Isoflurane, but Not Sevoflurane, Increases Transendothelial Albumin Permeability in the Isolated Rat Lung: Role for Enhanced Phosphorylation of Caveolin-1

The study was performed in the isolated rat lungs and cultured rat lung microvascular endothelial cells.

After approval from the Institutional Animal Care and Use Committee (Chicago, Illinois), studies were conducted in 35 adult Sprague-Dawley rats (Charles River, Wilmington, MA) of either sex (300–350 g). The rats were anesthetized with pentobarbital sodium (40 mg/kg intraperitoneal). After tracheal intubation, the lungs were mechanically ventilated with room air at a rate 60 breaths/min and a tidal volume of 2.8 ml. A midsternal thoracotomy was performed, and 400 U heparin was injected into the vena cava for anticoagulation. The pulmonary artery and left atrium were cannulated and the aorta was ligated so that the venous effluent could be drained into a left atrial cannula. The heart and lungs (with a segment of the trachea) were removed en bloc  from the thoracic cavity and mounted on a perfusion apparatus. The lungs were immediately perfused with Krebs solution at 37°C via  the pulmonary artery at a constant flow (0.035 ml/g) and venous pressure (−4.0 cm H²O). The composition of the solution was as follows: 118 mm NaCl, 4.7 mm KCl, 1.0 mm CaCl², 1.0 mm MgCl², 5.0 mm HEPES, 11 mm glucose, and 0.025 mm EDTA (pH 7.35–7.45) supplemented with bovine serum albumin (3 g/100 ml). A continuous record of pulmonary arterial pressure and lung weight was obtained via  an in-line pressure transducer and a counterweighted beam balance during experiments. These data were displayed on a personal computer video monitor (IBM PS2 model 50z; White Plains, NY) after being digitized by an analog-to-digital converter (μDAS 8PGA board; Keithley Metrabyte, Solon, OH). Data acquisition software (Notebook Pro for Windows; Labtech, Andover, MA) was used to analyze data. The lungs were ventilated at a constant peak inspiratory pressure (5 cm H²O) throughout and end-expiratory pressure of 2 cm H²O.12 

After a stabilization period of 15 min, the isolated lung was ventilated with room air (carrier gas) alone (control) or with 1.0 and 2.0 minimum alveolar concentration (MAC) isoflurane (1.4% and 2.8%) or sevoflurane (2.4% and 4.8%)13,14in air by means of a calibrated vaporizer (Dräger, Lübeck, Germany) for 30 min. Inspired concentrations of isoflurane and sevoflurane were confirmed by gas chromatography.

Albumin permeability–surface area product (PS) product was determined using the single-sample technique.15,16After 30 min of administration of either carrier gas or anesthetic, the isogravimetric lung preparation was rapidly switched to perfusate, which contained tracer ¹²⁵I-albumin (approximately 80,000 counts/ml) and 0.3 g/100 ml unlabeled albumin. After 3 min of exposure to ¹²⁵I-albumin, the lung was switched back to buffer without tracer but with 3.0 g/100 ml albumin for 6 min to remove cell surface and circulating ¹²⁵I-albumin, and then the perfusion was abruptly stopped. The lungs were immediately removed, cleaned of connective tissue, weighed, and counted for γ radioactivity. Samples of the instilled ¹²⁵I-albumin were also counted. PS product (in ml · min⁻¹· g⁻¹dry lung) was calculated with the formula A/(Cp · t), where A and Cp are concentrations of tracer albumin in the tissue (in counts/g) and in the perfusate (in counts/ml), respectively, and t is the perfusion time for tracer albumin (3 min).15–17Previous studies indicated that a 3-min infusion period was sufficient for ¹²⁵I-albumin to flux across the endothelium.15,16With this protocol, back flux of ¹²⁵I-albumin from the pulmonary interstitium could be considered negligible because of the transient infusion of the labeled albumin.15Therefore, the accumulation of ¹²⁵I-albumin is taken as a unidirectional protein flux across the endothelial barrier into the interstitium.

Capillary filtration coefficient (K^f,c) was determined in separate lung preparations using a method described previously.12Before and after 30 min of administration of either carrier gas or anesthetic, the K^f,cwas measured from the rate of lung wet weight gain more than 1 min after a step increase (+6 cm H²O) in venous pressure for 5 min. The rate of weight gain was normalized by the lung dry weight and step size to calculate K^f,c. The analytical procedures used for computing K^f,cfrom recordings of lung wet weight have been presented in detail.18Values are expressed as ml · min⁻¹· cm H²O⁻¹· g⁻¹dry lung tissue.

Wet/dry lung weight ratio served as an index of lung water content. For determination of lung dry weight, lung tissue was dried in an oven to a constant weight (60°C for 72 h).

Rat lung microvascular endothelial cells were obtained from Vec Technologies (Rensselaer, NY). For monolayer culture, the cells were resuspended at a density of 1 × 10⁵/ml in high-glucose Dulbecco's modified Eagle's medium (supplemented with 10% fetal bovine serum, 50 U/ml penicillin, and 50 μg/ml streptomycin) and incubated (37°C) under a humidified atmosphere of 5% CO²–95% air. Confluent monolayers formed on culture dishes, microporous Transwell filter inserts, or glass coverslips within 24–48 h. RLMVECs stably expressing Y14F-caveolin-1 were generated as described previously.3,19RLMVEC monolayers were serum deprived for 2 h before experiments.

The 6-well or 12-well cell culture plates containing RLMVEC monolayers in HEPES-buffered Hanks' balanced salt solution (HBSS, 1.0 or 0.5 ml/well, respectively) were placed in a layer of water within the airtight exposure chamber in a water bath heated to 37°C.20Carrier gas (air) was passed through a calibrated vaporizer (Dräger, Lübeck, Germany) providing 1.0 and 2.0 MAC of either isoflurane (1.4% and 2.8%) or sevoflurane (2.4% and 4.8%) at a flow rate of 5 l/min into the airtight chamber for 30 min. Preliminary experiments showed that complete equilibration of anesthetics between the liquid (HBSS) and gas phases was achieved within 2 min. The concentrations of volatile anesthetics were confirmed by sampling on both the inflow and the outflow side of the chamber and in the wells of the multiwell plates and assessing the anesthetic concentration by gas chromatography (mode 5890; Hewlett Packard, Wilmington, DE).

Uptake of ¹²⁵I-albumin into wild-type RLMVECs was assayed as described previously,21with minor modifications. Briefly, confluent RLMVEC monolayers were seeded in six-well plates (1.0–1.2 × 10⁶cells/35 mm). Unlabeled albumin was added at a final concentration of 0.1 mg/ml in four wells and 100 mg/ml in other two wells containing HBSS, respectively. ¹²⁵I-albumin was then added into each well at an activity of 1 × 10⁶counts per minute (cpm). Uptake of ¹²⁵I-albumin was allowed to proceed at 37°C for 30 min during exposure to either carrier gas (air) or 1.0 and 2.0 MAC isoflurane or sevoflurane. Albumin uptake was terminated by chilling the samples on ice. All samples were then washed three times in acid wash buffer (0.5 m NaCl and 0.2 m acetic acid, pH 2.5) and another three times in cold HBSS, to remove any ¹²⁵I-albumin that attached to the cell surface. Cells were finally lysed with 1 ml Tris-HCl buffer (0.05 m Tris-HCl, 1% Triton X-100, and 0.5% SDS, pH 7.4). The radioactivity in samples was determined using a γ counter (Packard Instruments, Downers Grove, IL). Specific uptake of ¹²⁵I-albumin was estimated by subtracting the nonspecific cell-associated activity (determined in the presence of 100 mg/ml unlabeled albumin) from the total (determined in the presence of 0.1 mg/ml unlabeled albumin).21Specific uptake was normalized to cell protein (determined by bicinchoninic acid method) and expressed in units of cpm/mg cell protein.

On the basis of our initial experimental results, additional experiments, using the same basic protocol, were performed to evaluate the roles of the caveolae and caveolin-1 in the effect of 2.0 MAC isoflurane on albumin uptake. In one group of studies, endothelial cells were pretreated with methyl-β-cyclodextrin, the cholesterol-binding agent that disassembles caveolae.21,22This pretreatment consisted of incubation of RLMVECs with 2.0 mm methyl-β-cyclodextrin in HBSS for 15 min at 37°C, followed by two washes with HBSS. In the second group of studies, Y14F-caveolin-1 mutant RLMVECs were used.

To confirm the effect of volatile anesthetic on uptake of ¹²⁵I-albumin, we performed studies on fluorescent albumin uptake in RLMVECs. For determinations of fluorescent albumin uptake into RLMVECs, we used confluent RLMVEC monolayers grown on glass coverslips. Monolayers were treated with fluorescently tagged albumin (Alexa 488-labeled bovine serum albumin, at 50 μg/ml, plus 0.5 mg/ml unlabeled albumin in HBSS) and simultaneously exposed to air (control) or 2.0 MAC isoflurane or sevoflurane for 30 min at 37°C.21Unincorporated probe was removed by rinsing with HBSS.

Internalized fluorescent albumin (green) was viewed by confocal microscopy in optical sections midway through the cell. Cell nuclei (blue) were labeled with 4′,6-diamidino-2-phenyl indole dihydrochloride (DAPI). Briefly, the monolayer cultures were rinsed and fixed with 4% paraformaldehyde in HBSS; DAPI (1 μg/ml) was added immediately after cell fixation for 15 min, and the cells were rinsed three times and finally mounted on glass slides using ProLong antifade mounting medium (Molecular Probes, Eugene, OR). Confocal images were acquired with a laser-scanning confocal microscopy (Zeiss LSM 510; Thornwood, NY) with a 488-nm excitation laser line to detect Alexa 488 (BP505–550 nm emission), and nonconfocal DAPI images were acquired using Hg lamp excitation with the ultraviolet filter set (BP385–470 nm emission) respectively. Optical sections had a thickness of less than 1 μm (pinhole set to achieve 1 Airy unit). Quantitative analysis of images (average whole cell fluorescence intensity) in the acquired confocal images (n = 6/treatment group) was determined using Zeiss LSM 510 META software. The average background fluorescence detected from areas where no cells were present in each experimental condition was subtracted from the total fluorescence to yield specific cellular fluorescence intensity.

Rat lung microvascular endothelial cells were grown on clear microporous polyester Transwell membranes (12 wells, 1 cm²growth area, 0.4 μm pore size; Corning Costar, Cambridge, MA). The membrane inserts (inner well) were filled with a total of 0.5 ml HBSS containing ¹²⁵I-albumin (1 × 10⁶cpm) in the presence of unlabeled albumin (0.1 or 100 mg/ml). The lower well was filled with 1.5 ml HBSS of the same osmolarity as the inner well. Thus, fluid levels and osmotic pressure in the “upper” and “lower” wells were equalized to eliminate hydrostatic and osmotic pressure differences across the monolayer. Aliquots of 500 μl were sampled from the lower chamber after exposure to air or 1.0 and 2.0 MAC isoflurane or sevoflurane at 37°C for 30 min, and then γ radioactivity of the samples was measured. Transendothelial ¹²⁵I-albumin permeability was calculated from the flux of radiolabeled albumin across the cell monolayer, as described previously.23,24Albumin permeability is expressed in units of μl · min⁻¹· cm⁻².

The role of caveolae in effects of 2.0 MAC isoflurane and sevoflurane on transendothelial albumin transport was also evaluated in RLMVECs with methyl-β-cyclodextrin using the same basic protocol. Before the administration of volatile anesthetic, endothelial cells were pretreated for 15 min with 2.0 mm methyl-β-cyclodextrin, followed by two washes.

Phosphorylation of caveolin-1 and Src  kinase in RLMVECs was determined by Western blot analysis. Confluent RLMVEC monolayers on six-well plates were treated with 2.0 MAC isoflurane or sevoflurane at 37°C for 30 min. Cell lysates were subsequently prepared as described,25,26and the protein concentration in the lysate was assayed. Equal amounts of protein (20 μg) were loaded on 12% acrylamide gels, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Nonspecific binding was blocked with 5% nonfat dry milk in Tris-buffered saline (TBS) for 1 h at room temperature. The membranes were washed three times with TBS-T solution (0.05% Tween 20 in TBS) and incubated with tubulin monoclonal antibody (1:5,000) in 5% nonfat dry milk, pY14-caveolin-1 (1:1,000), or pY416-Src  polyclonal antibodies (1:1,000) in TBS-T solution. Incubation was carried out overnight; the membranes were washed three times for 5 min each and then incubated for 60 min with goat anti-rabbit (polyclonal) or anti-mouse (monoclonal) immunoglobulin G conjugated to horseradish peroxidase (1:5,000). Membranes were washed three times for 5 min each, and the protein bands were detected using the ECL reagent (Pierce, Rockford, IL). Molecular mass of the proteins was determined using known marker proteins. Relative intensity of tubulin, phosphorylated caveolin-1, and Src  was measured using Scion Image (National Institutes of Health, Bethesda, MD).

The chemicals and reagents used were obtained from Sigma Chemical Company (St. Louis, MO) unless otherwise stated. Horseradish peroxidase–conjugated goat anti-rabbit and anti-mouse immunoglobulin G were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). pY14-caveolin-1 and pY416-Src  polyclonal antibody were from Chemicon (Temecula, CA) and Cell Signaling Technology, Inc. (Beverly, MA), respectively. Goat anti-mouse and anti-rabbit immunoglobulin G, Alexa 488-albumin, and DAPI were purchased from Molecular Probes, Inc. (Eugene, OR). HBSS containing NaHCO³(4.2 mm) and HEPES (10 mm) was adjusted to pH 7.4. Bovine serum albumin (fraction V, 99% pure, endotoxin free, cold alcohol precipitated) and methyl-β-cyclodextrin were dissolved in HBSS. Dulbecco's modified Eagle's medium and fetal bovine serum were obtained from Gibco BRL (Grand Island, NY) and Hyclone (Logan, UT), respectively. ¹²⁵I-labeled albumin was purchased from PerkinElmer, Inc. (Wellesley, MA). Trichloroacetic acid precipitation analysis was used to confirm the purity27; free ¹²⁵I was removed from ¹²⁵I-labeled albumin with a Sephadex G25 column such that the contaminant free ¹²⁵I in the tracer used contributed less than 0.3% of the total counts.

One-way analysis of variance and the Student-Newman-Keuls test for post hoc  comparisons were used to determine differences between control and experimental groups. The Student t  test was performed for paired samples. Data were expressed as mean ± SD. Differences were considered significant at P < 0.05.

Initial baseline values for pulmonary artery pressure were similar for the experimental groups. There was no change in pulmonary artery pressure (10–15 cm H²O) among different groups after administration of 1.0 or 2.0 MAC isoflurane or sevoflurane for 30 min.

Compared with the control group, isoflurane at 2.0 MAC increased ¹²⁵I-albumin PS product (a measure of protein permeability) in the isolated, perfused rat lung by 48% (fig. 1), whereas 1.0 MAC isoflurane and sevoflurane and 2.0 MAC sevoflurane had no effect on ¹²⁵I-albumin PS product. K^f,c(a measure of liquid permeability) at baseline was similar in all groups. Neither 1.0 and 2.0 MAC isoflurane nor sevoflurane altered the K^f,c(fig. 2A). In addition, the wet/dry lung weight ratio (lung fluid content) was not significantly different in the various groups (fig. 2B).

In wild-type RLMVECs, isoflurane at 2.0 MAC more than doubled the uptake of ¹²⁵I-albumin (4.33 ± 0.25 vs.  2.16 ± 0.12 cpm/mg cell protein × 10⁴; fig. 3A) and caused an approximately 54% increase in transendothelial transport of ¹²⁵I-albumin (8.78 ± 0.22 vs.  5.70 ± 0.21 ml · min⁻¹· cm⁻²× 10⁻²; fig. 3B). Isoflurane and sevoflurane at 1.0 MAC and sevoflurane at 2.0 MAC did not alter ¹²⁵I-albumin uptake and transport in cultured cells. Consistently, confocal images showed that 2.0 MAC isoflurane caused a fourfold increase in fluorescent albumin uptake, whereas 2.0 MAC sevoflurane had no effect on fluorescent albumin uptake in RLMVECs (fig. 4). The mean specific fluorescence intensity under control conditions was 37 ± 2 fluorescence units/pixel, which increased to 153 ± 27 in cells treated with 2 MAC isoflurane and to 47 ± 4 in cells treated with sevoflurane. Furthermore, the effect of isoflurane on fluorescent albumin uptake was abolished in cells expressing the phosphorylation defective caveolin-1 mutant, Y14F-Cav-1 (fig. 4); the specific fluorescence intensity of internalized albumin was less than that observed under control conditions (19 ± 6 fluorescence units/pixel; P < 0.05 vs.  control).

To test whether the isoflurane-induced increase in albumin uptake and transendothelial transport are mediated by caveolae, we pretreated RLMVECs for 15 min with methyl-β-cyclodextrin, a membrane cholesterol–depleting agent that thereby flattens caveolae, and then observed the effects of isoflurane on ¹²⁵I-albumin uptake and transport. Our results indicated that pretreatment of RLMVECs with methyl-β-cyclodextrin reduced uptake and transendothelial transport of albumin to less than 50% of control levels in the absence or presence of 2.0 MAC isoflurane or sevoflurane (fig. 5).

To address further the role of caveolin-1 in the isoflurane-induced increase in albumin permeability, we compared the effects of isoflurane on ¹²⁵I-albumin uptake in wild-type and Y14F-caveolin-1 mutant RLMVECs. Isoflurane caused an approximately 100% increase in ¹²⁵I-albumin uptake in wild-type RLMVECs, and interestingly, this effect was abolished in Y14F-caveolin-1 mutant expressing cells (fig. 6). This key observation demonstrates that the isoflurane-induced increase in albumin uptake is dependent on and requires caveolin-1 phosphorylation.

Figure 7shows that in comparison with the control group, 2.0 MAC isoflurane caused a twofold increase in phosphorylation of both caveolin-1 and Src  family kinase in wild-type RLMVECs (fig. 7A) but had no effect in Y14F-caveolin-1 mutant expressing cells (fig. 7B). Equal protein loading per lane was confirmed by reprobing the blots with antitubulin antibody, which showed similar tubulin protein bands in each groups. Increase in phospho-Src  (PY-416) immunoblot indicates Src  activation, and increased phosphorylation of caveolin-1, which we previously showed is mediated by activated Src  kinase,26,28is indicative and required for caveolae-mediated endocytosis and transcytosis.19Sevoflurane at 2.0 MAC did not alter phosphorylation of caveolin-1 and Src  family kinase (fig. 7A).

At identical minimum alveolar concentrations (corresponding to nearly equal aqueous molar concentrations at 37°C), isoflurane (0.6 mm), but not sevoflurane (0.7 mm), activated albumin transcytosis in endothelial cells, indicating that isoflurane has properties not shared by sevoflurane. The physiologic activator of albumin transcytosis is the albumin molecule, which engages a specific receptor molecule (gp60) on the endothelial cell surface.25–29Upon binding to gp60, albumin activates Src , which enhances the phosphorylation of caveolin-1.19,26–29Hence, Src  activation is a critical regulatory event required for the initiation of albumin transcytosis through endothelium.19,26–29We now show that the presence of isoflurane further promotes Src  activation and caveolin-1 phosphorylation, thereby resulting in augmented albumin transcytosis as observed.

The effect of volatile anesthetics on albumin transport across the vessel wall has not been previously investigated. The current findings show that isoflurane increases transendothelial transport of albumin (albumin PS product) in an ex vivo  lung preparation. Because no change in pulmonary artery pressure, venous pressure, or perfusion rate occurred during the experiments conducted, the increase in albumin transport cannot be attributed to increased capillary hydrostatic pressure and the resultant convective flux of albumin. Albumin transport across the endothelial barrier occurs by means of two distinct pathways, i.e. , a transcellular pathway (caveolae-mediated) and a paracellular pathway (diffusion).2,3In the current study, however, isoflurane increased albumin PS product without causing changes in K^f,cor wet/dry lung weight ratio, suggesting that isoflurane did not affect the integrity of interendothelial cell junctions. These results indicate that albumin transport stimulated by isoflurane used a nonhydraulic pathway to cross the vascular endothelial barrier.12,29 

The cellular barrier to proteins in lung microvessels consists of an endothelial cell layer with highly restrictive interendothelial junctions. To identify the albumin-transporting mechanism, we designed studies on rat pulmonary endothelial cell monolayers, which form a similarly restrictive barrier in vitro .21We first showed that isoflurane (but not sevoflurane) increased monolayer permeability, using the same anesthetic concentration effective in stimulating albumin uptake in the ex vivo  lung preparation. These experiments showed that endothelial cells indeed represent a significant site of action of isoflurane. The in vitro  approach was advantageous by allowing analysis of cellular internalization of iodinated or fluorescent albumin tracers. Analysis of cell lysates for iodinated albumin indicated that isoflurane (but not sevoflurane) activated the cellular uptake of albumin, and imaging studies confirmed that isoflurane stimulated fluorescent tracer uptake (fluorescent forms of albumin or cholera toxin) into punctate intracellular structures that we previously demonstrated were positive for caveolin-1.21,29Our previous confocal imaging studies showed that Alexa 488-albumin colocalizes in endocytic vesicles with Alexa 594-cholera toxin subunit B, a specific caveola marker in endothelial cells.21,29These observations indicate that isoflurane activated caveolae-mediated uptake of tracer protein molecules. The additional observation that isoflurane stimulated transendothelial transfer of tracer molecules across the entire endothelial monolayer further indicated that caveolae participated in a transendothelial transport process, i.e. , transcytosis. Blocking of the transendothelial transport process with the cholesterol-depleting agent cyclodextrin confirmed this expectation. Cyclodextrin was used for its ability to deplete endothelial caveolae.4,21Because cyclodextrin completely eliminated the isoflurane stimulated transport process, we are able to rule out the paracellular pathway as a significant mode of transendothelial transport in the case of isoflurane.

It is known that albumin binding to the 60-kd glycoprotein (gp60) on the endothelial cell surface induces clustering of gp60 and its physical interaction with caveolin-1.29,Src  tyrosine kinase is bound to caveolin-1 via  its scaffolding domain,30palmitoylated C-terminal cysteine residue, and N-terminal phosphorylated tyrosine residue,19and the kinase is activated upon albumin binding to gp60.27Activated Src , in turn, phosphorylates caveolin-1, gp60, and dynamin-2 to initiate plasmalemmal vesicle fission and transendothelial vesicular transport of albumin.26–28The current findings provide additional support for this hypothesis by indicating that isoflurane increases albumin uptake and transport in endothelial cells and that these effects are associated with enhanced Src  activation and phosphorylation of caveolin-1. We also observed that the increase in uptake of albumin (induced by isoflurane) was abolished in endothelial cells expressing the phosphorylation-defective Y14F-caveolin-1 mutant. These data show that caveolin-1 phosphorylation is a crucial event in isoflurane-induced increase in albumin uptake.

Our data showed that isoflurane increased the phosphorylation of Src  at tyrosine 416, which is required for activation of Src .28Moreover, our previous studies demonstrated that the phosphorylation of caveolin-1 and albumin uptake were abolished in RLMVECs transfected with dominant-negative Src .26,28Our current results dovetail with our previous observations by indicating that Src  activation (mediated by isoflurane) stimulates albumin uptake in endothelial cells. This conclusion is consistent with the pharmacologic evidence for Src  activation by isoflurane in the rat heart, using the Src  inhibitor 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolol[3,4-day]pyrimidine (PP1).31 

The mechanism by which isoflurane activates Src kinase remains to be clarified. One possibility is a direct activation of Src  by isoflurane. It is known that the potency or efficacy of volatile anesthetics correlates with lipid solubility, protein perturbation, or both. Because of its lipid solubility, anesthetic molecules can interact at the molecular level with lipids and proteins.32Therefore, we speculate that isoflurane may accumulate in caveolae and directly activate Src , which is normally present bound to caveolin-1.26Other potential mechanisms for Src  activation include (1) displacement of Src  from caveolin-1, which is thought to negatively regulate Src  activity33; (2) activation of an upstream stimulatory kinase34; or (3) inhibition of C-terminal Src kinase, which negatively regulates Src  activity by phosphorylating Src  on tyrosine 527.35 

Noteworthy was that sevoflurane, in contrast to isoflurane, did not alter albumin uptake and transport in RLMVECs. Consistent with these functional observations, we showed that sevoflurane lacks an ability to activate Src  or induce caveolin-1 phosphorylation in the RLMVECs. These observations reinforce the view that isoflurane acts by stimulating Src  activity and caveolin-1 phosphorylation.

There are several explanations for the absence of pulmonary edema (in the isolated rat lung preparation) despite increased transendothelial albumin permeability. Increased transport of albumin from blood into the interstitium may be compartmentalized in an “excluded volume” within the interstitial space, because of its large molecular size and/or charge density.36This interstitial albumin exclusion is thought to serve as an important buffering mechanism to prevent development of pulmonary edema.12,36Another possibility is that the increased interstitial pressure due to a slight increase in fluid filtration caused by isoflurane (secondary to extravasation of albumin) buffers or even nullifies the Starling pressure gradient that causes microvascular fluid filtration.37The normal lung has a low interstitial compliance provided by the inflexible structure of the subendothelial matrix, such that a small increase in interstitial volume provokes a marked increase in tissue pressure.37–39In the Starling equation, increased tissue pressure favors fluid reabsorption by capillaries and limits fluid filtration. Low compliance also represents an important “tissue safety factor” to counteract further progression of pulmonary edema.36–40Another possible explanation is that isoflurane may also affect albumin transport in the reverse direction, i.e. , from the interstitium to the lumen (abluminal to luminal transport).41Thus, isoflurane-induced increase in extravascular albumin clearance may limit the accumulation of interstitial albumin and an increase in oncotic pressure, and therefore, no increase in lung weight as a consequence of an increase in extravascular osmolarity would be observed.

Our findings are consistent with studies that evaluated the effect of volatile anesthetics on alveolar epithelial and capillary permeability in the isolated perfused rabbit lung8,10and in the rat9and human.11These studies indicated that isoflurane increased pulmonary capillary permeability but had no effect on tissue edema in the normal lung. Recently, other studies have been performed to test the effect of isoflurane on tissue edema during pathologic conditions and showed that isoflurane enhanced the development of neurogenic pulmonary edema in rats42and caused increased interstitial fluid accumulation in the crystalloid volume-loaded sheep.43Although neurohumoral regulatory mechanisms (e.g. , vascular endothelial growth factor, neuropeptide Y, atrial natriuretic peptide) have been suggested in these in vivo  studies, local and direct effects of isoflurane on pulmonary endothelial cells (i.e. , endothelial albumin permeability) could not be ruled out. Therefore, our study, when combined with these in vivo  results, suggests the possibility that isoflurane may decrease the threshold for alveolar edema in patients at high risk for pulmonary edema.

In summary, the current study demonstrated that isoflurane, but not sevoflurane, increases pulmonary endothelial albumin permeability and both albumin internalization and transendothelial albumin transport in cultured endothelial cells. Isoflurane increased Src  activation and caveolin-1 phosphorylation, which we previously showed were critical events regulating caveolae-mediated endocytosis and transcytosis.19,26–29The isoflurane-induced increase in albumin permeability was not associated with net fluid movement in the normal lungs, probably because several safety factors operate to prevent edema. Conditions such as sepsis and inflammation, known to reduce pulmonary safety factors, may complicate use of anesthetic agents with permeability increasing properties. Isoflurane-induced Src  activation, resulting in caveolin-1 phosphorylation, may be an underlying molecular mechanism critical to the regulation of vascular protein permeability.