INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bleomycin (BLM), an antineoplastic agent, induces acute lung injury (ALI) followed by chronic fibrotic alterations in a dose-dependent manner in both human and experimental animals (1, 2). BLM-elicited fibrotic lung injury has been extensively used to investigate the mechanisms involved in the pathogenesis of pulmonary fibrosis because of close histopathological similarities to human idiopathic pulmonary fibrosis (IPF) (1, 3, 4). Several studies have suggested that enhanced expression of intercellular adhesion molecule-1 (ICAM-1) on the endothelial cell surface contributes to the development of human IPF (5). Upregulation of ICAM-1 was seen in BLM-induced fibrotic lung injury of the rat, with altered ICAM-1 distribution along the alveolar epithelia and endothelia (8, 9). ICAM-1 is expected to mediate not only leukocyte adherence to the microvascular endothelium but also leukocyte invasion into regions with serious inflammation (10). The functional importance of endothelial ICAM-1 on leukocyte behavior has been principally analyzed in the systemic microcirculation. Whereas some studies suggested that ICAM-1 mediates firm adherence between leukocytes and endothelial cells in injured venules of the systemic circulation (10, 13), we recently demonstrated that ICAM-1 induces leukocyte rolling along, but not firm adherence of leukocytes to, venular walls as well as sustained capillary entrapment of leukocytes in acute lung injury (14), suggesting that the ICAM-1-related adhesive pathway functions differently between the systemic and pulmonary microcirculation. It has not been reliably elucidated, however, whether this peculiar function of ICAM-1 in the pulmonary microcirculation holds universally true and can be investigated in the lung with chronic inflammation as well. Using a newly developed real-time laser confocal observation system (14), we analyzed the specific transitional changes in endothelial ICAM-1 expression and its effects on leukocyte recruitment in the pulmonary microcirculation of the BLM- injured lung, in which essential roles of endothelial ICAM-1 can be concurrently assessed in ALI and chronic fibrosis (FIB).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation of Animals and Administration of BLM

Specific pathogen-free Sprague-Dawley male rats, 8 wk of age, weighing 230 to 280 g were anesthetized with halothane and received a single intratracheal injection of 10 mg/kg of BLM hydrochloride (Nippon Kayaku Co., Tokyo, Japan) in 0.3 ml of sterile saline through a 26-gauge needle to the left bronchus (1, 3, 8). Control animals were given an equal volume of sterile saline only. A total of 223 animals were used and divided into control and BLM-treated groups. The animals were anesthetized with pentobarbital sodium (50 mg/kg, intraperitoneally), and killed by exsanguination from the aorta 2, 7, 21, and 42 d after BLM or saline administration.

BLM concentration in the sera was estimated by biological assay according to the method of Ohnuma and colleagues (17). At 15, 30, 45, 60, 90, 120, and 480 min after BLM injection, blood samples were collected and sera were preserved at 80° C until use.

The differential cell counts in the peripheral blood were microscopically examined for the animals before, and 1, 2, 6, 12, and 24 h followed by 2, 7, 21, and 42 d after BLM administration.

Examination of Histopathology, Collagen Contents, and Bronchoalveolar Lavage (BAL)

The lungs harvested from control and BLM-treated animals were inflated with formalin at a pressure of 30 cm H₂O for 48 h (n = 5 for each). Six sections, cut at identical intervals from the apex to the base of the left lung, were embedded in paraffin and stained with hematoxylin-eosin. The severity of pulmonary fibrosis in these sections was scored according to the numerical fibrotic scale proposed by Ashcroft and colleagues (18). More than 30 fields of each lung section were assessed and allotted a score from 0 (normal lung) to 8 (total fibrous obliteration of the field). The mean value of the scores from all fields was taken as representative of the extent of fibrosis.

Whole collagen content of the left lung was estimated by hydroxyproline assay. The lung tissue was hydrolyzed by addition of 6 N HCl at 110° C for 16 h in a sealed glass tube (Iwaki Co., Tokyo, Japan), and hydroxyproline content in the lung was determined by high-performance liquid chromatography (19).

BAL was carried out by a three times rewash method, in which 5 ml phosphate-buffered saline (PBS) was used for each wash. Total numbers of cells and their differential counts were examined on Diff-Quik-stained preparations (Kokusai Shiyaku, Kobe, Japan).

Preparation of Isolated Perfused Lungs

The animals were anesthetized with pentobarbital (50 mg/kg, intraperitoneally) and artificially ventilated with room air at a tidal volume of 10 ml/kg and respiratory rate of 60/min through the tracheal tube. After a median sternotomy had been made and heparin (5 mg/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 perfused at a constant flow rate of 10 ml/min with recirculation. A modified Krebs-Henseleit solution was used as the perfusate and 3% bovine serum albumin was added to maintain isosmotic pressure. Perfusate hematocrit was adjusted to 6.4 ± 0.4% by the addition of fresh blood obtained from donor rats injected with 5 mg/kg heparin. Because the total volume of perfusate was adjusted to 100 ml, heparin concentration in the circulating system approximates 6.5 µg/ml. To avoid the movement caused by artificial ventilation, the trachea was ligated at the end-inspiratory position and gas exchange was maintained with an extracorporeal membrane oxygenator (ECMO; Merasilox-S, Senko, Tokyo, Japan). A gas mixture containing 21% O₂ and 5% CO₂ in N₂ flowed into the ECMO, allowing adjustment of the perfusate PO₂, PCO₂, and pH to 142 ± 6 mm Hg, 39 ± 4 mm Hg, and 7.39 ± 0.05, 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 to maintain a temperature of 37 ± 0.4° C and to avoid desiccation of the lung surface. Pulmonary arterial pressures were continuously monitored by force displacement of pressure transducers (TP-400T; Nihon Kohden, Tokyo, Japan).

Experimental Protocols

We designed the following five experiments, in which 1A29 (Serotec, Oxford, UK), the monoclonal antibody against rat ICAM-1 (20), was administered to nearly half of the animals [1A29(+) group], while the remainder was not [1A29() group]. (1) Control group (n = 12): animals were killed 7 d after administration of an equal volume of sterile saline [1A29(), n = 6; 1A29(+), n = 6]. Because preliminary experiments indicated that morphologic characteristics, differential cell counts in the BAL fluid (BALF), and microcirculatory cell kinetics did not differ among the lungs sacrificed at any time after receiving saline alone, we assigned the animal group killed 7 d after saline treatment as the control. (2) B2d group (n = 12): animals were killed 2 d after BLM treatment [1A29(), n = 6; 1A29(+), n = 6]. (3) B7d group (n = 15): animals were killed 7 d after BLM treatment [1A29(), n = 9; 1A29(+), n = 6]. (4) B21d group (n = 16): animals were killed 21 d after BLM treatment [1A29(), n = 9; 1A29(+), n = 7]. (5) B42d group (n = 12); animals were killed 42 d after BLM treatment [1A29(), n = 6; 1A29(+), n = 6]. The perfusate concentration of 1A29 was maintained at 20 µg/ml. After a 10-min perfusion with 1A29, the microcirculatory kinetics of leukocytes and erythrocytes were investigated.

Because 1A29 consists of a mouse IgG (20), we separately conducted a series of negative control experiments using isotype-matched mouse IgG (Sigma, St. Louis, MO) for evaluating the specificity of 1A29 inhibiting ICAM-1 functions. We performed the negative control experiments in the B2d group (n = 6), because this group exhibited a marked abnormality in microvascular leukocyte kinetics as well as a significant upregulation of ICAM-1 along microvascular walls (see RESULTS). The perfusate concentration of mouse IgG was adjusted to 20 µg/ml.

Observation of Microvascular Cell Kinetics by Confocal Luminescence Microscopy

We used a real-time confocal laser scanning luminescence optical microscope (CSU10; Yokogawa, Tokyo, Japan) incorporating a high-speed video analysis system (14). The reflected light or fluorescent emission from the specimen was imaged onto a high-sensitivity charge-coupled-device (CCD) camera with an image intensifier (VSG; Kodak, San Diego, CA). By incorporating an excitation wavelength of 488 nm emitted from a low-power air-cooled argon laser (532-BSA04; Omnichrome, Cino, CA), the present confocal system allowed us to obtain apparently instantaneous images at 1,000 frames/s. The final magnifying power of our system reached ×484 with the ×20 objective. We registered confocal images at a rate of 250 frames/s by a high-speed video analysis system (1000 Processor; Kodak) connected to the image-intensified CCD camera. All images were stored in a video cassette recorder (SVQ-260; Sony, Tokyo, Japan).

Leukocytes were stained with carboxyfluorescein diacetate succinimidyl ester (CFDASE; Molecular Probes, Eugene, OR). CFDASE solution was injected into the femoral vein of donor rats (in vivo concentration, 1 mg/kg). CFDASE forms carboxyfluorescein succinimidyl ester (CFSE) in leukocytes and platelets but not in erythrocytes (21). After a 45-min incubation in vivo, blood was aspirated from the left ventricle and administered into the perfusion reservoir, allowing the lung to be circulated with perfusate in which 29 ± 3% of leukocytes were polymorphonuclear cells (PMNs) and the remainder was mononuclear cells (lymphocytes, 67 ± 6%; monocytes, 4 ± 2%). Leukocytes and platelets stained with CFSE were distinguishable during the microscopic measurements because of their distinctive size difference.

To examine the contribution of mononuclear cells to microvascular cell kinetics, we isolated mononuclear cells from the donor rat blood stained with CFSE by applying a classic Ficoll density gradient procedure (22). Five milliliters of CFSE-stained blood were laid on 3 ml of Ficoll-Conray solution (Immunobiological Co., Gumma, Japan), adapted for use with rat cells (specific gravity, 1.09). After centrifugation at 1,800 rpm for 30 min, the mononuclear cell layer was washed with PBS and administered into the perfusion reservoir. Measurements were taken for lungs harvested from control or BLM-treated animals killed 2, 7, 21, and 42 d later (n = 5 for each). In this series of experiments, 1A29 was not administered into the reservoir. Preliminary experiments confirmed that the mononuclear cell layer separated by the Ficoll contained few PMNs and that was mostly composed of lymphocytes (95 ± 2%).

To examine the erythrocyte kinetics in the same microvessels used for assessment of leukocyte behavior, erythrocytes were stained with fluorescein isothiocyanate (FITC) (Sigma). Fresh rat blood was centrifuged at 1,000 rpm for 5 min and the buffy coat was discarded. The packed erythrocyte solution was diluted with PBS and FITC was added to give a final concentration of 0.1 mg/ml. This solution was added to the reservoir after the leukocyte kinetics had been measured.

To examine the architecture of microvessels in which leukocyte and erythrocyte behavior had been analyzed, 200 µl of 5% FITC-dextran with a molecular weight of 145,000 (Sigma) was added to the reservoir. The microvessel from which fluorescence-labeled blood cells entered the capillary networks was defined as the arteriole, while the microvessel into which blood cells flowed from the capillary networks was taken to be the venule.

Analysis of Blood Cell Kinetics in Pulmonary Arterioles and Venules

Replaying the videotapes obtained with a high-speed video analysis system at the normal video rate allowed the axial velocities of CFSE-labeled leukocytes (w) as well as FITC-labeled erythrocytes (r) to be determined by measuring the distance traveled between two or more successive video frames. The highest value of r was assumed to be equal to the centerline velocity (max). max was used to estimate mean fluid flow (mean) in each microvessel, where mean =  max/1.6 (23). Wall shear rate () was estimated from the equation;  = 8(mean/D), where D represents vessel diameter (24).

An adherent leukocyte in an arteriole or a venule was defined as a cell that 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 (23), who calculated the critical velocity (crit) of a freely flowing cell traveling close to, but not adhering to, the microvascular wall. Any cell flowing at a velocity below crit was assumed to be slowed by the interaction with the vascular endothelium and defined as a rolling cell. Firm adhesion and rolling of mononuclear cells in arterioles and venules were similarly analyzed.

Analysis of Blood Cell Kinetics in Pulmonary Capillaries

w and r in capillaries were measured with the same procedure as that applied for arterioles and venules. Leukocytes or mononuclear cells retained continuously in capillaries were excluded from the calculation of w. To quantify the cell transit in capillaries, we classified the observed leukocytes or mononuclear cells into three categories: ( 1) cells moving smoothly without interruption; (2) cells stopping transiently for more than 40 ms but rejoining the free-flowing stream within 2 s; and (3) impeded cells stopping completely within a capillary segment for more than 2 s.

Quantitation of ICAM-1 Expression Along Microvessel Walls

We examined ICAM-1 expression along microvessel walls by intravital luminescence immunochemistry (n = 3 for each). This novel method allows us to reliably separate arterioles from venules in association with quantitative determination of ICAM-1 along microvessel walls (14, 15, 25). 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 and 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 conjugation between endothelial ICAM-1, 1A29, and FITC-IgG antibody. Endothelial ICAM-1 bound to 1A29 and FITC-IgG antibody was determined under confocal scanning luminescence microscopy and displayed on a high-sensitivity CCD camera (TEC-470; Optronics, Goleta, CA). As 1A29 is 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. A quantitative analysis of the ICAM-1 fluorescent intensity along microvessel walls was performed by processing a confocal image with a digital image-analyzing system (Quadra 840AV/Image 1.58; Apple, Cupertino, CA), and normalized by the mean ICAM-1 fluorescent intensity along venule walls of saline-treated control lungs.

Statistical Analysis

The results are presented as means ± SD. Significant differences among the experimental 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 under 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 Characteristics of BLM-injured Lungs

The average serum BLM levels were 4.3, 5.4, 8.4, 3.8, 2.9, and 1.8 µg/ml at 15, 30, 45, 60, 90, and 120 min, respectively. The BLM levels were, however, less than 0.1 µg/ml 480 min later.

The relative numbers of PMNs in the blood before BLM administration were 29 ± 3%, and were markedly reduced within 2 h of treatment (1 h, 5 ± 5%; 2 h, 6 ± 6%). Blood PMN numbers were quickly restored to the control level 6 h later, and were maintained without any variation thereafter (6 h, 29 ± 4%; 12 h, 24 ± 12%; 24 h, 37 ± 11%; 2 d, 27 ± 1%; 7 d, 32 ± 7%; 21 d, 30 ± 4%; 42 d, 31 ± 4%).

The lung treated with saline only showed normal alveolar structure. Two days after BLM administration (B2d group), the lung exhibited infiltration of leukocytes in edematous perivascular and peribronchiolar regions and alveolar spaces, and slightly thickened alveolar walls. Seven days later (B7d group), the extent of infiltration of leukocytes and thickening of alveolar walls progressed further. In addition, hyperplasia of alveolar type II cells was evident. Twenty-one days later (B21d group), diffuse and moderate fibrosis of alveolar walls was observed. Forty-two days later (B42d group), the severity and extent of pulmonary fibrotic changes were augmented and honeycombing was clearly recognized.

The Ashcroft scores of BLM-injured lungs were much greater in the B21d (4.4 ± 1.1) and B42d (5.3 ± 1.0) groups than in the control (0.2 ± 0.4), B2d (1.6 ± 1.0), and B7d (1.8 ± 1.1) groups. The Ashcroft score of the B42d group was significantly greater than that of the B21d group. Hydroxyproline contents in lungs treated with saline averaged 1 mg/lung, with little variation regardless of the time of observation. Similarly, the hydroxyproline contents in BLM-treated lungs were found to be much greater in the B21d (2.0 ± 0.4 mg/lung) and B42d (2.1 ± 1.2 mg/lung) groups than in the control (1.0 ± 0.2 mg/ lung) and B2d (1.0 ± 0.3 mg/lung) groups.

The relative frequency of PMNs in the BALF in BLM-treated lungs was consistently higher than that in control lungs, and the maximal increase was observed in the B2d group (69 ± 7%). PMN numbers were reduced in the B7d group (39 ± 7%) and reached a plateau in the B21d (17 ± 4%) and B42d (17 ± 10%) groups, although PMN numbers in these three groups were still higher than those of the control group (0.3 ± 0.2%).

Changes in Hemodynamic Parameters in BLM-injured Lungs

Confocal luminescence microscopy provided images that facilitated assessment of the detailed architecture of the microvascular network in the pulmonary acinus (Figure 1). There was no difference in vascular diameter of precapillary arterioles or postcapillary venules between the control and BLM groups (ranging from 18 to 30 µm; Table 1). However, the capillary diameter was appreciably decreased in the B2d, B7d, and B21d groups, but increased in the B42d group as compared with the control (Table 1). The mean in either arterioles or venules did not differ between the control and BLM groups, indicating that the arterioles or venules of all experimental groups were exposed to equivalent flow-induced shear force. The mean values in arterioles averaged 0.9 ± 0.5 mm/s, corresponding to a wall shear rate of 327 ± 166 s¹. Venular mean averaged 1.6 ± 0.9 mm/s, giving a wall shear rate of 522 ± 285 s¹. Both the mean and shear rates obtained for venules were much greater than those for arterioles. The mean in capillaries did not differ between any of the experimental groups and averaged 0.7 ± 0.2 mm/s. Although the capillary diameter was altered during the development of BLM-induced lung injury, capillary wall shear rates were statistically identical among the groups, averaging 790 ± 269 s¹, owing to variations in mean.

View larger version (122K): [in this window] [in a new window]   Figure 1.   Confocal image of the in-focus plane at 25 µm from the lung surface. (A) Control. (B) B42d group. Original magnification: ×484. v = venule; c = capillary. Bar, 10 µm. In both images, the venule was in focus but the capillary was out of focus, thus allowing determination of the venular diameter from these images. The capillary diameter was similarly examined in the confocal images focused on the capillary boundaries.

Leukocyte Kinetics in Arterioles and Venules

We found no leukocytes adhering firmly to the endothelium of arteriolar and venular walls under any experimental condition.

In the control group, the relative frequency of rolling leukocytes was higher in arterioles than in venules. In the BLM groups, the numbers of rolling leukocytes were not changed in arterioles, but were much higher in venules of the B2d, B7d, and B21d groups than the control group (Figure 2). In the B42d group, however, the rolling leukocyte number in venules was reduced to a level comparable with that observed for the control group. Venular rolling was conspicuous in the B2d and B21d groups in comparison with that in the B7d group. The augmented leukocyte rolling in BLM-injured venules was inhibited by anti-ICAM-1 monoclonal antibody (Figure 2). The number of leukocytes moving slowly along arteriolar walls in BLM-injured lungs was not influenced by anti-ICAM-1 monoclonal antibody (Figure 2).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Relative frequency of rolling leukocytes along microvascular walls. (A) Arterioles. (B) Venules. Control: received saline alone; B2d: 2 d after BLM instillation; B7d: 7 d after BLM; B21d: 21 d after BLM; B42d: 42 d after BLM. Leukocyte kinetics were evaluated without (open bars) and with (solid bars) 1A29. *Larger than the value obtained for the control. Larger than the value obtained for the B7d. #Smaller than the value obtained for the corresponding group without 1A29.

Mononuclear cells did not show firm adhesion to arteriolar and venular walls in both control and BLM-injured lungs. In the control lung, the relative frequency of rolling mononuclear cells was 8% in arterioles and 5% in venules, with no significant difference between these values (Table 2). The numbers of rolling mononuclear cells in arterioles and venules in BLM-injured lungs did not differ from those in the control lungs.

Leukocyte Kinetics in Capillaries

The relative frequency of leukocytes transiently entrapped within capillary segments was enhanced in the B2d, B7d, and B21d groups, but not in the B42 group (Figure 3). The number of leukocytes transiently entrapped in capillaries was not influenced by monoclonal antibody against ICAM-1 (Figure 3). The frequency of firmly tethered leukocytes was also increased in the B2d, B7d, and B21d groups, but this increase was inhibited by the antibody against ICAM-1 (Figure 3). Although the frequency of capillary leukocytes with sustained arrest in the B42d group did not differ from that in the control group, anti-ICAM-1 antibody significantly reduced this frequency.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Leukocyte entrapment in capillaries. Groups studied are the same as those defined in Figure 2. (A) Relative frequency of leukocytes transiently stopped. (B) Relative frequency of leukocytes persistently stopped. *Larger than the value obtained for the control. #Smaller than the value obtained under a condition without 1A29.

In mononuclear cell experiments, the relative frequency of cells transiently entrapped within capillary segments did not differ between the control and any BLM-treated group (Table 2). The mononuclear cell numbers with sustained arrest were, however, slightly increased in the B21d group (Table 2).

Specificity of 1A29 for ICAM-1 Inhibition

There was no difference in rolling leukocyte numbers in arterioles of all the B2d groups, including B2d without any additional medication (27 ± 15%), B2d with 1A29 (20 ± 5%), and B2d with mouse IgG (25 ± 2%). The numbers of rolling leukocytes in venules did not differ between the B2d group without further medication (27 ± 10%) and that treated with mouse IgG (22 ± 6%), although 1A29 administration significantly inhibited the increase in rolling leukocytes (12 ± 13%). The enhanced frequency of leukocytes transiently entrapped within capillaries in the B2d group (33 ± 7%) was not corrected by either 1A29 (34 ± 3%) or mouse IgG (31 ± 8%), whereas the enhanced frequency of capillary leukocytes with sustained arrest (18 ± 5%) was inhibited by 1A29 (7 ± 2%), but not by mouse IgG (17 ± 3%).

ICAM-1 Expressions along Microvessel Walls

In negative control experiments involving the administration of mouse IgG and FITC-labeled anti-mouse IgG antibody, no luminescence was detectable along microvascular walls in any experimental group. Intravital confocal luminescent analysis confirmed that ICAM-1 was expressed along venular and capillary walls in both control and BLM-injured lungs, though little ICAM-1 was detected along arteriolar walls (Figure 4). With the exception of the B42d group, venular ICAM-1 was upregulated during the development of BLM-induced lung injury (Figure 5). Venular ICAM-1 in BLM-injured lungs exhibited two peaks at 2 and 21 d after BLM treatment. The quantitative ICAM-1 expression in capillaries differed from that in venules, in that capillary ICAM-1 enhancement was sustained up to 42 d after BLM administration, resulting in more distinct capillary ICAM-1 expression than that in venules in the B7d and B42d groups (Figure 5).

View larger version (151K): [in this window] [in a new window]   Figure 4.   Confocal view of ICAM-1 distribution along microvascular walls in the control lung (A) or in the BLM-injured lung (B). BLM- injured lung: 21 d after BLM administration. Objective lens, ×20. v = venule; c = capillary. Bar, 10 µm.

View larger version (15K): [in this window] [in a new window]   Figure 5.   Fluorescent intensity of ICAM-1 along microvessel walls. Groups are the same as those defined in Figure 2. The most intense fluorescence along the microvascular wall was taken as the representative of ICAM-1 expression in each microvessel. *Larger than the value obtained in the control venule. Larger than the value obtained for B7d venule. ¶Larger than the value obtained in the control capillary. §Larger than the value obtained in the venule of the same group.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critique of Methods

Kawamoto and Fukuda (26) reported that interstitial cells began to proliferate 2 d after BLM treatment, and endothelial cells proliferated in small vessels and in capillaries 7 d after BLM administration in the rat. They also showed that fibrotic areas in alveoli were covered with alveolar epithelial cells 14 d after BLM treatment, and numerous interstitial cells were tightly arranged in fibrotic areas 21 d later and persisted for 56 d. In addition, the amount of collagen, laminin, or fibronectin was increased 21 d after treatment (3, 27). Our findings obtained from morphological analysis and hydroxyproline determination were qualitatively consistent with the previous studies as described earlier. Taken together, conspicuous fibrotic alterations are expected to develop 2 or 3 wk after BLM administration. Furthermore, we found that inflammatory cell numbers in the airspace composed mainly of PMNs were markedly increased at the early phase (2 d), but gradually decreased at the later phases (21 and 42 d) after BLM administration, suggesting that BLM-induced lung injury may be divided into two stages involving acute injury and subsequent fibrosis. Based on the transitional changes in Ashcroft score, hydroxyproline content, and inflammatory cell counts in the BALF, we considered that the turning point between acute injury and FIB may fall between 7 and 21 d after administration of BLM. To simplify analyses, we therefore defined 2 and 7 d after BLM treatment as the phase of ALI, and 21 and 42 d after BLM as the phase of FIB. In support of this definition, Shen and coworkers (28) found that the severity of BLM- induced pulmonary fibrosis is closely related to that of ALI initiated by BLM treatment. Furthermore, in the BLM model of interstitial lung disease, an initial phase of injury represented by alveolitis was demonstrated to eventually progress to chronic inflammation characterized by fibrotic alterations (4).

Although our confocal method is extremely useful for estimating the dynamic behavior of leukocytes as a whole, it does not allow us to simply investigate the cell kinetics of each leukocyte population because both PMNs and mononuclear cells are stained with CFSE and are indiscernible during the microscopic measurements. We therefore investigated pure mononuclear cell dynamics in addition to the cell kinetics of the whole leukocyte population. Because relative numbers of PMNs and those of mononuclear cells (mainly lymphocytes) in the perfusate were 30% and 70%, respectively, mononuclear cell kinetics data (Table 2) indicated that 75% of rolling leukocytes in arterioles, 70% of rolling leukocytes in venules, and 85% of entrapped leukocytes in capillaries in control lung were PMNs where diluted whole blood was used for analysis. Similarly, 75% of abnormal leukocytes in BLM-injured arterioles in the whole leukocyte experiment were PMNs. In addition, 70% of abnormal leukocytes in venules and 85% of entrapped leukocytes in capillaries in the BLM-injured lung were thought to be PMNs in the whole cell experiment, indicating that the unusual leukocytes observed in the experiments in which whole leukocyte kinetics were analyzed may be primarily composed of abnormally behaving PMNs.

Although the specificity of 1A29, the monoclonal antibody against ICAM-1, was clearly demonstrated, it may not be concluded immediately that the abnormal leukocyte kinetics observed in isolated perfused lungs treated with BLM are closely related to upregulated ICAM-1 and resemble those under in vivo conditions. This is because a couple of artificial manipulations were made in the isolated lung preparation, including the addition of heparin and leukocytes obtained from intact donor rats. Heparin has been shown to have complicated effects on endothelium-leukocyte interactions (29). Miller and coworkers (29) reported that heparin would inhibit ICAM-1 expression in human endothelial cells, whereas Penc and coworkers (30) failed to demonstrate inhibitory effects of heparin on ICAM-1 expression in human microvascular endothelial cells. Nelson and coworkers (31) found that heparin would be L- and P-selectin inhibitors, thus reducing leukocyte rolling in injured systemic microvessels. Diamond and coworkers (32) suggested that heparin would inhibit endothelial-leukocyte interactions through the CD11b/CD18 pathway in a concentration-dependent manner.

Taken together, these findings may indicate that heparin appreciably inhibits endothelial-leukocyte interactions causing rolling and/or firm adherence. In our observation system, a small amount of heparin is indispensable for avoiding the aggregation of blood cells in the perfusion circuit connected with the isolated lung. To minimize the artificial effects of heparin on endothelial-leukocyte interactions, we adjusted heparin concentration in the perfusion circuit to 6.5 µg/ml, the concentration similar to that of heparinlike substances in normal human plasma (33). In addition, we should emphasize the fact that all experiments in the present study were conducted in the presence of the same concentration of heparin, indicating that comparison of changes in microvascular leukocyte kinetics under various experimental conditions would make sense, although the absolute frequency of abnormally behaving leukocytes might be underestimated by heparin. Furthermore, as the leukocytes used for analysis were obtained from intact donor rats, we cannot exclude the possibility that the leukocyte adhesion molecule expression might qualitatively differ in vivo, which could also affects leukocyte kinetics in the pulmonary microcirculation.

Transitional Changes in ICAM-1 Expression along BLM-injured Microvessel Walls

ICAM-1 was detected in the acute phase of BLM-induced pulmonary fibrosis (8). Kasper and coworkers (9) demonstrated that ICAM-1 was involved in BLM-induced rat pulmonary fibrosis. However, the importance of ICAM-1 at various time points during the development of BLM-induced lung injury, including ALI and FIB, has not been fully evaluated. Leeuwenberg and coworkers (34) showed that in vitro ICAM-1 expression in human endothelial cells activated by tumor necrosis factor remained elevated for 3 d, with a maximum at 8 to 10 h. Norris and coworkers (35) reported that in vivo ICAM-1 expression was elevated for 7 d in experimental cutaneous inflammation. These findings are qualitatively consistent with those obtained in the present study, though the primary insult inducing acute inflammation is different. We found that BLM-related ICAM-1 was predominantly upregulated in venules and capillaries, but not in arterioles (Figures 4 and 5). Furthermore, we demonstrated that the transitional changes in BLM-elicited ICAM-1 expression were highly vascular-specific, in that venular ICAM-1 showed upregulation with two peaks, whereas the elevation of capillary ICAM-1 was sustained over time after BLM treatment. We thought that venular ICAM-1 is enhanced at the early phase (2 d) probably due to acute, serious inflammation caused by BLM-mediated injury, but it declines as the acute inflammation is extinguished (7 d). However, venular ICAM-1 was again augmented at the early FIB phase (21 d). This suggests that there are specific stimuli that regulate venular ICAM-1 expression that may not be immediately related to the initial injurious effect of BLM, but are closely associated with factors inducing fibrotic alterations, because serum BLM concentrations were rapidly reduced to undetectable levels within 8 h after BLM treatment. However, capillary ICAM-1 expression was qualitatively different, showing sustained elevation during both the ALI and FIB phases (Figure 5), indicating that the sensitivity of ICAM-1 upregulation in responding to either acute or chronic inflammation is higher in capillaries than in venules.

Role of ICAM-1 in Mediating Leukocyte Behavior in BLM-injured Microvessels

Corresponding to ICAM-1 expression along microvessel walls, leukocyte kinetics in BLM-injured lungs were not distorted in arterioles, but were abnormal in venules (Figure 2). The transitional changes in rolling leukocyte numbers in BLM-injured venules were well correlated with those in ICAM-1 expression (Figures 2, 4, and 5), leading to the conclusion that rolling of venular leukocytes in BLM-related lung injury can be mainly attributed to ICAM-1 expressed on the venular surface. This is supported by the fact that leukocyte rolling in BLM-injured venules was significantly inhibited by a monoclonal antibody against ICAM-1 (Figure 2). We should give emphasis to the fact that leukocytes did not adhere firmly to the venular endothelium in BLM-injured lungs, though ICAM-1, the key adhesion molecule composed of the immunoglobulin superfamily causing firm leukocyte tethering in systemic venules (10, 13, 36), was significantly upregulated in BLM-treated lungs (Figures 4 and 5). These findings are qualitatively consistent with those observed in our previous studies (14, 15), in which venular ICAM-1 was demonstrated to contribute little to the firm adhesive interaction between leukocytes and venular endothelial cells, but was important for inducing leukocyte rolling along venular walls in ALI evoked by hyperoxia exposure. These findings strongly suggest that the leukocyte-endothelium interaction is quite different between the pulmonary and systemic microcirculation.

ICAM-1 expression alone can not account for the gamut of abnormal leukocyte behavior in BLM-injured capillaries (Figure 3). Although transient entrapment of leukocytes in capillaries was enhanced at the ALI and early FIB phases (2, 7, and 21 d after BLM), it was minimally inhibited by treatment with anti-ICAM-1 antibody (Figure 3), indicating that augmented capillary endothelial ICAM-1 plays little role in regulating abnormal leukocyte behavior represented by transient entrapment. This is also supported by the experimental data observed 42 d after BLM administration, in which the leukocyte number entrapped momentarily in capillaries was reduced to a control level, though endothelial ICAM-1 expression remained elevated (Figures 3 and 5). Changes in mechanical impact in BLM-injured lungs may be another mechanism of enhancement of transient entrapment of capillary leukocytes independent of endothelial ICAM-1 (Table 1). The extent of capillary entrapment of leukocytes partly depends on the geometry of the pulmonary capillary network in which a great fraction of capillary bed contains narrow portions for leukocyte transit (15, 16, 37). Serious inflammation triggered by BLM may evoke a considerable accumulation of the exudation fluid in the interstitium, which is expected to compress the capillary network, leading to further increases in the narrow regions. In fact, we found that the capillary diameter in BLM-injured lungs, except 42 d after BLM, was smaller than the control capillary diameter (Table 1). Interestingly, however, the capillary diameter at the established FIB stage with honeycombing (42 d later) was increased and the mean diameter was appreciably greater than that of the control diameter (Table 1). The increased vascular diameter may allow leukocytes to move more smoothly through the capillary network, thus decreasing numbers of transiently entrapped leukocytes (Figure 3).

In contrast, sustained leukocyte entrapment in BLM- injured capillaries appears to be largely mediated via an ICAM-1-related pathway, because the augmented frequency of leukocyte entrapment was significantly reduced by the addition of anti-ICAM-1 antibody (Figure 3). These findings suggest that enhanced ICAM-1 plays a major role in mediating a firm interaction between leukocyte and capillary endothelium in most stages of BLM-induced lung injury. However, numbers of firmly entrapped leukocytes were reduced at the established FIB stage even in the absence of anti-ICAM-1 antibody (Figure 3), further indicating the importance of mechanical impact caused by capillary diameter changes.

In conclusion, BLM elicited ALI followed by FIB, which were discriminated at approximately 2 wk after BLM treatment. Vascular diameter was significantly altered in capillaries, but not in arterioles or in venules of BLM-injured lungs. Endothelial ICAM-1 expression was highly vascular-specific, i.e., little enhancement in arterioles, two-peak upregulation in venules, and sustained enhancement in capillaries. Augmented ICAM-1 did not induce the firm adherence of leukocytes to venular endothelial cells, but did mediate the leukocyte rolling along venular walls. In BLM-injured capillaries, ICAM-1 was not the main factor responsible for transient leukocyte entrapment, which was predominantly attributed to changes in the capillary diameter. However, sustained leukocyte entrapment in BLM-injured capillaries appeared to be governed by changes in endothelial ICAM-1 expression as well as capillary diameter changes. These findings suggest that endothelial ICAM-1 is essential for abnormal leukocyte accumulation in capillaries and venules at both ALI and subsequent fibrosis in BLM-induced lung injury.