 inches outer diameter [O.D.]; Dow Corning, Midland, MI) and to attach vascular occluders to the right pulmonary veins. Precautions were taken to avoid systemic contamination of the lung lymph: the distal part of the caudal mediastinal lymph node was ligated. In addition, the lymph vessels approaching this node from the diaphragm and the posterior aspects of the right hemithorax were stained with Evans blue and subsequently cauterized. The right hemithorax was then closed and the left hemithorax was opened in the fifth intercostal space. A silastic catheter (0.062 inches I.D., 0.125 inches O.D.; Dow Corning) was placed in the left atrium for the measurement of left atrial pressure (PLA), and vascular occluders were attached to all left pulmonary veins. For recovery from the operation the sheep were kept in metabolic cages for 7 d where they had free access to food and water.

Data Collection

Twenty-four hours before the experiment, the catheters were connected to pressure transducers (P23 ID; Statham Gould, Oxnard, CA) with continuous flushing devices. Zero calibration of the catheters was performed at the level of the elbow joint while the animals were standing. All hemodynamic variables were later recorded with the animals standing. Blood pressures and heart rate were read from physiological recorders (Honeywell OMJ9; Electronics for Medicine, Pleasantville, NY). Pmv was derived from Ppcw tracings according to the technique of Holloway and colleagues (23). Pulmonary thermal dilution cardiac output (CO) was measured with a CO computer (model 9520 or 9530; American Edwards Laboratories, Santa Ana, CA). Cardiac index (CI) and systemic and pulmonary vascular resistance indexes (SVRI and PVRI, respectively) were calculated using standard formulas. Body surface area was derived from body weight according to Guyton and colleagues (24).

PLA was set at Ppcw level at baseline. PVRI and its distribution between a precapillary and a postcapillary component was then calculated according to the following equations:

(1)

(2)

(3)

Lung lymph was collected over a period of 15 min in graduated cylinders, and lung lymph flow (L) was calculated. Samples of lung lymph and arterial plasma were transferred to heparinized tubes and centrifuged. The supernatants were frozen at 80° C and later used to measure lymph protein concentration (CLP), plasma protein concentration (CPP), and albumin concentrations as well as colloid osmotic pressures. The Biuret technique was used to determine total protein concentration. Albumin concentrations were measured by high pressure liquid chromatography. Osmotic pressures were measured through the semipermeable membrane of a colloid osmometer (Model 4100; Wescor, Logan, UT) with a pore diameter of 0.0004 mm.

Permeability Analysis

The vascular protein clearance (VPC) was calculated as shown in Equation 4.

(4)

The reflection coefficients to total protein (_P) and albumin (_Alb) were determined at 48 h and at the end of either experiment. Sigma was determined using the technique of pulmonary venous occlusion that was previously described by this group (25). In short, all pulmonary venous occluders were inflated with saline to increase Pmv. As a result, L rose and CLP decreased with CPP remaining stable. All variables were measured every 30 min during venous occlusion. The occlusion was maintained until a further rise in L was not accompanied any more by a decline in CLP/CPP, for at least 2 h. This minimal CLP/ CPP is assumed to be filtration-independent and can be used to calculate _P and _Alb using Equations 5 and 6.

(5)

(6)

Experimental Protocol

In the first experiment, 20 animals were randomly assigned to one of two groups: endotoxin (n = 10) and endotoxin + L-NAME (n = 10). After baseline data had been collected all sheep were started on a continuous infusion of Escherichia coli endotoxin (10 ng/kg/min) in NaCl 0.9%, which was maintained for 35 h. After 24 h of endotoxemia, 10 animals received a bolus injection of the NO synthase inhibitor L-NAME in NaCl 0.9%, whereas the other group was given an equivalent volume of saline. The last 3 h of this experiment were used to elevate the pulmonary transvascular fluid filtration by pulmonary venous occlusion in order to measure .

In the second experiment, 10 animals were assigned to one of two groups: NaCl (n = 4) and L-NAME (n = 6). After collection of baseline data six animals received a bolus of L-NAME in NaCl 0.9% (25 mg/kg), and four animals were given an equivalent volume of NaCl 0.9%.

Data Analysis

Results are presented as means ± SEM. Differences from 0 h were tested by analysis of variance (ANOVA) for repeated measures with post hoc Dunnett's test. The Mann-Whitney U test was performed to test whether there were statistically significant differences between groups during the last 8 h of either experiment. Differences were regarded as statistically significant at p < 0.05.

The experiments were approved by the Animal Care and Use Committee of the University of Texas Medical Branch.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results in Experiment 1 are presented in two parts; the first describes the characteristic pattern of the systemic inflammatory response, focusing on the changes in the lung, and the second is concerned with the effects of L-NAME.

The Systemic Inflammatory Response during the First 24 Hours of Endotoxemia

Systemic vasculature. Early changes in the systemic circulation were characterized by a trend to systemic vasoconstriction (Figure 1). This was followed by pronounced systemic vasodilation and a decrease in by 24 h with a consecutive rise in CI.

View larger version (25K): [in this window] [in a new window]   Figure 1.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. SVRI = systemic vascular resistance index; CI = cardiac index; MAP = mean arterial pressure.

Pulmonary vasculature. After an initial peak at 1 h, remained significantly elevated (Figure 2). Pmv was significantly increased at 1 and at 4 h (p < 0.05 versus 0 h) when Ppa was highest. PVRI was significantly elevated both at 1 and at 4 h, and this increase was caused by elevations of both the precapillary and the postcapillary components of PVRI (Figure 3). PVRI remained at baseline level thereafter.

View larger version (30K): [in this window] [in a new window]   Figure 2.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. MPAP = mean pulmonary artery pressure; Pmv = effective pulmonary capillary pressure.

View larger version (25K): [in this window] [in a new window]   Figure 3.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. PVRI = pulmonary vascular resistance index.

L was elevated throughout the first 24 h in both groups (Figure 4). The gradient between CPP and CLP showed an initial rise at 1 h (p < 0.05). This was due to a decrease in CLP (p < 0.05) (Figure 5) during an elevated L. At 4 h, lymph flow remained elevated, but the gradient between CPP and CLP declined markedly to levels that were then even lower than at 0 h (p < 0.05) (Figure 4). The decrease in the gradient was caused by a novel increase in CLP to baseline levels, whereas CPP declined (p < 0.05) (Figure 5). Hematocrit was elevated during the decline in CPP at 4 h. Pulmonary VPC was elevated between 1 and 16 h (p < 0.05 versus 0 h) with a steep rise until 4 h and a subsequently gradual decrease (Figure 4).

View larger version (22K): [in this window] [in a new window]   Figure 4.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. L = lung lymph flow; CPP-LP = gradient of plasma and lymph protein concentrations; VPC = vascular protein clearance.

View larger version (23K): [in this window] [in a new window]   Figure 5.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. CLP = lymph protein concentration; CPP = plasma protein concentration.

Effects of L-NAME during the Hyperdynamic Phase of Endotoxemia

Systemic vasculature. SVRI was low and CI was high in endotoxin throughout the last 8 h of the experiment (p < 0.05 versus 0 h) (Figure 1). remained low until 28 h (p < 0.05 versus 0 h). After injection of L-NAME, SVRI was transiently elevated until 26 h (versus 0 h) and was then no longer different from baseline. Cardiac index was normalized and was elevated (p < 0.05 versus 0 h). Both SVRI and were higher and CI was lower in endotoxin + L-NAME than in endotoxin.

Pulmonary vasculature. remained elevated from baseline in both groups between 24 and 32 h, but it was even further increased after L-NAME versus endotoxin alone (Figure 2). Pmv became transiently elevated at 24 and 25 h in endotoxin + L-NAME (p < 0.05 versus 0 h) and was also once significantly elevated in endotoxin at 26 h. Pmv did not differ between groups. PVRI and its precapillary and postcapillary components were elevated after the administration of L-NAME (p < 0.05 versus 0 h and endotoxin) (Figure 3).

L remained elevated (p < 0.05 versus 0 h) in endotoxin and in endotoxin + L-NAME except at 28 h, when the elevation in L in endotoxin + L-NAME was no longer statistically significant (Figure 4). L was not different between the two groups between 24 and 32 h. The gradient between CPP and CLP was significantly elevated after the administration of L-NAME between 25 and 28 h (p < 0.05 versus 0 h). The gradient was also higher in the L-NAME-treated animals than in the endotoxin-treated animals. This was due to a higher CPP in the L-NAME-treated animals than in the endotoxin-treated ones (Figure 5). Hematocrits showed a similar trend as CPP with the changes, however, not reaching statistical significance. Pulmonary VPC had normalized by 24 h in both groups and did not differ between the two groups thereafter (Figure 4). Both the osmotic reflection coefficients to total protein and to albumin were normal before and after endotoxemia in both groups (Figure 6).

View larger version (51K): [in this window] [in a new window]   Figure 6.   All animals received endotoxin for 32 h. After 24 h, one group (endotoxin + L-NAME, n = 10) received a bolus injection of L-NAME (25 mg/kg), whereas the other group (endotoxin, n = 10) was given the carrier. Endotoxin: *p < 0.05 versus 0 h; endotoxin + L-NAME: p < 0.05 versus 0 h; p < 0.05 for a difference between the two groups during the last 8 h of the experiment. Sigma Prot. = lung osmotic reflection coefficient to total protein; Sigma Alb. = lung osmotic reflection coefficient to albumin.

L-NAME in Healthy Sheep

Systemic vasculature. The administration of L-NAME to healthy sheep resulted in systemic vasoconstriction and a decrease in CI throughout the observation period (Table 1). Both SVRI and CI were different between the two groups (p < 0.05). was elevated in the L-NAME-treated sheep until 4 h (p < 0.05 versus 0 h). , however, was not different between the two experimental groups.

Pulmonary vasculature. There was a transient increase in Ppa and Pmv versus 0 h, which reached statistical significance 30 min after L-NAME had been injected. Both Ppa and Pmv were higher in L-NAME than in NaCl (p < 0.05). PVRI as well as its precapillary and postcapillary components were higher in L-NAME than in NaCl. The increases from baseline of PVRI and postcapillary PVRI were still statistically significant at 2 h, whereas precapillary PVRI was only significantly elevated at 0.5 h. There was no obvious effect of L-NAME on lung L, VPC, the gradient of CPP and CLP or the reflection coefficients to total protein and albumin (Figure 6).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main findings in this study were as follows. (1) The systemic vascular response to the continuous infusion of endotoxin was characterized by early hemodynamic instability and the subsequent development of a hyperdynamic circulatory pattern with systemic vasodilation and elevated blood flow. (2) According to the systemic changes the pulmonary microvasculature reacted in three phases: after intense pulmonary vasoconstriction at 1 h, transient acute lung injury with increased endothelial permeability to large molecules occurred at 4 h and was followed by normalization of the endothelial barrier function to proteins during the hyperdynamic phase. (3) L-NAME increased SVRI and PVRI in healthy sheep and during systemic inflammation without increasing pulmonary microvascular permeability.

This model has repeatedly been studied (14, 26), particularly because of the late hyperdynamic response that mimics the reaction to endotoxin in healthy volunteers (27) and the hemodynamic state in septic patients (28). Less attention, however, has been paid to the phasic changes that occur in the pulmonary microvasculature.

The Phasic Response of the Pulmonary Microvasculature to Systemic Inflammation

The response can be divided into three phases. The first phase can be attributed to the liberation of thromboxane and is characterized by pulmonary vasoconstriction with increased and Pmv. Thromboxane has been shown to rise at 1 h in this model (14), and blockade of the thromboxane synthetase has attenuated pulmonary vasoconstriction after the injection of a bolus of endotoxin (29). L rose at 1 h and CLP declined. Together with the increase in the gradient between CPP and CLP this indicates a primarily hydrostatic origin of the elevated lymph flow.

During the second phase, pulmonary vasoconstriction, elevated , Pmv, and L were still present, but the quality of the lymph changed; there was now evidence of an increase in pulmonary microvascular permeability. The decrease in the gradient of CPP and CLP concomitant with a decrease in CPP alone would not be sufficient to prove an increase in permeability to large molecules. The fact, however, that the decrease in the gradient occurred simultaneously with a novel increase in CLP at declining CPP and stable L is regarded as a consequence of a transient increase in pulmonary endothelial permeability to large molecules. These phenomena also resulted in a maximal clearance of plasma protein through the lung lymph (VPC) at 4 h. The early increase in hematocrit is at least in part due to a concentration effect; fluid balance was negative during the first 4 h of endotoxemia in this model (data not published). This confirms that the decline in CPP may indicate leakage of protein into the interstitial space rather than an effect of dilution. The sheep, however, were not splenectomized and therefore it cannot finally be excluded that splenic contraction may have contributed to the increase in hematocrit, even if the animals were not in shock during the first 4 h of the experiment. Increased pulmonary endothelial permeability to large molecules had been reported approximately 4 h after a single bolus of endotoxin in sheep (30).

The characteristic changes of the third phase of the pulmonary microvascular response are best documented after 24 h of endotoxemia and thereafter and include persistent elevations in and L in the presence of an already normalized PVRI. The high L had been found to be due to an increased fluid filtration with the permeability to protein being at baseline level after 24 h of endotoxemia (33). Reflection coefficients not only to total protein but also to the small-sized fraction of serum albumin were at baseline level at 32 h in this study. Accordingly, pulmonary VPC had normalized by that time.

In conclusion, the continuous administration of endotoxin to sheep, first introduced by Morel and colleagues (34), constitutes a model in which three segments of the pulmonary inflammatory response can be identified. The two early segments, including the transient acute lung injury, correspond to the two phases that have been described after a bolus injection of endotoxin or bacteria. The third segment may represent the period when usually systemic inflammation becomes apparent in patients because of systemic vasodilation. At that time pulmonary microvascular permeability to large molecules had normalized in this model.

Pulmonary microvascular permeability was only transiently elevated in our model of systemic inflammation. There is evidence that this phenomenon may not be confined to our model. Early pulmonary endothelial barrier dysfunction was induced in previous investigations (30, 31). We are not aware of experimental models, however, in which prolonged periods of increased pulmonary microvascular permeability could reproducibly be induced by the inflammatory cascade of sepsis despite the large scale of models in which sepsis is studied. Nevertheless, sepsis is still one of the most significant risk factors for the development of the acute respiratory distress syndrome (ARDS) (35, 36). Only 40 to 50% of patients with the clinical syndrome of systemic inflammation that characterizes sepsis, however, develop ARDS (35, 36), and only a fraction of these patients with sepsis-associated ARDS have an increase in alveolar-capillary permeability (37). Such a heterogeneous response of the pulmonary microvasculature to the inflammatory cascade of sepsis does not seem very logical. As a consequence research should focus on possible other factors that may be etiologically responsible for long-lasting impaired barrier function in sepsis-associated ARDS. Stress failure of pulmonary capillaries with mechanical damage to the endothelium, basal membrane, and even pulmonary epithelium may occur in this patient population and may be due to aggressive strategies of ventilation, or strongly elevated microvascular pressures (20). Aspiration of gastric acid, pneumonia, or reperfusion effects after prolonged periods of hypoperfusion may be further etiologies contributing to sustained pulmonary barrier dysfunction.

L-NAME during Systemic Inflammation

The rationale for administering NO synthase inhibitors during conditions like sepsis would be to reverse systemic vasodilation as seen in our hyperdynamic model of endotoxemia. The dose of L-NAME used here was selected according to a previous study in which it reversed systemic vasodilation without selective impairment of regional blood flows (14).

The pulmonary vasculature reacted to L-NAME with vasoconstriction. The interpretation of PVRI is limited in the low pressure vasculature of the lung where may be more influenced by blood volume than changes in pulmonary blood flow. An increase in the calculated PVRI can therefore be due either to true pulmonary vasoconstriction or to reduced cardiac output in the absence of a proportionate decrease in , in which case pulmonary vasoconstriction may not even occur. Increased in the presence of decreased cardiac output, however, clearly indicated a state of vasoconstriction in the endotoxin + L-NAME animals.

The increase in PVRI induced by L-NAME occurred both on the precapillary and the postcapillary side of the pulmonary microvasculature. NO induces the formation of cGMP. Elevated cGMP levels have been found in both pulmonary arteries and veins of septic sheep during the third phase of the inflammatory response (38). The higher postcapillary PVRI may have contributed to the transient elevation in Pmv after NOS inhibition. The extent of changes in Pmv, however, did not reach statistical significance compared with the endotoxin control.

We did not observe changes in pulmonary endothelial permeability after the administration of L-NAME. There was even an increase in the gradient between CPP and CLP after L-NAME. This was accompanied by an increase in CPP and a trend of hematocrits into the same direction in the L-NAME-treated sheep. Again, the increase in CPP was most likely due to an effect of fluid loss from the intravascular space; this would be consistent with the finding that the increasingly positive fluid balance was counteracted by an increase in urine output after L-NAME in this model (26).

In summary, L-NAME did not impair barrier function or oxygenation in this model of systemic inflammation. Even after the measurement of the reflection coefficients, which includes the induction of pulmonary edema by pulmonary venous occlusion, the sheep recovered well. In the presence of ARDS with sustained permeability changes, however, the L-NAME-induced pulmonary hypertension may bear a significant risk for worsening pulmonary edema and oxygenation.

L-NAME in Healthy Sheep

Similar to the experiments in endotoxemic sheep the administration of the equivalent dose of L-NAME to healthy sheep resulted in systemic and transient pulmonary vasoconstriction. L-NAME is a nonselective NOS inhibitor that has similar affinities for the inducible and the constitutive NOS (21). The fact that pulmonary endothelial permeability was not even increased in healthy animals suggests that NO does not play an important role in the maintenance of proper pulmonary endothelial barrier function in healthy sheep or during endotoxin-induced systemic inflammation.

In a rodent model of early acute lung injury the administration of a NOS inhibitor even ameliorated the injury (17). Given before the systemic inflammatory response induced by ischemia and reperfusion the extravasation of Evans blue dye after reperfusion was increased by NOS inhibition (39). The increase in extravasation, however, may have occurred even if the NOS inhibitor elevated only the pulmonary hydrostatic pressure under the condition of an impaired barrier function. Still the underlying mechanisms leading to NO-related increased permeability in one but not in another model are unclear. Factors such as different species investigated, the vascular bed- systemic versus pulmonary, and the presence of a particular initial insult may influence the effect of NOS inhibition on the microvascular barrier.

In summary, the continuous administration of endotoxin to sheep resulted in a typical inflammatory response. The pulmonary response was characterized by early transient lung injury with increased pulmonary endothelial permeability to large molecules and a late hyperdynamic phase with normal pulmonary barrier function to large molecules. Endogenous NO production is involved in the regulation of pulmonary microvascular tone both in healthy sheep and during the hyperdynamic phase of systemic inflammation, but it does not seem to play a significant role in pulmonary microvascular barrier function under these conditions. Further studies are needed to elucidate possible side effects of NOS inhibition in cases of systemic inflammation when it is combined with severe cases of ARDS and persistent impairment of the endothelial barrier. ARDS with long-lasting impairment of the microvascular barrier function, however, seems to be a clinical entity different in etiology from the acute lung injury of sepsis and systemic inflammation.