Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Respiratory distress syndrome (RDS) is the most common neonatal respiratory disorder and is a major cause of mortality and morbidity. Pathophysiologically, RDS is characterized by increased permeability, pulmonary edema, and fibrin deposition within the lung's intravascular, interstitial, and intra-alveolar spaces (1). Studies suggest that fibrin deposition may contribute to the severity of RDS as well as the development of the chronic lung disease, bronchopulmonary dysplasia (BPD) (4). Fibrin monomer, produced by thrombin proteolysis of fibrinogen, leads to impaired surfactant function (7), increased vascular permeability, and further fibrin formation. The mechanisms controlling intra-alveolar fibrin formation are poorly understood.

Previous studies have demonstrated that rat fetal distal lung epithelium (FDLE) grown in primary culture possesses activities that regulate thrombin generation and function (8). Thrombin, a key enzyme in the coagulation cascade, cleaves fibrinogen to fibrin. Factor (F) Xa, in combination with F Va, Ca²⁺, and phospholipids, forms the prothrombinase complex that activates prothrombin to thrombin (9). Activation of F X occurs in vivo, either by direct reaction with F VIIa bound to tissue factor (TF) (10) or by other F VIIa/TF-mediated feedback loops (11). FDLE secrete three glycosaminoglycans (GAGs) with antithrombin activity: heparan sulfate (HS), chondroitin sulfate (CS), and dermatan sulfate (DS) (8). HS forms a complex with the physiological inhibitor antithrombin (AT) and augments AT's activity against several serine proteases, including F Xa and thrombin (14). DS inhibits thrombin by binding to the natural inhibitor heparin cofactor II (HCII), which specifically inhibits thrombin.

Our studies have shown that FDLE also express F VII-dependent TF activity (15). Thus, although FDLE produced anticoagulant GAGs, procoagulant TF was detected in the extracellular fluid and the FDLE surface was found to promote thrombin generation in plasma (15). A recent study has shown that the use of mechanical stretch to simulate fetal breathing movements (FBMs) increased GAG production by rat fetal mixed lung cells (FMLCs) (16). During late gestation, a normal human fetus spends approximately 30% of its time making breathing movements (17). An infant born prematurely would not experience late-gestation FBMs and might have impaired GAG production with enhanced fibrin deposition in the lung. Although simulated FBM has been shown to give rise to increased GAG mass by FMLCs, total antithrombin activity from cellular-produced GAGs varies, as does the relative activity from different GAG types (8). In addition, the effect of FBMs on TF production is unknown. Therefore, we investigated the effects of simulated respiratory FBMs on both anticoagulant GAG release and TF activity and expression by rat FMLCs.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Lung Organotypic Cultures

Isolation of FMLCs and their use in organotypic cultures have previously been described in detail (18). In brief, lung cells were isolated from fetal Wistar rats at 19 d gestation (term being 22 d). Cell suspensions were applied to the surfaces of 2 × 2 × 0.25 cm³ gelatin sponges (Gelfoam; Upjohn, Kalamazoo, MI) and incubated for 1 h at 37°C in room air containing 5% CO₂. Three milliliters of Eagle's minimal essential medium (MEM) (GIBCO, Burlington, ON, Canada) containing 10% (vol/vol) fetal bovine serum (FBS) (GIBCO) were added and the sponges were then allowed to incubate overnight, following which nonadherent cells were removed by washing three times in serum-free MEM. The medium was then substituted with 10% (vol/ vol) FBS in MEM.

Stretch Experiments

The procedure for mechanically stretching gelatin sponges containing lung organotypic cultures has previously been described in detail (19). In brief, a device consisting of a burst timer, a control unit, a dual regulated direct current power supply, and solenoids was used. A culture dish containing a sponge was placed in front of each solenoid. One end of the sponge was fixed to the culture dish and the other end was attached to a metal bar that was sealed in plastic tubing. Electrical current through the solenoid produced a magnetic field, which drew the metal bar towards it and stretched the sponge. When the current was shut off, the elasticity of the sponge caused the sponge to return to its original length. For all studies the frequency was set at 1 Hz, the duration was 15 min/h, and the degree of stretch was 5% elongation. The duration and degree of stretch used were similar to those reported for normal FBMs in vivo (20) and have been shown previously to be the values required to elicit a mitogenic response (19). All samples were stretched within a humidified incubator containing room air and 5% CO₂. Control cultures from the same cell harvest were subjected to the same magnetic field; however, the sponges had no attached metal bar and were not stretched. Samples of conditioned medium (CM) were collected from the cultures after 48 h. The CM were concentrated by pressure dialysis under N₂ against 0.15 M NaCl to one-thirtieth of their original volume.

GAG Antithrombin Activity Assay

A previously described antithrombin activity assay was used to measure the GAG content in various experiments (8). To determine the antithrombin activity of the concentrated CM, 40 µl of each sample was diluted to 80 µl with 0.15 M NaCl. The samples were then heated at 37°C for 1 min followed by the addition of 10 µl of defibrinated plasma to provide a source of AT and HCII. Defibrinated plasma was prepared by treating 500 µl of adult human plasma with 15 µl of Arvin (6 U/ml) (Connaught Laboratories, Toronto, Canada) and winding out the clot. After the plasma was heated for a further minute at 37°C, 10 µl of 15 U bovine thrombin (Parke-Davis, Scarborough, ON, Canada) in 0.15 M NaCl was added with mixing. Following a 1-min incubation, 100 µl of preheated (37°C) adult plasma was added to the sample and the time to the first appearance of a clot on the end of a wire loop used for agitation was recorded.

To determine the antithrombin activity of the different GAG types, some samples were treated with GAG-degrading enzymes prior to assay. Stock solutions of the following enzymes (ICN Immunobiologicals, Lisle, IL) were prepared: chondroitinase AC (ACase), 5 U/ml of 0.01 M sodium acetate, and 0.15 M NaCl; chondroitinase ABC (ABCase), 5 U/ml of 0.01 M sodium acetate, and 0.15 M NaCl; and heparitinase (heptase), 0.4 U/ml of 0.01 M CaCl₂, 0.01 M sodium acetate, and 0.15 M NaCl. Assay samples were treated by reacting 40 µl of the concentrated medium with 10 µl of the appropriate enzyme plus 0.15 M NaCl to make a total volume of 80 µl. The enzyme-containing medium samples were allowed to incubate for 5 h at 37°C, followed by the addition of defibrinated plasma as described previously. Any decrease in antithrombin activity was considered proportional to the activity of the inactivated GAG. Clot times attributed to CS were determined by subtracting the ACase-treated medium values from the untreated medium values of the same trial group. Clot times attributed to DS were evaluated by subtracting the ABCase-treated medium clot times from the ACase-treated medium clot times of the same trial group. Clot times attributed to HS were determined by subtracting the ABCase plus heptase-treated medium clot times from the ABCase-treated medium clot times of the same trial group. Inclusion of GAG-degrading enzymes in antithrombin assays with no GAG present (0.15 M NaCl control) had no effect.

Polyacrylamide Gel Electrophoresis Proteoglycan Analysis

Stretch-conditioned medium (S-CM) and control-conditioned medium (C-CM) samples were analyzed by gradient polyacrylamide electrophoresis (PAGE) to quantitate the proteoglycan (PG) content. Diethylaminoethyl (DEAE) Sepharose beads (Pharmacia, Uppsala, Sweden) were first used to separate PGs and any other negatively charged material from the concentrated CM. Forty microliters of a 50% vol/vol suspension of DEAE Sepharose equilibrated with 0.01 M Tris HCl, pH 8.0, was taken, and the wet packed beads were gently mixed for 10 min with 9 µl of the concentrated CM in 31 µl of 0.01 M Tris HCl, pH 8.0. After centrifugation at 2,000 × g, the resultant supernatant was discarded and the beads were washed twice with 200 µl of 0.25 M NaCl in 0.01 M Tris HCl, pH 8.0, followed by two elutions with 200 µl of 1.0 M NaCl in 0.01 M Tris HCl, pH 8.0. The elutions were dialyzed against 0.01 M NaCl in tubing with a 12,000- to 14,000-D cutoff. After dialysis, the isolated material was concentrated by freeze drying followed by resuspension in 40 µl H₂O. Ten-microliter subsamples were digested with either ACase (0.45 U/ml final concentration), ABCase (0.45 U/ml final concentration), or ABCase plus heptase (0.009 U/ml final concentration). All enzyme treatments were for 3 h at 37°C. The samples were analyzed on a 7.5%-2% gradient sodium dodecyl sulfate (SDS)-PAGE with a 3% stacking gel. The gels were stained with alcian blue 8GX followed by silver staining (21). To determine the relative amount of stained PG and GAG material in each lane, the gels were dried on clear membranes and densitometry performed on an LKB Bromma-enhanced laser densitometer (Pharmacia, Uppsala, Sweden). The amount of PG was recorded as the area of the traced peak under the densitometry curve minus the baseline determined using an empty (control) lane. The percent increase of total GAG was determined by subtracting the densitometry value for the amount of stained material in the stretched samples from the control sample values. CS levels were determined by subtracting the value for the material remaining in the ACase-treated samples from the material in the nonenzymatic treated sample. HS levels were determined by subtracting the value for the material remaining in the ABCase + heptase samples from the material in the ABCase samples.

Procoagulant Activity

The S-CM and C-CM samples were tested for their ability to accelerate thrombin generation using an assay previously described by our laboratory (8). In brief, 8.3 µl of the concentrated CM was diluted with 41.7 µl of 0.15 M NaCl. The diluted CM was placed in a 5 mm × 5 cm borosilicate glass tube, followed by the addition of 50 µl of 0.04 M CaCl₂, 0.036 M sodium acetate, 0.036 M sodium barbital, and 0.145 M NaCl, pH 7.40. After a 1-min incubation period at 37°C, 100 µl of prewarmed adult human plasma was added to the sample. The time to the first appearance of a clot on the end of a wire loop was recorded. Differences in clotting time were assumed to be due to procoagulant material present in the various samples. Pretreatment of CM samples with ABCase + heptase had no effect on this assay. Clot times were converted to theoretical TF concentrations by comparison with a standard curve constructed using phospholipids and varying amounts of recombinant TF (a kind gift from Dr. George Vlasuk, Corvas International Inc., San Diego, CA) in place of CM diluted with 0.15 M NaCl. Analyses of cell-associated procoagulant activity were carried out by placing a measured mass of FMLC containing a sponge from stretch or control experiments (stored at 20°C until use) into a glass tube containing 50 µl of 0.15 M NaCl, followed by CaCl₂ and plasma, to obtain clot times as described previously. The effect of CM on the generation of free thrombin activity in recalcified defibrinated adult human plasma (15) was determined using the chromogenic substrate S-2238 (Helena, Mississauga, ON, Canada). The presence of TF protein in CM was detected by SDS-PAGE, followed by Western blotting (22) using a polyclonal sheep antihuman TF antibody (Affinity Biologicals, Hamilton, ON, Canada) that cross-reacts with rat TF.

RNA Extraction and Northern Analysis

After the stretch procedure, the gelfoams were placed in 8 ml of GTC buffer (4 M guanidium isothiocyanate, 25 mM sodium citrate, and 0.5% sarkozyl) and the total RNA was extracted using the method of Chomcyznski and Sacchi (23). Fifteen micrograms of total RNA was size-fractioned on a formaldehyde-agarose gel and transferred onto a Hybond N⁺ nylon membrane (Amersham, Oakville, ON, Canada) and fixed using an ultraviolet Stratalinker 1800 (Stratagene, La Jolla, CA). The membrane was probed with a ³²P-labeled 149-bp PCR fragment cDNA probe against the rat tissue factor mRNA. Hybridization and washing conditions have been described previously (24). Autoradiography was performed at 80°C using a Kodak X-OMAT-AR film (Rochester, NY), and band intensities were quantitated using an LKB Ultrascan XL Enhanced Laser Densitometer (Pharmacia, Montreal, PQ, Canada). TF RNA expression was normalized to -tubulin expression.

Statistical Analysis

Results are reported as mean ± 1 SEM unless otherwise indicated. Comparisons between control and stretched groups on different parameters were assessed by paired Student's t test, with P 0.05 determined to be significant. When multiple comparisons were performed, a Bonferroni correction was used.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Influence of Mechanical Stretch on Rat FMLC Antithrombin Activity

Antithrombin activities were calculated as the increase in clot time above that of a saline control, as there is no standard reference GAG for assay of different unknown GAG mixtures. The antithrombin activity produced by S-CM and C-CM were compared using paired Student's t test. S-CM had substantial antithrombin activity leading to prolonged clotting times compared with C-CM (P < 0.01; Table 1).

Increase of Antithrombin Activity in CM Was Due to Chondroitin Sulfate

Because stretch-induced antithrombin activity in CM and previous studies have shown that stretching increases GAG production (16), we next determined which of the different types of GAGs were responsible for the increase in antithrombin activity. To determine the contribution of each major GAG component to the antithrombin activity, concentrated CM was subjected to enzymatic treatment with GAG-specific enzymes before assay.

Antithrombin activity due to CS in S-CM was significantly increased compared with C-CM (P < 0.01; Table 1). Although both S-CM and C-CM contained significant amounts of DS and HS antithrombin activities, there was no significant difference between S-CM and C-CM for thrombin inhibition due to these GAGs. Thus, CS was responsible for the majority of the additional antithrombin activity (Table 1).

Characterization of Different GAGs by Mass

We have demonstrated that stretch-induced antithrombin activity in CM was mainly due to CS. Because GAG antithrombin activity does not necessarily correlate with GAG mass (8), we characterized the PGs in concentrated CM according to the mass of GAG. Concentrated samples were partially purified and analyzed using SDS-PAGE. Results were calculated as the percent increase in mass of GAG in conditioned media from stretched compared with control cells because there is no standard reference GAG for staining of different unknown GAG mixtures.

Overall, the S-CM had a small increase (14 ± 6% ) of GAGs by mass compared with C-CM. As expected, the GAG material quantified by PAGE showed little correlation with the antithrombin activity (8). The activity assay demonstrated that the most significant increase in clot time prolongation due to stretching was attributed to CS (Table 1). The gel analysis demonstrated only a 26 ± 21% increase in the amount of CS but there was a 220 ± 49% increase in the amount of HS in S-CM (Figure 1).

View larger version (27K): [in this window] [in a new window]   Figure 1.   Gradient polyacrylamide gel electrophoretic analysis of glycosaminoglycans from proteoglycans in media conditioned by fetal mixed lung cell cultures. Material isolated by chromatography on DEAE Sepharose was incubated with either 0.15 M NaCl (panel a; lanes 1 and 5), chondroitinase AC (panel a; lanes 2 and 6), chondroitinase ABC (panel a; lanes 3 and 7) or chondroitinase ABC + heparitinase (panel a; lanes 4 and 8) prior to gradient gel electrophoresis (2%-7%) and sequential staining with alcian blue and silver. Molecular weights are indicated on the left margin. Samples containing conditioned media from fetal mixed lung cells that had undergone mechanical stretch for 48 h (panel a; lanes 1-4) or control cell-conditioned media (panel a; lanes 5-8) are shown. Stretch-induced increases in glycosaminoglycan mass (panel b) were calculated from densitometry scans of stained glycosaminoglycan from stretch cell-conditioned media and control cell-conditioned media (error = standard error of the mean; *P < 0.05).

We could not analyze the amount of DS by mass using enzymatic degradation. Treatment with ABCase, which contained the enzymes to degrade C4S, C6S, and DS, removed less GAG material than did treatment with ACase, which had the enzymes to degrade C4S and C6S but not DS. This result has been observed previously for rat FDLE (8) and is due to the fact that PG contained copolymer chains of CS and DS (25). When CS- and DS-degrading enzymes were present with copolymers, the DS section of the chain could be degraded by the base enzyme until a region of CS sequence was encountered, after which the DS-degrading enzyme could not react further but sterically inhibited attack by the CS-degrading enzyme.

Influence of Mechanical Stretch on Rat FMLC Procoagulant Activity

All CM samples exhibited significant procoagulant activity when tested in recalcified plasma. In both S-CM and C-CM, procoagulant activity was increased compared with a 0.15-M NaCl control (Table 2). Comparison of the CM showed that S-CM yielded significantly less procoagulant activity compared with C-CM in terms of clot time (P < 0.01). TF protein was detected in C-CM at a 2.2-fold greater concentration than in S-CM (Figure 2), in agreement with the relatively higher procoagulant activity of C-CM. Clot time values were converted to theoretical TF concentrations by performing assays containing 2 µM phosphatidylcholine and 4 µM phosphatidylserine, along with varying amounts of recombinant TF. Data from recombinant TF experiments resulted in the standard curve Y = Ae(^BX) + Ce(^DX) + E (where Y = clot time in seconds, X = recombinant [TF] in the final assay mixture in µM, A = 64.21, B = 8,136, C = 644.2, D = 5.916, and E = 704.5 [r² = 0.975]), which was used to calculate the equivalent TF concentrations in CM. The reduced TF procoagulant activity in S-CM compared with C-CM was confirmed by the measurement of free thrombin generation in recalcified defibrinated plasma, which was supplemented with conditioned media. The area under the curve for free thrombin activity over time was significantly decreased (P < 0.05; paired t test) in the presence of S-CM (1.55 ± 0.02 µM · min) compared with C-CM (1.96 ± 0.06 µM · min). Analyses of procoagulant activity from sponges that contained cellular material showed that stretched cell sponges were significantly less active than control cell sponges (P < 0.05; Table 2).

View larger version (51K): [in this window] [in a new window]   Figure 2.   Detection of tissue factor in conditioned media by Western immunoblotting. Loads were molecular weight standards (lane 1), 10 µl of Thromborel® S (Behring, Marburg, Germany) human placental thromboplastin (TF) reagent (lane 2), 9.35 µl of 20.5-fold concentrated control conditioned medium (lane 3) and 10 µl of 19.1-fold concentrated stretch-conditioned medium (lane 4). Blots were probed with a polyclonal sheep antihuman TF antibody (Affinity Biologicals, Hamilton, ON) that cross-reacts with rat TF.

To determine if the expression of procoagulant activity by cells was due to TF, we substituted F VII-deficient plasma for normal plasma in some of the procoagulant activity assays. The clot times in assays with F VII-deficient plasma were prolonged (whether CM was present or not) compared with those with normal plasma. This loss in procoagulant activity in the absence of F VII indicated that F VII-tissue factor complex was probably responsible for the generation of thrombin and subsequent shortening of the clot time.

Tissue Factor mRNA Levels in FMLC Undergoing Mechanical Stretch

The effect of stretch on TF mRNA expression was assessed as the ratio of FMLC-derived TF mRNA to -tubulin mRNA. Stretch/control comparisons of the data showed a correlation with the procoagulant activity observed in the calcium add-back assay. There was no significant difference in TF mRNA levels in stretched (0.26 ± 0.072; n = 5) compared with control (0.34 ± 0.037; n = 5) cells (P = 0.36).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Although the severity of BPD has decreased, the incidence of BPD is unchanged since the advent of surfactant therapy for neonatal RDS. Some studies suggest that fibrin deposition may contribute to the severity of RDS as well as the development of BPD (4). This concept is supported by the fact that fibrin monomer, produced by thrombin proteolysis of fibrinogen, impairs surfactant function (7), and fibrin stimulates fibroblast proliferation within regions of extravascular fibrin deposition (6). GAGs have catalytic activity toward thrombin inhibitors and a recent study has demonstrated that FBM increases GAG production by FMLCs (16). Alternatively, FDLE has previously been shown to express TF and facilitate plasma thrombin generation on its apical surface (15). GAG from conditioned media significantly decreases plasma thrombin generation on procoagulant FDLE (8). In vivo, the anticoagulant effect of conditioned media (lung fluid) would depend on GAG and TF concentrations as well as the total volume of liquid phase to alveolar surface area ratio. We hypothesized that FBM altered the antithrombin and procoagulant properties of FMLC. In this study, we showed that mechanical stretch, which simulates FBM, induced production of GAGs with antithrombin activity and downregulated TF-related procoagulant activity.

Thrombin, a key enzyme in the coagulation pathway, cleaves fibrinogen into fibrin. Generation of thrombin occurs through the activation of prothrombin by F Xa in the presence of F Va, phospholipid and Ca²⁺ (9). TF binds to F VIIa and, in the presence of Ca²⁺ and a phospholipid surface, initiates coagulation as F VIIa-TF activates both F X and F IX. TF is produced by several cell types and is present in almost all tissues. The generation of thrombin and thrombin's activities are regulated by several inhibitory systems. AT, HCII, and ₂macroglobulin (₂M) are plasma inhibitors that complex with thrombin, neutralizing its coagulant activities (26, 27). GAG antithrombin activities are modulated by these plasma inhibitors (8). HS contains a specific pentasaccharide sequence necessary to complex with AT. HS catalyzes AT activity against several serine proteases, including F Xa and thrombin (14). DS inhibits thrombin's activity by complexing with HCII, which specifically inactivates thrombin (28).

Our studies showed that organotypic lung cultures, when subjected to simulated FBM, released greater amounts of GAGs with antithrombin activity. Incubation of either stretched or control cells for 48 h led to secretion of antithrombin GAG into the extracellular media. Total GAG antithrombin activity was greater in S-CM than in C-CM. The majority of the increase in antithrombin activity in the S-CM was due to CS. Interestingly, production of GAGs based on mass did not match the type of GAGs with antithrombin activity. The mass of HS increased by 220% without any significant increase in antithrombin activity, suggesting that most HS did not contain AT binding sequences. FDLE are known to express GAGs with antithrombin catalytic activities (8). Also, it has been reported previously that FBM induces GAG production by FMLC (16), but it was not determined whether the GAGs produced under the influence of stretch had any antithrombin activity. Previous studies have shown that increased mass of GAG does not necessarily correlate with increased GAG antithrombin activities, because not all GAGs have binding sequences for antithrombin inhibitors (8). The data reported here suggest that the increased antithrombin activity is not a nonspecific consequence of increased GAG production but a highly regulated process resulting from synthesis of CS with increased antithrombin activity. In addition, the detection of native CS with significant antithrombin activity is an unusual finding, which may be the result of increased sulfation (29).

Simulated FBM also had an effect on rat FMLC procoagulant activity. Incubation of either stretched or control cells for 48 h led to significant procoagulant activity either associated with the cells or in the CM. Procoagulant activity in S-CM was lower than that in C-CM. It was likely that this activity was due to TF secreted by the cells. Thus, decreased coagulant activity in S-CM may be a result of a downregulation of TF biologic activity by the FMLC when stretched. Recently, it has been shown that systemic levels of TF in plasma from patients with acute myocardial infarction can range from 4.3-19 pM (30). Local TF concentrations at the site(s) of TF production in vivo are likely to be several times greater than the average circulating levels, which would give values that are similar to the TF concentrations calculated for CM in this study. Thus, the difference in TF concentration for S-CM and C-CM may have biological significance. Cell-associated TF may also be significant given that procoagulant activity from the cells (sponges) was higher than that in the conditioned media. Because previous work has shown that the surface of FDLE is procoagulant and promotes thrombin generation (15), FBM may provide important regulation of thrombin production by decreasing expression of TF-related activity.

It is possible that GAG production may be induced by lung stretch induced after birth. However, the rate and degree of stretch of FBMs as mimicked in our model system (1 Hz and 5% elongation) are faster and more moderate than assisted ventilation (20). Thus, any effect on GAG or TF production by breath motion simulated by standard ventilation protocols may be different. In fact, in a newborn piglet model, it has been shown that increased procoagulant activity occurs with ventilation set at a low respiratory rate and a large peak pressure (31).

In summary, we showed that mechanical stretch, which mimics FBM, induced rat FMLCs to upregulate GAGs with antithrombin activity. Furthermore, mechanical stretch decreased procoagulant activity of rat FMLCs, which may be due to decreased FMLC expression of TF. Mechanical stretch of rat FMLCs results in a significantly reduced procoagulant milieu, which may lead to an anticoagulant environment on the surface of these cells. The anticoagulant effect of stretch may explain in part that, despite the successful use of surfactant to treat neonatal RDS, there is persistence of BPD in premature infants. The persistent presence of BPD is likely multifactorial, to which dysregulation of thrombin's activities leading to fibrin deposition, with subsequent fibrosis, may be important. The latter would have important implications for the treatment of RDS as recent advances in anticoagulant therapy can be used safely to suppress thrombin activity and fibrin deposition in neonatal RDS.