INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Acute lung injury (ALI) frequently arises after severe infections as bacterial endotoxin is released into the blood stream (1). It is still unsettled in what direction this condition is modified by endogenous nitric oxide (NO). Produced from the amino acid L-arginine by different isoforms of NO synthase (NOS), NO stimulates soluble guanylate cyclase, resulting in intracellular accumulation of cyclic guanosine 3'-5'monophosphate (cGMP). The synthesis of small amounts of NO in the vascular endothelium, catalyzed by a constitutive calcium- dependent isoform (eNOS), is regarded as a key regulator of vascular tone and integrity, and of regional blood flow (2). From 2 to 12 h after exposure to inflammatory stimuli such as cytokines and endotoxin, upregulation of an inducible calcium-independent NOS isoform (iNOS) starts in various organs, including the lungs. Produced in excess by iNOS, NO may cause circulatory shock, microvascular damage, and multiple organ failure. The mechanism of NO-induced tissue injury includes reaction with superoxide to form the potent oxidant peroxynitrite, and activation of cyclooxygenases (3). On the other hand, physiologic NO formation by eNOS is believed to protect against endotoxin-induced damages by scavenging superoxide and by counteracting aggregation of leukocytes and platelets (2, 9). Stimulation of cGMP production is suggested to enhance the barrier function of endothelial cells (10). Recent investigations also report favorable effects of low concentrations of inhaled NO on lung microvascular pressure and permeability in endotoxemic sheep and in patients with ALI (11, 12).

Nonselective inhibition of NOS has been shown to cause an additional increase in pulmonary hydrostatic pressure (13) and aggravation of pulmonary edema (17). Because damage to microvascular barriers could be caused by pulmonary hypertension per se, selective inactivation of iNOS eventually could be favorable because NO production still prevails by eNOS. Aminoguanidine (AG), containing the guanido-group of L-arginine linked to hydrazine, in vitro, displays 10- to 100-fold higher potency as inhibitor of iNOS than of eNOS (18). In endotoxemic rodents and dogs, AG suppresses activation of iNOS and peroxynitrite production in the lungs and decreases plasma levels of the stable NO metabolites, nitrites and nitrates (NO_X). Moreover, AG reduces lung edema, improves gas exchange, and increases survival by counteracting circulatory failure (4). So far, the influence of AG on dynamic changes in lung fluid filtration and pulmonary hemodynamics has not been specifically addressed.

After endotoxin (lipopolysaccharide) (LPS), sheep respond with respiratory distress and increments in lung vascular resistance, lymph flow and protein clearance, the latter as a consequence of enhanced lung microvascular pressure and permeability (15, 19, 20). Our aim was to investigate whether AG administered from 2 h after LPS influences pulmonary hemodynamics and fluid filtration and improves gas exchange in awake sheep.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The experiments were approved by the Norwegian Experimental Animal Board. Eighteen yearling sheep of both sexes weighing 34.0 ± 1.5 kg (mean ± SEM) were instrumented under endotracheal anesthesia with halothane 0.8 to 1.25% (Zeneca, Cheshire, UK) as a modification of a method introduced by Staub and colleagues (21). Through a right thoracotomy, the efferent duct of the caudal mediastinal lymph node was cannulated with a medical-grade catheter (602-155; Dow Corning, Midland, MI). To prevent contamination with nonpulmonary lymph, all lymphatic tissue traversing the diaphragm was interrupted with ligating clips (LS-200; Ethicon, Somerville, NJ) via a second right thoracotomy. Via a left thoracotomy, medical-grade catheters were implanted into the pulmonary artery and the left atrium. After surgery, the animals were allowed 6 to 7 d of recovery. Twelve sheep had open and well-functioning lymph catheters at the end of the recovery period. The day before the experiment began, a 16-G catheter (1715-2; Ohmeda, Swindon, UK) and a 5-Fr introducer (CP-07511-P; Arrow International, Reading, PA) were inserted into the left common carotid artery, and an 8.5-F introducer (CC-350B; Baxter Healthcare, Irvine, CA) was placed in the ipsilateral external jugular vein under intravenously administered anesthesia with ketamine 10 mg/kg (Nycomed, Oslo, Norway).

The following day, the awake sheep was placed in an experimental pen. A 4-Fr fiberoptic thermistor catheter (PV2024L; Pulsion Medical Systems, München, Germany) was advanced into the thoracic aorta. A 7-Fr flow-directed thermal dilution catheter (131HF7; Baxter Healthcare) was introduced into the pulmonary artery. The catheters were connected to pressure transducers (Transpac III; Abbott Critical Care Systems, North Chicago, IL) and continuously flushed with heparin 10 IU/kg/h (Nycomed). Hemodynamic and lung lymph flow (L) measurements were carried out at 30-min intervals. Mean systemic arterial pressure (), mean pulmonary artery pressure (), pulmonary artery occlusion pressure (Ppao), and mean left atrial pressure () were displayed on a 565A Patient Data Monitor (Kone, Espoo, Finland) and recorded on a 79 Polygraph (Grass Instruments, Quincy, MA) with the zero reference level at the shoulder of the front leg of the standing animal. Effective pulmonary capillary pressure (Pc) was derived from the Ppao tracing according to the technique of Holloway and colleagues (22).

Cardiac index (CI), extravascular lung water content (EVLW), and pulmonary blood volume index (PBVI) were determined by using a thermal-dye dilution technique (23), as assessed by a Cold Z-021 (Pulsion Medical Systems). Every value was calculated as a mean of 5 measurements, each consisting of a 5-ml bolus of indocyanine green 0.5 mg/ ml (Pulsion Medical Systems) in ice-cold 5% glucose injected into the right atrium. Total pulmonary vascular resistance index (PVRI) was calculated as (  )/CI, upstream resistance index (PVRI_UP) as (  Pc)/CI, downstream resistance index (PVRI_DWN) as (Pc  )/CI, and systemic vascular resistance index (SVRI) as /CI.

Blood samples were drawn at hourly intervals from the systemic artery (a) and pulmonary artery (v) lines and analyzed for pH, PO₂, and PCO₂ (Ciba-Corning 288 Blood Gas System; Corning Medical, Medfield, MA), as well as oxygen saturation (SO₂) and hemoglobin (OSM3 Hemoximeter; Radiometer, Copenhagen, Denmark). Alveolar-arterial oxygen tension difference (AaPO₂), oxygen delivery (O₂), oxygen consumption (O₂), and venous admixture (S/T) were calculated using standard equations (11). In addition, samples were collected in heparinized tubes (Venoject VT-100SHL; Terumo, Leuven, Belgium) for analysis of plasma and lung lymph protein and plasma NO_X, and in EDTA tubes (Vacutainer 367655; Becton Dickinson, Meylan Cedex, France) for analysis of plasma cGMP. Immediately after sampling, blood was centrifuged at 4° C (2,000 g for 10 min). The plasma and lymph samples were stored at 70° C.

After 2 to 2.5 h of stable hemodynamic and L baseline measurements, 18 sheep (including 12 lymphing sheep) received LPS 1 µg/kg (Escherichia coli O26:B6; Sigma Chemical, St. Louis, MO) dissolved in 30 ml of NaCl 0.9% and infused intravenously for 20 min from time 0. Two hours later, sheep were randomly assigned either to receive intravenously a bolus of AG 10 mg/kg (aminoguanidine bicarbonate; Sigma Chemical) followed by infusion of 1 mg/kg/h (LPS/AG group), or a corresponding volume of NaCl 0.9% (LPS group) for the remaining 4 h of the experiment. After 1 wk of recovery, five sheep (including four lymphing sheep) were exposed to AG alone for the same period of time (AG group), and five animals received intravenous injections of AG in cumulative doses of 1, 5, 10, and 20 mg/kg at 1-h intervals. After the final measurements, the animals were killed.

Protein concentrations were determined with reagents from Bio-Rad Laboratories (München, Germany) and bovine serum albumin (Sigma Chemical). Lung lymph-to-plasma protein concentration ratio (L/P) and lung lymph protein clearance (CL = L × L/P) were calculated. The concentration ratio of interstitial fluid to plasma protein has been shown to fall hyperbolically with increasing net fluid filtration and lymph flow, according to the equation L/P = PS/(PS + L), where PS is the permeability-surface area product for protein. A prerequisite for this assumption is that transcapillary protein transport is purely dissipative (i.e., diffusion or vesicle transport is proportional to the transcapillary protein concentration difference) (24). We calculated mean PS at baseline, at 2 h, and from 3 through 6 h of endotoxemia. Iso-PS lines were drawn by keeping the estimated PS at the selected time intervals constant while varying L between 0 and 40 ml/h. The derived L/P values were plotted against L (11).

Plasma samples or standards (NaNO₃ in distilled water) for NO_X analysis were incubated in 96-well microassay plates for 3 h at room temperature with nitrate reductase from Aspergillus species and NADPH. After incubation, the Griess reagent (1% sulphanilamide + 0.1% N-(1-naphthyl)-ethylenediamine in 2.5% phosphoric acid) was added. The plates were incubated for a second time for 10 min at room temperature. Absorbances were measured at 540 nm using a microplate reader (25). Nitrate reductase, NADPH, and Griess reagent were purchased from Sigma Chemical. The plasma concentrations of cGMP were determined by an enzyme immunoassay (Biotrak RPN226; Amersham International, Buckinghamshire, UK).

Data are expressed as mean ± SEM and analyzed by a repeated measure ANOVA, followed by, when appropriate, Bonferroni's correction for multiple comparison. A p value of < 0.05 was regarded as statistically significant, or otherwise, not significant (NS).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As depicted in Table 1, hemodynamics changed after LPS without intergroup differences until AG was superimposed. Then, the LPS-induced rise in SVRI was enhanced because CI declined (p < 0.05) at unchanged . Moreover, and Pc increased by as much as 28 and 26%, respectively, and PVRI by as much as 60% above that of the LPS group (p < 0.04). ascended slightly, albeit without significant intergroup difference (not shown). PVRI_DWN rose by as much as 72% above that of the LPS group, from 3.5 h through 6 h, and PVRI_UP correspondingly increased by 56% at 6 h (p < 0.05), tending to reduce the PVRI_UP-to-PVRI_DWN ratio (PVRI_UP/ PVRI_DWN) (NS). In addition, PBVI decreased (p < 0.05). In the AG group, AG enhanced and Pc to above baseline beginning at 3 h and continuing throughout the experiment, whereas PVRI and PVRI_UP increased transiently (p < 0.05). When AG was administered in cumulative doses of 10 and 20 mg/kg, SVRI, , Pc, PVRI, PVRI_UP, and PVRI_DWN all increased (p < 0.05), as demonstrated in Table 2. Moreover, all cumulative doses of AG reduced PVRI_UP/PVRI_DWN (p < 0.05). We also observed a tendency for CI to decrease and for to increase with a dose of 20 mg/kg (p = 0.08).

It can be seen in Figure 1 that EVLW increased approximately twofold and stayed above baseline from 0.5 h after LPS (p < 0.05), displaying no intergroup difference for the remainder of the experiment. Thirty minutes later, L and CL started to gradually increase sixfold and remained above baseline for the duration of the experiment (p < 0.05), as depicted in the upper and lower panels of Figure 2. AG caused additional increments in L and CL from 5 through 6 h, peaking above the LPS group at 6 h by 31 and 39%, respectively (p < 0.05), and leaving L/P unchanged. In the AG group, L, L/P, and CL demonstrated no intragroup differences (not shown). The relationship between the mean L/P and mean L, combined with the calculated mean iso-PS lines at baseline, at 2 h, and from 3 through 6 h are depicted in Figure 3. Iso-PS lines at baseline were different from those at and after 2 h (p < 0.05). AG shifted the L/P-L relationship to the right when compared with the time-corresponding LPS group value (p = 0.18).

View larger version (14K): [in this window] [in a new window]   Figure 1.   The effect of aminoguanidine (AG) on extravascular lung water (EVLW) in awake, endotoxemic sheep. LPS/AG = endotoxin-aminoguanidine group; LPS = endotoxin group. Data are expressed as mean ± SEM, n = 9. *p < 0.05 from intragroup baseline in both groups.

View larger version (13K): [in this window] [in a new window]   Figure 2.   The effect of aminoguanidine (AG) on lung lymph variables in awake, endotoxemic sheep. LPS/AG = endotoxin-aminoguanidine group; LPS = endotoxin group; L = lung lymph flow; L/P = lymph-to-plasma protein concentration ratio; CL = lymph protein clearance (CL =  L × L/P). Data are expressed as mean ± SEM, n = 6. *p < 0.05 from intragroup baseline in both groups; p < 0.05 between the groups.

View larger version (14K): [in this window] [in a new window]   Figure 3.   The relationship between mean lymph-to-plasma protein concentration ratio (L/P) and mean lung lymph flow ( L) at baseline (0 h), at 2 h and from 3 to 6 h (presented together) after endotoxin in awake sheep. LPS/AG = endotoxin-aminoguanidine group; LPS = endotoxin group. The corresponding mean permeability-surface area product iso-lines, dashed for the LPS/AG group and solid for the LPS group, are superimposed. Data are expressed as mean ± SEM, n = 6. *p < 0.05 from intragroup baseline in both groups.

As shown in Figure 4, Pa_O2 increased and AaPO₂ decreased, from 4 to 6 h in the LPS/AG group (p < 0.04). Moreover, s/ T decreased from 3 to 6 h (p < 0.03). Arterial oxygen saturation also increased from 3 to 4 h in the LPS/AG group (not shown) (p < 0.05), whereas no intergroup differences were noted in Sv_O2, O₂, O₂, and Pa_CO2 (not shown). In the AG group, O₂ and O₂ declined transiently at 3 h (p < 0.05), whereas other gas exchange variables remained unchanged (not shown).

View larger version (15K): [in this window] [in a new window]   Figure 4.   The effect of aminoguanidine (AG) on pulmonary gas exchange in awake, endotoxemic sheep. LPS/AG = endotoxin-aminoguanidine group; LPS = endotoxin group; PaO2 = arterial oxygen tension; AaPO2 = alveolar-arterial oxygen tension difference; S/ T = venous admixture. Data are expressed as mean ± SEM, n = 9. *p < 0.05 from intragroup baseline in both groups; p < 0.05 between the groups.

Plasma NO_X concentrations (Table 3) increased gradually above baseline by a maximum of approximately 30% in both endotoxemic groups, starting at 1 h (p < 0.01) and displaying no intergroup difference. In the AG group, NO_X remained unchanged, but a rise was seen after AG was administered in cumulative doses of 10 and 20 mg/kg (p < 0.03), as demonstrated in Table 2. Plasma cGMP concentrations (Table 3) rose markedly in both endotoxemic groups (p < 0.03). From 3 to 4 h, cGMP was slightly higher in the LPS/AG group (NS).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study revealed that, in sheep, the endotoxin-induced increase in EVLW remained unchanged after administration of AG, despite additional rises in pulmonary capillary pressure and, probably, also in permeability. Apparently, the enhanced lung fluid filtration after AG was balanced by a simultaneous increase in lung lymph drainage. In addition, AG improved gas exchange.

After exposure to endotoxin, activation of iNOS may occur in the lungs, leading to microvascular injury with plasma leakage (3). Nonselective NOS inhibitors such as N-nitro-L-arginine methyl ester (L-NAME) and N-monomethyl-L-arginine (L-NMMA) may further aggravate endotoxin-induced pulmonary hypertension and increase EVLW in sheep (13, 17), whereas AG may reduce lung edema, as determined by wet-to-dry weight ratio and extravasation of a tracer dye in rodents and dogs (4). In the present investigation, AG caused additional increments in PVRI and Pc without altering EVLW. The surprising finding that during AG, PVRI_DWN rose above that of the endotoxemic controls 2.5 h earlier and more markedly than PVRI_UP, may have contributed to the increase in Pc. This observation is consistent with the notion that in ovine septicemia, iNOS is expressed more in the venous than in the arterial segment of the pulmonary circulation (26). In contrast, L-NAME causes a proportionate increase in PVRI_UP and PVRI_DWN (14, 15, 17). The observed increase in SVRI and further reduction in CI after AG is in agreement with other reports on inhibition of NOS in sheep (13). We also found that pulmonary vasoconstriction occurred in animals receiving AG alone, and that cumulative doses of 10 mg/kg or above additionally caused changes in the systemic circulation. In rodents, AG suppresses endogenous release of proinflammatory prostanoids after endotoxin (6, 8). On the basis of the latter observations, it is unlikely that AG induces pulmonary hypertension by activating the cyclooxygenase pathway. A recent study has shown that at least in rat intestinal tissue, the selectivity of AG for iNOS is only twofold higher than for eNOS (27). Thus, the present findings support the contention that in vivo, AG most likely exerts a dose-dependent inhibition of eNOS, in addition to that of iNOS.

In sheep, pulmonary microvascular permeability increases from approximately 4 h after start of endotoxin and subsequently returns towards baseline during the hyperdynamic phase, as reported in previous studies (15, 19). Permeability can be evaluated by observing changes in calculated PS, provided surface area remains unaltered (24). Under the influence of AG, we noticed a tendency for a rightward shift of both the L/P-L relationship and the iso-PS lines. Although PVRI_DWN increased, the reductions in CI, PBVI, and S/T make vascular recruitment, and thus increased surface area, an unlikely explanation for the rightward displacement of PS. More likely, AG decreased the surface area available for filtration, leaving L/P unchanged at simultaneous 30% rises in L and CL. This indicates that during the phase of increased microvascular leakage, AG may further enhance damage to microvascular integrity (27). In contrast, inhalation of gaseous NO has been reported to produce a nearly 60% reduction in L by completely returning Pc to baseline, probably in combination with a preclusion of the endotoxin-induced rise in microvascular permeability (11). In fact, in vitro, low concentrations of NO may even act as an endogenous tool to counteract increment in permeability by raising vascular endothelial cGMP. At variance, L-NAME, which ultimately reduces cGMP, may enhance permeability (10). However, L-NAME does not increase pulmonary microvascular permeability in healthy sheep, which is consistent with the observed effect of AG (15).

In the present study, AG increased the factors determining lung fluid filtration. Consequently, we expected EVLW to rise rather than to remain unchanged. Apparently, the enhanced fluid filtration was balanced by a simultaneous increase in L. Investigations in ovine mesenteric lymphatic endothelial cells and porcine isolated tracheobronchial lymph vessels suggest that NO is involved in the local regulation of lymph flow (28, 29). Thus, the increased lung lymph drainage after AG may be secondary to a reduction of locally generated NO resulting in enhanced activity of lymph vessels. Therefore, AG may prevent the lungs from contracting edema despite the increases in Pc and, most likely, also in microvascular permeability. This is the first report addressing a role for the L-arginine/NO pathway in the control of EVLW after endotoxin. Although previous investigators have noticed an additional transient increase in L after a bolus injection of L-NAME, they have not focused on simultaneous changes in EVLW (14).

In endotoxemia, redistribution of intrapulmonary blood flow from poorly to better oxygenated areas by means of hypoxic pulmonary vasoconstriction (HPV) is blunted, at least in part, by increased endogenous NO production, leading to hypoxemia (16, 20). Whereas improved gas exchange has been shown after inhaled gaseous NO (11, 12), reports about favorable effects of inhibiting NOS remain inconsistent. Bolus injection of L-NAME decreased S/T in sheep exposed to continuously infused endotoxin, but did not improve Pa_O2 (13). In another recent study, L-NAME reduced Pa_O2 even further because of increasing lung edema (17). It is unlikely that the observed increases in Pa_O2 and, correspondingly, in Sa_O2 after AG resulted from the decrease in CI since AaPO₂ and S/T decreased in parallel, and O₂ and O₂ remained unchanged. In septicemic sheep exposed to unilateral airway hypoxia, a reduction of HPV was recovered to more than 60% of its baseline value by L-NMMA (16). Although the present study was not designed for demonstration of HPV, the improvement of gas exchange indicates that the increase in PVRI after AG, at least in part, was due to reinforcement of HPV. The observed reduction in PBVI supports this assumption. An alternative explanation could be that the most poorly ventilated areas were also those where iNOS was expressed most and that consequently had the strongest response to AG. Preferential inhibition of iNOS by AG, combined with less influence on eNOS than L-NAME (18), could eventually explain the different effects of AG and L-NAME on gas exchange. Our results confirm findings in anesthetized dogs and rabbits, showing that AG ameliorates derangement of gas exchange after endotoxin (5, 6).

Although the hemodynamic response to AG was consistent with other studies employing NOS inhibitors in sheep (13), it was surprising that the increase in plasma NO_X was not prevented. Still more surprising, the increment in plasma cGMP turned out to be slightly, albeit not significantly, larger in the LPS/AG group, and cumulative doses of AG increased plasma NO_X by as much as 40%. It is possible that these unexpected effects of AG could be caused by a slow onset of iNOS inhibition with reciprocal activation of eNOS (7, 18). However, previous studies have shown that AG in a dose of 2 mg/kg has been capable of suppressing pulmonary activation of iNOS and peroxynitrite formation after endotoxin in dogs (5), and in doses as great as 20 mg/kg, of reducing plasma NO_X in endotoxemic rodents (4, 6). It is, therefore, possible that in ovine endotoxemia, large doses of AG would inhibit plasma NO_X and cGMP more efficiently. At the same time, such doses also could raise Pc secondary to reduced local NO production, thereby escalating pulmonary edema beyond the lymphatic drainage capacity. Unfortunately, plasma assays do not necessary reflect local liberation of NO and cGMP, particularly in the lungs.

In summary, infusion of AG in ovine endotoxemia produces a further rise in lung fluid filtration at unchanged EVLW. These findings may result from increased lung capillary pressure and permeability balanced by enhanced drainage of lymph that prevents lung edema from developing. This indicates that AG may reduce pulmonary NO production. Because gas exchange improves, AG may act preferentially to constrict dilated blood vessels within poorly ventilated areas of the lungs. Although the latter effect is beneficial, clinical use of inhibitors of iNOS such as AG remains speculative.