Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The general problem of O₂ toxicity involves several organs, but the lungs are directly exposed and are therefore the principal site of injury (1). Therapy with supraphysiological concentrations of O₂ (hyperoxia) is required in a number of clinical situations but may result in lung tissue damage (1). Pulmonary O₂ toxicity has been the subject of many investigations over nearly 30 yr (2), and this work has consistently shown that O₂ toxicity gives rise to specific lung injury (5, 6). Different cell types of the lung vary greatly in their susceptibility to hyperoxia. Capillary endothelial cells are a primary target of O₂ toxicity. The first signs of hyperoxic lung injury involve subtle changes in endothelial cell ultrastructure, which lead to pericapillary accumulation of fluid (5). With further hyperoxic exposure, platelets are retained in the microvasculature of the lung. Following platelet accumulation, neutrophils are rapidly recruited. The appearance of neutrophils is associated with marked accentuation of lung injury and a rapid decline in lung function. However, the endogenous regulation of inflammation during hyperoxic lung injury is still poorly understood.

The recruitment of leukocytes into areas of inflammation is dependent on cell adhesion molecules (CAMs) (7, 8), which mediate an adhesion cascade involving rolling, activation, binding, and transmigration through endothelial cells. There are at least 10 different CAMs, each involved in a specific step of leukocyte-endothelial cell interactions (7). However, the expression of CAMs during organ development suggests that their roles are not confined to leukocyte-endothelial cell interactions (9). Recently, increased intercellular adhesion molecule-1 (ICAM-1) expression was described during hyperoxic lung injury (13, 14). Both in vitro (15) and in vivo (13, 16) evidence suggests that alveolar epithelial cells, not endothelial cells, are the major site of ICAM-1 production during O₂ injury. Therefore, changes in any single CAM are not sufficient to explain the complex process of inflammation in response to hyperoxia.

The platelet-endothelial cell adhesion molecule-1 (PECAM-1) is a 130-kD molecular weight protein and a member of the immunoglobulin gene superfamily. Molecular cloning has revealed the presence of six external immunoglobulinlike domains, a short transmembrane domain encoded by a single exon, and a relatively long cytoplasmic tail encoded by six different exons (17, 18). PECAM-1 is found in large amounts on endothelial cells and is less abundant on platelets and most leukocytes (19, 20). It is localized at the junctions between endothelial cells, and when fibroblasts were transfected with PECAM-1 complementary DNA (cDNA), it was found at the junctions of these cells as well (12). PECAM-1 is thought to play a role during thrombosis (12) and appears to be required for neutrophil transmigration (21). Intravenous injection of anti-PECAM-1 antibody resulted in a 75% decrease in neutrophil sequestration in the lung in a model of immune-complex injury (20). PECAM-1 is also expressed at specific stages of lung development (10). Moreover, alternatively spliced isoforms of PECAM-1 have been documented both in humans (17) and mice (18, 22). However, the regulation and the role of these different isoforms are still unknown.

Because pulmonary inflammation is a notable component of pulmonary O₂ toxicity and PECAM-1 has an established role in the lung and inflammation, we hypothesized that PECAM-1 expression is regulated in the response to hyperoxic lung injury. In this report, we examine increases in the expression of PECAM-1 and its isoforms in mouse lungs during exposure to 100% O₂.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals, O₂ Exposure, and RNA Isolation

Adult (8 wk old) male mice of C3H/HeJ strains (Charles River Canada, St. Constant, PQ, Canada) were housed three to four per cage (28 × 17 × 12 cm each) with food and water available ad libitum. Up to 12 cages were put in a large Plexiglas box that was flushed with 3 L/min of 100% O₂. The concentration of O₂ was determined with a MiniOX III O₂ monitor (MSA Catalyst Research, Owings Mills, MD), and remained above 95% for the entire experiment. The mice were kept in the chamber for up to 96 h, when they showed signs of lung injury but were still alive (23). All animals were killed by cervical dislocation immediately upon return to room air.

Membrane Hybridization

The whole lungs were perfused with cold buffered saline, flash frozen in liquid N₂, and homogenized in guanidine isothiocyanate solution. Total RNA was isolated by sedimentation through 5.7 M CsCl. RNA concentrations were determined by spectrophotometric absorption at 260 nm. Total RNA was fractionated by formaldehyde gel electrophoresis and transferred onto nylon membranes (MagnaGraph; MSI, Westboro, MA), or slot blots were prepared from a single dilution of glyoxal-denatured RNA as described previously (24). Each slot was sliced in half, and one half of each slot was hybridized to a control cDNA probe encoding mouse CuZn superoxide dismutase (CuZn SOD; gift from Dr. Y. S. Ho, Institute of Chemical Toxicology, Wayne State University, Detroit, MI). CuZn SOD mRNA was chosen because its mRNA abundance in whole-lung homogenates does not change during hyperoxic lung injury (our observations and Reference 25). The other half of each slot was hybridized with ³²P-labeled cDNA probes for the full-length mouse PECAM-1 (18). Blots were hybridized at 68°C for approximately 16 h in 1× SSPE (10 mM NaH₂PO₄, 150 mM NaCl, 10 mM EDTA), 2× Denhardt's solution (2% wt/vol ficoll, polyvinylpyrolidone, bovine serum albumin, and 20 mM EDTA), 10% dextran sulfate, and 2% sodium dodecyl sulfate plus 200 µg/ml of sheared and denatured salmon sperm DNA with 2 × 10⁶ cpm probe/ml hybridization buffer. The membranes were then washed, 30 min each time, once in 0.1× saline sodium citrate, 0.5% sodium dodecyl sulfate at room temperature with shaking, followed by four washes in the same buffer at 68°C without shaking. After washing, the membranes were blot dried, wrapped in Saran Wrap, and exposed for autoradiography at 80°C using X-AR X-ray film (Eastman Kodak Co., Rochester, NY) and MCI Lightning Plus intensifying screens (MCI, Cedar Knolls, NJ).

Alternative Splicing Studies

To determine if alternatively spliced forms of PECAM-1 are expressed in mouse lungs (17) and to see if exposure to hyperoxia affects the distribution of the different isoforms, message-specific primers were used on RT-PCR assays. For the cytoplasmic domains, the sense primer was chosen from exon 10 (CCAGCTGCTCCACTTCTGAA), and the antisense primer was complementary to part of the exon 16 (GCACTGCCTTGACTGTCTTA) (18). For the transmembrane domain, the sense primer was chosen from the last immunoglobulinlike domain (CGACAGTATGAGGACCAGTCCCAGAAGCAG), and the antisense primer was complementary to part of the exon 10 (TTCAGAAGTGGAGCAGCTAA). To increase the sensitivity of detection, the antisense primer was labeled in 5' with [-³²P]-ATP using a T4 polynucleotide kinase (Boehringer Mannheim, Laval, PQ, Canada).

The reverse transcription (RT) and the polymerase chain reaction (PCR) were performed in one step using the Tth DNA polymerase (Boehringer Mannheim). The RT was performed with 200 ng of total RNA in 1× RT-buffer (Boehringer Mannheim), 9 mM MnCl₂, 10 mM dNTP mix, 30 µM of labeled antisense primer, and 4 U Tth DNA polymerase per 20 µl of reaction mix at 65°C for 20 min. The volume of reaction was increased to 100 µl to obtain a final concentration of 1× PCR buffer (Boehringer Mannheim), 7.5 mM ethyleneglycol tetraacetic acid (EGTA), and 30 µM of the sense and antisense primers. Twenty-nine cycles were performed using a program that included incubation at 94°C for 30 s, 55°C (for the cytoplasmic domain) or 65°C (for the transmembrane domain) for 30 s, and at 72°C for 1 to 6 min (20 s increase at each cycle). Samples were taken at cycles 20, 25, and 30. The PCR products were separated on 6.5% polyacrylamide gels. The gels were dried under vacuum and exposed for autoradiography. The relative abundance of each band was quantified using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The four major bands resulting from the amplification of the cytoplasmic domain were excised, solubilized in water, and reamplified (26). The final products were sequenced using the ABI Prism Dye terminatorcycle sequencing radioactive kit with the amplitaq polymerase SS (Perkin Elmer, Mississauga, ON, Canada) and analyzed using the ABI model 373 DNA sequencing stretch (Perkin Elmer). Additionally, for the transmembrane domain, the unlabeled PCR products were separated on agarose gel and transferred onto nylon membranes (MagnaGraph; MSI) for two Southern blot hybridizations using either a radioactively labeled full length cDNA clone or an oligonucleotide probe corresponding to a region of the transmembrane domain (CGATGACCACTCCAATGACAACCACCGCAA). Hybridization was detected by autoradiography (26).

In Situ Hybridization at the Light Microscope Level

The trachea was exposed and cannulated with a 22-gauge catheter, and the lungs were instilled with 10% phosphate-buffered formalin at 20 cm H₂O pressure. After 10 min of instillation, the lungs were removed and immersed in 10% phosphate-buffered formalin (for up to 18 h) until they were embedded in paraffin. Four-micron sections were used for hybridization.

Complementary RNA probes (from a full-length cDNA) were labeled with [-³³P]-UTP (DuPont Canada, Markham, ON, Canada) as previously described (27). The DNA template was digested with RNAse-free DNAse I (Promega, Madison, WI), and the probe was extracted with an equal volume of phenol:chloroform (1:1), precipitated out of ethanol, and resuspended in diethylpyrocarbonate-treated water. Before hybridization, limited alkaline hydrolysis of the RNA probes was performed to reduce the transcript length to 0.1 to 0.3 kb. Partially hydrolyzed transcripts were sized by agarose gel electrophoresis. Hybridization methods were as described previously (27, 28). Hybridization was performed overnight at 54°C in 50% formamide, 0.3 M NaCl, 10 mM Tris-Cl (pH 8.0), 1 mM EDTA, 1× Denhardt's solution, 10% dextran sulfate, 0.5 mg/ml yeast tRNA, and 48 ng/ml of cRNA probe in a moist chamber containing 0.3 M NaCl, 50% formamide. After hybridization, the sections were treated with RNAse (27). The final wash was performed in 0.1× SSC for 30 min at 68°C. Slides were dipped in NBT-2 photographic emulsion (Eastman Kodak Co.), exposed for 10 d, developed, and counterstained with hematoxylin and eosin before photomicrography. At least three animals were studied at each time point. Nonspecific binding and background were evaluated using sense-strand probes. Photomicrographs were recorded using a Nikon Optiphot microscope (Nikon Corp., Melville, NY) containing a Darklite stage illuminator (Micro Video Instruments, Inc., Avon, MA). The Darklite illuminates silver grains in the emulsion, yielding an effect similar to conventional dark-field illumination, in which silver grains appear bright against a dark background. This was used simultaneously with sufficient conventional bright-field illumination, so that histologic features were also apparent. Photomicrographs were recorded on Ektachrome 35-mm film (Eastman Kodak Co.) and images transferred to a Photo CD (Eastman Kodak Co.). Images were digitally adjusted for contrast and color balance using Adobe Photoshop 3.0a (Adobe Systems Inc., Mountain View, CA) prior to printing. A semiquantitative analysis of the in situ hybridization was performed using an adapted scale (Table 1). The expression of PECAM-1 mRNA in vessels and at the alveolar level was evaluated separately. We used continuous values ranging from 0 (no signal) to 4 (intense signal). Three animals were studied by time point and results were confirmed by two independent observers.

In Situ Hybridization at the Electron Microscope Level

The trachea was exposed and cannulated with a 22-gauge catheter, and the lungs were instilled with 1.6% glutaraldehyde in 0.1 M phosphate-buffered pH 7.3 (Sjorensen's buffer) at 20 cm H₂O pressure. After 10 min of instillation, the lungs were removed, cut in 1-mm³ pieces, and fixed in the same solution for 90 min at 4°C. After being rinsed in Sjörensen's buffer (3 × 10 min), the blocks were dehydrated in an ethanol series and embedded in Lowicryl K4M (Chemishe Werke Lowi, Walduzeiburg, Germany) following the protocol recommended by the supplier. Polymerization of the capsules was carried out under ultraviolet light (360 nm wavelength) at a distance of 20 cm at 4°C for 72 h. Ultrathin sections (30-40 nm) were cut with a diamond knife on a Reichert-Jung ultramicrotome (Wien, Austria). Sections were collected on nickel grids (300 mesh) coated with Formvar membranes, and stored at 4°C in the dark until use.

All of the following steps were carried out in a moist chamber on the surface of Parafilm M laboratory film (American National Can, Greenwich, CT). Grids with ultrathin sections were first inverted and floated over 25 µl of proteinase K solution (2 µg/ml in H₂O) (Sigma, Mississauga, ON, Canada) for 15 min at 37°C. The grids were further rinsed over three successive drops of H₂O, and air dried. Complementary RNA probes were labeled with digoxigenin-11-dUTP (Boehringer Mannheim) as previously described (27). Hybridization buffer was prepared as described for in situ hybridization at the light microscope level using 480 ng/ml of PECAM-1 cRNA. Grids were inverted over 10-µl drops of hybridization buffer for 90 min at 37°C, washed twice in H₂O (1 min each), once for 5 min in phosphate buffered saline (PBS) (130 mM NaCl, 7 mM Na₂HPO₄, 3 mM Na₂HPO₄ × 1 H₂O), and washed for 5 to 10 s in freshly prepared 1% ovalbumin (grade V; Sigma) in PBS. Grids were transferred to a drop of sheep antidigoxigenin antibody conjugated with immunogold (10 nm particle size; Bio-Cell, Cedar Lane Ltd., Hornby, ON, Canada) diluted 1:80 in PBS for 60 min. Afterwards, the grids were transferred successively to three drops of PBS, rinsed with H₂O and air dried. Finally, the sections were stained in the dark for 10 min with 5% aqueous solution of uranyl acetate. Sections were viewed with a JEOL-1200 CX electron microscope operating at 60 kV.

Western Blot

The antibody used for Western blotting is a polyclonal antibody generated by injecting rabbits with a synthetic peptide corresponding to the 18 amino acids present in exon 14 (LGTRATETVYSEIRKVDPK). This peptide was produced by Research Genetic (Huntsville, AL) and linked to a multiple antigen peptide before injection into rabbits. Serum was then immunopurified using a protein-G Sepharose column and specificity for exon 14 was confirmed by preferential immunoprecipitation and Western blotting of stably transfected L cell containing full-length mouse PECAM-1 compared with stably transfected L cell containing PECAM construct missing the exon 14. The presence of PECAM-1 expression by the L cells was confirmed by immunoprecipitation and fluorescent-activated cell sorter analysis using the mAb 390 as detailed previously (22).

Lungs were flash-frozen in liquid nitrogen, ground, and solubilized in Laemmli sample buffer (2% SDS, 10% glycerol, 100 mM DTT, 0.4 mM Tris-HCl, pH 6.8). The samples were immediately boiled for 10 min, passed 10 times through an 18-g syringe, and centrifuged at 13,000 × g for 10 min. Supernatant protein was measured directly after a TCA precipitation-resuspension step (29). SDS-PAGE (15 µg of protein, 5% stacking and 10% resolving gel) was done under reducing conditions (30). Electrophoretic protein transfer to nitrocellulose membranes (Nitrocellulose-1; GIBCO BRL, Gaithersburg, MD) was carried out overnight at 4°C at 35 V in a buffer containing 20% methanol, 192 mM glycine, 25 mM Tris-HCl pH 8.0, and 0.1% SDS. The membrane was blocked in Tris-buffered saline containing 5% filtered nonfat milk, 0.1% Tween-20 for 1 h at room temperature. Then the blocking solution was exchanged for the same buffer containing 10 µg/ml of antibody and allowed to incubate overnight at 4°C. After two 5-min washes in the blocking solution, the membrane was incubated in the same buffer containing 2 × 10⁵ cpm/µl of ¹²⁵I-labeled goat antirat IgG (specific activity of 7.5 × 10⁸ dpm/µg of protein) for 1 h at room temperature. After two 5-min washes in 0.1% Tween in Tris-buffered saline, the membrane was exposed for autoradiography.

Immunohistochemistry

The trachea was exposed and cannulated with a 22-gauge catheter, and the lungs were instilled with 20 µM of iodoacetamine (ICN Pharmaceuticals, Montreal, PQ, Canada) and 2 µM phenylmethylsulfonylfluoride (ICN Pharmaceuticals) in cold acetone. After 15 min of instillation, the lungs were removed and fixed for an additional 15 min in acetone. The lungs were then minced into 2-mm³ pieces, which were immersed in methyl benzoate (Fisher, Ottawa, ON, Canada) for another 15 min. Lung pieces were transferred in 5% glycol methacrylate (JBS Supplies, Pointe-Claire, PQ, Canada) in methyl benzoate for 24 h at 4°C. Lung pieces were then embedded in glycol methacrylate resin (10 ml of glycol methacrylate monomer, 250 µl of N,N-dimethylanilline in PEG 400, 45 mg of benzoyl peroxide), allowed to polymerize at 4°C for 2 d, and stored at 20°C. Sections of 1 µm lung were used for immunohistochemical localization of PECAM-1 protein following a protocol adapted from a previously published method (31). The primary antibody was an affinity-purified rat antimouse PECAM-1 monoclonal antibody (18). The secondary reagents were purchased from Vector Laboratories (Burlingame, CA). The slides were incubated with a 1:20 (24 µg/ ml) dilution of the primary antibody in 1% rabbit normal serum overnight at 4°C. The secondary antibody, a biotinylated rabbit antirat IgG, was incubated with a dilution of 1:500 (2 µg/ml) in 1% rabbit normal serum at room temperature for 30 min and detected with peroxidase-conjugated avidin-biotin complex reacting with 3-amino-9-ethylcarbazole. At least three animals were studied at each time point. Negative controls for the nonspecific binding were as before but without primary or secondary antibody on both unexposed and hyperoxic exposed lungs. Photomicrographs were recorded using conventional bright field illumination, as described for in situ hybridizations.

Data Analysis

To control for small differences in membrane-bound target RNA, hybridization values for PECAM-1 mRNA were normalized to CuZn SOD hybridization, which is constitutive in mouse lungs under the conditions of these experiments. The data were evaluated by analysis of variance for independent measure (Fisher) using StatView II (Abacus Concepts, Berkeley, CA) on an Apple Macintosh computer. The two-tailed level of significance was set at P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

To test the hypothesis that PECAM-1 is induced during pulmonary O₂ toxicity, mice were exposed to 100% O₂ for up to 96 h. During this time, typical hyperoxic lung injury occurs, accompanied by specific alterations in gene expression (27). PECAM-1 mRNA was easily detected by Northern blot hybridization, both in control and hyperoxic lung (Figure 1A). To quantify the relative abundance of PECAM-1 mRNA, slot-blot hybridization was performed with RNA purified from whole-lung homogenates. PECAM-1 hybridization was normalized to CuZn SOD mRNA abundance because CuZn SOD mRNA abundance is unchanged during hyperoxia (Reference 21 and our unpublished results). Figure 1 (panel B) shows that by 2 d, there was a 35% increase (P < 0.05) in PECAM-1 mRNA levels, and a 66% increase by 4 d (P < 0.0001). These observations indicate that hyperoxia-induced increases in PECAM-1 expression are relatively small but reproducibly measurable in whole-lung homogenates.

View larger version (15K): [in this window] [in a new window]   View larger version (79K): [in this window] [in a new window]   Figure 1.   PECAM-1 mRNA abundance in whole-lung homogenates is increased following hyperoxic exposure. (Panel A) Northern blots hybridized for PECAM-1 mRNA and constitutively expressed CuZn SOD mRNA. Small differences in hybridization intensity for CuZn SOD mRNA are attributable to slight inconsistencies in RNA loading or transfer efficiency. (Panel B) Slot blots from hyperoxic and unexposed lungs were quantified with PhosphorImager, as described in MATERIALS AND METHODS. All values were normalized to hybridization to the constitutively expressed mRNA for the CuZn superoxide-dismutase (CuZn SOD). Data are expressed as a percentage of the mean value of unexposed animals (mean ± SD). Each column represents the mean value of four mice. The asterisk denotes statistical significance relative to control (P < 0.05).

To determine if alternative splicing of PECAM-1 occurs in lungs of the adult mouse, RT-PCR assays using primers flanking either the cytoplasmic or the transmembrane domain were performed. Figure 2 shows the presence of at least seven splice variants of the cytoplasmic domain in lung homogenate. The presence and the relative abundance of these isoforms did not change with exposure of the lungs to hyperoxia or when compared to other organs (Figure 2). The relative abundance of each band was compared and the four major bands were sequenced to confirm their identity. Table 3 summarizes the results. The isoform encoding the full-length cytoplasmic domain accounts for 16% of total PECAM-1 mRNA, the isoform missing exon 15 for 19%, the isoform missing exon 14 for 13%, and the isoform missing both exon 14 and 15 for 33%. The amplification of the transmembrane domain showed the presence of only one isoform. Southern blots (Figure 3) using either the full length cDNA (Figure 3A) or an oligonucleotide from a region of the transmembrane domain (Figure 3B) shows that the isoform(s) of PECAM-1 expressed in mouse lungs encoded the transmembrane domain.

View larger version (115K): [in this window] [in a new window]   Figure 2.   Amplification of the cytoplasmic domain of the mouse PECAM-1 showing the presence of multiple isoforms. Total RNA from lung before (unexposed) and after 96 h of O2 exposure, as well as from other mouse organs were used for the RT- PCR amplification of the cytoplasmic domain. The downstream primer was radioactively labeled and the PCR product separated on acrylamide gel before autoradiography. For each sample, aliquots of cycles 20, 25, and 30 were run through the gel. Numbers on the right side of the figure represent the sizes of the different bands of the molecular weight marker.

View larger version (87K): [in this window] [in a new window]   Figure 3.   Amplification of the transmembrane domain of the mouse PECAM-1 showing the presence of a single isoform. Total RNA from lung exposed for various periods to hyperoxia was used for the RT-PCR amplification of the transmembrane domain. The PCR products were separated on an agarose gel, transferred on nylon membrane, and hybridized with the full length cDNA (A) or an oligonucleotide, corresponding to the transmembrane domain (B).

The modest increase in bulk PECAM-1 mRNA suggested that there may be limited number of cell types expressing it. To test this idea, we performed in situ hybridizations (Figure 4). In unexposed lungs (Figures 4A and 4B), basal levels of PECAM-1 transcripts were readily detectable in endothelial cells of large and medium size pulmonary arteries. By 2 d of hyperoxia, there were obvious increases in PECAM-1 transcripts in these cells (Figures 4C and 4D), and also in endothelial cells of smaller, less muscularized vessels, which are identifiable as arteries by their proximity to airways and the thickness of the muscular wall (Figure 4E). The pulmonary veins did not appear to accumulate PECAM-1 mRNA. Moreover, cells in the gas exchange region accumulated PECAM-1 mRNA (Figure 4F). Unfortunately, the limited resolution of light microscopy makes identification of these cells equivocal. With longer exposure to hyperoxia, no further changes in the cell-specific pattern of mRNA accumulation were apparent, but the abundance of the transcript appears slightly increased in small vessels (Figures 4F and 4G). Hybridization to the sense-strand probe resulted in a low background (Figure 4H). Table 2 shows results of the semiquantitative analysis of the in situ hybridization. After 1 d of hyperoxia, a general decrease in PECAM-1 expression was observed, but it was only significant for the expression on the endothelial surface of arteries. By 2 d of exposure, both the expression in arteries and the alveoli level were significantly increased above controls. With longer exposure, no further significant changes were observed.

View larger version (145K): [in this window] [in a new window]   Figure 4.   Localization of PECAM-1 mRNA in control and hyperoxic lungs by in situ hybridization. Lung sections (4 µm each) were hybridized with radiolabeled cRNA antisense RNA probes (panels A through G) and sense-strand control probes (panel F ) and hybridization was detected by autoradiography (white grains). Panels A and B show a pulmonary artery and vein, respectively, from unexposed mice. Arrows in these and other panels point to hybridization in endothelium. Panels C through F are from mice exposed to hyperoxia for 48 h. Panels C and D show a pulmonary artery and vein, respectively. Panel E shows hybridization to endothelium of smaller blood vessels and within alveolar walls. Panel F shows hybridization in alveolar walls in the distal tip of a lung. Panels G and H are serial sections from a mouse exposed to hyperoxia for 96 h. Panel G shows hybridization in alveolar regions. Panel H shows that sense-strand background was low. aw = airway, sm = smooth muscle, lu = lumen of blood vessel, al = alveolus. Magnification bar = 200 µm.

To identify the cells in the gas exchange region that express PECAM-1 mRNA, we adapted a technique of lung in situ hybridization at the EM level (Figure 5). PECAM-1 mRNA was barely detectable in alveolar cells of unexposed animals (Figure 5c). By 48 h of hyperoxia, PECAM-1 mRNA was clearly expressed in the capillary endothelial cells of alveolar septa (Figure 5a). Two-thirds of the cells expressing PECAM-1 at the alveolar level after 2 d of hyperoxia could be identified as endothelial cells. In contrast, type I cells (Figure 4B) or connective tissue cells each counted for only 10% of the cells expressing the message. PECAM-1 transcripts were also occasionally detected in a few scattered platelets, but that finding was inconsistent. Cell organelles were generally difficult to identify, perhaps because of proteinase K digestion. In some cases, however, PECAM-1 mRNA was evident in endoplasmic reticulum, suggesting active translation of the message (Figure 4A). The specificity of localization was confirmed by hybridization with the sense probe (not shown), which yielded essentially no signal. Under the conditions used, PECAM-1 was not detected in type 2 pneumocytes or in leukocytes. These data show that endothelial cells in capillaries of gas exchange regions accumulate PECAM-1 mRNA.

View larger version (50K): [in this window] [in a new window]   View larger version (103K): [in this window] [in a new window]   View larger version (112K): [in this window] [in a new window]   Figure 5.   Localization of PECAM-1 mRNA in hyperoxic lungs by EM in situ hybridization. Tissue sections of animals unexposed (panel c) or exposed to hyperoxia for 2 d (panels a and b) are shown. PECAM mRNA was detected with gold-conjugated antibodies to digoxigenin-labeled cRNA probes (small black grains). Panel a is a longitudinal section of capillary showing the endoplasmic reticulum of an endothelial cell with gold particles close to the ribosomes (arrows). Panel b shows the expression in a type I pneumocyte. ER = erythrocyte within a capillary; Alv = alveolar space. Bar = 0.5 µm.

To determine whether the observed increases in the PECAM-1 message were accompanied by corresponding increases in the protein, we used both Western blot and immunohistochemical techniques using the available antibodies. Figure 6 shows an example of Western blot using an antibody directed against exon 14 of the protein. No significant change was apparent. The absence of changes in bulk PECAM-1 protein suggested either there may be limited number of cell types expressing it or the antibody we used missed the important isoforms. To test these ideas, we used a monoclonal antibody recognizing the 12,15 PECAM-1 (18) for immunolocalization on tissue sections (see Figure 7). Figure 7A shows the presence of immunoreactive PECAM-1 protein in lung endothelial cells of small vessels and, to a lesser extent, the alveolar walls (presumptive capillaries) of unexposed animals. Exposure to hyperoxia yielded an obvious increase in immunoreactive protein at both these sites: by 4 d of exposure, abundant PECAM-1 was detected throughout the alveolar walls as well as in endothelial cells of small vessels (Figures 7B and 7C). Moreover, PECAM-1 was abundant within lumen of alveoli despite very small increases in lavageable inflammatory cells (23). The specificity of the antibody was confirmed by controls consisting of omitting either the primary antibody (not shown) or by using nonreactive primary serum (Figure 7D), either of which resulted in low background.

View larger version (75K): [in this window] [in a new window]   Figure 6.   Western blot showing the abundance of PECAM-1 protein in lung extract. Protein extracts from animals raised in room air (control) or exposed to hyperoxia for 2 or 4 d (2 d O2 and 4 d O2) were separated by SDS gel electrophoresis. The PECAM-1 protein was detected with antibody made against exon 14 of the cytoplasmic tail. The secondary antibody was 125I-labeled goat anti-rat IgG. Molecular weights are indicated on the right (×103 kD).

View larger version (134K): [in this window] [in a new window]   Figure 7.   Immunolocalization of PECAM-I protein in control and hyperoxic lungs. Lung sections (4 µm each) were incubated with anti-PECAM-1 antibody and detected as described in MATERIALS AND METHODS. Panel A is from an unexposed mouse. Arrows point to endothelium of medium-size blood vessels. Arrowheads point to alveolar cells. Panels B through D are from a mouse exposed to hyperoxia for 4 d. Panel C is a higher magnification of a portion of panel B; lu = alveolar lumen. Panel D is a serial section incubated with nonreactive serum in place of primary antibody. Magnification bar = 50 µm for panel A and 100 µm for panel C.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we have described the cell-specific induction of PECAM-1 during hyperoxic injury in mouse lungs. The time frame of increased PECAM-1 is relatively early with respect to the onset of demonstrable pulmonary damage, and occurs prior to leukocyte infiltration (23). Moreover, PECAM-1 was induced in capillaries where leukocytes are known to invade during injury (32). Additionally, PECAM-1 was also increased in arteries, similar to what we observed with P-selectin (33). These results contrast with the increased expression of ICAM-1, which occurs on the surface of type II cells (15, 16), but not endothelial cells (13) during hyperoxic lung injury. We speculate that leukocyte recruitment in the lung requires not only the coordinated timing of CAM activation at the cell surface, but also an orchestrated cell type-specific distribution of CAMs at different times to ensure the migration of leukocytes from the capillary lumen to alveolar lumen. The increased expression of CAMs at the endothelial surface of arteries seems unique to the lung. This is certainly related to the difference in the site of leukocyte migration between lungs and other tissues, and it may be necessary to activate leukocytes before they reach the capillary bed.

Identification of the capillary endothelium is not possible using conventional light microscopy with 4- to 5-µm sections. We therefore adapted an electron microscopy in situ hybridization technique, which has the potential to allow unequivocal cell identification. This method is rapid, requiring less than 1 d to process the grid, and is also relatively sensitive and completely specific. It should be noted however, that the fixation technique employed and the use of proteinase K to enhance target accessibility can affect cell morphology, sometimes obscuring cell type identification. For example, lamellar bodies of type II pneumocytes are not well preserved by this technique. Another caveat is that the rigors of sampling make it difficult to identify the entire set of the approximately 60 cell types of the lung (34) that might express a particular gene of interest. Nevertheless, when conventional in situ hybridization yields preliminary clues, this approach should provide a solution to the problem of precisely identifying many of the cell types that are involved. Advancements in our understanding of cell-specific lung gene expression should therefore be accelerated by using this approach. In addition, because the gas-exchange region is often an early site of injury, a better understanding of the first events of alveolar injury could lead to new ideas for the preservation of alveolar integrity and protection of lung function. However, at that point the sensitivity of the technique is known. Therefore, it is possible that PECAM-1 mRNA is expressed in other cell types such as leukocytes and platelets at a lower but still biologically significant level.

The presence of multiple isoforms of PECAM-1 resulting from alternative splicing of the cytoplasmic domain was previously demonstrated in mouse embryo heart (18). Here, we show that those isoforms persist into adulthood with a similar distribution among different organs. Moreover, the predominant isoform is the 14-15, the "full-length PECAM-1" accounting for less than 20% of the total PECAM-1 (Table 1). It has been shown that PECAM-1 isoforms have different adhesive function in vitro (22). The full-length PECAM-1 mediates calcium- and heparin-dependent heterophilic aggregation. In contrast, isoforms missing the exon 14 mediate calcium- and heparin-independent homophilic aggregation.

Recently, a soluble form of human PECAM-1 was described, which seems to be encoded by an alternative splicing of the transmembrane domain (17). Soluble PECAM-1 has been found in cell culture fluids as well as in normal human plasma. One could suspect that the presence of immunodetectable PECAM-1 in the lumen of alveoli after 96 h of hyperoxia represent an accumulation of a secreted form. This seems unlikely, however, because using RT-PCR techniques, we did not find the presence of PECAM-1 isoform missing the exon encoding for the transmembrane domain in mouse lungs. Alternatively, PECAM-1 in the alveolar lumen might be derived from a cleavage of the protein at the surface of the cell, resulting in a protein without its cytoplasmic tail (17). Actually, the antibody we used for Western blots was raised against the peptide encoded by exon 14, which corresponds to the cytoplasmic tail, and therefore was unable to detect a truncated form. Finally, the presence of abundant PECAM-1 protein within the lumen of alveoli could also result from the cell surface of damaged neutrophils, or platelets, that migrated within the alveoli. However, it is improbable because the hyperoxic lung injury is characterized in the mouse by a relative limited increase of inflammatory cells in the alveoli lumen (23).

We observed no increase in PECAM-1 by Western blot analysis. A likely explanation for this apparent discrepancy is that the antibody was raised against the cytoplasmic tail, which might have been cleaved. This fits with the observation that intraluminal PECAM-1 was detected, whereas the mRNA encoding soluble form was not. Because the antibody used for immunohistochemistry does not react with the denatured, blotted protein, this issue cannot be resolved. Alternatively, the increase at the whole lung mRNA level was relatively modest, and perhaps the Western blot analysis is not sensitive enough to show similar changes at the protein level. On the other hand, we observed a tight correlation in spatial and temporal increases in PECAM-1 mRNA and protein seen by immunohistochemistry and in situ hybridization indicating that the increase was much greater on a cellular basis, but for a limited number of cell types. Finally, the presence of abundant PECAM-1 mRNA in the endoplasmic reticulum after 48 h of hyperoxia (seen at the electron microscope level) suggests active translation of the message.

Cell adhesion molecules play a key role in modulating the inflammatory response of the lung. For example, a role for different CAMs has been implied in acute and chronic airways inflammation (8, 35, 36). Blocking ICAM-1 function also reduced airway responsiveness and eosinophil infiltration after acute antigen challenge in animal models, but had little effect on neutrophil infiltration induced by repeated antigen challenges (8, 37, 38). On the other hand, anti-PECAM-1 injection caused a 75% reduction in lung neutrophil accumulation in a model of lung injury induced by the deposition of immune complexes (20). Antibodies to ICAM-1 produced a significant, but only partial, attenuation of lung damage and neutrophil influx induced by breathing 100% O₂ in mice (39). The cellular sites of synthesis of CAMs vary with the type of injury. To date, ICAM-1 has been localized to lung endothelium during acute antigen challenge (40, 41) or bacterial infection, but to epithelium and not endothelium during hyperoxic lung injury (13, 15, 16). In contrast, PECAM-1 has been shown in this report to be localized to blood vessels and expressed in endothelium. Because capillary endothelial cells are hypersensitive to O₂ injury (5), this finding suggests that PECAM-1 expression is a relatively early step in the inflammation cascade. Perhaps interfering with the expression of PECAM-1 together with ICAM-1 during the early phase of injury may prevent further damage.

The molecular basis of PECAM-1 regulation in the lung is still largely unknown. Judging from studies of other CAMs (7), specific PECAM-1 isoform expression might have importance in particular types of inflammation. Further studies on PECAM-1 regulation during different types of lung injury will shed more light on this issue and improve our understanding of CAM interaction.