Published ahead of print on March 27, 2003, doi:10.1164/rccm.200210-1216OC
© 2003 American Thoracic Society
Systemic Microvascular Leak in an In Vivo Rat Model of Ventilator-induced Lung InjuryDepartment of Medicine, Pulmonary/Critical Care Unit and Renal Unit, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts Correspondence and requests for reprints should be addressed to Charles A. Hales, M.D., Massachusetts General Hospital, Pulmonary and Critical Care Medicine, 55 Fruit Street, Bullfinch 148, Boston, MA 02114. E-mail: chales{at}partners.org
Positive pressure mechanical ventilation has significant systemic effects, but the systemic effects associated with ventilator-induced lung injury (VILI) are unexplored. We hypothesized that VILI would cause systemic microvascular leak that is dependent on nitric oxide synthase (NOS) expression. Rats were ventilated with room air at 85 breaths/minute for 2 hours with either VT 7 or 20 ml/kg. Kidney microvascular leak, which was assessed by measuring 24-hour urine protein and Evans blue dye, was used as a marker of systemic microvascular leak. A significant microvascular leak occurred in both lung and kidney with large VT (20 ml/kg) ventilation. Injection of 0.9% NaCl corrected the hypotension and the decreased cardiac output related to large VT, but it did not attenuate microvascular leak of lung and kidney. Serum vascular endothelial growth factor was significantly elevated in large VT groups. Endothelial NOS expression significantly increased in the lung and kidney tissue with large VT ventilation but not inducible NOS. The NOS inhibitor, N-nitro-L-arginine methyl ester, attenuated the microvascular leak of lung and kidney and the proteinuria with large VT ventilation. Endothelial NOS may mediate the systemic microvascular leak of the present model of VILI.
Key Words: microvascular leak • permeability • proteinuria • endothelial nitric oxide synthase • N-nitro-L-arginine methyl ester The cause of death in most patients with acute respiratory distress syndrome has been found to be multiple-organ failure rather than hypoxia (1). A large multicenter randomized trial showed that mechanical ventilation with a lower VT (6 ml/kg) resulted in lower mortality than conventional (12 ml/kg) ventilation. The number of days without nonpulmonary organ failure was significantly higher in the group treated with lower VTs (2). Positive pressure mechanical ventilation caused a significant increase in renal failure in trauma patients (3). Altogether, these data support the idea that mechanical ventilation may have significant systemic effects. Nitric oxide (NO) has been shown to be produced from L-arginine by a family of NO synthase (NOS) isoforms that are expressed in a large variety of tissues and cells (4). NO has been shown to regulate multiple cellular functions, including smooth muscle cell relaxation, neurotransmission, macrophage-induced cytotoxicity, and apoptosis, as well as direct induction of vascular and epithelial hyperpermeability (5, 6). In the isolated rabbit lung ventilator-induced lung injury (VILI) model investigators have reported that the nitrite + nitrate concentrations of bronchoalveolar lavage fluid in high-pressure ventilation (30 cm H2O) significantly increased in comparison with control group (15 cm H2O) (7). A recent study suggested that NO has the ability to impair the glomerular permeability barrier and increase albumin permeability in ex vivo rat glomeruli model (8). Inhibition of NOS in the peritoneum during acute peritonitis significantly improved ultrafiltration and reduced protein loss in dialysate (9). We hypothesized that VILI would cause an increase in microvascular permeability in the systemic circulation, which is dependent on increased NOS expression. Some of the results of these studies have been previously reported in the form of an abstract (10).
Additional detail on the methods is provided in the online supplement.
Ventilator Protocol
Inhibition of NOS
Measurement of Lung and Kidney Microvascular Leak by Evans Blue Dye
Creatinine Clearance and Urine Protein Measurement
Vascular Endothelial Growth Factor Measurements
Immunoblot Analysis for NOS
Statistical Methods
Peak Inspiratory Pressure Peak inspiratory pressure of the rats ventilated at a VT of 20 ml/kg was 31.5 ± 2 cm H2O, and VT of 7 ml/kg was 10.5 ± 1 cm H2O.
Microvascular Leak
Hemodynamic Data Mean arterial pressure (MAP) was monitored to evaluate hemodynamic changes related to mechanical ventilation and L-NAME treatment. At baseline, there were no significant differences in MAP among the study groups. One hour after starting mechanical ventilation, there was a significant decrease in MAP in rats ventilated with VT20 compared with rats ventilated with VT7 (Figure 2A) . At the end of mechanical ventilation, MAP of the VT20 group was significantly lower than in all of the other groups (p < 0.01). NS injection (10 ml/kg) in the VT 20 ml/kg (VT20NS) group increased MAP to a level comparable to that of the ventilated rats with lower VT. All groups treated with L-NAME had significantly higher MAP when compared with untreated groups throughout mechanical ventilation (p < 0.01). At baseline, there was no significant difference in CO between VT20 and VT20NS groups. NS injection (10 ml/kg) in the VT20NS group significantly increased CO throughout mechanical ventilation compared with VT20 and the L-NAME–treated groups (p < 0.01) (Figure 2B). There was no significant difference in CO between the baseline and the end of mechanical ventilation in VT20NS group. L-NAME caused a significant fall in CO in all groups (Figure 2B). Saline infusion (VT20 + L-NAME) significantly improved CO but not to baseline level. CO in VT20NS + L-NAME was, however, similar to VT20 without saline or L-NAME.
Renal Function in Rats with or without L-NAME Treatment To determine the influence of large VT ventilation and L-NAME on kidney function in this model of VILI, we measured creatinine clearance, serum creatinine, urine volume, and fractional excretion of sodium (Figure 3) . There were no significant differences in creatinine clearance, urine volume, and fractional excretion of sodium among VT7, VT20, and VT20NS groups. There was a trend toward a higher creatinine clearance and lower serum creatinine with L-NAME treatment in all three groups (Figures 3A and 3B), but there was only a significant difference (p < 0.02) between VT20 and VT20 + L-NAME in serum creatinine (Figure 3B). There was no significant change in urine volume and fractional excretion of sodium with or without L-NAME treatment (Figures 3C and 3D).
Urine Protein To evaluate physiologic significance of renal vascular leak, urine total protein and albumin were measured. Twenty-four–hour total urine protein-to-creatinine ratio (24-hour TP/Cr) increased significantly in VT20 (p < 0.05) and VT20NS (p < 0.05) compared with VT7 (Figure 4A) . After L-NAME, 24-hour TP/Cr was significantly decreased in VT20 + L-NAME and VT20NS + L-NAME (p < 0.05 and p < 0.05, respectively) compared with VT20 and VT20NS groups. Twenty-four–hour urine albumin was increased in VT20 (p < 0.05) but not significantly in VT20NS (p = 0.2) compared with VT7 (Figure 4B). L-NAME prevented a significant increase in albuminuria in VT20 + L-NAME and VT20NS + L-NAME compared with untreated VT7 (Figure 4B). There were no significant changes in 24-hour TP/Cr and urine albumin between VT7 and VT7 + L-NAME groups. There was no difference in urine ß2-microglobulin excretion among the groups normalized by urine creatinine (data not shown).
Serum VEGF Levels Serum VEGF levels were significantly elevated in VT20 (p < 0.05) and VT20NS groups (p < 0.05) compared with the VT7 group (Figure 5) . There was no significant difference between VT20 and VT20NS groups.
eNOS and iNOS Expression To study the effect of mechanical ventilation on eNOS and iNOS, iNOS and total eNOS were measured by immunoblot analysis. eNOS was detected at 140 kD in the lung and kidney tissue from the VT7, VT20, and VT20NS groups, but iNOS was not detected (data not shown). Densitometric analysis showed that the protein level of eNOS was significantly increased in VT20 and VT20NS compared with VT7 in both lung (p < 0.05) and kidney (p < 0.01) (Figure 6C) . Again, iNOS was not detected in either the VT20 or VT20NS groups.
We have found that mechanical ventilation with a moderately large VT of 20 ml/kg for only 2 hours resulted in lung and renal microvascular leak of protein, as measured by the EBD technique (Figure 1). Twenty-four–hour urine total protein as a ratio of urinary creatinine was also increased (Figure 4), as was urinary albumin. Creatinine clearance was not changed (Figure 3). This technique, EBD concentration per wet lung weight, probably underestimated the degree of increased permeability in the large VT groups as the wet weight of these lungs was increased (Figure 1C). The renal weight did not change (Figure 1D). Systemic blood pressure and CO fell significantly in the VT20 group and thus could have been responsible for the abnormal renal and lung changes (Figure 2). Restoration of systemic blood pressure to that of the VT7 rats and also restoration of CO during mechanical ventilation to that of the baseline, however, did not reverse the microvascular lung and renal increased permeability to protein (Figure 1) and did not correct the increase in urinary total protein to creatinine ratio (Figure 4). It did, however, partially reduce the urinary albumin toward the VT7 group, although L-NAME completely reversed it (Figure 4). Thus, albuminuria may have been partially related to the fall in systemic blood pressure and CO as volume replacement had some impact on the albuminuria, but the other changes in renal function caused by VT20 were not corrected by volume. We found that the protein level for eNOS rose in the lungs and kidneys of the VT20 and VT20NS groups (Figure 6), but iNOS was not detected in the lung or kidney in any group. L-NAME, a NOS inhibitor, blunted the increase in EBD protein leak in the kidney and lung microvessels (Figure 1). It also inhibited the VILI-associated increase in 24-hour urine protein and albumin (Figure 4). Thus, the detection of an increased level of eNOS in the lungs and kidneys of the VT20 and VT20NS groups combined with the reversal of the permeability leak in kidney and lung supports a causative role for eNOS in the lung and kidney injury secondary to large VT breaths. The action of L-NAME to blunt the increase in lung and renal microvascular permeability to protein can be complex as the effect may relate to a direct action of loss of NO on vascular permeability or to a change in microvascular surface area (13). A number of agents that increase the permeability of venular microvessels via calcium-dependent release of NO also act on the endothelial cells in arterioles to cause endothelium-dependent vasodilation (14). Thus, by increasing the number of vessels perfused and/or the pressure within microvessels, NO-dependent processes can increase the total amount of solute transported into tissue with or without a true increase in permeability (13). Administration of L-NAME may thus cause arteriolar vasocontriction and reduce microvascular surface area. In this model, L-NAME caused a significant fall in CO, which was only partially reversed with saline infusion (Figure 2B). This fall in CO could result in a significant reduction in pulmonary flow and thus a derecruitment of pulmonary vessels with loss of vascular surface area leading to decreased EBD uptake. However, we do not believe this occurred, as significantly increasing CO in the L-NAME group (VT20 + L-NAME) by saline infusion (VT20NS + L-NAME) did not increase dye uptake in the lung and because VT20 and VT20NS + L-NAME groups had similar COs, although the L-NAME–treated group (VT20NS + L-NAME) had significantly less EBD accumulation compared with VT20 without L-NAME. CO was unlikely to influence renal retention of EBD or glomerular filtration rate, as renal blood flow tends to remain constant by intrinsic mechanisms despite fluctuations in perfusion pressure (15). Under ordinary conditions, basal NO release from endothelial cells has important physiologic properties, such as inhibition of leukocyte attachment, maintenance of mast cell stability, and reduction of platelet aggregation (16). NO also has a protective effect on microvascular permeability in intestinal vessels (17). Furthermore, NO has a protective role in shear stress-induced thrombosis and inflammation in carotid arteries (18). However, NO also has the ability to impair the glomerular permeability barrier and to increase albumin permeability in an ex vivo rat glomeruli model (8). An increase in NO production can also induce systemic microvascular hyperpermeability in caveolin-1 knockout mice (19). Therefore, NO may have a different role in different vascular beds depending on its concentration, its relative contribution to hemodynamic changes, and its leukocyte attachment. Large VT ventilation has been found to cause detachment of endothelial cells from the basement membrane within 5 minutes (20). During mechanical stress, the endothelium also has been shown to respond by forming paracellular gaps, resulting in increased permeability (21). Caveolin-1–deficient lung capillaries have defects in tight junction morphology, abnormalities in capillary endothelial cell adhesion to the basement membrane, and increased permeability. Microvascular hyperpermeability of caveolin-1 knockout mice can be reversed by L-NAME, which suggests the permeability change in the caveolin-1 knockout mice was dependent on NO overproduction (19). These data all together lend circumstantial support that excess NO might be important in the hyperpermeability, as we have noted in our model of VILI using VT 20 ml/kg. The pathway by which NO increases vascular permeability may be through extracellular signal-regulated kinases 1 and 2 and VEGF. Inhibition of extracellular signal-regulated kinases 1 and 2 in human umbilical vein endothelial cells blocked cGMP and VEGF-induced hyperpermeability (22, 23). Extracellular signal-regulated kinases 1 and 2 may activate eNOS during in vitro ischemia in bovine pulmonary artery endothelial cells (24). NO modulates VEGF-induced vascular permeability, and eNOS predominantly mediates this process (25). Knockout mice for eNOS show decreased vascular permeability in response to VEGF infusions, suggesting that eNOS may be important in VEGF-induced permeability change (25). Thus, the inhibition of eNOS by L-NAME might work to decrease microvascular permeability in lung and kidney by inhibiting the effect of VEGF. We have not identified the mechanism that links lung stretch with microvascular leak in the kidney. Mechanical stretch stimulated VEGF mRNA expression and protein in sheep pulmonary artery smooth muscle cells (26). Plasma levels of VEGF have been shown to be predictors of the severity of postoperative capillary leak syndrome in humans (27). Plasma VEGF levels have been related to the degree of damage in the alveolar epithelium (28, 29). Plasma VEGF levels are elevated in patients with acute respiratory distress syndrome, which may contribute to increased endothelial permeability (28). Taken together, a possible explanation for systemic leak may be VILI-induced release of VEGF from injured lung, which affects distant organs and causes systemic leak via a NO-mediated process. Our observation that serum VEGF was significantly elevated in large VT groups compared with VT7 would support this interpretation. Circulating inflammatory cytokines are another possible mechanism of systemic leak in VILI. Mechanical ventilation with a large VT as compared with a small VT in patients with acute respiratory distress syndrome increases the concentration of several cytokines in bronchoalveolar lavage fluid and plasma (30). Although the role of eNOS in inflammation is not well understood, eNOS expression is downregulated in the lung of rats treated with endotoxin (31). Cyclic stretch upregulates transforming growth factor-ß1 production (32), and transforming growth factor-ß directly increases alveolar epithelial permeability by depletion of intracellular glutathione in an acute lung injury model (33). Transforming growth factor-ß is also able to stimulate VEGF release in an epithelial cell line (34). VEGF is able to increase eNOS protein expression (35). Thus, we speculate that there may be an interplay of eNOS and as yet unidentified cytokines such as transforming growth factor-ß in the permeability change in this model. Shear stress could be another mechanism causing systemic microvascular leak. Shear stress caused by tangential shear force generated by flow across the endothelial cell surface triggers the generation of NO in endothelial cells (36). Large VT probably increases mean intrathoracic pressure during inspiration, which falls repeatedly during exhalation, undoubtedly decreasing and increasing venous flow, possibly generating shear force on the vessels. Shear stress can cause an increase in NO production not only by an increase in eNOS expression but also by phosphorylation of eNOS (37). eNOS phosphorylation by shear stress may occur independent of changes in intracellular calcium levels (38). Shear stress-related NO production can occur in a few minutes (36). Therefore, in this model, L-NAME may block excessive NO production induced by both eNOS expression and eNOS phosphorylation. NOS inhibitors, which reverse permeability change, mainly act on the venular endothelial cells (13). We did not measure shear stress in our model, but we speculate that venular microvessels could be an important site for microvascular leak in the present model of VILI. There was no significant difference in urine volume among the groups in this model. Rats were placed in metabolic cage after mechanical ventilation and then were allowed to take water without limitation. Therefore, 24-hour urine volume may not have accurately assessed the status of acute intravascular volume change. In addition, the response of urine volume might have varied depending on the L-NAME dose. L-NAME at pressor doses tends to decrease urine volume compared with control (39), whereas L-NAME at a subpressor dose induces diuresis (40). L-NAME tended to increase the glomerular filtration rate in this model (Figure 3A). In an observation similar to ours, an injection of L-NAME caused a decrease in renal blood flow without a decrement in glomerular filtration rate (41), leading Treeck and Aukland (41) to conclude that NO may have predominantly regulated postglomerular vasodilator tone. The observed increase in the 24-hour TP/Cr ratio may have resulted from an increase in glomerular permeability, a dysfunction in tubular reabsorption, or a combination of the two. Increased albuminuria is likely due to increased glomerular permeability. This study showed that 24-hour TP/Cr ratio increased in the large VT groups compared with the VT7 group. An analysis of risk factors for acute renal failure in trauma patients has shown that mechanical ventilation increases the odds ratio of developing renal failure (3). This result suggests that mechanical ventilation can result in injury to the tubules of the kidney either due to factors that are released into the circulation or a hemodynamic effect. However, increased urine protein excretion does not appear to be related to tubule dysfunction in this model as there was no significant difference in urine ß2-microglobulin excretion and fractional excretion of sodium among the groups. L-NAME did reduce albuminuria to levels similar to the VT7 and significantly reduced 24-hour TP/Cr ratio in large VT groups. These findings support the thesis that proteinuria may be dependent on NO production and due to a glomerular lesion. We conclude that increased systemic microvascular leak with enhanced albuminuria occurs in a rat model of VILI. This microvascular leak is attenuated by L-NAME. eNOS may mediate the systemic effects of the present model of VILI.
The authors thank John Beagle and Bin Ouyang, M.D., for their technical assistance and helpful comments.
Supported by K08 HL03920, HL39150, DR 39773, DR 38452, and NS 10828. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form October 24, 2002; accepted in final form March 19, 2003
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
p < 0.05 versus no treatment at the same VT.
p < 0.01 versus VT20; 
p < 0.05 versus VT20 + L-NAME (closed circle = VT7; closed square = VT20; closed triangle = VT20NS; open circle = VT7 + L-NAME; open square = VT20 + L-NAME; open triangle = VT20NS + L-NAME).
