INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

TUMOR NECROSIS FACTOR (TNF)- is a major inflammatory cytokine released from macrophages and monocytes activated during inflammation and/or gram-negative infection, especially after trauma, burns, and major surgery (10, 49). The endothelium is one of the primary targets for such inflammatory cytokines (31), and high levels of TNF- are believed to contribute to altered integrity of the lung endothelial barrier, resulting in excessive transendothelial leakage of plasma proteins into the interstitium and interstitial pulmonary edema (16, 17, 46, 47). Purified TNF- added to the culture medium of confluent calf pulmonary artery endothelial (CPAE) monolayers increases their protein permeability (6, 50, 51); and the intravenous infusion of TNF- into postsurgical sheep can also increase lung endothelial protein permeability (39), very similar to that observed in postsurgical sheep after Pseudomonas or endotoxin challenge (7, 8, 11, 14). Such observations have suggested that high concentrations of TNF- in the lung microcirculation during pulmonary inflammation and/or bacterial sepsis after surgery or trauma may contribute to the loss of lung endothelial integrity (42).

Studies (16, 50) on the TNF--induced increase in transendothelial protein permeability with immunofluorescence microscopy have confirmed a dramatic disruption and/or reorganization of the fine fibrillar fibronectin (FN) network in the subendothelial extracellular matrix (ECM) of lung endothelial cell monolayers exposed to TNF-. This includes an apparent aggregation of FN in the matrix temporally associated with both an increase in protein permeability and the formation of gaps between cells in previously confluent monolayers (16, 17, 51). A very similar disruption of the FN matrix is seen after the addition of either monoclonal antibodies to the ₅₁-integrin complex or soluble Arg-Gly-Asp (RGD)-containing peptides to the culture medium (16, 17).

Addition of endotoxin-free and highly purified human plasma FN to the culture medium can both attenuate and reverse the TNF--induced increase in endothelial monolayer protein permeability, a protective response that appears to be dependent on the incorporation of the added soluble FN into the ECM (16, 17, 51). Similarly, the intravenous infusion of such commercially purified human FN into bacteremic sheep after surgical trauma to significantly elevate the circulating plasma FN concentration (by 20-40%) can also attenuate the increase in lung protein permeability once the infused human FN becomes incorporated into the lung interstitial ECM (11, 14, 40). Although disruption of the FN subendothelial matrix after TNF- exposure is now well documented, the mechanisms mediating this change in the fibrillar FN matrix are not known. However, an alteration of the FN matrix has the potential to either reduce endothelial cell adhesion or spreading, interrupt integrin-mediated ECM signaling to the cell cytoskeleton, or perhaps modify the integrity of the vascular barrier, including the exclusion properties of the matrix.

The organization of FN in the ECM of lung endothelial cell layers is influenced by several processes that include its assembly into fine FN fibrils within the ECM, its turnover potentially mediated by proteolysis, and the biochemical structuring of the FN lattice within the ECM. Structuring of FN in the normal ECM is influenced by disulfide exchange in the amino-terminal domain of FN and perhaps by transglutaminase-mediated cross-linking of FN based on experiments primarily with fibroblasts (3, 4). We determined whether TNF- could induce changes in the properties of FN within the ECM to explain its influence on both the fine fibrillar structure of the FN matrix and the protein permeability of previously confluent endothelial monolayers. In this regard, we tested the concept that TNF- may have enhanced extracellular transglutaminase activity, resulting in FN multimerization leading to disruption of the FN matrix and increased protein permeability of endothelial monolayers. Our observations indicate that the total amount of FN in the matrix as well as the FN mRNA levels were unchanged after TNF- exposure. Moreover, we were unable to detect the release of FN fragments from the cell layer after TNF- exposure. In contrast, there was a marked increase in the amount of nonreducible high molecular mass (HMW) multimers of FN detected in the ECM of the lung endothelial cell monolayers after TNF- exposure that was associated with enhanced extracellular endothelial cell surface transglutaminase activity. Blocking such enhanced transglutaminase activity attenuated the TNF--enhanced multimerization of the FN matrix in endothelial monolayers but did not prevent either the rearrangement of the fibrillar FN matrix as analyzed by immunofluorescence microscopy or the increase in protein permeability as quantified by transendothelial ¹²⁵I-albumin clearance.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Fibroblast and endothelial cell cultures. CPAE cells (American Type Culture Collection, Manassas, VA) were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal bovine serum (FBS; HyClone, Logan, UT), 100 U/ml of penicillin, and 100 µg/ml of streptomycin. Before TNF- treatment, the culture medium was replaced with DMEM supplemented with 5% FBS plus the antibiotics. For analysis of the effect on TNF- on the release of FN from the endothelial cell layers, the CPAE monolayers were cultured to confluence in 3-4 days and then washed four times with phosphate-buffered saline (PBS) before the addition of DMEM containing FN-deficient FBS prepared by gelatin-Sepharose affinity chromatography (20). A1-F human foreskin fibroblasts (originally obtained from Dr. Lynn Allen-Hoffman, University of Wisconsin, Madison) were cultured in DMEM supplemented with 10% FBS, 100 U/ml of penicillin, and 100 µg/ml of streptomycin. Both fibroblast and endothelial cell layers were exposed to recombinant human TNF- at a concentration of 200 U/ml medium for 18 h before analysis.

Phase-contrast and immunofluorescent microscopy. Endothelial cells cultured on coverslips were treated with and without 200 U/ml of TNF- for 18 h. The cells were examined and photographed with an inverted microscope (Olympus IX50). Immunofluorescent staining for FN in the cell layers was performed as described by Curtis and colleagues (16, 17) with an Olympus BX60 microscope. CPAE cells cultured on coverslips were fixed, permeabilized, and stained with a rabbit polyclonal antibody to bovine FN (Chemicon International, Temecula, CA) and a fluorescein goat anti-rabbit IgG secondary antibody (Molecular Probes, Eugene, OR).

Enzyme-linked immunosorbent assay and immunoblot analysis. Standard enzyme-linked immunosorbent assay (ELISA), dot blot, and Western blot analysis of FN were done with a rabbit polyclonal antibody to bovine FN (Calbiochem, San Diego, CA) and a secondary antibody of horseradish peroxidase-conjugated goat anti-rabbit IgG (Calbiochem). For ELISA, FN was detected with the 3,3',5,5'-tetramethylbenzidine microwell peroxidase substrate system (Kirkegaard & Perry Laboratories, Gaithersburg, MD). Dot blot and Western blot analyses were performed with a SuperSignal ULTRA chemiluminescent substrate (Pierce, Rockford, IL) with subsequent exposure to Fuji X-ray films (Fuji Medical Systems USA, Stamford, CT).

RNA isolation and Northern blot analysis. Poly(A)⁺ RNA was isolated directly from cell extracts of CPAE cells with the FastTrack kit (Invitrogen, San Diego, CA). RNA samples (1 µg) were electrophoresed on a 1.3% agarose gel containing 0.66 M formaldehyde and transferred to nitrocellulose as described by Davis et al. (18). DNA probes were radiolabeled with [-³²P]dCTP with the Random Primer Plus Extension Labeling System (New Life Science Products, Boston, MA), and hybridization was performed as previously described (19). Equal loading of mRNA samples on Northern blots was verified by simultaneous probing with a cDNA specific to mouse -actin.

Detection of HMW multimers of ¹²⁵I-FN incorporated in the matrix of endothelial cells. FN purified from fresh human plasma by gelatin-Sepharose affinity chromatography (20, 38) was iodinated as described by Rebres et al. (38). The iodinated FN was confirmed to be intact (440 kDa) by SDS-PAGE and autoradiography. At 48 h after CPAE cells were seeded at a density of 10,000 cells/cm², the medium of the preconfluent cells was changed to FN-deficient medium supplemented with 3 µg/ml of ¹²⁵I-FN. After an 18-h incubation with and without TNF-, the cells were washed four times with PBS and solubilized in SDS-PAGE sample buffer [0.125 M Tris (pH 6.8), 4% SDS, and 10% glycerol]. Samples with equal radioactivity were analyzed by SDS-PAGE and autoradiography. Polyacrylamide gradient gels (4-15%) with a 3% stacking gel were used for PAGE analysis. When analyzed under reducing conditions, the concentration of -mercaptoethanol in the sample buffer was 2.5% (vol/vol). The gels were then dried and exposed to Kodak Bio-Max films with an intensifying screen. When the effect of transglutaminase inhibitors on the formation of nonreducible HMW FN was examined, monodansylcadaverine (MDC; Sigma) at 0.1 mM or cystamine (Cys; Sigma) was added during TNF- treatment. Both MDC and Cys are primary amine substrates that have been routinely used to inhibit transglutaminase activity (24, 41, 44).

Assay of transglutaminase activity in cell lysate. Transglutaminase activity was determined on the basis of [³H]putrescine incorporation into dimethylated casein as previously described by Korner et al. (29). Cells were washed twice with cold PBS, once with Tris-EDTA-dithiothreitol (TED) buffer (50 mM Tris · HCl, pH 7.4, 1 mM EDTA, and 1 mM dithiothreitol) supplemented with 150 mM NaCl (TED-buffered saline), and then scraped into 1 ml TED-buffered saline/dish. The cells were collected by centrifugation; resuspended in 200 µl of TED buffer containing 0.5 mM phenylmethylsulfonyl fluoride, 5 µg/ml of pepstatin, and 5 µg/ml of leupeptin; and then lysed by three cycles of freezing and thawing. The final reaction mixture (120 µl) consisted of 50 mM Tris · HCl, pH 7.4, 10 mM CaCl₂, 15 mM dithiothreitol, 0.5 mM (1 µCi) putrescine, 50 mM NaCl, and 0.8 mg of dimethylated casein, and the cell lysates that contained 80-130 µg of total protein. After incubation at 37°C for 9-15 min, the proteins were precipitated with TCA, and free putrescine was removed by passage through a filter paper as described by Korner et al. Each sample had a negative control in which CaCl₂ in the reaction mixture was replaced with 50 mM EDTA. Putrescine incorporation, used as a measure of transglutaminase activity (29), was quantified by scintillation counting.

Assay of cell surface transglutaminase activity. Cell surface transglutaminase activity was specifically determined by measuring the incorporation of biotinylated cadaverine (Pierce) into FN on precoated 96-well plates as described by Verderio et al. (48). Briefly, the plates were first coated with purified human FN at 10 µg/ml. Then endothelial cells that were either not exposed to TNF- (control) or pretreated with TNF- (200 U/ml) were seeded at various densities from 10,000 to 50,000 cells/well and incubated at 37°C for 3 h in the presence of 0.5 mM biotinylated cadaverine. The reaction was first stopped by rinsing the wells with 2 mM EDTA in PBS, and then the cell layers were removed by incubation for 20 min with 0.1% sodium deoxycholate in PBS at room temperature. Cadaverine incorporation was quantified with horseradish peroxidase-conjugated streptavidin (Amersham, Piscataway, NJ) and the 3,3',5,5'-tetramethylbenzidine color-developing kit (Kirkegaard & Perry Laboratories).

Preparation of preformed endothelial cell matrices and assay of cell adhesion and spreading. Confluent endothelial monolayers cultured on glass coverslips in 12-well plates were treated for 18 h with either TNF- (200 U/ml) alone or TNF- in conjunction with the transglutaminase inhibitor Cys (0.05 mM) to obtain two preformed matrices: one in which the fine fibrillar FN matrix was disrupted or reorganized with the parallel formation of the HMW FN multimers inhibited (TNF- and cystamine treated), and the other in which the FN matrix was disrupted in parallel with the TNF--induced formation of nonreducible HMW FN multimers (TNF- treated).

Radiolabeling of albumin and assay of transendothelial protein permeability. Transendothelial protein permeability was assessed by measuring the diffusive protein (¹²⁵I-albumin) permeability across the endothelial monolayer with a dual-chamber monolayer system as previously described (15-17, 50, 51). This technique allows for the measurement of transendothelial albumin flux in the absence of a hydrostatic or oncotic pressure gradient (15). This in vitro model system consists of a luminal chamber containing the "tissue culture ready" filter that is covered with a confluent monolayer of CPAE cells. A styrofoam collar around the luminal chamber allows it to float in a larger abluminal chamber. This allows the fluid height in both chambers to be maintained at an identical level to eliminate the convective flux of albumin across the monolayers. During each experiment, the fluid in the abluminal chamber was stirred constantly, and both chambers were kept at 37°C with a thermostatically controlled water bath.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

TNF- causes a decrease in monomeric FN detectable under reducing conditions. Previous immunofluorescent studies (16, 50) revealed a dramatic reorganization of the FN matrix under the endothelial cell monolayers after TNF- exposure. The matrix FN changed from a fibrillar meshwork of fine FN fibers into a matrix with predominantly thick bundles of FN after TNF- exposure. To determine whether this reorganization of the FN matrix was related to a change in the quality and/or quantity of FN in the matrix after TNF- exposure, we analyzed the FN in cell lysates by Western blotting under reducing conditions. Unexpectedly, with limited film exposure (Fig. 1A), we observed that the lysate of TNF--treated endothelial cells contained almost no monomeric FN detectable at 220 kDa. More extensive exposure (Fig. 1B) revealed the presence of monomeric FN that was still much less after TNF- exposure than in control cells. We initially speculated that the reduction in FN monomers could be due to a drastic detachment of cells, a pronounced increase in FN degradation, and/or a decrease in FN synthesis. We therefore examined the endothelial cell monolayers by both phase-contrast and immunofluorescent microscopy using an antibody to FN. Figure 2 shows that after an 18-h exposure to TNF- (200 U/ml), the endothelial cell layer was still intact, although intercellular gaps in the previously confluent monolayer became very noticeable (compare Fig. 2, A and B), and the fine FN subendothelial matrix was disrupted, with thick bundles of FN apparent (Fig. 2, C and D). Thus the negligible detection of monomeric FN under reducing conditions was not due to the loss of CPAE cells from the monolayer after TNF- exposure.

View larger version (102K): [in this window] [in a new window]   Fig. 1.   Effect of tumor necrosis factor (TNF)- on the amount of monomeric fibronectin (FN) detected under reducing conditions by Western blot analysis. Confluent calf pulmonary artery endothelial (CPAE) cells were incubated with and without 200 U/ml of TNF- for 18 h (n = 4 wells/group). Cells were washed and scraped into SDS-PAGE sample buffer. Aliquots of cell lysates containing equal amounts of total protein were analyzed for FN by Western blotting under reducing conditions with a polyclonal antibody to bovine FN. Purified plasma FN standard (Std) was used to indicate the position of the well-characterized 220-kDa monomer of FN. A: limited exposure of the film for development. B: exposure for a more extensive interval.

View larger version (172K): [in this window] [in a new window]   Fig. 2.   Endothelial cell monolayers were still visually intact and matrix FN was mostly in the form of thick bundles after TNF- exposure. Confluent endothelial cells were treated with (B and D) and without (A and C) 200 U/ml of TNF- for 18 h. Cells were either photographed under an inverted microscope (A and B) or fixed, permeabilized, and stained for FN (C and D). TNF- caused the formation of gaps between endothelial cells (B) and the loss of the fine fibrillar FN matrix (D) but did not elicit significant cell loss from the monolayer.

View larger version (33K): [in this window] [in a new window]   Fig. 3.   TNF- did not change either the FN content or FN mRNA levels in CPAE cell layers. Confluent CPAE cells were incubated in FN-deficient medium with and without (control) 200 U/ml of TNF- for 18 h. A: at 3-, 6-, or 18-h time points, 15 µl of the culture medium were collected and assayed for FN content by ELISA with a polyclonal antibody against bovine FN. n, No. of samples. B: reduced samples of the culture medium harvested from both control (CON) and TNF--treated cells (done in duplicate) were analyzed by Western blot with a polyclonal antibody against bovine FN. C: confluent CPAE monolayers (n = 4) that had been incubated with and without TNF- were washed and scraped into SDS-PAGE sample buffer, and cell lysates (containing an equal amount of total protein) were assayed for FN content by dot blot analysis under reducing conditions. D: CPAE monolayers were assayed at 18 h for mRNA content by Northern blot with a rat FN cDNA probe, with actin mRNA used as a loading control.

TNF- caused an increase in nonreducible HMW FN multimers in endothelial cell layers. Because the total FN content in the cell lysate was also unchanged after TNF- (Fig. 3C), we hypothesized that the reduced monomeric FN in TNF--treated cells could mean that much of the FN may have existed as nonreducible HMW complexes, which were perhaps not transferred onto the nitrocellulose membrane and therefore not detectable by subsequent antibody probing.

View larger version (52K): [in this window] [in a new window]   Fig. 4.   TNF- increases the formation of nonreducible high molecular mass (HMW) FN complexes in lung endothelial monolayers. Preconfluent endothelial cells were incubated in FN-deficient medium containing 3 µg/ml of 125I-FN with and without 200 U/ml of TNF- for 18 h. Cell layers were washed 4 times and scraped into SDS-PAGE sample buffer. Aliquots of cell lysates that contained equal 125I radioactivity were analyzed by SDS-PAGE and autoradiography under both nonreducing (A) and reducing (B) conditions. The autoradiographs from a representative experiment were scanned, and the percentage of total FN that existed as HMW FN was quantified (C). Percentage of total FN retained as HMW FN multimers under reducing conditions was 28.3 ± 3.4% in control monolayers (n = 3) and 57.0 ± 0.5% in TNF- treated monolayers (n = 3). The amount of HMW FN multimers detected under reducing conditions (C) in the TNF--treated group was significantly greater than that in control group (*P < 0.05).

The increase in nonreducible HMW FN multimers is associated with an increase in transglutaminase activity in TNF--treated endothelial cells. FN can be processed into HMW complexes by a transglutaminase-mediated cross-linking mechanism (29, 32). Thus one possible explanation for the observed increase in nonreducible HMW FN complexes detected after exposure of the endothelial cell layers to TNF- is an increase in transglutaminase activity, although this has not been previously documented. Accordingly, we measured transglutaminase activity in whole endothelial cell lysates by quantifying the incorporation of [³H]putrescine into dimethylated casein (29). Transglutaminase activity increased in the endothelial cell layers in a dose-dependent manner after TNF- exposure (P < 0.05; Fig. 5B) in association with the appearance of nonreducible HMW FN multimers (Fig. 5A), suggesting that the larger amounts of nonreducible HMW FN complexes detected in TNF--treated endothelial cells may be due to an unexpected elevation in transglutaminase activity.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Ability of TNF- treatment of endothelial cells to cause a dose-dependent increase in both HMW FN multimers and transglutaminase activity. Preconfluent endothelial cells incubated in FN-deficient medium supplemented with 125I-FN (3 µg/ml) were exposed to increasing concentrations of TNF- for 18 h. After the 18-h incubation, the cell layer was washed 4 times and scraped into reducing SDS-PAGE sample buffer. Aliquots of the cell lysates that contained equal radioactivity were analyzed by SDS-PAGE and autoradiography under reducing conditions. A: representative autoradiographic film. B: lysed cell layers assayed for transglutaminase activity by measurement of [3H]putrescine incorporation. Values are means ± SE of 3 wells/TNF- dose. Transglutaminase activity was greater after 50 or 800 U/ml of TNF- compared with that in control layers (P < 0.05).

View larger version (62K): [in this window] [in a new window]   Fig. 6.   Inhibitors of transglutaminase can attenuate TNF--induced formation of HMW FN in endothelial cell layers. Preconfluent CPAE cell layers were incubated in FN-deficient medium containing 3 µg/ml of 125I-FN in the absence and presence of 2 different transglutaminase inhibitors, 0.10 mM monodansylcadaverine (MDC) and 0.05 mM cystamine (Cys). After 18 h of incubation with TNF- (200 U/ml) at 37°C, the cells were washed 4 times and scraped into SDS-PAGE sample buffer, and cell lysates were analyzed by SDS-PAGE and autoradiography under reducing conditions (A). For quantification, the autoradiographic films in A were scanned, and the percentage of the total FN that existed as nonreducible HMW FN multimers was calculated (B). Values are means ± SE of 3 repeat experiments, each of which had 2 wells containing no inhibitor and 2 wells with each inhibitor (total n = 6/each).

The amount of nonreducible FN multimers corresponds to the level of transglutaminase activity in both endothelial cells and fibroblasts. If a TNF--induced increase in transglutaminase activity was actually the basis for the significant increase in nonreducible HMW complexes of FN detected in the endothelial matrix, we predicted that there should be very little formation of such HMW FN complexes with fibroblast cell layers because fibroblasts have limited transglutaminase activity. In a side-by-side comparative analysis, we observed that nearly all of the FN in the fibroblast ECM existed as HMW species under nonreducing condition (data not shown). More importantly, the basal level of transglutaminase activity in control fibroblast cell layers was only ~20% of that detected in endothelial cell layers (Fig. 7). Furthermore, the transglutaminase activity in the fibroblast cell layers remained low even after treatment with TNF- for 18 h, whereas with endothelial cell layers, transglutaminase activity increased in a dose-dependent manner in response to TNF- (P < 0.05; Fig. 7). This finding was consistent with two additional observations; first, most of the HMW FN multimers in the fibroblast cultures were readily reducible to FN monomers (Fig. 8A), and second, TNF- was unable to further increase the formation of nonreducible HMW FN multimers in the ECM of the fibroblasts, although it could readily do so with endothelial cells (Fig. 8B). These results suggest a correlation between the formation of HMW nonreducible complexes of FN in the ECM and the level of transglutaminase activity. These findings also support the conclusion that a TNF--induced increase in transglutaminase activity in the CPAE cell layers may have contributed to the increased formation of such nonreducible HMW FN complexes in the matrix.

View larger version (26K): [in this window] [in a new window]   Fig. 7.   Comparative analysis of transglutaminase activity in fibroblast and endothelial cell layers as influenced by TNF-. Confluent cell layers of endothelial cells and fibroblasts were incubated with TNF- for 18 h and then assayed for transglutaminase activity by measurement of [3H]putrescine incorporation. Values are means ± SE; nos. in parentheses, total no. of determinations. For endothelial cells, there were 3 experiments, each with triplicate wells. For fibroblasts, there were 3 experiments, each with duplicate wells. Transglutaminase activity in normal fibroblasts was ~20% of that in endothelial cells. TNF- caused a dose-dependent increase in transglutaminase activity in endothelial monolayers (P < 0.05) but had no significant effect on transglutaminase activity in fibroblasts (P > 0.05).

View larger version (38K): [in this window] [in a new window]   Fig. 8.   TNF- did not stimulate the formation of increased amounts of nonreducible HMW FN multimers in fibroblast cell layers. Confluent fibroblasts were incubated in FN-deficient medium containing 3 µg/ml of 125I-FN in the presence and absence of 200 U/ml of TNF- for 18 h. After incubation at 37°C, the cell layers were washed 4 times and scraped into SDS-PAGE sample buffer. Aliquots of the fibroblast cell lysate (containing equal amounts of radioactivity) were analyzed by SDS-PAGE and autoradiography under reducing conditions (A). The autoradiograph was scanned, and the percentage of total FN that existed as HMW FN in the fibroblast cell layers was quantified and compared with that in endothelial cells (B). Values are means ± SE from 2 experiments, each of which had 3 control and 3 TNF- wells. TNF- increased the formation of nonreducible HMW FN in CPAE cells (P < 0.01) but not in fibroblasts.

TNF--induced, transglutaminase-mediated FN cross-linking activity is located on the surface of the endothelial cell. If the enhanced formation of nonreducible HMW complexes of FN in the ECM was indeed caused by TNF--induced transglutaminase activity, we would expect to see an increase in extracellular FN cross-linking activity on the surface of the endothelial cells and/or in the culture medium. Thus we examined the ability of both the endothelial cells and their conditioned medium to cross-link ¹²⁵I-FN that was precoated on a cultured surface. Lung endothelial cells were first preincubated with and without TNF- (200 U/ml) for 18 h, gently detached by mild trypsinization, washed in FN-deficient medium, and then plated onto the ¹²⁵I-FN coated wells and allowed to adhere. After 3 h, the cells and their ¹²⁵I-labeled matrices were dissolved and analyzed for the presence of HMW ¹²⁵I-FN complexes.

View larger version (53K): [in this window] [in a new window]   Fig. 9.   TNF--treated endothelial cells can enhance the cross-linking and multimerization of FN precoated on a culture surface in the absence of conditioned medium. CPAE cells with and without TNF- (200 U/ml) pretreatment for 18 h were trypsinized, washed once with FN-deficient medium, and seeded onto 12-well culture plates precoated with 125I-FN. After 3 h of incubation, cells were washed 4 times, scraped into SDS-PAGE sample buffer, and then analyzed by SDS-PAGE and autoradiography under nonreducing (A) or reducing (B) conditions. The autoradiographs were scanned, and the percentage of total cell layer FN that existed as nonreducible HMW FN was calculated (C). Values were derived from 2 repeat experiments.

View larger version (14K): [in this window] [in a new window]   Fig. 10.   TNF- caused an increase in cell surface transglutaminase activity on endothelial cells as determined by analysis of the incorporation of biotinylated cadaverine into FN precoated on culture surfaces of 96-well plates with human FN at 10 µg/ml. Endothelial cells pretreated with and without (control) TNF- were seeded at the indicated no. of cells/well and incubated at 37°C in the presence of 0.5 mM biotinylated cadaverine for 3 h. The incorporated cadaverine was quantified by measuring optical density (OD) at 450 nm after treatment with horseradish peroxidase-streptavidin and ELISA color development. Values are means ± SD from 6 experiments. Activity in TNF--treated groups was greater than in control groups at all 3 levels of cell seeding (P < 0.05).

Effect of the inhibitors of transglutaminase on TNF--induced reorganization of the FN matrix, cell adhesion and spreading on the matrix, and endothelial protein permeability. Tissue transglutaminase activity has been found in many cell types including endothelial cells (13, 26). Regulated expression of transglutaminase may also influence FN incorporation and play an important role in the regulation of cell attachment and apoptosis (48). Because exposure of the endothelial monolayers to TNF- can induce both a disruption of the fine FN fibers in the matrix and an increase in endothelial protein permeability, we determined whether these events were dependent on the increase in transglutaminase activity and the enhanced formation of nonreducible HMW FN complexes in the ECM. Accordingly, we treated the lung endothelial cell layers with TNF- (200 U/ml) in the absence and presence of the transglutaminase inhibitor Cys to attenuate the formation of nonreducible HMW FN multimers in the ECM of TNF--treated cells. Then we used these two different matrices to study cell adhesion and cell spreading. First, as shown in Fig. 11, Cys at concentrations that blocked the formation of nonreducible HMW FN complexes was unable to prevent the TNF--induced reorganization of the FN matrix as studied by immunofluorescent microscopy (Fig. 11, A and B). We next studied cell adhesion and cell spreading on these preformed matrices. To do this, the original cell layers were gently removed with 0.5% sodium deoxycholate, and normal CPAE endothelial cells were then seeded onto these preformed matrices. This allowed us to determine whether CPAE cells adhere or spread differently on the cross-linked FN matrix containing the nonreducible HMW multimers compared with the non-cross-linked FN matrix. As shown in Fig. 11, C-F, endothelial cell adhesion and cell spreading were essentially identical on the subendothelial matrices preformed by TNF--treated CPAE cells in the presence and absence of the transglutaminase inhibitor (Fig. 11, C and D). Cell adhesion to the FN network is shown by overlays of staining for ₅₁-integrins (green) and the matrix FN (red; Fig. 11, E and F). Direct microscopic counting of 12 areas/slide confirmed that cell adhesion and cell spreading were basically similar on these two different matrices (Fig. 12).

View larger version (117K): [in this window] [in a new window]   Fig. 11.   Matrices preformed by CPAE cell layers exposed to TNF- in the presence and absence of a transglutaminase inhibitor supported similar adhesion and spreading of normal CPAE cells. Cell matrices were prepared over 3 days in culture by confluent endothelial cell monolayers treated with TNF- (A) or TNF- plus 0.05 mM Cys (B). Cell layers were removed with 0.5% sodium deoxycholate. Control cells were then seeded on both matrices and incubated at 37°C for 30 min in the adhesion spreading assay. After removal of nonadhered cells, the coverslips were fixed and stained with both an anti-FN (red; A and B) and an anti-51-integrin (green; C and D) antibody. A and C: same viewing field of cells adhering to TNF--treated FN matrix. E: computer overlay of A and C to show cell spreading on FN. B and D: same viewing field of cells adhering to TNF- plus Cys-treated FN matrix. F: computer overlay of B and D to show cell spreading on FN.

View larger version (24K): [in this window] [in a new window]   Fig. 12.   Comparative analysis of CPAE cell adhesion and cell spreading on subendothelial matrices preformed by TNF--treated endothelial monolayers in the presence and absence of Cys. Both matrices manifest a reorganization of their fine fibrillar FN network but with reduced HMW FN multimerization in those supplemented with Cys. Cell adhesion and cell spreading were assayed by microscopy with a fluorescent microscope. Cells were stained with the anti-51-integrin antibody. Values are means ± SD of 12 areas counted.

View larger version (24K): [in this window] [in a new window]   Fig. 13.   Effect of inhibitors of transglutaminase on the TNF--induced increase in transendothelial protein permeability. Confluent CPAE cell monolayers on tissue culture-ready filter wells were treated with and without TNF- (200 U/ml) for 18 h in the presence and absence of 2 transglutaminase inhibitors, 0.10 mM MDC and 0.05 mM Cys. Transendothelial protein permeability was determined by measuring the clearance of 125I-labeled albumin across endothelial monolayers. Values are means ± SE of 4 separate wells that were typically completed in each separate experiment. The experiment was repeated 5 separate times. Protein clearance was significantly increased (P < 0.05) after TNF- and not blocked (P > 0.05) by either inhibitor.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The present study indicates that TNF- causes an increase in the formation of nonreducible HMW FN complexes in endothelial cell layers, potentially due to an enhanced transglutaminase activity on the surface of endothelial cells. This conclusion is supported by at least three observations. First, both the increased formation of nonreducible HMW matrix FN complexes and the increase in cellular transglutaminase activity in the endothelial cell layers was TNF- dose dependent. Indeed, fibroblast cell layers with low basal levels of transglutaminase activity contained a matrix with little nonreducible HMW FN, and we observed no increase in transglutaminase activity in fibroblast cell layers after TNF- exposure. Second, inhibition of transglutaminase activity abolished the TNF--induced increase in nonreducible HMW matrix FN while also reducing the constitutive amounts of such HMW FN complexes detected in control endothelial cells not exposed to TNF-. Third, TNF--treated endothelial cells displayed high levels of cell surface transglutaminase activity and were capable of cross-linking substrate FN precoated on culture plates at a much greater level compared with that of untreated control endothelial cell layers.

TNF- is released from activated mobile monocytes and macrophages, from sessile macrophages such as hepatic Kupffer cells (12), and from cells within the gut during bacteremia and/or endotoxemia (30). TNF- is believed to make a positive contribution to the inflammatory and generalized host response to sepsis (12, 30) by increasing endothelial permeability, thus allowing certain plasma proteins important for both wound repair and immune function to enter the extravascular space. However, high levels of TNF- can also elicit pathophysiological effects in that they can disrupt the integrity of the lung endothelial barrier (16, 50, 51) and elicit apoptosis of endothelial as well as of tumor cells (36). Intracellular transglutaminase activity is elevated when cells undergo apoptosis, and transglutaminase-mediated cross-linking of proteins has been suggested to contribute to the regulation of programmed cell death (21, 27).

The present study documents an increase in extracellular transglutaminase activity in the bovine lung endothelial cell layer after exposure to TNF-. It is not likely that this increase in transglutaminase activity was due to the nonspecific release of the enzyme from apoptotic endothelial cells because it was only found associated with the harvested CPAE cells and not in their conditioned medium. Indeed, our findings indicate that the increased transglutaminase activity that was likely mediating the enhanced FN multimerization after TNF- exposure was on the cell surface and not in the culture medium. Extracellular tissue transglutaminase has been clearly documented in many cell types including endothelial cells (1, 2, 5, 21, 32, 43). A recent study by Gaudry et al. (22) has shown for the first time the secretion and expression of transglutaminase on the surface of human umbilical vein endothelial cells. Our studies not only confirm these novel findings but also establish their pathological relevance, demonstrating an increased surface expression of transglutaminase activity on endothelial cells after exposure to the inflammatory cytokine TNF-.

For most adherent cell types, FN in the ECM is organized into HMW complexes primarily by interchain disulfide bonds. However, the cross-linking of FN by extracellular transglutaminase appears to provide an additional mechanism especially available to endothelial cells, although it can be seen to a limited degree in fibroblast cell layers. Transglutaminase-mediated cross-linking of FN (26, 32, 48), fibrinogen (33), and proteoglycans (28) into HMW complexes can take place constitutively in endothelial cells. TNF- has also been shown to upregulate transglutaminase expression in other cell types such as brain astrocytes (34). Our present findings, which suggest increased cell surface transglutaminase in TNF--treated lung endothelial cell layers, provide a mechanism for the enhanced FN multimerization observed in the monolayers after exposure to this cytokine.

Our results show that the majority of matrix FN (~80%) was in the form of HMW complexes in the endothelial cell layers, with ~30% of these HMW FN multimers being stable under reducing conditions. This is in contrast to cultured fibroblasts where, although a major portion of the FN in the ECM also exists in HMW form, most of the HMW FN complexes in the ECM of fibroblasts will readily become monomeric (220 kDa) under reducing conditions. Consistent with this observation is the fact that it appears that fibroblast cell layers have very low levels of transglutaminase activity to mediate such FN multimerization compared with CPAE cells. However, Barry and Mosher (3, 4) have shown that activated factor XIII, also known as plasma transglutaminase, can stimulate the cross-linking of FN in the ECM of fibroblasts. They speculated that both FN already assembled in the ECM and FN in the process of being incorporated can be acted on by activated factor XIII. Based on studies with fibroblast cell layers, it has been proposed that the cross-linking of FN by transglutaminase may work in conjunction with disulfide-bonded FN multimerization to stabilize the assembly of the FN matrix (3).

An unexpected observation was that the increased transglutaminase activity induced by TNF- did not cause additional HMW FN complexes to be seen under nonreducing conditions. This finding suggests that transglutaminase-mediated cross-linking of FN in CPAE cell layers is primarily influencing those FN molecules already incorporated within the matrix due to disulfide linkage. This potential extracellular function of transglutaminase, which has been released and localized to the cell surface of endothelial cells after exposure to TNF-, may enable the endothelium to stabilize its subendothelial matrix especially in response to local vascular inflammation, causing the release of TNF-, a conclusion consistent with the report by Gentile et al. (23) that indicates that cell-matrix interactions can be stabilized by transglutaminase activity. The speculation that constitutively expressed extracellular transglutaminase can stabilize cell-matrix interactions is further supported by the findings that cells deficient in transglutaminase are less able to adhere to an ECM compared with normal cells. However, it is also possible that the enhanced transglutaminase cross-linking of matrix FN may have pathological effects, including alteration of the signaling properties of the ECM or perhaps a reduction in FN turnover in the ECM.

TNF- has been shown to alter the subendothelial FN matrix and reduce the barrier function of the endothelial cell monolayers (16, 17, 50, 51). After exposure of CPAE cells to TNF-, the matrix FN is characterized by the formation of nonreducible HMW FN multimers and the appearance of thick FN bundles deep within the matrix (16, 17). There is also reduced colocalization of endothelial cell surface FN ₅₁-integrins with the fine FN fibers in the matrix (16, 17). Our current findings suggest an important role for enhanced endothelial cell surface transglutaminase activity in the formation of nonreducible HMW complexes of FN in the matrix after TNF- exposure. However, our findings also suggest that such cross-linking of matrix FN is not the basis for either the increase in endothelial protein permeability or the reorganization of the fine fibrillar FN matrix because transglutaminase inhibitors did not prevent either of these TNF--induced changes. Moreover, endothelial cell adhesion and cell spreading on the FN-rich matrix was also not altered by Cys treatment to block transglutaminase.

Another possible factor contributing to the altered ECM is the degradation of matrix FN by proteases. Partridge et al. (35) showed by zymography that medium from TNF--treated microvessel endothelial cells contains metalloproteinase that can cleave FN and other matrix proteins. TNF- has also been shown to induce the release of metalloproteinases such as gelatinase and collagenase (35), which have the potential to cause matrix degradation and/or matrix FN reorganization. But TNF- also causes endothelial cells to release protease inhibitors such as plasminogen activator inhibitor-1 and -2 and tissue inhibitor of metalloproteinases (25, 45), which can also block matrix proteolysis. In this regard, the recent findings by Curtis et al. (17) clearly document that local proteolysis of matrix FN is not the biochemical mechanism causing disruption and/or reorganization of FN in the ECM of CPAE monolayers after their 18-h exposure to TNF-, a protocol also used in the present study.

In the present study, we did not detect by Western blot analysis or tracer ¹²⁵I-FN experiments any significant increase in the amount of FN fragments in CPAE cell layers or their culture medium after TNF- exposure. These findings confirm and extend the recent study from our laboratory (17) documenting that although conditioned medium from TNF--treated CPAE cells contains proteolytic activity that can degrade both gelatin (denatured collagen) as well as FN, no FN fragments were found in either the cell lysate or the conditioned medium. Obviously, the proteolytic balance within the medium and/or matrix will determine the presence or absence of FN fragments after TNF- exposure.

The possibility that the increase in the cross-linking of FN matrix could be a protective response of endothelial cells to the presence of TNF- warrants consideration. For example, cross-linking of matrix proteins by extracellular transglutaminase can promote cell-matrix interaction (17) and potentially stabilize matrix integrity. Accordingly, increased extracellular transglutaminase activity leading to the enhanced cross-linking of FN assembled in the matrix via disulfide-bonded multimer formation may provide endothelial cells an additional mechanism to resist the cytotoxic effects of TNF-, especially in regard to the integrity of the ECM, which is vital to cell adhesion and endothelial barrier function. The rapid normalization of barrier function that can be seen after the addition of excess soluble FN to the culture medium followed by its ECM incorporation (16, 17, 50, 51) may reflect the fact that a large portion of the endogenous FN already assembled in the ECM became cross-linked after TNF- exposure.