Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The accumulation of fluid within the pleural cavity is a hallmark of pleural disease. When effusions develop because of an imbalance in the hydrostatic and/or oncotic pressure, the pleural membrane per se is intact and the fluid has a low protein content typical of a transudate (1). In contrast, inflammation or injury to the pleura characteristically results in an exudative effusion with a high protein content implying a loss of integrity of the underlying pleural membrane. When severe, such injury is further characterized by the intrapleural deposition of fibrin which, if unchecked by the fibrinolytic system, may lead to pleural sclerosis and obliteration of the pleural cavity (2).

Tissue factor (TF), an integral membrane protein, is the major cellular initiator of the extrinsic coagulation cascade (6). Previous in vitro studies have demonstrated that human mesothelial cells (HMC) are capable of expressing TF (7, 8). Procoagulant activity has been found in pleural fluid from patients with exudative effusions (4, 9), but not in pleural fluid of patients with transudative effusions (9). However, cultured HMC from both types of effusion have been shown to activate prothrombin, thus suggesting that HMC expression of TF is constitutive (8). Extravascularly, TF is expressed in an apparently constitutive manner by many cells including the cells of the perivascular adventitia, epidermis, and mucosal epithelium (10, 11). However, in other cell types, such as quiescent fibroblasts, TF expression is not constitutive but can be induced by exposure to serum (12) or various cytokines (15). Since all the soluble, plasma-derived coagulation factors are normally present in pleural fluid (19), constitutive TF expression by HMC would be expected to promote fibrin deposition, even in the absence of disease. For example, increased vascular permeability alone is reported to cause fibrin deposition in skin and muscle tissues (20). Since fibrin is not normally present on pleura (3) and procoagulant activity is not detectable in transudative effusions (9), we hypothesized that TF is not constitutively expressed by HMC but can be induced by exposure to serum or inflammatory mediators.

To define the extent of TF expression and activity in pleura in vivo, we used epitope-defined monoclonal antibodies (MAbs) to human TF for immuno-localization of TF in tissue specimens. We found that the mesothelial monolayer of light microscopically normal tissue specimens did not express TF, while the mesothelial monolayer of tissue specimens with acute pleural inflammation did express TF. We further demonstrate that the in vitro expression of TF in quiescent HMC is serum-inducible both at the mRNA and protein level, but is not inducible by plasma.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Cell Isolation and Culture

Transudative pleural effusions were harvested by sterile technique from 46 patients according to a protocol approved by the University of Oklahoma Human Studies Committee. Mesothelial cells were isolated as previously described (21). In brief, pleural fluid was subjected to centrifugation, erythrocytes and platelets were then lysed, and the remaining cells were suspended in complete medium (M-199 medium [Gibco BRL, Gaithersburg, MD] supplemented with 10% heat-inactivated fetal bovine serum [FBS, Hyclone, Logan, UT], and 50 µg/ml gentamicin [Gibco BRL]) for propagation at 37°C in a humidified 5% CO₂ atmosphere. Fresh complete medium was replaced every 2-3 days until cells were confluent. Upon confluence the cells were lifted by 1× trypsin-EDTA (Gibco BRL) and subcultured at a 1:2 dilution. Morphologically, the cells formed a confluent monolayer of polygonal cells with a cobblestone appearance typical of mesothelial cells. HMC identity was regularly confirmed by intermediate filament typing as previously described (21). The presence of contaminating macrophages was excluded by verifying the cells were peroxidase and esterase negative.

Two human fetal lung fibroblast (FLF) lines (GM01604 and GM05387, NIGMS Mutant Cell Repository) were grown in their complete medium (Dulbecco's Modified Eagle's Medium [DMEM] with high glucose [Gibco BRL], supplemented with 20% heat-inactivated FBS, penicillin [200 U/ml], and streptomycin [200 µg/ml] [1× P + S]). Cells were grown to 80-85% confluence, lifted with 1× trypsin-EDTA, and subcultured at a 1:15 dilution.

For studies, HMC or FLF was subcultured into either 12- or 24-well culture plates (for ELISA or TF activity assays) or 75 cm² cell culture flasks (for cellular RNA studies), and grown to confluence in their respective complete medium. Upon confluence, quiescence was induced by serum deprivation as previously described; complete medium was replaced with serum-free medium (M-199 medium supplemented with 50 µg/ml gentamicin [M199-SF] for HMC and DMEM supplemented with 1× P + S [DMEM-SF] for FLF) for 48 h to induce cell quiescence (22). Third to fifth passage HMC or seventh to eleventh passage FLF were utilized in these studies. For ELISA and TF activity assays, cell lysates were prepared, following incubation, by washing the plated cells twice with 1 ml/well TBS, then adding 0.5 ml TBS to each well and subjecting the cells to three cycles of rapid freeze-thaw lysis.

For immunohistochemistry, HMC were seeded into eight-well Lab-Tek Chamber Slides (NUNC, Naperville, IL), grown to confluence and quiescence induced as above.

Tissue Procurement

Samples of parietal pericardium and both visceral and parietal pleura were obtained from fresh autopsy or surgical specimens from the Pathology Department of the University of Oklahoma Health Sciences Center. Upon removal, tissue samples were embedded in Tissue-Tek O.C.T. compound (Miles, Kankakee, IL) snap-frozen in liquid nitrogen, and stored at 70°C.

Immunohistochemistry

Cells grown in vitro on chamber slides and cryostat sections (4-6 µm) of pleura and pericardium were fixed for 10 min in acetone at 20°C and air dried. Detection of TF utilized three murine anti-human MAb against TF (TF9-9C3, TF9-10H10, and TF9-9B4), derived from purified human brain TF immunizations as previously reported (23) which were diluted with PBS containing 1% BSA to 50 µg/ml, 60 µg/ml, and 15 µg/ml, respectively. Sequential sections of each tissue were incubated for 60 min with either the primary MAb against TF or mouse monoclonal IgG₁ (Bethyl Labs, Montgomery, TX) to serve as negative control. Following incubation with primary antibody, the sections were incubated for 20 min with a secondary biotinylated horse anti-mouse antibody (DAKO, Carpinteria, CA), then exposed for an additional 20 min to avidin-peroxidase complex, then exposed for 5 to 10 min to either 3-amino-9-ethylcarbazole (AEC; BioGenex Lab, San Ramon, CA) or 3,3'-diaminobenzidine (DAB; Sigma, St. Louis, MO), and finally counter-stained with Mayer's hematoxylin. Tissue staining was evaluated as signal:noise (image:background) positivity, and staining intensity was scored as () for absence of staining and 1+ to 3+ for positive staining. The positive scores were assigned as follows: 1+ (minimal but recognizable staining); 2+ (moderate staining); and 3+ (intense staining).

Tissue Factor ELISA

HMC lysates from each well were diluted 1:5 with TBS containing 0.1% BSA (TBSA) and 0.1% Triton X-100 (TBSA/Triton). Human TF was purified as previously described (24) and diluted with TBSA/Triton to concentrations between 0 and 1.0 ng/ml for use as standards. A sandwich ELISA was performed as follows using three different anti-TF MAbs (23). ELISA plates (Corning Glass Works, Corning, NY) were incubated at 4°C overnight with 100 µl/well of capture MAb (TF8-11D12 at 5 µg/ml in 0.1 M sodium carbonate [pH 9.2], washed (×2) with TBS, blocked with TBS containing 5% non-fat powdered milk for 2 h at 37°C, and washed (×3) with TBSA/Triton. Diluted cell lysates or TF standards were added (100 µl/well), incubated at 37°C for 90 min, and washed (×3) with TBSA/ Triton. The presence of TF was detected by incubation with 100 µl/well of the combined biotinylated detection MAbs (TF9-1B8 and TF9-5B7, utilized at a concentration of 0.2 µg/ml and 0.1 µg/ml, respectively, in TBS/Triton) at 37°C for 60 min, washing (×3) with TBSA/Triton, incubation with 0.025 µg/well streptavidin/alkaline phosphatase (Pierce, Rockford, IL) at room temperature for 30 min, washing (×1) with TBSA/Triton and (×3) with TBSA, incubation with 50 µl/well ELISA substrate (ELISA Amplification System, Gibco BRL) for 15 min at room temperature followed by 50 µl/well ELISA amplifier (ELISA Amplification System) for 10 to 15 min at room temperature. The reaction was stopped by addition of 50 µl/well 0.3 M H₂SO₄ and absorbance at 495 nm was measured on a thermodyne plate-reader (Molecular Devices, Sunnyvale, CA). TF levels of the TBS diluted cell lysates were derived from the linear portion of the standard curve and normalized to total protein of cell lysates measured using the Coomassie Blue dye-binding assay (Bio-Rad Protein Assay; Bio-Rad, Richmond, CA).

Tissue Factor Activity

TF activity was assessed by the rate of factor Xa generation. For these studies, HMC and FLF lysates were diluted as for the TF ELISA. TF was reconstituted into vesicles composed of mixed brain phospholipid (TF-MBPL) as previously described (25) and diluted with TBSA to concentrations ranging from 0 to 20 pM TF for use as activity standards. Diluted cell lysates or TF-MBPL standards were incubated at room temperature for 10 min with an equal volume (25 µl) of factor VII/X reaction mix (4 nM factor VII (25), 100 nM factor X (25), 50 mM Tris-HCl (pH 7.4), 100 mM NaCl, 10 mm CaCl₂, 0.1% BSA, and 0.2% NaN₃) after which 25 µl of stop solution (40 mM EDTA, 200 mM Tris-HC [pH 9.1]), and 4% Lubrol-PX [Sigma] was added. Immediately upon adding 25 µl of 0.5 mM chromogenic substrate (S-2765; Chromogenix, Mölndal, Sweden), absorbance at 405 nm was monitored for 10 min using a thermodyne plate-reader. The TF activity (rate of Factor X_a generation) of the diluted cell lysates was derived from the linear portion of the standard curve and normalized to total protein content.

To ensure factor X activation was the result of TF, cell lysates from some studies were incubated at room temperature for 30 min, prior to TF assessment, with either the murine anti-human TF MAb, TF8-11D12, which recognizes TF in the presence or absence of factor VII and inhibits the TF:factor VIIa complex activation of factor X (26), or the irrelevant antibody MK-D6 (ATCC HB3; American Type Culture Collection, Rockville, MD). Both antibodies were diluted with TBSA and utilized at a final concentration of 1 µg/ml.

Cellular RNA Isolation

Total cellular RNA was extracted by a modification of the acid guanidinium thiocyanate-phenol-chloroform method, as previously described (27). Briefly, cell monolayers were washed (×2) with ice-cold TBS, subject to cellular disruption by addition of lysis buffer (4 M guanidinium thiocyanate HCl [pH 4.0]), 1% 2-mercaptoethanol, and 0.5% N-lauroylsarcosinate) and scraping with a rubber policeman. RNA was extracted with one volume phenol (pH 5.0):chloroform (5:1 v/v), re-extracted with one volume chloroform:isoamyl alcohol (49:1 v/v), and the RNA precipitated with two volumes absolute ethanol at 20°C. RNA was washed (×2) with 70% EtOH and stored in purified H₂O at 80°C.

Northern Blotting and Hybridization

Except where noted, 10 µg total cellular RNA per lane was subjected to electrophoresis (Fisher Biotech Electrophoresis System; Fisher Scientific, Pittsburgh, PA) in formaldehyde/agarose (28), transferred to nylon membranes (GeneScreen; New England Nuclear, Boston, MA), and hybridized as previously described (27). Membranes were probed sequentially using random primer-labeled (Boehringer Mannheim Biochemicals, Indianapolis, IN) (29) TF cDNA and glyceraldehyde phosphate dehydrogenase (GAPDH) cDNA (ATCC HHCMC32). The TF cDNA clone used in these studies was prepared by removing the 5' non-coding region (base pairs 1-138) and the Alu-family repeat (base pairs 1287-1925) of clone pcTF554 (30). Radioactivity was quantitated by linear densitometric analysis of the autoradiograph image using a Ultrascan XL Enhanced Laser Densitometer (Pharmacia LKB Biotechnology, Uppsala, Sweden). TF mRNA signals were normalized to GAPDH signals to correct for any variations in gel loading.

Statistical Analysis

Unless stated otherwise, all experiments were performed on cultures from 4 or more different donors. All assays on cell lysates were performed in triplicate, and all RNA gels were run in duplicate. The data are presented as mean (± SEM). The significance of differences between experimental conditions was determined by analysis of variance and statistical significance was accepted when P < 0.05.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Mesothelial Cell Expression of TF In Vivo

Immunohistochemical staining of sections of normal human lung and pericardium with anti-TF MAb were performed to see if the mesothelial monolayer constitutively expressed TF in vivo. The age of the 26 patients from whom tissue was obtained ranged from 46 to 88 years of age, except for one patient that was 7 days of age. Fifteen specimens (10 pleura and 5 pericardium) were normal by light microscopy. Thirteen of these were surgical specimens from patients undergoing CABG (n = 6), repair of an aneurysm (n = 2), or resection for lung cancer (n = 3). Autopsy specimens were obtained from a patient with sudden cardiac death and a patient who died after myocardial infarction. Except for mild emphysema (n = 1) and acute pulmonary edema (n = 1), the underlying lung parenchyma was normal in these individuals. Normal pleural and pericardial tissues from different individuals showed essentially identical staining characteristics. Anti-TF MAb staining of morphologically normal sections of lung (Figure 1a) and parietal pericardium revealed the mesothelium to be negative () for detectable TF expression while the lung parenchyma was moderately (2+) positive, and the perivascular adventitia of small arteries embedded in fat tissue beneath the pericardium was strongly (3+) positive. The negative control antibody did not demonstrate any positive staining in either the lung or pericardium.

View larger version (128K): [in this window] [in a new window]   Figure 1.   Immunohistochemical detection of TF in cryosections of human tissues and in vitro cultured HMC. (Panel a) TF staining of normal lung (ABC technique, DAB). Note the parenchyma is moderately positive (brown) while the visceral pleura remains negative () (×400). (Panel b) TF staining of inflamed lung (ABC technique, DAB). Note the intensely positive (3+) staining of the mesothelial monolayer and connective tissue of the pleura overlying the parenchymal inflammation (×400). (Panels c and d) TF staining of HMC (ABC technique, AEC). Note that HMC maintained in complete medium (c) stain intensely positive (3+) (red) while HMC maintained for 48 h in serum-free medium (d) are negative () to minimally positive (1+) (×400).

Eleven specimens were obtained from patients with acute inflammation in the underlying lung. Five were surgical specimens from patients undergoing lobectomy for cancer with post-obstructive pneumonitis (n = 3) or decortication after empyema (n = 2), while 6 autopsy specimens were obtained from patients with pneumonia. In three specimens with extensive pleural fibrosis a distinct mesothelial cell monolayer could not be identified. The other 8 specimens all had an acute pleuritis and TF immunohistochemical staining demonstrated intense (3+) positive staining of both the mesothelial monolayer and the pleural connective tissue (Figure 1b). No differences were noted between staining patterns for any of the three different anti-human TF MAbs.

Mesothelial Cell Expression of TF In Vitro

Immunohistochemical studies of fixed HMC maintained in vitro in complete medium exhibited intensely positive (3+) staining with the anti-TF MAb (Figure 1c). However, after being maintained in serum-free medium for 48 h, quiescent HMC expression of TF was absent () to minimally positive (1+) upon immunohistochemical staining with anti-TF MAb (Figure 1d).

To quantitatively evaluate this loss of TF expression when HMC were incubated under serum-free conditions, cell lysate levels of TF activity and TF protein expression, normalized to cell lysate protein content, were measured serially from time of serum withdrawal. Normalized TF activity remained relatively unchanged over time for HMC maintained in the presence of 10% FBS (Figure 2). However, normalized TF activity levels decreased progressively over time upon withdraw of serum, which was statistically significant (P < 0.005) by 12 h (Figure 2). Similar findings were seen for levels of normalized TF protein expression (data not shown).

View larger version (12K): [in this window] [in a new window]   Figure 2.   TF activity of HMC lysates normalized to total protein content. TF activity levels were measured for HMC maintained in either serum-free medium (open circle) or complete medium (closed circle). All data are mean ± SEM for four experiments, each performed in triplicate.

Serum Induction of TF Expression in Quiescent Mesothelial Cells

To see if serum was capable of inducing TF expression in quiescent HMC, confluent cells were rendered quiescent by serum-deprivation for 48 h, then stimulated with either fresh M199-SF or complete medium. Following stimulation, cell lysates were harvested serially over time and both TF activity and TF protein levels were measured and normalized to cell lysate protein content. Stimulation with fresh serum-free medium did not significantly affect either TF activity (Figure 3a) or protein levels (Figure 3b) in quiescent HMC during the ensuing 24 h. In contrast, stimulation with complete medium led to a greater than 2-fold rise in TF activity (P < 0.0001) which peaked at 9 h; by 24 h levels were not significantly different from baseline (Figure 3a). Similarly, stimulation with complete medium resulted in a significant (P < 0.00001), greater than 3.5-fold, increase in TF protein levels, again peaking 9 h following stimulation (Figure 3b). However, unlike TF activity, TF protein levels did not return to baseline but remained significantly elevated (P < 0.0005) at approximately twice baseline values for up to 24 h (Figure 3b).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Time course curve for TF activity and protein expression in quiescent HMC following serum stimulation. (Panel A) Induction of TF activity in quiescent HMC stimulated with either serum-free medium (closed circle) or complete medium (open circle). (Panel B) Induction of TF protein expression in quiescent HMC stimulated with either serum-free medium (closed circle) or complete medium (open circle). All data are mean ± SEM for ten experiments, each performed in triplicate.

To compare these findings in HMC with those of other cell types, such as fibroblasts, in which serum-induction of TF activity (by procoagulant activity) and protein expression have been reported, levels of TF activity (by rate of factor Xa generation) were determined in quiescent FLF following serum-induction. Confluent FLF were rendered quiescent by serum-deprivation for 48 h, then stimulated with either DMEM-SF or complete medium. In a manner parallel to quiescent HMC, complete medium induced quiescent FLF basal TF activity (280 ± 40 fmol/mg), as measured by rate of factor Xa generation, to transiently increase 4-fold (1,095 ± 220 fmol/mg) at the time of peak activity, 12 h post-stimulation.

To ensure that the 2-fold increase in Factor X activation we were witnessing with HMC was due to TF, quiescent HMC were stimulated for 9 h by either serum-free medium (315.9 ± 13.1 fmol/mg) or complete medium (689.3 ± 11.4 fmol/mg). Lysates from HMC stimulated by complete medium were pre-incubated with either no antibody, an anti-TF MAb (TF8-11D12), or an irrelevant Ab (ATCC HB3). Anti-TF MAb completely inhibited the serum-induced increase and inhibited basal activity by greater than 70% (91.8 ± 0.4 fmol/mg); the irrelevant Ab did not significantly alter either the induced or basal levels (705.0 ± 7.2 fmol/mg).

To determine whether the response of quiescent HMC to serum was a graded or an all-or-nothing response, a serum dose-response curve was constructed. Quiescent HMC were unresponsive to serum concentrations less than 1% (vol/vol) but above this concentration HMC TF activity became progressively greater (P < 0.0001) until reaching a plateau at a serum concentration between 10% and 20% (Figure 4).

View larger version (10K): [in this window] [in a new window]   Figure 4.   Dose-response curve for the induction of TF activity in quiescent HMC by serum-supplemented medium. Quiescent HMC were stimulated for 9 h with medium supplemented with various concentrations of FBS. All data are mean ± SEM for five experiments, each performed in triplicate.

Effect of Plasma on TF Expression in Quiescent Mesothelial Cells

The mesothelial monolayer is normally continuously bathed in a plasma ultra-filtrate in vivo. Since we demonstrated TF expression appears to be normally absent in vivo, but is inducible by FBS in vitro, we evaluated the response of quiescent HMC to human plasma and human serum in vitro. Quiescent HMC were stimulated for 9 h with either fresh serum-free medium (negative control), complete medium (positive control), medium supplemented with 10% heat-inactivated human plasma (HP), or medium supplemented with 10% heat-inactivated human serum (HS) (Figure 5). Medium supplemented with either 10% FBS (complete medium) or 10% HS resulted in comparable degrees of increased normalized TF activity levels. In contrast, the addition of medium supplemented with 10% HP resulted in a slight but significant 20% decrease (P = 0.01) in normalized TF activity. To ensure that the failure of plasma to induce TF activity was not due to the presence of heparin in the plasma, the study was repeated using complete medium in which the FBS had received the same level of heparin. Medium supplemented with 10% heparinized FBS induced a rise in normalized TF activity (896 ± 40 fmol/mg) that was comparable (P < 0.1) to that seen with complete medium in which the FBS has not been heparinized (876 ± 23 fmol/mg).

View larger version (45K): [in this window] [in a new window]   Figure 5.   Induction of TF activity in quiescent HMC by either human plasma, human serum, or FBS. TF activity levels were measured in quiescent HMC stimulated with either: serum-free medium, medium supplemented with 10% human plasma, medium supplemented with 10% human serum, or complete medium. All data are mean ± SEM for three experiments, each performed in triplicate.

TF mRNA Levels in Quiescent Mesothelial Cells

TF expression by at least some cell types has previously been reported to be entirely on the cell surface (31) and thus should qualitatively reflect the 2.2 kb TF mRNA species (32). To determine whether the induction of TF activity seen with serum was reflected at the mRNA level, Northern blot analysis of TF mRNA and GAPDH mRNA levels was performed (Figure 6a). Serum stimulation of quiescent HMC resulted in a rapid, transient, 4.5-fold rise (P < 0.000001) in normalized TF mRNA which peaked by 2 h post-stimulation (Figure 6b). Serum stimulation did not, however, significantly alter GAPDH mRNA levels.

View larger version (21K): [in this window] [in a new window]   Figure 6.   Effect of serum stimulation on TF mRNA levels in quiescent HMC. (Panel A) Representative Northern blot of 10 µg total cellular RNA per lane extracted from quiescent HMC stimulated by complete medium for the indicated times. (Panel B) Time course curve for TF mRNA normalized to GAPDH mRNA in quiescent HMC following stimulation by complete medium. All data are mean ± SEM for six experiments.

TF has been reported to be an immediate-early gene (in certain cell types such as fibroblasts) whose transcription does not require the synthesis of transcription factors following induction (33). To determine if serum-induced activation of the TF gene in HMC was dependent on de novo protein synthesis, quiescent HMC were stimulated with serum and/or the protein synthesis inhibitor cycloheximide (10 µg/ml) for 2 h before mRNA was harvested for Northern blot analysis (Figure 7a). Inhibition of protein synthesis alone resulted in a greater than 3.5-fold increase (P = 0.005) in normalized TF mRNA, as did stimulation with complete medium (Figure 7b). However, stimulation with complete medium in the presence of cycloheximide resulted in an even greater increase (P < 0.05) in normalized TF mRNA (5.5-fold rise over baseline) as shown in Figure 7b.

View larger version (32K): [in this window] [in a new window]   Figure 7.   Effect of protein synthesis inhibition on serum induction of TF mRNA in quiescent HMC. (Panel A) Representative Northern blot of 10 µg total cellular RNA per lane extracted from quiescent HMC stimulated for 2 h by either: (1) serum-free medium, (2) serum-free medium and 10 µg/ml cycloheximide, (3) complete medium, or (4) complete medium supplemented with 10 µg/ml cycloheximide. (Panel B) TF mRNA normalized to GAPDH mRNA for quiescent HMC stimulated, in either the absence (solid bars) or presence (hatched bars) of 10 µg/ml cycloheximide, by either serum-free medium or medium supplemented with 10% serum (complete medium). All data are mean ± SEM for six experiments.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

In health, the opposing surfaces of the parietal and visceral pleural mesothelium are separated by a thin layer of fluid which is formed from the submesothelial capillaries. This fluid has many characteristics of a plasma ultrafiltate with a protein concentration of 1-3 g/dL. Although albumin is the principal protein species in pleural fluid, fibrinogen and other coagulation proteins including prothrombin and factors V, VII, VIII, IX, X, XI, XII, and XIII are present (19). If TF expression by the pleura were constitutive as previously suggested (8), one would expect some degree of fibrin deposition to be detectable on normal mesothelium. Yet, despite the presence of all the soluble components required for fibrin production, there is no evidence of fibrin deposition on the normal mesothelium (34).

The absence of pleural fibrin deposition implies the coagulation cascade is either not activated under normal conditions or it is inhibited at some state. Transudative pleural effusions which develop because of an imbalance in the hydrostatic and/or oncotic pressures across the pleura are not associated with fibrin deposition and do not have procoagulant activity in vitro (5). In the present study, tissue factor was not demonstrable by immunohistochemical staining of the mesothelium of the patient who died with pulmonary edema and transudative pleural effusions after myocardial infarction. In contrast, the exudative effusions which develop as a consequence of pleural inflammation possess significant procoagulant activity and are associated with fibrin deposition on the pleural membranes (4, 5). This procoagulant activity is felt to be a consequence of the extrinsic coagulation protease cascade since the effusions contain no demonstrable intrinsic clotting activity (4).

The mesothelial cells lining the pleural cavity are ideally situated to initiate the extrinsic coagulation cascade. Previous investigators have shown that the mesothelium is capable of expressing tissue factor in vitro (7) and have speculated that HMC TF expression is constitutive both in vitro and in vivo (9). The present study suggests that in vivo HMC tissue factor expression is not constitutive but is inducible. In the absence of pleural disease, morphologically normal mesothelium did not express detectable tissue factor in vivo. However, in the presence of acute inflammation, immunohistochemical staining demonstrated marked TF expression by the mesothelium. Thus, activated mesothelial cells are excellent candidates for triggering the coagulation system as observed in exudative, but not transudative, effusions.

In contrast to in vivo studies, we found that HMC express TF when grown in the presence of serum in vitro. To reconcile the apparent discrepancy between our observations that normal mesothelium does not express TF in vivo and the constitutive expression of TF in vitro, we examined the effect of serum on mesothelial cell TF expression. Serum deprivation resulted in a progressive decline in normalized TF activity and protein levels, similar to that seen in other quiescent cells, such as fibroblasts.

Serum simulation of quiescent HMC markedly enhanced TF expression. This serum-induction of TF protein and activity in HMC was proceeded by a parallel increase in TF message, similar to that shown to occur upon serum-stimulation of quiescent human fibroblasts (12). Many other serum inducible genes do not require de novo protein synthesis for induction (33). Our cycloheximide studies similarly indicate de novo protein synthesis is not required for the serum induction of TF mRNA in quiescent HMC. TF gene expression was, in fact, superinduced by cycloheximide, as has been observed previously in fibroblasts (35).

Failure to detect TF by immunohistochemistry does not absolutely exclude the possibility that the mesothelium expresses extremely low levels of TF in vivo since the sensitivity of the technique is unknown. However, TF activity in vivo must have been lower than that of quiescent HMC in vitro since tissue factor expression was detected in these cells by immunohistochemical techniques. In a recent report, Verhagen and colleagues found minimal tissue factory activity in explants from omental peritoneum but constitute TF expression by harvested HMC grown in vitro (36). Although both human and fetal bovine serum stimulated TF message expression in vitro, only the former increased TF activity. These investigators attributed the lack of response to FBS stimulation to a translational or post-translational failure in protein synthesis. We found no evidence for such a defect in the present study, but differences in the source of mesothelial cells, culture conditions, and/or experimental techniques may account for the discrepancy.

In contrast to serum, supplementing the medium with plasma did not induce quiescent HMC to express TF in vitro, but instead resulted in a further decrease in TF expression. This observation is consistent with our in vivo finding that normal HMC do not express detectable TF since, in the absence of pleural injury, pleural fluid resembles an ultrafiltrate of plasma. In pleural inflammation fibrin deposition is characteristic (3) and the exudative effusions contain macrophages and other inflammatory cells that are capable of promoting the extrinsic coagulation cascade by induction or enhancement of TF expression (14, 17, 27, 37). Previous studies have found 2-4-fold increases in TF expression by macrophages and monocytes stimulated by inflammatory mediators such as LPS, TNF, and gamma IFN (16, 37, 38). We speculate that once intrapleural fibrin deposition is initiated by pleural macrophages, the conversion of the plasma ultrafiltrate to serum further enhances the process by inducing mesothelial cell TF expression. Alternatively, activation of platelets in the effusion may elaborate substances that directly induce TF expression, by either HMC or macrophages, which is then further enhanced by the subsequent conversion of the plasma ultrafiltrate to serum (37). This hypothesis is supported by our observation that, in the presence of acute pleural inflammation, mesothelial cells express TF in vivo.

HMC TF expression would be expected to result in thrombin generation which has recently been shown to stimulate HMC proliferation and act as a chemotactic factor for HMC (39). Thus, the inducible expression of TF by HMC after injury or inflammation to the mesothelium may represent a two-edge sword in the process of wound healing. When the pleural injury is severe enough to cause desquamation of the mesothelial cells lining the serosal surface, the denuded areas are covered by fibrin while thrombin stimulates HMC proliferation thus facilitating repopulation of the denuded area by HMC from the wound edge and opposing pleural surface (40). However, if left unchecked by fibrinolytic processes, fibrin may serve as a temporary matrix for fibroblast proliferation, and may develop into a thick peel covering the pleura and leading to the development of adhesions, and eventually, to a fibrothorax or trapped lung (3). Currently, we are attempting to identify the factor or factors present in serum which lead to the induction of HMC TF expression.