Human Endothelial Cell Damage by Neutrophil-Derived Cathepsin G
Role of Cytoskeleton Rearrangement and Matrix-Bound Plasminogen Activator Inhibitor-1
Abstract Cathepsin G, a major protease released by activated neutrophils, induces functional and morphological damage to human endothelial cells. We studied the mechanisms involved and ways to reverse this damage. Cathepsin G induced a concentration- and time-dependent injury to human umbilical vein endothelial cell (HUVEC) morphology simultaneous with cytoskeleton rearrangement. Preincubation of the endothelial monolayer with phallacidin completely prevented damage to cell morphology by cathepsin G, whereas preincubation with cytochalasin B potentiated its activity. Damage to cell shape and F-actin cytoskeleton were prevented by eglin C, an inhibitor of the active site of cathepsin G. Furthermore, cathepsin G increased transcellular permeability to albumin and induced a time-dependent detachment of PAI-1 from the extracellular matrix of a cell-free system. The inhibition of matrix-bound PAI-1 activity by specific antibodies induced changes in HUVEC monolayers similar to those observed after cathepsin G. However, although stabilization of F-actin microfilaments by phallacidin prevented changes in cell shape, it did not prevent the ability of cathepsin G to increase cell permeability and release matrix PAI-1. The damage of cathepsin G to cell morphology and cytoskeleton arrangement was reversed within 12 hours if the deendothelialization area was <50% to 55% and the subendothelial matrix was still able to bind the newly synthesized PAI-1. Thrombin, whose role in the thrombotic process is well known, also induced changes in cell morphology and cytoskeleton arrangement of HUVEC. Cathepsin G reaches the subendothelial matrix through an increase in cell permeability and injures endothelial cell morphology by detaching matrix-bound PAI-1. These events expose a highly thrombogenic surface to which platelets can adhere, become activated, attract further neutrophils, and trigger thrombus formation.
- Received May 19, 1995.
- Accepted August 3, 1995.
Neutrophils are involved in several pathological processes such as inflammation, thrombosis, and tumor cell metastasis.1 2 3 4 5 With cell activation, proteolytic enzymes that are able to affect many cellular functions are released from neutrophil azurophilic granules.6 Among these proteolytic enzymes, cathepsin G acts as a potent agonist of platelet aggregation and alters the functional and morphological integrity of endothelial cells by increasing vascular permeability and promoting growth factor synthesis.7 8 9 10
The endothelial lining is a nonthrombogenic surface formed by a continuous layer of cells that actively modulates critical functions of the vascular wall and the cellular components of blood.11 To ensure endothelial integrity, cells firmly adhere to the subendothelial matrix and to adjacent cells. The exposure of the endothelial monolayer to cathepsin G induced cell contraction with loss of cell-cell contact and formation of intercellular gaps.9 12 These phenomena can expose the subendothelial matrix, leading to a loss in the so-called “nonthrombogenic properties” of the vascular wall.
It is not known whether the cell cytoskeleton, important in maintaining cell shape and allowing cell contraction and motility,13 14 is involved in the effect of cathepsin G on endothelial cell integrity. We have previously shown that human cathepsin G increased PAI-1 levels in culture medium by detaching the main inhibitor of plasminogen activator from the subendothelial matrix.12 PAI-1 bound to vitronectin in the matrix is known to be a spreading factor for several cell types.15 Although matrix composition is considered important for cell growth and migration,16 17 it is not yet known to what extent matrix proteins such as PAI-1 are involved in modulating the noxious effect of cathepsin G on the endothelial monolayer. Finally, quick repair of vascular integrity after injury can limit the extent of the thrombotic process and modulate further evolution of atherosclerosis.13 18
The aim of the present study, therefore, was to evaluate cellular events after treatment of endothelial cells and the underlying matrix with cathepsin G. Our results show that cathepsin G affects the morphological integrity of endothelial cells through a mechanism involving vitronectin-bound PAI-1. Presumably, the detachment of PAI-1 from the subendothelial matrix induces F-actin cytoskeleton rearrangement with consequent changes in endothelial cell shape.
Medium 199, glutamine, penicillin, streptomycin, NCS, and trypsin were obtained from GIBCO laboratories; cathepsin G, purified from human PMN, was purchased from Calbiochem Corp; collagenase was obtained from Boehringer Mannheim Corp; and Triton X-100 was obtained from Aldrich Chimica S.r.l.. Human thrombin, phallacidin, cytochalasin B, bovine serum albumin, gelatin, collagenase, and rhodamine-phalloidin were purchased from Sigma Chemical Co; glutaraldehyde, formaldehyde, osmium tetroxide, and tannic acid were purchased from Fluka. Polycarbonate filters and cylinders (transwell-clear) were purchased from Costar Corp and rabbit anti–human PAI-1 antibody was obtained from Kabru. Antiserum to human vitronectin was a kind gift from Dr D.J. Loskutoff (Research Institute of Scripps Clinics). Recombinant eglin C (CGP 32968) was kindly provided by Ciba Geigy AG. Reagents for the determination of LDH and PAI-1 levels were obtained from Sigma Chemical Co and Biopool, respectively.
Endothelial Cell Cultures
Endothelial cells were isolated from human umbilical cord veins by 0.1% collagenase (type I) perfusion (10 minutes, 37°C) and cultured in medium E 199 containing 27 mmol/L NaHCO3, 2 mmol/L glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin, supplemented with 20% NCS. Cells were identified as endothelial cells by their typical cobblestone morphology and by immunostaining for von Willebrand factor. Primary cultures were plated in 25-cm2 plastic flasks, washed, and refed with the same medium every day. After 4 to 6 days, the cultures were at confluence and cells were detached from the flasks by incubation for 1 minute at 22°C with 2 mL of 0.05% trypsin in 0.02% EDTA. Cells were washed once with medium E199 containing 20% NCS. Cells (1×105 cells per well) were plated on glass coverslips or on 12-well plates coated with 1.5% (wt/vol) gelatin and grown for 3 to 4 days to obtain a confluent monolayer.
Confluent cultures grown on 12-well plates (2.5×105 cells per well, 22 mm diameter) or on glass coverslips were washed twice with a serum-free medium and subsequently incubated with a serum-free medium (0.5 mL per well) with or without the addition of different concentrations of cathepsin G (0.04, 0.2, and 0.4 μmol/L) or thrombin (0.06 μmol/L) for different time periods. The same area of each monolayer was examined under the microscope at time intervals of 15, 30, and 45 minutes and 1, 2, 3, 4, and 6 hours.
Phase-Contrast Microscopy and Morphometric Studies
Cells were washed twice with serum-containing medium and fixed with 4% formaldehyde in PBS (pH 7.6) for 10 minutes at room temperature. Changes in cell shape were studied under a phase-contrast Zeiss microscope (objective ×10, ocular ×10). For each concentration at all time periods, deendothelialization indexes were evaluated as percentages of the entire monolayer.
Transmission Electron Microscopy
Specimens were fixed in 2.5% buffered glutaraldehyde, postfixed in 1% osmium tetroxide for 1 hour, dehydrated, and embedded in araldite. Sections were cut and prestained with uranyl acetate and Reynolds citrate. Specimens were examined under a transmission electron microscope (Zeiss 109).
Coverslip-attached cells were fixed in 4% formaldehyde in PBS (pH 7.6) for 10 minutes at room temperature. After thorough rinsing in PBS, cells were made permeable to large molecules by soaking coverslips for 3 minutes at 0°C in 1% triton X-100 in PBS. Cells that were fixed and made permeable were stained with 2 μg/mL rhodamine-phalloidin for 30 minutes at 37°C and mounted in 50% PBS glycerol. Observations were carried out under a Zeiss Axiophot microscope equipped for fluorescence microscopy.
Evaluation of Cell Damage
Cell viability was assessed by trypan blue exclusion test and cell damage was tested by measuring the amount of LDH released in the culture medium.
Effect of Cytochalasin B and Phallacidin on Cathepsin G Activity
HUVEC confluent monolayers were preincubated with cytochalasin B (0.1, 0.25, and 1.25 μg/mL) for 1 hour at 37°C or with phallacidin (0.3 μmol/L) for 3 hours at 37°C, then cathepsin G (0.2 μmol/L) was added. Cells were washed and fixed at different times, as described above for microscopic examination.
Evaluation of Endothelial Albumin Permeability
To evaluate endothelial albumin permeability, 0.2 mL of a suspension of endothelial cells (106 cells per mL) was placed on 0.4-μm pore-size polycarbonate filters (2×105 cells per filter, final concentration), glued to polycarbonate cylinders. The filters were coated with 1.5% (wt/vol) gelatin in PBS, pH 7.4. Cylinders were placed in 24-well tissue culture plates containing 0.8 mL E199 medium with 20% NCS and incubated at 37°C in 95% air and 5% CO2 until confluence.
Filters were washed twice with serum-free medium, placed in 24-well plates containing 0.7 mL medium 199 only, and filled with 0.15 mL medium 199 containing 10 mg/mL bovine serum albumin. These volumes assured equal fluid height on either side of the filter, resulting in no hydraulic pressure across the monolayer. The filters were incubated for 1 hour at 37°C in 95% air and 5% CO2 in culture conditions described above. The medium above the monolayer was then carefully aspirated and the filters were replaced in the culture plates containing 20% NCS medium 199.
Albumin concentration above and below the filters was measured colorimetrically by Bradford assay.9 Results for each monolayer were expressed as an albumin ratio: Only filters with an albumin ratio ≤0.02 were used for further experiments.
Filters were washed again with serum-free medium 199 and incubated with cathepsin G (0.2 μmol/L) for 1 hour at 37°C, with or without preincubation with phallacidin (0.3 μmol/L) for 3 hours at 37°C. Albumin flux was evaluated as described above.
Effect of PAI-1 and Vitronectin Inhibition on HUVEC Morphology
HUVEC preconfluent monolayers were incubated with either rabbit anti–human PAI-1 (10 μg/mL in serum-free medium), antiserum to human vitronectin (diluted 1:100 in serum-free medium), or both for 5 hours at 37°C. Cells were washed and fixed for morphological evaluations.
Extracellular Matrix Preparation and PAI-1 Antigen Evaluation
After cell extraction with 0.25% triton X-100, the remaining extracellular matrix was washed three times with distilled water until no cell debris was detected by light microscopy in the wells. Serum-free medium (0.5 mL per well) with or without cathepsin G (0.04 μmol/L) was added at various time intervals. After incubation, supernatants were collected on ice, centrifuged at 2000 rpm in an Eppendorf centrifuge for 10 minutes at 4°C, divided into aliquots, and stored at −80°C until assay. The subendothelial matrix was then washed three times with distilled water, collected with 150 μL per well of PBS containing 0.1% SDS, and stored at −80°C until use.
PAI-1 antigen in samples was measured by means of a double monoclonal antibody ELISA (Imulyse PAI-1, Biopool). SDS-lysed material (HUVEC extracellular matrix), prepared as indicated above, was incubated in PBS-EDTA-Tween 20 buffer containing 1% triton X-100 to allow antigen recognition by monoclonal antibody.
In some experiments, cathepsin G (0.2 μmol/L) was added to the endothelial monolayer directly or after preincubation with phallacidin (0.3 μmol/L for 3 hours at 37°C). PAI-1 antigen levels were then evaluated in conditioned medium and extracellular matrix as described above.
Endothelial Repair Experiments
Cells were washed three times with serum-containing medium to inhibit cathepsin G activity and were incubated in the presence of 20% NCS for 24 hours. At different time intervals, HUVEC monolayers were observed for microscopy studies and PAI-1 antigen levels were evaluated in extracellular matrix and in culture medium.
Data are expressed as mean±SEM of at least three experiments, and each experiment was performed in triplicate wells. Statistical analysis was performed by use of one-way ANOVA followed by multiple-comparison tests. In some experiments, a regression analysis was performed.
Effect of Cathepsin G on HUVEC Morphology
The morphological changes induced by cathepsin G (0.04, 0.2, or 0.4 μmol/L) were studied at different time intervals, ranging from 15 minutes to 6 hours. The same area of the monolayer was focused by phase-contrast microscopy. Nontreated confluent monolayers showed the typical cobblestone appearance; some giant cells were also observed (Fig 1A⇓). After treatment with cathepsin G (0.2 μmol/L), endothelial cells looked contracted and started to lose contact with their neighbors. Little gaps were formed such that the cells tended to become star-shaped until totally detached (Fig 1B⇓ through 1F).9 12 19 20 These effects depended on concentration and length of exposure to cathepsin G (Fig 2⇓). At a concentration of 0.04 μmol/L, cathepsin G did not induce significant morphological changes during the first hour of incubation; in the subsequent period, cells started to contract, with formation of little gaps among them. In either case, cell detachment was never observed at this concentration (data not shown). Giant cells were much more resistant to cathepsin G treatment; when all the other cells were practically detached, they looked only contracted (Fig 1F⇓).
With use of transmission electron microscopy, a disruption of the cell-cell contact areas could be observed, whereas the gap-junction regions appeared more resistant to treatment. No significant changes in cellular ultrastructure were detected after treatment with cathepsin G (0.04 to 0.4 μmol/L, from 15 minutes to 6 hours, data not shown).
We quantified the effect of cathepsin G on HUVEC morphology by calculating the deendothelialization index (Fig 2⇑). Regression analysis showed a significant linear relation between the deendothelialization indexes and both the concentration of cathepsin G and the incubation time. Cathepsin G at 0.04 μmol/L did not induce a significant increase in deendothelialized area, even after 6 hours of incubation (12% of the confluent monolayer area). The highest concentration of cathepsin G used detached all cells after 6 hours of incubation (Fig 2⇑).
Eglin C, an inhibitor of cathepsin G, completely prevented the morphological changes induced by cathepsin G in endothelial cells (Fig 3A⇓ and 3B⇓) and showed no effect on cell morphology by itself (Fig 3C⇓).
Even at concentrations of 0.2 or 0.4 μmol/L, cathepsin G did not alter cell viability as assessed by trypan blue exclusion test, even after the longest incubation time when practically all of the cells were detached (93% viable cells both in control and in cultures treated for 6 hours). Moreover, the levels of LDH, a sensitive biochemical marker of cellular integrity, were not significantly increased by treatment with cathepsin G for 6 hours (4.9±1.5 U/L in control and 4.9±1.5, 4.4±1.7, and 10.5±3.7 U/L after 0.04, 0.2, and 0.4 μmol/L cathepsin G, respectively; P=NS, n=4).
Effect of Cathepsin G on F-Actin Cytoskeleton Arrangement
Treatment with cathepsin G caused rearrangement of cytoskeleton, which indeed accompanied the endothelial cell–shape changes described above.
In the control monolayer, when the endothelial cells showed typical cobblestone appearance (Fig 4A⇓), F-actin microfilaments were organized in central bundles, whereas at the periphery of the cells, they formed microfilament bundles lining the cellular border (the so-called DPB) (Fig 4B⇓). When the cells were confluent (Fig 4C⇓), cathepsin G (0.04 μmol/L for up to 1 hour, 0.2 μmol/L for up to 30 minutes, and 0.4 μmol/L for up to 15 minutes) induced loss of the DPB, and an elaborate array of microfilament bundles of the stress-fiber type (Fig 4D⇓) appeared that was located mainly under the luminal plasmalemma, as indicated by transmission electron microscopy (data not shown).
Under treatment conditions that induced marked changes in cell shape, endothelial cell retraction was observed with a tendency to loose reciprocal contacts (Fig 4E⇑). Stress fibers partially disappeared or became less regular, and F-actin microfilaments appeared diffused in the cytoplasm (Fig 4F⇑). At later stages, when the endothelial cells formed a network and were in contact mainly through their lamellipodia, disappearance of the stress fibers was almost complete; we could only observe a few actin fibers arranged in the thin lamellipodia of surviving cells. When the cells were detaching (Fig 4G⇑), F-actin was condensed, making it impossible to see the microfilament arrangement (Fig 4H⇑).
We compared the effect of cathepsin G with that of thrombin. Thrombin (0.06 μmol/L), like cathepsin G (0.2 μmol/L), affected HUVEC morphology, inducing a time-dependent increase in the deendothelialization area. After 2 hours of treatment, cells appeared contracted and had lost their reciprocal contacts (Fig 5A⇓). Similarly, the cytoskeleton rearrangement observed after treatment with thrombin did not differ from that induced by cathepsin G: DPB disappeared and stress fibers could be observed (Fig 5B⇓). Moreover, thrombin was effective at lower concentrations than cathepsin G.
To prove the relevance of F-actin rearrangement in cell-shape changes induced by cathepsin G, we first stabilized F-actin microfilaments by using phallacidin and then destroyed them with cytochalasin B.
Preincubation of confluent HUVEC monolayer with phallacidin (0.3 μmol/L, 3 hours), an inhibitor of F-actin depolymerization, did not induce changes in cell shape or cell viability (Fig 6A⇓). However, it completely prevented the effect of cathepsin G (0.2 μmol/L, 30 minutes) on cell morphology and F-actin arrangement (Fig 6B⇓ and 6C⇓). Moreover, at a dose not affecting cell confluence, cytochalasin B, an inhibitor of actin polymerization, potentiated the effect of cathepsin G (0.2 μmol/L, 30 minutes) (Fig 7⇓).
Repair of Endothelial Damage Induced by Cathepsin G
The effect of cathepsin G on endothelial cell morphology was reversible. By removing cathepsin G and adding a serum-containing medium, we observed spreading of attached cells. Nevertheless, only when the deendothelialization index was less than 50% to 55% was it possible to observe full restoration of the endothelial monolayer, which appeared confluent after about 12 hours.
After inhibition of cathepsin G activity (0.2 μmol/L, 1 hour), while endothelial cells spread and started to restore their confluence (Fig 8A⇓ and 8C⇓), numerous front ruffles were observed (Fig 8B⇓), and stress-fiber arrangement of actin filaments reappeared (Fig 8D⇓). After full restoration of HUVEC confluence (Fig 8E⇓), the actin cytoskeleton looked identical to that observed in control, with F-actin filaments arranged in DPB and in central, short microfilaments (Fig 8F⇓). Thrombin-induced damage was also reversible after addition of serum to endothelial cells (data not shown).
Effect of Cathepsin G on PAI-1 Levels in Extracellular Matrix
Cathepsin G (0.04 μmol/L) induced a time-dependent decrease in PAI-1 levels in the extracellular matrix of a cell-free system, simultaneously increasing these levels in the supernatant (Fig 9a⇓) as compared with time zero and with control (Fig 9b⇓). The effect reached a plateau after 2 hours of incubation, when about 90% of PAI-1 was detached from the matrix. After treatment of the HUVEC monolayer with cathepsin G, we also observed a decrease in PAI-1 content of extracellular matrix without reaching a balance between loss of PAI-1 in extracellular matrix and its yield in conditioned medium. Indeed, during the procedure of cell extraction, extracellular matrix can be lost, with a consequent underestimation of its PAI-1 content before cathepsin G treatment. In both conditions, cell-free system and HUVEC monolayer, coincubation with eglin C totally prevented this effect of cathepsin G (data not shown).
To verify whether PAI-1 depletion of extracellular matrix could account for the damaging effect of cathepsin G on HUVEC, we incubated confluent monolayers with polyclonal antibodies against human PAI-1 (0.04 μmol/L for 5 hours). Morphological changes similar to those induced by cathepsin G were observed: endothelial cells were contracted, the contact between cells was lost, and intercellular gaps appeared (Fig 10A⇓). Only minor changes in cell shape were observed after incubation with an anti-vitronectin serum (Fig 10B⇓). However, the coincubation of both antibodies induced the same effects as the anti–PAI-1 antibody alone (Fig 10C⇓).
To verify whether the effect of cathepsin G on cell morphology depended primarily on PAI-1 depletion of extracellular matrix, we inhibited the effect of cathepsin G on F-actin cytoskeleton using phallacidin, and we evaluated PAI-1 antigen levels in the conditioned medium and in the subendothelial matrix after cathepsin G treatment (0.2 μmol/L, 1 hour). Preincubation of confluent endothelial cells with phallacidin (0.3 μmol/L, 3 hours) did not modify the effect of cathepsin G on PAI-1 levels, either in cultured medium or in subendothelial matrix (Table⇓).
To explain how cathepsin G could act on subendothelial matrix PAI-1 in the absence of cytoskeletal changes, we assessed transcellular permeability to albumin. Preincubation of confluent endothelial cells, grown on gelatin-coated polycarbonate filters, with phallacidin (0.3 μmol/L, 3 hours) did not modify the increase in albumin flux across the endothelial monolayer induced by cathepsin G (0.2 μmol/L, 1 hour) (Fig 11⇓).
Restoration of Matrix-Bound PAI-1 and Endothelial Damage Repair
To verify whether restoration of matrix-bound PAI-1 was associated with recovery from the morphological changes induced by cathepsin G, we inhibited the activity of cathepsin G on HUVEC monolayer after 2 hours of incubation by washing cells with serum-containing medium. Then, cells were incubated in the presence of serum, and PAI-1 antigen levels were measured in the culture medium and in the extracellular matrix at different time intervals, ranging from 6 to 24 hours.
As expected, in control cells, a time-dependent increase in PAI-1 antigen levels was observed both in culture medium and in extracellular matrix (Fig 12a⇓ and 12b⇓). In cathepsin G–treated cells, PAI-1 levels increased in the culture medium to the same extent as in the control, independent of the concentration and the duration of treatment with cathepsin G. In the extracellular matrix, however, the increase in PAI-1 levels was significantly lower than in the control and was observed only in the first hour of incubation. The degree of this effect was dependent on the concentration of cathepsin G used in the first part of the experiment (Fig 12b⇓).
The same results were observed after restoration of normal culture conditions after 6 hours of incubation with cathepsin G (data not shown).
Cathepsin G, a protease released by human polymorphonuclear cells when activated, affects the morphological integrity of endothelial monolayers. Depending on the dose and the duration of exposure to cathepsin G, endothelial cells retract and lose contact, forming cellular gaps until complete detachment9 12 19 20 ; as a consequence, progressively larger areas of matrix are uncovered.
Eglin C, a potent granulocytic proteinase inhibitor,21 prevented the damage caused by cathepsin G, suggesting that the proteolytic active site is essential for the enzyme activity on endothelial cell morphology; moreover, this activity seems specifically directed to some cellular components and does not seem to be due to a nonspecific cytotoxic effect of the enzyme.
Adhesion and spreading on the subendothelial matrix and, ultimately, the shape of endothelial cells are regulated by complex structures including the cytoskeleton microfilament system, components of extracellular matrix, and transmembrane receptors.13 Cell cytoskeleton has been shown to be critical to maintaining the shape integrity of endothelial cells exposed to blood flow.22
Simultaneously with the changes in cellular shape, cathepsin G induced a progressive rearrangement of the F-actin cytoskeleton until its complete disaggregation at the higher concentrations used. The early stages of this reorganization were very similar to those observed after treatment with thrombin23 or cytokines24 or in conditions of hyperoxia25 and were not associated with changes in cellular shape.
Stabilization of F-actin microfilaments by phallacidin26 completely prevented the effect of cathepsin G on cell morphology. In contrast, cytochalasin B, at concentrations inducing only small changes in microfilament distribution and not affecting cell confluence,18 27 potentiated the effect of cathepsin G. These data support the hypothesis that cathepsin G alters endothelial cell integrity by affecting F-actin microfilament arrangement, which is important in maintaining both the shape and the adhesion of endothelial cells to their substrate.28 29 When this structure is destroyed, cell adhesion to the substrate is impaired and cells detach. These late-occurring phenomena expose large areas of subendothelial matrix, a highly thrombogenic surface on which platelets can promptly adhere, localizing the complex sequence of events that leads to thrombus formation.30 Activated platelets, in turn, expose P-selectin, which may favor further PMN accumulation at the injured site,31 32 thus amplifying the endothelial damage.
Interestingly, the described effects on cell morphology and cytoskeleton arrangement are similar for cathepsin G and thrombin at comparable concentrations for enzymatic activity. The peculiarity of cathepsin G activity results from its localization at sites of close contact between activated neutrophils and the endothelial layer, where a microenvironment could be formed protecting the proteolytic enzyme from inactivation by plasma protease inhibitors.7
In addition to cathepsin G, other neutrophil proteases, such as elastase, might induce morphological changes to endothelial monolayers20 ; therefore, the potential in vivo damage of PMN to endothelial cells might result from interaction among different neutrophil mediators and could replace or be potentiated by thrombin generation on the thrombus surface.
Neutrophil proteases are also enzymatically active on subendothelial matrix,2 12 33 which plays an important role as an active substrate in regulating cell shape, migration, and growth.16 29 34 Elastase degrades matrix fibronectin,2 19 inducing contraction and detachment of endothelial cells. We have recently demonstrated that cathepsin G released PAI-1 from the extracellular matrix to the culture medium.12 PAI-1, known as the major inhibitor of fibrinolysis activators, is constitutively secreted by endothelial cells and stored in the extracellular matrix, bound to vitronectin.35 36 Vitronectin/PAI-1 complexes, however, act not only as antifibrinolytic factors but also as spreading factors for several types of cells.15 35 In the present study, anti-vitronectin antibodies did not affect endothelial cell morphology. In contrast, the inhibition of matrix-bound PAI-1 activity by polyclonal antibodies impaired cell monolayer integrity similarly to cathepsin G, indicating that the detachment of PAI-1 from extracellular matrix after treatment with cathepsin G can account, at least in part, for the damaging activity of cathepsin G on endothelial cells. In endothelial cells, loss of contacts with matrix PAI-1 induced F-actin cytoskeleton rearrangement and subsequent shape changes. The changes in cell morphology and cytoskeleton reported in the present study may also be related to degradation of other extracellular matrix proteins, such as fibronectin, which is often seen in association with actin stress fibers.13 34 37
The effect of cathepsin G on cell morphology could also be explained by a direct effect of the enzyme on cells, possibly via a specific receptor. Rounding and detachment of cells would then allow access to the matrix. However, the inhibition of cell-shape changes by phallacidin did not prevent the effect of cathepsin G on matrix-bound PAI-1, suggesting that the latter is a primary event for cathepsin G activity on cell morphology; cell morphology changes, in turn, occur through cathepsin G–induced cytoskeleton rearrangement.
Phallacidin did not affect the increase in transcellular permeability induced by cathepsin G. These results could explain how cathepsin G reaches the subendothelial matrix in the absence of cell-shape changes, since endothelial cells form an impermeable barrier under normal conditions. We suppose that by acting as a proteolytic enzyme for intercellular proteins, cathepsin G reaches the extracellular matrix by passing through the intercellular junctions. A concomitant direct effect on endothelial cells, however, may not be excluded. Peterson9 showed that at high concentrations, cathepsin G increased calcium release in pig endothelial cells. The use of Ca2+ antagonists, however, which partially prevented the increase in albumin flux, did not modify the changes in cell shape and the increase in intercellular gaps induced by cathepsin G.9 Therefore, the effect on calcium homeostasis could be attributed to the changes in cell shape induced by cathepsin G activity on extracellular or intercellular proteins. Indeed, Weksler et al19 did not find any mobilization of calcium after addition of cathepsin G to HUVEC suspensions.
The effects of cathepsin G on cell morphology and on F-actin microfilament arrangement were partially reversible. When the damage was very severe, such as after high concentrations and long exposures to cathepsin G activity, the changes appeared to be irreversible, probably because the surface to which cells needed to adhere was destroyed.
Reversibility of endothelial shape and cytoskeleton changes was also reported after damage by oxygen radicals24 or exposure to cytokines,25 and endothelial cell viability appeared involved in these phenomena.
Under our conditions, we cannot exclude the possibility that reversibility of the damage could be ascribed to factors such as PAI-1 added to the cells with the culture medium or secreted by the same cells during migration or proliferation processes.17
The restoration of PAI-1/vitronectin complexes in the subendothelial matrix after cathepsin G injury could favor the respreading of contracted endothelial cells and the repair of cathepsin G–induced damage.
Endothelial cells, even contracted, still are able to synthesize and release PAI-1 independent of the activity of cathepsin G; however, PAI-1 can be bound to the matrix only when cathepsin G is used at lower concentrations. Higher concentrations of the enzyme not only release PAI-1 from the matrix but also affect its ability to bind the inhibitor, probably by a direct proteolytic action on matrix components, such as vitronectin. Thus, the ability of the subendothelial matrix to bind PAI-1 is associated with repair of the cellular damage induced by cathepsin G.
In conclusion, the mechanism by which cathepsin G alters the morphological and functional integrity of human endothelial monolayer relies primarily on matrix-bound PAI-1, which induces F-actin cytoskeleton rearrangement and consequent changes in cell shape.
These findings add new evidence for the thrombogenic role of neutrophils. Cathepsin G, when released in vivo in a microenvironment protected from antiprotease inhibition,7 31 such as in areas of close cell-cell contact, may alter endothelial morphology by chronically exposing large areas of subendothelial matrix, a highly thrombogenic surface on which platelets may adhere, become activated, attract further neutrophils, and trigger thrombus formation.29
This may represent an amplification loop to the clotting cascade. On the other hand, thrombin generated during this process may contribute to endothelial damage, amplifying and prolonging the prothrombotic condition. Finally, the active PAI-1 released from the subendothelial matrix by neutrophil cathepsin G may further contribute to thrombus growth through inhibition of the fibrinolytic process.
Selected Abbreviations and Acronyms
|DPB||=||dense peripheral band|
|HUVEC||=||human umbilical vein endothelial cells|
|PAI-1||=||plasminogen activator inhibitor-1|
This work was partially supported by the Italian National Research Council (Convenzione-CNR-Consorzio Mario Negri Sud and Progetto Finalizzato FATMA, contract No. 94.00694.41). Dr Pintucci was a recipient of a fellowship from the Regional Council of Abruzzo and the Commission of European Communities (CEE/Abruzzo). We wish to thank Dr Chiara Cerletti and Dr Alexander Mironov for helpful discussions and Raffaella Bertazzi, Rosanna Tucci, and the G. A. Pfeiffer Memorial Library staff for valuable assistance in preparation of the manuscript.
Henson PM, Jonston JR. Tissue injury in inflammation: oxidants, proteinases and cationic proteins. J Clin Invest. 1987;79:669-674.
Campbell EJ. Human leukocyte elastase, cathepsin G and lactoferrin: family of neutrophil granule glycoproteins that bind to an alveolar macrophage receptor. Proc Natl Acad Sci U S A. 1982;79:6941-6945.
Evangelista V, Rajtar G, de Gaetano G, White JG, Cerletti C. Platelet activation by fMLP-stimulated polymorphonuclear leukocytes: the activity of cathepsin G is not prevented by antiproteinases. Blood. 1991;77:2379-2388.
Harlan JM, Killen PD, Harker LA, Striker GE, Wright DG. Neutrophil-mediated endothelial injury in vitro: mechanisms of cell detachment. J Clin Invest. 1981;68:1394-1403.
Totani L, Piccoli A, Pellegrini G, Di Santo A, Lorenzet R. Polymorphonuclear leukocytes enhance release of growth factors by cultured endothelial cells. Arterioscler Thromb. 1994;14:125-132.
Gotlieb AI, Spector W, Wong MKK, Lacey C. In vitro reendothelialization: microfilament bundle reorganization in migrating porcine endothelial cells. Arteriosclerosis. 1984;4:91-96.
Ciambrone GJ, McKeown-Longo PJ. Plasminogen activator inhibitor type I stabilizes vitronectin-dependent adhesions in HT-1080 cells. J Cell Physiol. 1990;11:2183-2195.
Weksler BB, Jaffe EA, Brower MS, Cole OF. Human leukocyte cathepsin G and elastase specifically suppress thrombin-induced prostacyclin production in human endothelial cells. Blood. 1989;74:1627-1634.
Seemuller U, Fritz H, Eulitz M. Eglin C: elastase-cathepsin G inhibitor from leeches. In: Lorand L, ed. Methods in Enzymology. London: Academic Press Inc; 1981:804.
Phillips PG, Higgins PJ, Malik AB, Tsan MF. Effect of hyperoxia on the cytoarchitecture of cultured endothelial cells. Am J Pathol. 1988;32:59-72.
Phillips PG, Lum H, Malik AB, Tsan MF. Phallacidin prevents thrombin-induced increases in endothelial permeability to albumin. Cell Physiol. 1989;26:C562-C570.
Evangelista V, Piccardoni P, White JG, de Gaetano G, Cerletti C. Cathepsin G-dependent platelet stimulation by activated polymorphonuclear leukocytes and its inhibition by antiproteases: role of P-selectin–mediated cell-cell adhesion. Blood. 1993;81:2947-2957.
Weiss S, Regiani S. Neutrophils degrade subendothelial matrices in the presence of alpha-I-proteinase inhibitor: cooperative use of lysosomal proteinases and oxygen metabolites. J Clin Invest. 1984;73:1297-1303.
Knudsen BS, Harpel PC, Nachman RL. Plasminogen activator inhibitor is associated with the extracellular matrix of cultured bovine smooth muscle cells. J Clin Invest. 1987;80:1082-1089.
Mimuro J, Loskutoff DL. Purification of a protein from bovine plasma that binds to type 1 plasminogen activator inhibitor and prevents its interaction with extracellular matrix: evidence that the protein is vitronectin. J Biol Chem. 1989;264:936-943.