INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lung injury caused by exposure to a hyperoxic environment, which is extremely important in a clinical setting, is generally induced by a long-term supplementation of a gas mixture containing a high concentration of O₂ to patients with various respiratory diseases. Hyperoxic lung injury is characterized by endothelial cell damage, leukocyte accumulation within pulmonary microvessels, and subsequent leukocyte transmigration into the airspace associated with serious tissue injury (1, 2). Over the last several years, leukocyte accumulation in the systemic circulation during inflammation has been shown to be mediated by different classes of adhesion molecules (3, 4). However, the issue of whether the same adhesion molecules as specified in the systemic circulation are actually required for leukocyte accumulation in inflamed lungs is still controversial (5). Specifically, it has not been conclusively settled what sorts of adhesion molecules, if any, are involved in accumulating leukocytes within the pulmonary microcirculation, which is the initial event for a succeeding invasion of leukocytes into the lung parenchyma. We have recently demonstrated that intercellular adhesion molecule-1 (ICAM-1), bound to leukocyte ₂-integrins, is conspicuously upregulated in the pulmonary endothelial cells exposed to a hyperoxic environment (9). Furthermore, selectin members including P- and L-selectins have been shown to mediate leukocyte sequestration and lung injury induced by a specific stimulus such as cobra venom factor (10). Based on these facts, we attempted in the current study to assess whether ICAM-1, P-selectin, and L-selectin would play an important role in interfering with leukocyte kinetics in the pulmonary microvasculature including precapillary arterioles, postcapillary venules, and capillaries in hyperoxia-exposed lungs harvested at the time when infiltration of leukocytes into the airspace and tissue injury with alveolar flooding were not evident. We used hyperoxia- exposed lungs prior to outbreak of remarkable tissue injury, as our main purpose was to elucidate a possible role of various adhesion molecules in accumulating leukocytes within the pulmonary microvasculature, rather than to clarify the mechanism causing leukocyte transmigration into the airspace.

Recently, several groups of investigators (11) attempted to observe microcirculatory behavior of leukocytes in intact lungs by applying the classic epiluminescence nonconfocal microscopy, which had generally been used to observe blood cell kinetics in the systemic microcirculation such as the mesentery and skeletal muscle venules (15). However, classic epiluminescence microscopy without the confocal system may not be appropriate for determining the precise architecture of the pulmonary microcirculation and blood cell behavior in it, as the pulmonary microvascular network is exceedingly interwoven and arterioles, venules, and capillaries are densely convoluted in the acini. Therefore, we applied, in the current study, real-time confocal laser scanning luminescence optical microscopy, allowing precise discrimination of individual microvessels from neighboring vessels (16).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation of Isolated Perfused Lungs

In order to quantitatively assess leukocyte kinetics in hyperoxia- exposed pulmonary microvessels, Sprague-Dawley rats, 8 wk of age, weighing 250 to 300 g were exposed to either 90% O₂ (n = 32) or 21% O₂ (n = 12) for 48 h. The animals thus treated were used for isolated perfused lung preparation, a detailed description of which is provided elsewhere (17, 18). Briefly, the animals exposed to either a normoxic or hyperoxic environment were anesthetized intraperitoneally with pentobarbital sodium (50 mg/kg) and artificially ventilated with room air at a tidal volume of 10 ml/kg and respiratory rate of 60 breaths/min through the tracheal tube. After a median sternotomy had been made and heparin (1,000 U/kg) had been injected into the left ventricle, cannulas were inserted into the pulmonary artery and left atrium. Both cannulas were secured with strings. A ligature was placed around the aorta to prevent loss of perfusate into the systemic circulation. Isolated lungs were fixed on a microscopic stage in the supine position and then perfused at a constant flow rate of 10 ml/min in a recirculating manner with a roller pump (Rotor 1500N; Taitec, Tokyo, Japan). As the perfusate, a modified Krebs-Henseleit solution was used and 3% bovine serum albumin was added to maintain iso-osmotic pressure. Perfusate hematocrit was adjusted to 6.5 ± 0.2% by the addition of fresh blood obtained from donor rats and solutions containing either leukocytes or erythrocytes stained with appropriate fluoresceins as described subsequently. The hematocrit was kept low so as to approximately maintain the Newtonian nature of the fluid, yielding a parabolic velocity profile in a laminar flow regime in the pulmonary microcirculation (19). To avoid the movement caused by artificial ventilation, which would markedly hinder precise observations of microvascular architecture and blood cell kinetics, the trachea was ligated at the end-inspiratory position and gas exchange was adequately maintained with an extracorporeal membrane oxygenator (ECMO; Merasilox-S, Senko, Tokyo, Japan) inserted between the isolated lung and the roller pump. A gas mixture containing 21% O₂ and 5% CO₂ in N₂ was used as the gas flowing into the ECMO, allowing adjustment of the perfusate PO₂, PCO₂, and pH to 145 ± 3 mm Hg, 38 ± 2 mm Hg, and 7.38 ± 0.02, respectively. A warmed and humidified gas mixture containing the same composition of gases as those used for the ECMO was continuously supplied to the lung surface so as to maintain a temperature of 37 ± 0.2° C and to avoid desiccation of the lung surface. Pulmonary arterial pressures (Ppa) were continuously monitored by force displacement of pressure transducers (TP-400T; Nihon Kohden Co., Tokyo, Japan). Perfusate PO₂, PCO₂ as well as pH were measured with electrodes (Model 1306; I.L., Lexington, MA).

Experimental Protocols

To investigate leukocyte behavior and the role of adhesion molecules in hyperoxia-exposed lungs, we designed five experiments: (1) Control group (n = 12); the animals were housed in a normoxic environment for 48 h and assumed to have little lung injury. (2) Hyperoxic group (n = 9); animals were housed for 48 h in an isolated chamber in which O₂ concentration was maintained at 90%. (3) ICAM-1 group (n = 8); monoclonal antibody against rat ICAM-1 (1A29; donated by Dr. Miyasaka, Osaka University Medical School, Osaka, Japan) was administered to isolated perfused lungs prepared from the animals exposed to a hyperoxic environment. Perfusate concentration of 1A29 was maintained at 20 µg/ml. After a 10-min perfusion with 1A29, microcirculatory behavior of leukocytes and erythrocytes was investigated. The efficiency of 1A29 in blocking the rat ICAM-1 was previously discussed by Tamatani and Miyasaka (20). (4) P-selectin group (n = 8); monoclonal antibody against rat P-selectin (ARP2-4) was administered at a final concentration of 20 µg/ml into the perfusion circuit prepared from the animals exposed to hyperoxia. Ten minutes later, measurements of blood cell kinetics were made. The specificity concerning ARP2-4 in inhibiting the rat P-selectin is described elsewhere (21). (5) Fucoidin group (n = 7); leukocytes treated with fucoidin (Sigma, St. Louis, MO) were added to the perfused lungs obtained from the hyperoxia-exposed animals. In order to obtain leukocytes with attenuated L-selectin functions, donor rats were intravenously injected with fucoidin at a dose of 25 mg/kg. After a 10-min incubation in vivo, blood was gently taken by heart puncture and was administered into the perfusion reservoir. Fucoidin causes a generalized loss of selectin binding to counterpart ligands, leading to inhibition of not only L-selectin but also P-selectin (22, 23). Therefore, we also added fucoidin (300 µg/ml) to the perfusion circuit so that our fucoidin experiment would allow assessment of leukocyte kinetics in the pulmonary microcirculation under conditions in which L-and P-selectins are simultaneously inhibited.

Observations of Microvascular Leukocyte and Erythrocyte Kinetics by Confocal Luminescence Microscopy

We used a real-time confocal laser scanning luminescence optical microscope (CSU10; Yokogawa Electronics, Tokyo, Japan) incorporating a high-speed video analysis system, allowing precise discrimination of individual microvessels and measurement of blood cell velocities in the highly complex pulmonary microcirculation. The reflected light or fluorescent emission from the specimen was imaged onto a high-sensitivity charge-coupled device (CCD) camera with an image intensifier (EktaPro Intensified Imager VSG; Kodak, San Diego, CA) which can detect even very low fluorescence signals. By incorporating an excitation wavelength of 488 nm emitted from a low-power air-cooled argon (Ar) ion laser (532-BSA04; Omnichrome, Chino, CA) with appropriate fluoresceins, the present confocal system allowed us to obtain apparently instantaneous images at 1,000 frames/s, with a two-point spatial resolution of 0.2 µm. The final magnifying power of our system reached ×484, with the ×20 objective, on the video screen. The size of the resulting field of view was 210 × 210 µm. We registered confocal images at a rate of 250 frames/s by means of a high-speed video analysis system (EktaPro 1000 Processor, Kodak) connected with the image intensified CCD camera. All images obtained were monitored on a color video television (PVM-1444Q; Sony, Tokyo, Japan) and recorded in a video cassette recorder (SVQ-260; Sony).

Following the previously described method (15), we stained leukocytes with carboxyfluorescein diacetate succinimidyl ester (CFDASE; Molecular Probes, Eugene, OR). CFDASE solution diluted with 1.5 ml of normal saline was injected into the femoral veins of donor rats (in vivo concentration: 1 mg/kg). CFDASE forms carboxyfluorescein succinimidyl ester (CFSE) in leukocytes and platelets but not in erythrocytes. After a 30-min incubation in vivo, blood was aspirated from the left ventricle and administered into the perfusion circuit. These procedures allowed us to obtain the blood samples for perfusion in which 29.8 ± 2.1% (mean ± SEM) of leukocytes were polymorphonuclear cells (PMN) and the remainder was mononuclear cells (lymphocytes: 66.6 ± 3.5%; monocytes: 3.6 ± 1.3%, values are mean ± SEM). Leukocytes and platelets stained with CFSE were distinguishable during the microscopic measurements because of their distinctive size difference.

To examine relative contribution of mononuclear cells, predominantly consisting of lymphocytes, to adhesive or rolling phenomena in arterioles and venules of hyperoxia-exposed lungs, we isolated mononuclear cells from the donor rat blood stained with CFSE as described previously. Mononuclear cell isolation was accomplished with a classic Ficoll density gradient procedure. Five milliliters of CFSE-stained rat blood were diluted in 5 ml of phosphate-buffered saline (PBS) and carefully overlayed on 3 ml of Ficoll-Conray solution (Immunobiological Co., Gunma, Japan), adapted for use with rat cells (specific gravity: 1.09). After centrifugation at 1,800 rpm for 30 min at room temperature, the mononuclear cell layer was washed three times with PBS and administered into the reservoir of perfusion circuit connected with lungs exposed to a hyperoxic environment (n = 6). Preliminary experiments exhibited that the mononuclear cell layer separated by means of the Ficoll density gradient procedure was not contaminated by PMN and its cell constituents were mostly lymphocytes with a minimal quantity of monocytes.

Erythrocytes were stained with fluorescein isothiocyanate (FITC) (Sigma). Fresh rat blood was centrifuged at 1,000 rpm and the buffy coat was discarded. The packed erythrocyte solution was diluted with PBS to adjust the hematocrit to 10% and FITC was added to give a concentration of 0.1 mg/ml. This solution was then incubated at 37° C for 30 min. The FITC-labeled erythrocyte samples were washed three times in PBS and 1 ml was added to the reservoir after the measurements of leukocyte kinetics had been completed.

To determine the architecture of microvessels in which leukocyte and erythrocyte behaviors had been examined, we added 200 µl of 5% FITC-dextran with a molecular weight of 145,000 (Sigma) to the reservoir. We used FITC-dextran with a large molecular weight to prevent its leakage through injured microvascular walls. Defining the edge of each microvessel as the portion exhibiting a steep change in fluorescent signal, we quantified the vessel diameter by processing a confocal video image with a computer-assisted digital image analyzing system (Quadra 840AV/Image 1.58; Apple, Cupertino, CA). An example of confocal images of the microvascular architecture was presented in Figure 1, in which a diagonally running arteriole meeting numerous capillaries was seen.

View larger version (127K): [in this window] [in a new window]   Figure 1.   Confocal image of the pulmonary microcirculation in the lung exposed to hyperoxia. Precapillary arteriole joining capillaries observed on the plane in focus at a 25 µm distance from the lung surface (objective lens: ×20). a = arteriole. c = capillary.

Analysis of Leukocyte and Erythrocyte Kinetics in Pulmonary Arterioles and Venules

Replaying the videotapes obtained with the high-speed video analysis system at the normal video rate allowed the axial velocities of CFSE-labeled leukocytes (Vw) as well as FITC-labeled erythrocytes (Vr) to be determined by measuring the distance traveled between two or more successive video frames. The highest value of Vr was determined in all microvessels studied and was assumed to be equal to the centerline velocity (Vmax). Vmax was used to estimate mean fluid flow (Vb) in each microvessel on the basis of the relation Vb = Vmax/ 1.6 (19). Wall shear rate () was estimated by applying Poiseuille's law on the assumption that the nature of a Newtonian fluid is basically preserved under the present experimental conditions in which cell densities were low. Thus,  = 8 (Vb/D), where D represents vessel diameter (24).

An adherent leukocyte in an arteriole or a venule was defined as a cell which was firmly attached to the vascular endothelium and did not move during the observation period. Rolling leukocytes in arterioles and venules were defined as cells transiently interacting with the vascular endothelium and thus traveling much more slowly than centerline erythrocytes. These cells were identified by applying the velocity criterion proposed by Gaehtgens and colleagues (19, 25), who calculated the critical velocity (Vcrit) of a freely flowing cell traveling close to, but not adhering to, the vascular surface under a parabolic velocity profile in microvessels. Vcrit is given as (Vb)(2-), where  is the ratio of the leukocyte diameter to the microvessel diameter. Any cell flowing at a velocity below Vcrit was assumed to be slowed by the interaction with the vascular endothelium. Averaging Vw values of rolling leukocytes, we defined mean leukocyte rolling velocity (Vroll). In addition, as the index of fracture stress serving as a measure of adhesive force between leukocytes and the vascular endothelium (26), we calculated mean Vw/Vr, averaging the axial velocities obtained for all of the leukocytes (Vw) and erythrocytes (Vr) in respective microvessels.

Mononuclear cell kinetics (adhesion and rolling) in arterioles and venules was analyzed by applying qualitatively the same procedures as mentioned previously.

Analysis of Leukocyte and Erythrocyte Kinetics in Pulmonary Capillaries

Velocities of CFSE-leukocytes (Vw) and those of FITC-erythrocytes (Vr) in capillaries were measured with the same procedure as that applied for arterioles and venules. Leukocytes retained lastingly in capillaries were excluded from the calculation of Vw.

To quantify the characteristics of leukocyte transit in capillaries, we classified the observed leukocytes into three categories: cells moving smoothly without interruption; cells stopping transiently for more than 4 ms but rejoining the free-flowing stream within 2 s; and impeded cells stopping completely within a capillary segment for more than 2 s.

Histologic Analysis of Leukocyte Accumulation in the Alveolar Septa

Although confocal luminescence microscopy applied in the current study was extremely useful for assessing dynamic blood cell behavior including rolling and/or firm adhesion of leukocytes in pulmonary microcirculation, its observations were restricted to very small regions in the lung field. Furthermore, we could not estimate cell differentials from luminescence microscopic images, because both PMN and mononuclear cells were stained with CFSE. Considering these facts, we histologically examined leukocyte accumulation within the alveolar septa in lungs prepared from the animals exposed to either normoxia or hyperoxia, and used it as a measure of intracapillary sequestration of leukocytes (n = 5 for each condition). After completion of all measurements, a catheter was inserted into the trachea, and the isolated perfused lung was fixed with formalin and embedded in paraffin. Six sections, cut at identical intervals from the apex to the base of the left lung, were stained with hematoxylin-eosin (HE). In each section, 10 fields were randomly chosen and leukocytes localized within the alveolar septa in which the capillary network is located were counted at a magnification of ×1,000 with an oil immersion lens. The density of leukocytes in the septa was expressed as numbers per a single alveolus. The same lung sections were used for estimating leukocyte differentials, i.e., distinction between PMN and mononuclear cells.

Determination of ICAM-1 and P-selectin Expressions by Immunohistochemistry

The lungs exposed to either normoxia or hyperoxia (n = 5 for each) were fixed by instillation of periodate-lysine-paraformaldehyde solution through the trachea, embedded in optimum cutting temperature (OCT) compound (Miles Inc., Elkhart, IN), and then frozen in dry ice and acetone. Six cryostat sections (6 µm each), prepared from the lung fields cut at identical intervals from the apex to the base, were dried in air for 1 h at room temperature. The sections were washed with PBS, and incubated in 10% normal goat serum dissolved with PBS. To inhibit endogenous peroxidase activity, the sections were treated by methanol with 3% hydrogen peroxide for 20 min according to the method of Streefkerk (27). Anti-ICAM-1 (1A29) and anti-P- selectin monoclonal antibody (ARP2-4) were respectively diluted 100 and 50 times with PBS and absorbed to normal rat serum. Nonimmunized mouse serum was used as the negative control. These sera were layered on the section for 2 h. The sections were incubated with peroxidase-labeled goat anti-mouse IgG (H + L) (Zymed Laboratories, South San Francisco, CA) for 30 min at room temperature, and were rinsed three times with PBS. The labeled peroxidase was detected by reaction with 3,3'-diaminobenzidine tetrahydrochloride in 3% hydrogen peroxide-Tris (hydroxylmethyl) aminomethane buffer for 10 min. Counterstaining of nuclei was performed with methyl green. Immunoreactivity for ICAM-1 and P-selectin in each lung section was determined by means of a light microscope.

Determination of ICAM-1 Expression by Confocal Luminescence Microscopy

The distribution of ICAM-1 along the pulmonary microvessel endothelia in lungs exposed to hyperoxia was additionally examined by an intravital luminescence immunochemistry (n = 3). 1A29 (4 µg/g body weight) was injected into an anesthetized ventilated rat through the femoral vein. Five minutes later, an isolated perfused lung was prepared from the animal, and was injected with 0.5 ml of FITC-labeled anti-mouse IgG antibody (Sigma), the secondary antibody to 1A29. Subsequently, flow was ceased for 15 min to allow conjugations among endothelial ICAM-1, 1A29, and FITC-IgG antibody. Thereafter, perfusion was resumed for another 15 min. Endothelial ICAM-1 bound to 1A29 and FITC-IgG antibody was determined under confocal scanning luminescence microscopy and displayed with a high-sensitivity CCD camera (TEC-470; Optronics, Goleta, CA). Because 1A29 consists of a mouse IgG, IgG raised in mice (4 µg/g body weight, Sigma), and FITC-labeled anti-mouse IgG antibody were used as negative controls. Arterioles, venules, and capillaries were distinguished by administering both FITC-dextran and FITC-labeled erythrocytes at the end of each measurement.

Statistical Judgment

Unless otherwise specified, the results are presented as meas ± SEM. Differences in microvascular diameters, Vb, and shear rates between the control and hyperoxic groups were judged by unpaired t test. Significant differences in rolling leukocyte frequency, capillary entrapment, and leukocyte accumulation into alveolar septa among experimental groups (i.e., control, hyperoxic, ICAM-1, P-selectin, and fucoidin groups) were determined by applying one-way analysis of variance (ANOVA) followed by the Scheffé multiple comparison analysis. The values obtained for arterioles, venules, and capillaries in respective experimental conditions were statistically examined in terms of two-way ANOVA associated with the Scheffé analysis of multiple comparison. A p value less than 0.05 was deemed to be statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Basic Hemodynamic Parameters in Hyperoxia-exposed Lungs

There were no significant differences in vascular diameters of precapillary arterioles, postcapillary venules, or capillaries between the control (normoxia-exposed lungs) and the hyperoxic group (hyperoxia-exposed lungs with no medications) (Table 1). The diameters of arterioles and venules ranged from 18 to 35 µm, while those of capillaries were from 5 to 8 µm. The mean fluid flow, Vb, in either arterioles, venules or capillaries did not differ between the control and the hyperoxic group as well, leading to the equivalent flow-induced shear force conditions applied to each microvessel of the two groups. In the control and the hyperoxic group, Vb values in arterioles were 0.75 and 0.86 mm/s, respectively, corresponding to wall shear rates of 244 and 284 s¹. Venular Vb ranged between 1.1 and 1.3 mm/s in both groups, yielding wall shear rates in the range between 360 and 400 s¹. Capillary Vb averaged 0.57 mm/s in the control group and 0.64 mm/s in the hyperoxic group. We also found no significant differences in the diameters, Vb, or wall shear rates in any microvessels among the hyperoxic, ICAM-1, P-selectin, and fucoidin groups. Although vascular diameters were statistically identical, Vb as well as wall shear rates were much smaller in arterioles than those in venules of both the control and the hyperoxic groups. Qualitatively the same findings were observed for the ICAM-1, P-selectin, and fucoidin groups, resulting in no differences in microcirculatory hemodynamic characteristics among any experimental groups studied.

Leukocyte Kinetics in Arterioles and Venules

We found no leukocytes adhering firmly to the endothelium of arteriolar and venular walls in any experimental conditions employed.

In the hyperoxic group, relative frequency of rolling leukocytes was consistently higher in venules (30 ± 5%) than that in arterioles (16 ± 6%), despite being much lower in venules in the control group (Figure 2). Augmented number of rolling leukocytes in venules was certainly reduced in the hyperoxia-exposed lungs treated with either anti-ICAM-1 monoclonal antibody or fucoidin but not in those treated with monoclonal antibody to P-selectin. On the other hand, the number of leukocytes moving slowly along the arteriolar walls in hyperoxia-exposed lungs was significantly decreased by the treatment with fucoidin, but not by administration of anti-ICAM-1 and P-selectin monoclonal antibodies.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Relative frequency of leukocytes rolling along arteriolar or venular walls. (A) arterioles, (B) venules. Rolling leukocytes were defined as the cells moving with the axial velocity less than the critical velocity. Control = normoxia-exposed lungs. Hyperoxic = hyperoxia-exposed lungs with no medications. ICAM-1 = hyperoxia-exposed lungs treated with monoclonal antibody to ICAM-1. P-selectin = hyperoxia-exposed lungs treated with monoclonal antibody to P-selectin. Fucoidin = administration of fucoidin-treated leukocytes into hyperoxia-exposed lungs. *Differing from the control group. Smaller than the value obtained in the hyperoxic group.

Although mean Vw/Vr in arterioles did not differ among any experimental conditions, the venular Vw/Vr was significantly low in the hyperoxic group compared with that in the control group (Figure 3). The diminished Vw/Vr in hyperoxia-exposed lungs was securely restored to the control level by the treatment with either monoclonal antibody to ICAM-1 or fucoidin, though inhibition of P-selectin did not eliminate the hyperoxia-induced reduction of venular Vw/Vr. These findings were qualitatively in accordance with those obtained for relative frequency of rolling leukocytes.

View larger version (17K): [in this window] [in a new window]   Figure 3.   Mean leukocyte velocities relative to mean erythrocyte velocities (Vw/Vr). (A) arterioles, (B) venules, (C ) capillaries. Groups studied are the same as those defined in Figure 2. *Significantly smaller than the value obtained for the control. Significantly larger than the value observed for the hyperoxic group.

Vroll values of leukocytes in arterioles were respectively 0.30 and 0.28 mm/s in the control and in the hyperoxic group, not significantly differing between the two. These values corresponded to a quarter of centerline erythrocyte velocities. Venular Vroll averaged 0.48 mm/s in both the groups, much larger than arteriolar Vroll. These values again corresponded to a quarter of centerline erythrocyte velocities. Treatment with various inhibitors of adhesion molecules exerted no influence upon the Vroll values in hyperoxia-exposed lungs.

Addition of mononuclear cell solution to the reservoir did not induce firm cell adhesion to arteriolar and venular endothelia in hyperoxia-exposed lungs. Relative frequency of mononuclear cells with rolling was 4.9 ± 2.6% in arterioles and 3.7 ± 1.8% in venules, both values being significantly smaller than those obtained under conditions in which the whole leukocyte population was used for analysis (Figure 2). Because relative numbers of PMN and mononuclear cells (mainly lymphocytes) in the perfusate were, respectively, 30% and 70% (see METHODS), mononuclear cell kinetics data may lead one to deduce that 79% of arteriolar rolling leukocytes and 91% of venular rolling leukocytes were PMN under conditions in which nonseparated leukocyte solution (i.e., diluted whole blood) was used for analyzing microvascular cell kinetics in lungs exposed to hyperoxia.

Leukocyte Kinetics in Capillaries

Although there was no significant difference between the control and the hyperoxic groups in the relative number of leukocytes entrapped transiently within capillary segments, the frequency of leukocytes tethered firmly and thus located there without any discernible motion was remarkably enhanced in the hyperoxic group (Figure 4). The number of leukocytes with sustained arrest in capillaries was distinctly reduced by administering anti-ICAM-1 monoclonal antibody into the perfusion circuit. However, the addition of monoclonal antibody to P-selectin or fucoidin did not ameliorate the stationary leukocyte count in capillaries.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Leukocyte entrapment in capillaries. Groups studied are the same as those defined in Figure 2. From the left, relative frequency of leukocytes moving smoothly without any interruption (open bars), that stopped transiently, i.e., for more than 4 ms but less than 2 s (hatched bars), and that stopped persistently, i.e., for more than 2 s (solid bars). *Significantly larger than the value obtained for the control. Smaller than the value observed for the hyperoxic group.

Mean Vw/Vr values in capillaries of the control group were found to be 0.95, whereas those of the hyperoxic group were 0.67, with a significant difference in the values (Figure 3). Administration of anti-ICAM-1 monoclonal antibody, but neither anti-P-selectin antibody nor fucoidin, successfully restored the deteriorated Vw/Vr observed in pulmonary capillaries of lungs exposed to a hyperoxic environment.

Histologic Examination of Leukocyte Accumulation in the Alveolar Septa

Each microscopic field used for determining leukocyte counts contained at least five alveoli. Furthermore, we examined 10 fields in each lung section cut at identical intervals throughout the whole left lung; as a result, the total number of cells counted reached 300 or more for one lung. Because we analyzed five different lungs, the final leukocyte numbers examined were more than 1,500 for the respective experimental condition. We found very little leukocyte infiltration into the airspace of the hyperoxic lungs, i.e., leukocytes emigrated into the airspace amounted to 0.09 ± 0.04/alveolus in lungs exposed to a hyperoxic environment. The leukocyte number accumulated within the alveolar wall septa of hyperoxia-exposed lungs averaged 1.30 ± 0.15/alveolus without serious tissue injury such as alveolar flooding, while that in lungs exposed to normoxia for the same time period was 0.38 ± 0.04/alveolus, the former being much more abundant than the latter (Figure 5). In the hyperoxic group, the leukocyte density within the alveolar septa was 15 times as much as that observed within the airspace. The augmented accumulation of leukocytes within the alveolar septa of hyperoxia- exposed lungs was remarkably decreased by the treatment with monoclonal antibody against ICAM-1 (0.79 ± 0.08/alveolus), but the cell density did not reach the value obtained in the control. Administration of anti-P-selectin antibody (1.32 ± 0.14/ alveolus) or fucoidin (1.18 ± 0.16/alveolus) did not ameliorate the leukocyte accumulation in the alveolar septa of hyperoxia-treated lungs, thus supporting the findings obtained from the luminescence microscopic measurements.

View larger version (10K): [in this window] [in a new window]   Figure 5.   Accumulated leukocytes in the alveolar septa determined by histologic examinations. Categories studied are the same as those given in Figure 2. *Differing from the value obtained for the control group. Significantly smaller than the value observed for the hyperoxic group.

Hyperoxia exposure significantly elevated the proportion of PMN sequestered within the alveolar septa, i.e., the PMN percentage in capillaries of hyperoxia-exposed lungs was 76 ± 5% and the remainder was mononuclear leukocytes. The PMN fraction in hyperoxia-exposed lungs was much higher than that observed for normal control lungs, i.e., 27 ± 4%. The cell differentials of leukocytes sequestered within the septa in lungs exposed to hyperoxia were not modified by the treatment with anti-P-selectin antibody (73 ± 4%) or fucoidin (72 ± 3%), whereas inhibition of endothelial ICAM-1 slightly but significantly reduced the PMN fraction to 60 ± 5%, the value being smaller than that obtained in the hyperoxic, P-selectin, and fucoidin groups but larger than that observed in normal control lungs.

ICAM-1 and P-selectin Expressions along the Microvascular Endothelium

In total, 30 sections obtained from the upper, middle, and lower lung fields of five different animals were used for immunohistochemical assessment of ICAM-1 and P-selectin expressions under each experimental condition. In normoxia-exposed lungs, the alveolar walls showed faint but positive immunoreactivity for ICAM-1 (Figure 6). Furthermore, expression of ICAM-1 was found to be discernible along the venular and capillary segments, but not along the arteriolar walls. When lungs were exposed to hyperoxia, ICAM-1 expression seemed to be appreciably enhanced along the alveolar walls and the capillary lumens (Figure 6). ICAM-1 was upregulated in venules but not detectable along the arteriolar walls in hyperoxia-exposed lungs.

View larger version (121K): [in this window] [in a new window]   View larger version (118K): [in this window] [in a new window]   View larger version (109K): [in this window] [in a new window]   Figure 6.   Immunohistochemical examination of ICAM-1 expression. (A) ICAM-1 expression in the normoxia-exposed lung (magnification: ×200). (B) ICAM-1 expression in the hyperoxia-exposed lung (original magnification: ×200). (C ) ×1,000 magnification photograph of the part depicted in (B). Br = bronchiole (no ICAM-1 expression). Alv = ICAM-1 expression along the alveolar walls. (C ) ICAM-1 expression along the capillary lumens. Brown color indicates the positive immunoreactivity for ICAM-1, which is much stronger in the hyperoxia-exposed lung (B and C ) than that in the normoxia-exposed lung (A).

Because a histochemical approach has difficulty in distinguishing the capillary endothelium from the alveolar epithelium and also arterioles from venules, we additionally examined ICAM-1 distribution along hyperoxia-exposed microvessels in terms of intravital microscopy under the condition in which pulmonary perfusion was sufficiently maintained. The results indicated that ICAM-1 was conspicuously expressed along venular and capillary walls but not along arteriolar walls (Figure 7), supporting the findings obtained by the immunohistochemistry.

View larger version (149K): [in this window] [in a new window]   Figure 7.   Confocal view of ICAM-1 distribution along microvascular walls in hyperoxia-exposed lungs. Objective lens: ×20. V = venule; C = capillary. Although ICAM-1 was appreciably expressed along venular and capillary walls, the definite evidence for significant ICAM-1 expression along arteriolar walls was not obtained.

Although P-selectin was not expressed along any microvessels of the control lungs, it was sparsely but perceptibly enhanced along the arteriolar endothelium, but not along the venular and capillary endothelia, in lungs exposed to hyperoxia (Figure 8). There was no detectable immunoreactivity for P-selectin in the alveolar epithelium of both the control and the hyperoxic groups.

View larger version (138K): [in this window] [in a new window]   View larger version (139K): [in this window] [in a new window]   Figure 8.   Immunohistochemical examination of P-selectin expression. (A) No expression of P-selectin in the normoxia-exposed lung. (B) Scattering induction of P-selectin along the arteriolar wall in the lung exposed to hyperoxia (strong brown color indicated by arrow). Original magnification: ×200.

The experimental findings including abnormal behavior of leukocytes in, and ICAM-1 as well as P-selectin expressions along, the pulmonary microvasculature in hyperoxia-exposed lungs are summarized in Table 2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critique of Methods

Previous studies concerning pulmonary oxygen toxicity demonstrated that exposure to a gas mixture containing high O₂ for more than 24 h but less than 48 h would predominantly induce PMN retention within microvessels without notable morphologic injury, whereas exposure for more than 48 h would cause PMN infiltration into the airspace with discernible lung injury (1, 2, 28). Furthermore, we have recently demonstrated that the level of ICAM-1 expression significantly increases in cultured human pulmonary artery (HPAEC) and umbilical vein endothelial cells (HUVEC) exposed to a hyperoxic condition for the period ranging from 48 to 72 h, but decreases at 96-h exposure (9). Based on these facts, we selected 48-h exposure of the animals to a hyperoxic environment. This is because we attempted to determine whether various adhesion molecules on the vascular endothelium and leukocyte surface would modify pulmonary microcirculatory leukocyte behavior in hyperoxia-exposed lungs prior to occurrence of significant emigration of inflammatory cells into the airspace with serious lung injury, i.e., we did not try to elucidate the essential mechanism causing leukocyte transmigration into the airspace, which might qualitatively differ from that involved in its sequestration within pulmonary microvessels. In fact, histologic examination of lungs exposed to 90% O₂ for 48 h demonstrated no remarkable injury and minimal infiltration of leukocytes into the airspace but associated with augmented accumulation of leukocytes within the alveolar septa in which the capillary network is located (Figure 5), indicating that the 48-h exposure lung model is sufficient for accomplishing our main purpose.

Several groups of investigators (11) have observed leukocyte kinetics in the pulmonary microcirculation in terms of nonconfocal luminescence microscopes, which have been generally used for assessing leukocyte kinetics in the systemic microcirculation, e.g., mesenteric venules (15). However, the pulmonary microvascular network is exceedingly intricate due to the presence of alveoli with round shape, suggesting that most arteriolar and/or venular segments do not exist in the same plane as the capillary network. We recognized that nonconfocal luminescence microscopy occasionally misses the capillaries which run beside arterioles or venules (data not shown). This raises the possibility of misinterpreting the movement of blood cells in capillaries as taking place in arterioles or venules. To overcome the difficulties associated with classic nonconfocal microscopic observations, we used a real-time confocal laser scanning microscopic system allowing sufficient discrimination of pulmonary microvessel architecture (Figure 1).

Although our novel method is exceedingly useful for estimating dynamic behavior of leukocytes as a whole, it does not allow us to simply investigate the cell kinetics of each ingredient of leukocytes because both PMN and mononuclear cells are stained with CFSE and indiscernible during the microscopic measurements. Therefore, we conducted the experiments focused upon arteriolar and venular mononuclear cell dynamics (M-experiment) in addition to those upon the cell kinetics of the whole leukocyte population (W-experiment). The M-experiment suggested that about 80% of abnormal leukocytes in hyperoxia-exposed arterioles in the W-experiment could be identified as PMN. Similarly, more than 90% of abnormal leukocytes in venules in hyperoxia-treated lungs might be PMN. These findings may indicate that the unusual leukocyte kinetics observed in the W-experiment of hyperoxia-exposed lungs is interpreted as that caused primarily by abnormally behaving PMN but minimally by mononuclear cells.

Significance of Adhesion Molecules for Leukocyte Kinetics in Arterioles and Venules

We found no leukocytes adhering resolutely to arteriolar or venular endothelium even in the lungs exposed to the hyperoxic condition under which endothelial ICAM-1, causing firm adhesion of leukocytes in the systemic microvessels (3, 29), was upregulated at least in the pulmonary venules (Figure 7). Our findings are at variance with those obtained by Kuebler and associates (13) and Kuhnle and associates (14), who reported that leukocytes adhered firmly not just to venules but also to arterioles in intact rabbit lungs. On the other hand, Lien and colleagues (11, 12) found no adhesive leukocytes in either arterioles or venules in normal canine lungs, consistent with the results of the current study. Our findings may indicate that, in sharp contrast to the systemic circulation, firm attachment of leukocytes does not occur along either venular or arteriolar endothelium in both intact and injured lungs.

There were considerable quantities of rolling leukocytes in arterioles with reduced velocities in both the control and the hyperoxic group (Figures 2 and 3), though leukocyte-endothelial interactions leading to rolling were generally confined to the venular segments in the systemic microcirculation (26, 30). Supporting our findings, Gebb and coworkers (23) demonstrated leukocyte rolling along the arteriolar walls in canine lungs. In normal control lungs, the rolling frequency was significantly higher in arterioles in which the flow-induced shear rates were much lower than those in venules with a comparable size (Table 1). Exposure to hyperoxia did not alter arteriolar hemodynamic properties including wall shear rates (Table 1) and did not increase the endothelial ICAM-1 expression (Figures 6 and 7), thus supporting the findings of rolling leukocyte numbers and velocities in pulmonary arterioles exposed to the hyperoxic condition being not normalized by inhibiting ICAM-1 (Figures 2 and 3). Although endothelial P-selectin expression was sparsely augmented in arterioles of hyperoxia-treated lungs (Figure 8), P-selectin inhibition did not embellish rolling leukocyte numbers or their velocities in arterioles exposed to a hyperoxic environment (Figures 2 and 3). These findings may not be inconsistent with each other, because of a faint enhancement in P-selectin expression in pulmonary arterioles of hyperoxia-treated lungs (Figure 8). Further investigation is certainly required to elucidate the functional significance of P-selectin in hyperoxia-exposed arterioles. Administration of fucoidin, a competitor for both P- and L-selectins, remarkably reduced the number of rolling leukocytes in hyperoxia-exposed arterioles (Figure 2). Taken together, these observations may suggest that arteriolar leukocyte rolling is preferentially regulated by an adhesive force elicited via leukocyte L-selectin in addition to relatively low shear rates in arterioles (Tables 1 and 2, and Figure 2). Thus, the ICAM-1- and P-selectin-independent but L-selectin-dependent adhesive mechanism is thought to be involved in leukocyte rolling in pulmonary arterioles.

Rolling leukocyte frequency in venules of hyperoxia-exposed lungs was significantly increased and exceeded that in arterioles (Figure 2). This took place in combination with velocity reduction of leukocytes (Figure 3) and upregulation of ICAM-1 (Figure 7), indicating that adhesive interactions of leukocytes with endothelial ICAM-1 plays an important role for leukocyte rolling in venules of hyperoxia-exposed lungs, because shear rate serving as the force to disrupt tethering of leukocytes to the endothelium was distinctly larger in venules than that in arterioles (Table 1). Because venular rolling in hyperoxia-exposed lungs was comparably restrained by anti-ICAM-1 antibody and by fucoidin (Figures 2 and 3), pathways related to endothelial ICAM-1 and leukocyte L-selectin appear to be of equal importance for mediating venular rolling. P-selectin inhibition did not improve leukocyte rolling in venules (Figure 2), compatible with no unregulation of venular P-selectin in lungs treated with hyperoxia exposure (Figure 8). We suppose that leukocyte rolling related to the ICAM-1 pathway is unique to pulmonary venules, despite qualitatively the same phenomenon being observed in the mesenteric venules under a morbidly decreased shear force (31). Thus, P-selectin-independent but ICAM-1- and L-selectin-dependent mechanisms may be responsible for leukocyte rolling in pulmonary venules exposed to a hyperoxic environment (Table 2). These results also suggest that sequential multistep leukocyte-endothelium interactions (3, 29), thought to be a central mechanism inducing leukocyte margination in the systemic venules, are not applicable to pulmonary venules. Furthermore, leukocyte rolling velocities in pulmonary venules were about 25% of centerline erythrocyte velocity (see RESULTS), again indicating that a leukocyte rolling phenomenon in the pulmonary venules would be quite different from that observed in the systemic venules in which rolling leukocytes revealed a clumsy motion with velocities less than 10% of those of erythrocytes flowing at the centerline portion (15, 25, 30).

Significance of Adhesion Molecules for Leukocyte Kinetics in Capillaries

Several groups of investigators have reported that sustained entrapment of leukocytes in intact pulmonary capillaries is attributable to less deformability of the leukocyte (32, 33) and to the geometry of the pulmonary capillary network in which a great fraction of capillary bed contains narrow portions for leukocyte transit (33, 34). However, luminescence microscopic study revealed that stationary leukocyte frequency increased by hyperoxia exposure was significantly diminished by inhibiting ICAM-1 but not by administration of anti-P-selectin antibody and fucoidin (Figure 4), indicating that, in addition to mechanical impacts as described previously, ICAM-1, but neither P- nor L-selectins, serves as one of the important factors distorting leukocyte kinetics in capillaries of hyperoxia-exposed lungs. These findings were further supported by histologic examinations in which increased accumulation of leukocytes, consisting mainly of PMN, within the alveolar septa in lungs exposed to hyperoxia was significantly reduced by inhibition of ICAM-1, but not P- and L-selectins (Figure 5). These results are also in accordance with the findings that treatment of lungs with hyperoxia enhanced ICAM-1, but not P-selectin, expression in capillaries (Figures 6-8). Keeney and coworkers (35) demonstrated that leukocyte infiltration into the airspace was not ameliorated by the treatment of monoclonal antibody against CD18, the counterpart ligand for ICAM-1 expressed on the leukocyte surface, in injured lungs prepared from guinea pigs exposed to the hyperoxic condition for 72 h. Their findings (35) may suggest that ICAM-1-CD18-associated adhesive pathway is not involved in infiltration of inflammatory leukocytes into the airspace in hyperoxia-exposed lungs. However, we can not judge, from their measurements (35), whether anti-CD18 monoclonal antibody veritably failed in restraining capillary leukocyte entrapment, as they did not examine leukocyte dynamics in the pulmonary microcirculation. Although the difference in animal species should be seriously taken into account, the apparent discrepancy between the study of Keeny and coworkers (35) and ours may point to a potentially weighty matter concerning leukocyte kinetics in injured lungs induced by exposure to hyperoxia, i.e., ICAM-1-related adhesive pathway is of primary importance for eliciting an early-stage leukocyte accumulation within pulmonary capillaries (current experimental results), but not for leukocyte transmigration into the airspace occurring in a relatively late stage after exposure to hyperoxia (Keeney and coworkers [35]). ICAM-1-CD18-independent mechanisms including a variety of chemoattractants released by endothelial cells, epithelial cells, or other phagocytes (macrophage in particular) appear to be responsible for leukocyte transmigration from capillaries into the airspace in hyperoxia-treated lungs (2, 7, 8). The importance of ICAM-1-CD18-independent mechanisms has recently been confirmed in leukocyte invasion upon the airspace in the case of severe infection of gram-positive organisms (7, 8).

In conclusion (Table 2), the exposure of rat lungs to hyperoxia for 48 h did not cause serious tissue injury with alveolar flooding but did interfere with leukocyte kinetics especially in pulmonary venules and capillaries. Endothelial P-selectin expression was scatteringly augmented in hyperoxia-exposed arterioles, while ICAM-1 was obviously induced in hyperoxia-exposed venules and capillaries. Firm adhesion of leukocytes was not found in either arterioles or venules even in lungs harvested from animals exposed to hyperoxia. Leukocyte rolling was observed in both hyperoxia-exposed arterioles and venules, in which arteriolar rolling was conspicuously regulated via P-selectin- and ICAM-1-independent but L-selectin-dependent mechanisms, whereas venular rolling was predominantly mediated via P-selectin-independent but ICAM-1- and L-selectin-dependent mechanisms. Leukocyte accumulation within capillaries in lungs exposed to hyperoxia was considerably augmented via the pathway related to ICAM-1 even under a condition in which microvascular shear force is maintained. These findings may suggest that, in hyperoxia-exposed lungs, induction of adhesion molecules and their contribution to distorted leukocyte behavior are vascular specific. Furthermore, leukocyte kinetics appear to be quite different between the pulmonary and systemic circulation in several ways.