Recent advances in the ventilation of patients with acute respiratory distress syndrome (ARDS), including ventilation at low lung volumes, have resulted in a decreased mortality rate. However, even low-lung volume ventilation may exacerbate lung injury due to the cyclic opening and closing of fluid-occluded airways. Specifically, the hydrodynamic stresses generated during airway reopening may result in epithelial cell (EpC) injury. We utilized an in vitro cell culture model of airway reopening to investigate the effect of reopening velocity, airway diameter, cell confluence, and cyclic closure/reopening on cellular injury. Reopening dynamics were simulated by propagating a constant-velocity air bubble in an adjustable-height parallel-plate flow chamber. This chamber was occluded with different types of fluids and contained either a confluent or a subconfluent monolayer of EpC. Fluorescence microscopy was used to quantify morphological properties and percentage of dead cells under different experimental conditions. Decreasing channel height and reopening velocity resulted in a larger percentage of dead cells due to an increase in the spatial pressure gradient applied to the EpC. These results indicate that distal regions of the lung are more prone to injury and that rapid inflation may be cytoprotective. Repeated reopening events and subconfluent conditions resulted in significant cellular detachment. In addition, we observed a larger percentage of dead cells under subconfluent conditions. Analysis of this data suggests that in addition to the magnitude of the hydrodynamic stresses generated during reopening, EpC morphological, biomechanical, and microstructural properties may also be important determinants of cell injury.
- ventilator-induced lung injury
- pulmonary edema
- microbubble flows
- cellular injury
- cell mechanics
- surface tension
- pressure gradients
- shear stress
acute respiratory distress syndrome (ARDS) is characterized by the influx of protein-rich edema fluid into the lung because of an increased permeability of the alveolar-capillary barrier (25). In addition to hindering gas exchange, the presence of this edema fluid can disrupt the function of surfactant molecules that normally stabilizes pulmonary alveoli (16, 32). As a result, patients suffering from ARDS are unable to adequately inflate their lungs and must be mechanically ventilated to survive. However, the mechanical stresses associated with the reopening of fluid-filled and/or collapsed airways may damage the delicate epithelium. As a result, ventilation treatments can potentially exacerbate the underlying lung injury (10, 11).
Historically, ARDS patients were ventilated with large tidal volumes (i.e., VT = 10–12 ml/kg) to ensure adequate blood oxygenation. However, several studies (6, 7, 29–31) have demonstrated that the stretching deformations imparted to lung epithelial cells (EpC) during large-VT ventilation result in significant cellular injury and further disruption of the alveolar-capillary barrier. As a result, several investigators have used low-VT ventilation strategies (i.e., VT = 5–7 ml/kg) to minimize cellular injury (1, 3, 11). In particular, a recent clinical trial demonstrated a 22% reduction in patient mortality when VT was reduced from 12 ml/kg to 6 ml/kg (1). Despite this decrease in mortality, several studies indicate that significant lung damage may also occur at low lung volumes because of the reopening of collapsed/fluid-filled airways (8, 9, 12, 24).
At low lung volumes, the mechanical stresses generated during the reopening of closed pulmonary airways and alveoli may exacerbate the existing lung injury. However, the mechanisms responsible for airway/alveolar closure, and therefore the types of mechanical forces generated during reopening, have recently been debated (18). On the basis of computer tomographic images of injured lungs, Gattinoni and colleagues (13) proposed a closure mechanism in which the increased weight of the edematous lung results in the compliant collapse of small airways and alveoli. However, Hubmayr and colleagues (22, 23) have recently challenged the collapse mechanism based on direct experimental measurements of lung volumes with a parenchymal marker technique. These authors proposed an alternative mechanism of closure in which noncollapsed airways/alveoli become fluid filled due to the disruption of the alveolar-capillary barrier. Although the collapse and fluid-filling mechanisms may both occur in vivo, the present study focuses on the effect of reopening in a noncollapsed, fluid-filled airway.
Several investigators (4, 14, 15, 17, 19) have used computational fluid dynamic (CFD) techniques to investigate the mechanics of reopening a fluid-filled airway. In these models, a “finger of air” with surface tension γ is used to reopen the airway with velocity U as shown in Fig. 1. These models have provided quantitative information about the hydrodynamic stresses exerted on the EpC that line airway walls (arrows in Fig. 1). Specifically, cells downstream from the air bubble experience a nominal shear stress, cells near the bubble tip experience a combination of shear and normal stresses, and cells in the thin liquid film experience a normal stress (i.e., pressure). The cells near the bubble cap also experience significant spatial and temporal gradients in shear and normal stress. The maximum shear stress, normal stress gradient, and shear stress gradient experienced by EpC during bubble propagation have been quantified by Bilek et al. (4). Finally, a computational study by Jacob and Gaver (19) indicates that the morphology of EpC may amplify the magnitude of the hydrodynamic stresses under certain conditions.
Several investigators have also used experimental techniques to investigate the influence of airway reopening on lung injury. Although several animal studies indicate that low-volume ventilation techniques may generate significant lung damage (8, 9, 24), it is difficult to identify the consequences of a particular mechanical stimulus in whole lung models because of the spatial and temporal diversity of acute lung injury and the complex morphology of the lung. By contrast, in vitro cell culture models allow for a greater degree of control over the mechanical environment and therefore a more efficient means of evaluating how EpC respond to a given mechanical stimulus. For example, in vitro cell stretching devices have been successfully used to investigate the consequences of the large stretching deformations that occur during high-VT ventilation (27, 29, 33). To investigate the consequences of reopening a fluid-filled airway, Gaver and colleagues (4, 20) propagated an air bubble in a constant-height flow chamber. These studies report an increase in cellular necrosis at low reopening velocities and attribute this increase to the large normal stress gradients that develop at low bubble velocities. Previous experimental studies also indicate that pulmonary surfactants can significantly reduce the mechanical stresses generated during reopening (15) and the degree of cellular necrosis (4).
Although several parameters may influence the degree of EpC injury during the reopening of fluid-filled airways, the present study focuses on the effects of bubble velocity, airway diameter, repeated reopening events, and the degree of confluence in the EpC monolayer. The first motivation for this study is that the bifurcating structure of the lung results in a wide range of airway diameters. In addition, the total cross-sectional area of the lung increases as the airway generation number increases (21). As a result, the bubble velocity responsible for reopening will be different in different lung regions. Although the effect of bubble velocity has been investigated (4), the influence of airway diameter on cellular injury has not been quantified. Previous computational studies (15) indicate that the magnitude of the fluid mechanical stresses exerted on the EpC is inversely proportional to airway diameter. We therefore hypothesize that changes in both airway diameter and bubble velocity will alter the degree of cellular injury observed during reopening.
The second motivation for this study is that during ARDS the detachment of EpC will result in subconfluent monolayers and the disruption of cell-to-cell communications/attachments (32). Normally, EpC are tightly bound together into sheets called epithelia, and cell-cell contacts between adjacent EpC bear most of the mechanical stress (2). As a result, the loss of cell-cell contacts may make EpC more vulnerable to the hydrodynamic forces generated during reopening. Furthermore, EpC in subconfluent monolayers may exhibit different morphological, structural, and biomechanical properties than EpC in confluent monolayers (29). Changes in these properties may alter EpC susceptibility to the hydrodynamic forces generated during reopening. We therefore hypothesize that the level of cellular confluence will alter the degree of EpC injury observed during the reopening of fluid-filled airways.
To test these hypotheses, we have designed and utilized an experimental cell culture system that mimics the reopening of fluid-filled airways. In the first set of experiments, we exposed a 100% confluent monolayer of EpC to different bubble velocities in an adjustable-height, rigid-walled flow channel. In this system, channel height represents the diameter of an airway. We also exposed the confluent monolayer of EpC to repeated bubble passages to investigate the effect of cyclic closure and reopening. In a second set of studies, we cultured EpC to both 100% and ∼25% confluence and exposed both systems to equivalent reopening conditions. Cellular necrosis was monitored with standard live/dead staining protocols and quantified as the percentage of dead cells in each experiment. An analysis of the results indicates that, in addition to the hydrodynamic stresses generated during reopening, the biomechanical and biostructural properties of the EpC may also be important determinants of cellular injury during airway reopening.
MATERIALS AND METHODS
The rat pulmonary EpC line CCL-149 (American Type Culture Collection, Manassas, VA) was used in this study (passage number 20–30). Cells were maintained in a culture medium consisting of Ham's F-12K medium with 10% fetal bovine serum, 1% penicillin-streptomycin, and 1% fungizone (Invitrogen, Carlsbad, CA). The cell culture medium was changed every 2–3 days. For each experiment, cells were harvested with 0.125% trypsin (Invitrogen), counted with Trypan blue exclusion (Invitrogen), and seeded onto 30-mm-diameter coverslips placed in six-well plates by adding 2 ml of a 3 × 104 cells/ml solution to each well. Cells were grown under standard culture conditions (37°C, 5% CO2-95% air) for 1 day to obtain a subconfluent monolayer (∼25%) or for 4 days to obtain a confluent monolayer (100%).
Parallel-plate flow chamber.
EpC were exposed to reopening conditions in the modified perfusion chamber shown in Fig. 2. Here, the POC mini chamber system (Hemogenix, Colorado Springs, CO), which was specifically designed for high-resolution live-cell microscopy, was modified to create an adjustable-height parallel-plate flow chamber (Fig. 2A). Specifically, a rigid silicone gasket of variable thickness was sandwiched between a 0.17-mm-thick upper cover glass and a 0.17-mm-thick lower cover glass seeded with cells (see above) (Fig. 2B). Silicon gaskets (McMaster-Carr, New Brunswick, NJ) were cut to the desired flow geometry (Fig. 2C) with a water jet cutter, to form the flow channel. Although a variety of flow channel geometries can be created, this study utilized gaskets that formed a flow channel 10 mm in width and 25 mm in length. To assemble the chamber, the lower cover glass, containing cells, was washed with phosphate-buffered saline (PBS) to remove nonadherent cells and placed onto the chamber base. The silicon gasket, aluminum perfusion adapter with inlet and outlet tubes, upper cover glass, and flow chamber top adapter were then placed in sequential order to complete the assembly (Fig. 2, A and B). The channel height was governed by the thickness of the silicone gasket, and channel heights of 0.5, 0.8, and 1.7 mm were used in this study. These channel heights were selected based on the diameters of terminal and respiratory bronchioles (34).
The complete experimental setup is shown in Fig. 2D. The flow chamber was connected to a PHD 2000 high-accuracy syringe pump (Harvard Apparatus, Holliston, MA) and was mounted on an IX-71 inverted microscope (Olympus, Melville, NY) with fluorescence capabilities. A cooled SPOT RT-KI 7–33 Shot Color charge-coupled device camera (Diagnostic Instruments, Sterling Heights, MI) was used to obtain fluorescence live/dead images from several center line locations within the flow channel. Note that disassembly of the system is not required to obtain high-resolution images. All experiments were performed at room temperature (25°C).
Generation of airway reopening conditions.
Two occlusion fluids were used in the study. Cell culture medium was used for experiments related to the effect of cellular confluence, while PBS with 0.1 mg/ml CaCl2 and MgSO4 was used for experiments related to the effect of airway diameter. PBS represents a surfactant-deficient, high-surface tension airway-lining fluid, and since surfactant deficiency is a characteristic of ARDS (32), this fluid was used in most of our experiments. However, use of PBS in the experiments with a subconfluent monolayer of EpC resulted in significant cell detachment (see Fig. 6). In contrast, cell culture medium did not result in cell detachment for confluent or subconfluent monolayers (see Fig. 7) and was therefore used in the experiments related to the effect of cellular confluence.
After assembly, the flow chamber did not contain any fluid (i.e., it was air filled). A syringe pump was then used to fill the flow chamber with the occlusion fluid at a rate of 3 mm/s. Once the channel was filled, the same syringe pump was operated in reverse to create a constant-velocity “forward”-propagating air bubble that removed the occlusion fluid at a constant rate. Once the air bubble had displaced the occlusion fluid, the chamber was refilled with a live/dead stain at 3 mm/s and incubated before visualization (see below). On average, each experimental run (i.e., chamber assembly, bubble propagation, and obtaining fluorescence images), was performed in ∼10 min for experiments using PBS and in ∼20 min for experiments using cell culture medium (differences in duration were due to the different incubation times required by the live/dead stains in PBS or medium). The initial filling of the chamber as well as the refilling of the chamber with the live/dead stain involve the “backward” propagation of an air-liquid interface. Control experiments were therefore conducted to determine whether this “backward” air-liquid interface propagation and/or maintenance of cells at room temperature for 10 or 20 min causes any cell necrosis. Specifically, control conditions were established by pumping in an occlusion fluid containing the live/dead stain at 3 mm/s and maintaining the system at room temperature for 10 or 20 min. Almost no dead cells were observed under control conditions, which demonstrates that neither the “backward” air-liquid interface propagation during chamber filling nor maintenance of the cells at room temperature for 10 or 20 min results in cellular necrosis.
Bubble velocities used in this study were selected based on the range one would expect to find in the terminal and respiratory bronchioles. For normal breathing conditions of VT = 500 ml and 12 breaths/min, the range of total cross-sectional areas found in the terminal and respiratory bronchioles (34) (100–1,000 cm2) results in an expected range of reopening velocities from 1 to 10 mm/s. A baseline reopening velocity of 3 mm/s was selected to correspond to this range. In addition, we also investigated how an order of magnitude change in reopening velocity (i.e., reopening velocities of 30 and 0.3 mm/s) influences cellular injury. The flow rates required to create these reopening velocities were calculated based on the cross-sectional areas of the flow channel, Q = UA, where Q is the flow rate specified by the syringe pump, U is the velocity of the air bubble, and A is the cross-sectional area of the flow channel. This formula assumes that the thickness of the film deposited during reopening is negligible, and we utilized a well-established fluid-dynamic model of airflow in a fluid-filled channel (5) to verify this assumption. Specifically, film thickness is a function of a dimensionless velocity known as the capillary number, Ca = μU/γ, where μ is the viscosity of the medium and γ is the surface tension. Ca relates the relative importance of viscous forces to surface tension forces on bubble propagation. For the bubble velocities used in our experiments the capillary number range is Ca = 3.7 × 10−6 to 3.7 × 10−4. For this range of Ca values, the maximum theoretical film thickness is 0.5% of the channel height. Therefore, neglecting the film thickness in the flow rate calculation is justified.
Quantification of cellular injury.
Cell viability after bubble propagation was determined by live/dead staining. The live/dead stain (Invitrogen) consisted of 1.2 μM ethidium homodimer 1 and 1.2 μM calcein AM in PBS for the airway diameter study or 1 μM ethidium homodimer 1 and 2 μM calcein AM in serum-free cell culture medium for the confluence study. The working fluid for the stain was selected based on the occlusion fluid used in the experiment. For experiments conducted with PBS, the cells were incubated in the dark, at room temperature, for 2 min. This short incubation time was used to prevent any additional cellular injury that may occur during prolonged exposure to PBS. In contrast, cells were incubated in the dark for 10 min in the experiments that utilized cell culture medium as the working fluid. Calcein AM (494-nm excitation, 517-nm emission) and ethidium homodimer 1 (528-nm excitation, 617-nm emission) have been used extensively to differentiate live and dead cells. Calcein AM freely diffuses into living cells, where it is converted into a membrane insoluble product. As it accumulates inside the cell, the cytoplasm of the cell can be viewed in green color. On the other hand, ethidium homodimer 1 cannot enter through membranes of living cells. If the plasma membrane is ruptured, this dye enters the cell and binds to nuclear DNA. As a result, the nucleus of these “dead” cells can be viewed in red color.
For each control and experimental condition, 5–10 fluorescent images were obtained from the center line of the channel. The reason for acquiring images only along the center line was to eliminate any possible stress magnification effects near the side walls of the channel. The number of live cells (i.e., calcein positive) and dead cells (i.e., ethidium positive) in each image were counted with Metamorph image analysis software (Molecular Devices, Downington, PA). The percentage of dead cells was calculated as the number of dead cells divided by the total number of cells in each image. For each experimental condition, the percentage of dead cells was quantified and the mean and SE of these values were calculated. An analysis of variance (ANOVA) with P < 0.05 was used to document statistically significant differences.
Cyclic reopening experiments.
We also conducted a set of experiments to investigate the effect of cyclic reopening on EpC injury. These experiments were conducted by exposing confluent monolayers of EpC to repeated bubble propagations and used PBS as the occlusion fluid. For these experiments, we assembled the chamber in the normal fashion and filled it with the occlusion fluid. We then propagated a bubble in the “forward” direction at a velocity of 3 mm/s to remove the occlusion fluid. We then refilled the chamber with the occlusion fluid at a rate of 3 mm/s and repeated the “forward” bubble propagation at the same velocity. This procedure was repeated several times to produce one, three, or five “forward” bubble propagations over the EpC. Finally, a live/dead stain was pumped into the chamber, and fluorescent images were obtained at 5–10 random center line locations. In addition to quantifying the necrosis rate for adhered cells, we also quantified cell detachment by counting the number of attached cells in each fluorescent image and calculating the percentage of cells that remained adhered relative to the average number of cells counted under control conditions (i.e., no “forward” bubble propagation). All images were obtained with a ×10 objective.
In this study, confocal microscopy was used to assess the morphology of EpC in both low- and high-confluence monolayers. Cells were stained with calcein AM (494-nm excitation, 517-nm emission) and were viewed with a LSM 510 Meta confocal microscope (Zeiss, Thornwood, NY). After cells were grown to the desired degree of confluence, the flow chamber was assembled and serum-free cell culture medium with 2 μM calcein AM was pumped into the chamber. Calcein diffuses into the entire cytoplasm, which allows for the visualization of whole cell morphology. Cells were incubated in the dark for 10 min before a confocal scan was obtained, which consisted of 20–40 confocal slices depending on the thickness of the cells. Each confocal image/scan was 0.2 μm in thickness, and the total scanning time was ∼2 min.
To visualize actin filaments, cells were seeded on coverslips and grown to the desired degree of confluence. Cells were then washed with PBS and fixed in 10% neutral buffered formalin (3.7% formaldehyde in PBS) for 10 min at room temperature. After being washed with PBS, cells were permeabilized with a solution of 0.1% Triton X-100 in PBS for 5 min at room temperature and then rinsed with PBS. After nonspecific binding was blocked with 1% bovine serum albumin in PBS for 10 min at room temperature, cells were incubated with 2.5% Alexa 488-labeled phalloidin (Invitrogen) in PBS for 20 min at room temperature to localize actin filaments. Finally, cells were washed with PBS, and the coverslips were flooded with 0.002 mg/ml 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St. Louis, MO) in aqueous solution to label nuclei. Coverslips were then mounted on glass slides with Gelmount mounting medium (Biomeda, Foster City, CA) and visualized by epifluorescence with an IX-71 inverted microscope (Olympus, Melville, NY) with fluorescence capabilities.
To visualize microtubules, cell monolayers were washed with PBS and fixed in cold methanol for 10 min. After a wash with PBS nonspecific binding was blocked with 10% normal goat serum in PBS for 15 min at room temperature, and cells were subsequently rinsed with PBS. To localize tubulin, cells were first incubated with a mouse monoclonal anti-tubulin antibody (DM1A, 1:40; Sigma) in PBS at 37°C for 20 min. After incubation with the primary antibody, cells were incubated with a rhodamine-labeled goat anti-mouse IgG (1:40, Jackson Immunologicals, West Grove, PA) in PBS at 37°C for 20 min. Finally, cells were washed with PBS, and the coverslips were flooded with 0.002 mg/ml DAPI in aqueous solution to label nuclei. As above, coverslips were mounted on glass slides with Gelmount mounting medium and visualized by epifluorescence with an IX-71 inverted microscope (Olympus) with fluorescence capabilities.
Effect of channel height and bubble velocity on cellular injury.
Figure 3 shows representative fluorescent pictures obtained under control conditions and as a function of bubble velocity and channel height. Ethidium-stained cells (red) are dead, and calcein-stained cells (green) are alive. If the cell contained both colors, it was accepted as dead. As can be seen, the number of dead cells increases as the velocity and the channel height decrease.
Figure 4 shows the percentage of cell death (mean ± SE) for the various conditions used in this study. For all channel heights, control experiments resulted in almost no injury (<1%). The highest injury level was observed at a channel height of 0.5 mm and a bubble velocity of 0.3 mm/s (40.4 ± 3.6%), while the lowest injury was observed at a channel height of 1.7 mm and a bubble velocity of 30 mm/s (0.5 ± 0.01%). A two-way ANOVA indicated that increasing the bubble velocity resulted in a statistically significant decrease in cell death (F = 46.0, P < 0.01). In addition, propagating an air bubble in channels with larger heights resulted in a statistically significant decrease in cell death (F = 27.3, P < 0.01). Note that the decrease in cell injury with increasing bubble velocity is consistent with previous studies (4). A post hoc comparison using the Holm-Sidak method indicates that all mean values in Fig. 4, except for values under control conditions, are statistically different (P < 0.01).
Effect of repeated bubble passes on cell adhesion and injury.
Figure 5A shows how changing the number of repeated bubble passages influenced the percentage of cells that remained adhered relative to control conditions. ANOVA indicates that although one bubble pass does not result in cell detachment, three and five bubble passes result in a statistically significant reduction in the number of adhered cells (F = 26.7, P < 0.01). In addition, five bubble passes results in more cell detachment than three bubble passes (P < 0.01). Figure 5B shows how repeated bubble passages influence the necrosis rate of the adhered cells. ANOVA indicates that compared with one bubble passage, three bubble passages increases the cell necrosis rate (F = 53.3, P < 0.01). However, there was no statistically significant difference in the necrosis rate between three and five bubble passages (P = 0.8). These results suggest that cyclic airway reopening may cause significant lung damage and a denuded epithelium.
Effect of cell confluence on cellular injury.
During ARDS, the epithelium in the lung is disrupted because of the detachment of EpC from the basement membrane. As a result, the EpC monolayer becomes less confluent, cell-cell attachments are disrupted, and the ratio of type I to type II cells may be altered (32). To investigate these conditions, we cultured the CCL-149 cell line for 1 and 4 days to obtain low- and high-confluence monolayers of EpC. High-confluence monolayers represent uninjured airways, whereas low-confluence monolayers represent injured airways. Cells were then exposed to bubble velocities of 0.3, 3, and 30 mm/s in a 0.5-mm-high channel to investigate the effect of confluence on cellular viability. Initially, we propagated air bubbles in channels occluded with PBS that contained a low-confluence monolayer of EpC. However, under these conditions, all bubble velocities resulted in the detachment of a large percentage of the cells (see Fig. 6A). It should be noted that we did not observe this cell detachment behavior when PBS was used as the occlusion fluid for the high-confluence monolayer system (Fig. 3). To quantify the amount of cell detachment, we counted the number of attached cells in each fluorescent image (∼5–10 images per experiment) and calculated the percentage of cells that remained adhered relative to the average number of cells counted under control conditions. As shown in Fig. 6B, all three velocities produced statistically significant detachment relative to control conditions (ANOVA, F = 39.4, P < 0.01), while differences in the amount of cell detachment at the three different velocities were not statistically significant (P > 0.22). These results indicate that subconfluent monolayers undergo ∼80–95% cell detachment when exposed to bubble propagation in PBS. Since cellular detachment could lead to significant errors in quantifying cellular injury, we used an alternative occlusion fluid, i.e., cell culture medium, which did not result in significant cellular detachment in either high- or low-confluence monolayers.
Figure 7 shows representative fluorescent pictures of the live/dead staining obtained for both high- and low-confluence monolayers at various bubble velocities when cell culture medium was used as the occlusion fluid. Ethidium-stained cells (red) are dead, and calcein-stained cells (green) are living. Figure 8 shows the percentage of cell death (mean ± SE) for the various conditions. For both high- and low-confluence monolayers, control experiments resulted in almost no death (<1%). Two-way ANOVA indicated that there was significantly less death in the high-confluence monolayers than in the low-confluence monolayers at all three velocities (F = 111, P < 0.01). In addition, two-way ANOVA indicated that the degrees of cellular injury observed at different velocities in the low-confluence monolayers were not statistically different (F = 0.91).
Effect of cell confluence on cellular morphology.
In this study, we quantified the morphological properties of EpC in high- and low-confluence monolayers by confocal microscopy. Examples of the three-dimensional images obtained with this technique are shown in Fig. 9. Here, a cytoplasmic live stain, calcein AM, is used to visualize the entire cell body. In these images, the top section represents a side view of the cross sections obtained at the location marked with a horizontal line. Most cells are elongated in one direction, and the length of the cell (L) was defined as the maximum distance between the cell boundaries in the elongated direction. The maximum perpendicular distance to the length was defined as the width (W). For subconfluent cells, we defined the maximum distance from apical to basal surface of the cell as the cell height (hc). For confluent cells, we measured hc as the distance from the apical surface of cell-cell contact regions to the maximum apical surface of the cell. Note that hc was measured by using the side views. This measurement procedure was repeated for 28 cells in high-confluence monolayers and 19 cells in low-confluence monolayers. As shown in Fig. 10, the average dimensions measured for EpC in the high-confluence monolayer were L = 44 ± 1.6 μm, W = 28 ± 1.0 μm, and hc = 3.1 ± 0.14 μm, while the average dimensions in the low-confluence monolayer were L = 78 ± 5.7 μm, W = 48 ± 4.3 μm, and hc = 5.8 ± 0.32 μm. ANOVA indicates that the differences in the average L, W, and hc values were statistically significant (P < 0.01 for all cases).
Effect of cellular confluence on microstructural properties.
In this study, we also visualized the actin and tubulin cytoskeleton network for EpC in high- and low-confluence monolayers. As shown in Fig. 11A, for high-confluence monolayers, actin filaments display a normal structure in which filaments exhibit fiberlike structure, provide attachments between neighboring cells, and allow the EpC to form a tight sheet (i.e., an epithelium). In the low-confluence monolayer, however, actin forms a less fiberlike network and displays a circumferential organization of filaments at the cell periphery. Figure 11B shows the network of microtubules observed in both high- and low-confluence monolayers. Tubulin forms a relatively uniform network in the high-confluence cells, while in the low-confluence cells the density of tubulin appears to decrease at the cell periphery.
It is well established that the ventilation of ARDS patients at large VT imparts injurious stretching deformation to lung EpC (10, 32). However, the reopening of fluid-filled airways at low lung volumes may also impart injurious mechanical stresses (18). The goal of this study was to develop and utilize an in vitro cell culture model of airway reopening to investigate how several parameters including reopening velocity, airway diameter, repeated reopening events, and cellular confluence influence cellular injury. In this model, we propagated a long, continuous air bubble at different velocities in an adjustable-height parallel-plate flow chamber. This flow chamber was occluded with two different types of fluids (PBS and cell culture medium), and the lower wall was lined with alveolar EpC cultured to either confluent or subconfluent states. Fluorescent live/dead stains were used to quantify the percentage of dead cells, confocal microscopy was used to quantify cellular morphology, and cytoskeletal stains were used to visualize changes in EpC microstructure.
Although previous investigators (4, 20) utilized parallel-plate flow chambers to investigate the influence of two-phase air-liquid flows on alveolar EpC, the flow chamber used in this study offers some unique advantages. First, unlike previous systems, high-resolution fluorescence microscopy with ×63 oil-immersion objectives can be performed in situ without any disassembly of the system. The ability to visualize cells in situ allowed for the specification of well-defined control conditions (i.e., filling the chamber with fluid) that did not result in significant cell death. Second, although a simple flow channel geometry was used in this study, arbitrary flow geometries, including bifurcating patterns, can easily be accomplished by changing the pattern cut into the silicone-rubber flow channel gasket. Finally, the bottom coverslip of this system can be replaced with a transparent, flexible membrane to more accurately simulate the collapsibility of pulmonary airways.
During the ventilation of fluid-filled lungs, different-diameter airways will be exposed to different reopening velocities. In this study, we accounted for these differences by varying both the bubble velocity and the height of the parallel-plate flow chamber. As shown in Figs. 3 and 4, a reduction in both airway diameter and/or bubble velocity resulted in an increase in cellular necrosis for cells cultured to 100% confluence. These results indicate that airways in distal lung regions are most prone to injury and that ventilation strategies with rapid reopening dynamics may help prevent cellular and tissue damage. The magnitude of the hydrodynamic stresses exerted on the EpC will vary because of the different bubble velocities and channel heights used in each experiment. To investigate the relationship between hydrodynamic stress magnitude and cell death, we utilize the relationships developed by Bilek et al. (4) to calculate the maximum pressure gradient [(dP/dx)max], shear stress [(τs)max], and shear stress gradient [(dτs/dx)max] exerted on the EpC during bubble passage. (1) (2) (3) Here, γ is the air-liquid surface tension, H is the channel half-height, Ca = μU/γ is the capillary number, U is the bubble velocity, and μ is the fluid viscosity. For each experimental condition, i.e., different U and H, in which PBS was used as the occlusion fluid, the percentage of dead cells was plotted as a function of the different hydrodynamics stress parameters in Fig. 12. Figure 12 specifically demarcates changes in velocity and channel height, and all data in Fig. 12 are for 100% confluent monolayers of EpC. The linear regression analysis of the PBS data only (dashed line in Fig. 12) demonstrates that there is a strong (r2 = 0.91) and statistically significant (P < 0.01) correlation of cell death with (dP/dx)max. In addition, there was a weaker (r2 = 0.42) but statistically significant (P < 0.05) correlation of cell death with (dτs/dx)max. Finally, there was no statistically significant correlation of cell death with the applied shear stress (r2 = 0.03, P = 0.664). This analysis provides strong statistical evidence that the amount of cell injury is not a function of the maximum shear stress applied to the cell. In addition, although there is a statistically significant correlation with shear stress gradients, the divergence of the data in Fig. 12C as well as the nearly vertical relationships for isopleths in height suggest that shear stress gradients play a minor role. Therefore, the results of this study strongly indicate that the development of large spatial gradients in pressure is required to rupture the plasma membrane. Although the exact mechanisms are unknown, we hypothesize that spatial gradients in pressure may cause inhomogeneous cell deformation and/or the development of nonuniform plasma membrane strains (i.e., strain concentrations) that promote membrane rupture. Future computational models of this system that account for cell mechanics and deformability may be useful in testing this hypothesis.
In addition to the effect of a single reopening event, this study investigated how multiple bubble passes, which mimic cyclic airway reopening, influence cell injury. These studies utilized a confluent monolayer of EpC, PBS as the occlusion fluid, and a reopening velocity of 3 mm/s. As shown in Fig. 5B, although additional bubble passes result in an increase in cell necrosis (1 vs. 3 bubble passes), after a critical number of bubble passages (i.e., 3), the bubble does not induce additional cell necrosis. The data in Fig. 5A demonstrate that although EpC remain adhered to the surface after a single bubble pass, repeated bubble passes cause a significant amount of cell detachment. We therefore conclude that the major implication of repeated airway reopening events is a denuded epithelium. Cell detachment may be an important mechanism of lung injury since an intact epithelium is required for normal lung function and EpC detachment would lead to an increased permeability of the alveolar-capillary barrier and the flooding of lung air spaces. Future studies should therefore focus on the potential mechanisms of cell adhesion to identify cell-based properties that would prevent lung injury by promoting firm EpC adhesion.
In addition to the effect of reopening dynamics, data from the present study can be used to investigate the effect of different occlusion fluids on EpC injury during reopening. Specifically, in the present study 100% confluent monolayers of EpC were exposed to equivalent reopening conditions (i.e., bubble velocity and channel height of 0.5 mm) in two different occlusion fluids, PBS and cell culture medium. As shown in Fig. 13, for a given bubble velocity bubble propagation in cell culture medium resulted in significantly less cell death compared with bubble propagation in PBS (ANOVA, F = 121, P < 0.01). It should be noted that for both fluids, control experiments, which consisted of maintaining cells in PBS or cell culture medium for 10 or 20 min with no bubble propagation, resulted in almost no dead cells (<1%). The highest injury level was observed at a bubble velocity of 0.3 mm/s in PBS (40.4 ± 3.6%), and the lowest injury level was observed at a bubble velocity of 30 mm/s in cell culture medium (0.7% ± 0.2%). Two-way ANOVA also indicated that for cell culture medium conditions, increasing the bubble velocity resulted in a statistically significant decrease in cell death (F = 51.8, P < 0.01), which is consistent with data obtained for PBS conditions and with previous studies (4). Although PBS and cell culture medium have equivalent viscosities of μ = 0.007 g/(cm·s), measurements of the surface tension via the pendent drop technique indicate that the surface tension of cell culture medium is lower than the surface tension of PBS (i.e., γmedium = 57 dyn/cm vs. γPBS = 72 dyn/cm). To investigate whether the lower γmedium value was responsible for the decreased cell necrosis, the percentage of cell death measured in the cell culture medium experiments was plotted as a function of the different stress components as shown in Fig. 12; note that for these data points, stress components were calculated by using γmedium = 57 dyn/cm in Eqs. 1–3. All culture medium data points were significantly lower than the value predicted by a linear regression of the PBS data (dashed lines in Fig. 12). In addition, a linear regression on all data points in Fig. 12 indicates that the inclusion of the cell culture medium points significantly reduces the strength of the correlation [i.e., r2 = 0.66 for (dP/dx)max and r2 = 0.21 for (dτs/dx)max]. This result indicates that the decrease in cell death in the cell culture medium experiments cannot be explained by a decrease in hydrodynamic stress magnitude alone. We therefore hypothesize that differences in other fluid properties, such as osmolarity, may induce a rapid change in cell properties, which in turn alters the cell's susceptibility to bubble-induced injury. For example, Quadri et al. (26) indicate that exposure to hyperosmolar solutions induces changes in the cytoskeleton of pulmonary endothelial cells. Although PBS and cell culture medium exhibit different osmolarities (275 mosM vs. ∼330 mosM), these fluids also exhibit differences in several other parameters, i.e., pH, which may also influence EpC susceptibility to bubble-induced injury. Therefore, future studies that are specifically designed to investigate the effect of individual fluid properties (i.e., osmolarity or pH) are required to ascertain the exact role of these parameters in cell injury during bubble propagation.
In the present study, we demonstrated that, for a constant bubble velocity and channel height, subconfluent monolayers of EpC are more susceptible to bubble-induced injury than confluent monolayers (see Fig. 8). This increased susceptibility of subconfluent cells to mechanically induced injury is consistent with a study by Tschumperlin and Margulies (29), who demonstrated that for a constant stretch magnitude subconfluent cells experience more necrosis than confluent cells. In this study, we also demonstrated significant differences in cell morphology between the subconfluent and confluent monolayers. Recently, a computational study by Jacob and Gaver (19) demonstrated that that the magnitude of the hydrodynamic stresses applied to the EpC may be a function of cell morphology. Specifically, for constant surface tension, channel height, and flow rate, these authors predict that the magnitude of the hydrodynamic stress applied to the EpC is only a function of the ratio between hc and W, i.e., mW = hc/W. It should be noted that these authors utilized a two-dimensional model, and therefore this ratio may also be defined based on the cell length, i.e., mL = hc/L. To investigate whether the change in morphology shown in Fig. 10 would result in changes in the magnitude of the hydrodynamic stresses applied to the EpC, we calculated mW and mL for both confluent and subconfluent cells and report the means ± SE in Table 1. Although subconfluent cells are taller and wider than confluent cells (Fig. 10), mean mW and mL are not statistically different (ANOVA, F = 0.118). For constant mW or mL values, Jacob and Gaver (19) predict that the magnitude of the hydrodynamic stress applied to the EpC is nearly equivalent. Therefore, the differences in cell death between confluent and subconfluent cells (see Fig. 8) are not likely due to a change in the magnitude of the applied hydrodynamic stresses. This result suggests that factors other than the magnitude of the hydrodynamic stresses generated during reopening may be important for determining the degree of cellular injury.
In this study, we investigated some of the potential determinants of cell injury by visualizing microstructural differences between subconfluent and confluent cells. As shown in Fig. 11, these differences are significant and indicate that subconfluent and confluent cells likely exhibit different mechanical properties. For example, the redistribution of actin in the subconfluent cells, i.e., reduced actin density in the cell interior and increased concentration of actin in the periphery, would suggest heterogeneous mechanical properties. while EpC in 100% confluent monolayers exhibited relatively uniform actin distribution. In addition, the presence of cell-cell contacts in the confluent monolayer may provide the epithelium with the mechanical integrity required to withstand the bubble-induced hydrodynamic forces. We note that numerous biophysical parameters are changing simultaneously in the confluent versus subconfluent experiments, and it is therefore difficult to identify the specific mechanisms responsible for injury in these experiments. The determination of how various cell-based factors influence injury will therefore require more controlled studies that investigate the effect of individual cell-based factors. For example, future studies could investigate injury mechanisms by altering a specific biophysical property, such as cytoskeletal structure, and coupling observations of cell necrosis with quantitative measurements of cytoskeletal structure/mechanics.
One potential physiological limitation of the present system is that cells were cultured on a rigid glass substrate. To better mimic collapsible pulmonary airways, future studies should incorporate a flexible substrate and should investigate how altering the elasticity of this substrate influences cell injury during reopening. In addition, the effect of more complex airway geometries, i.e., bifurcations, on cell injury should also be investigated. Tubular channel geometries may also be used in future to better reflect the airway structure. Another nonphysiological parameter of the present system is the temperature at which the experiments were performed, 25°C. Future experiments should be performed at 37°C. Finally, ARDS typically involves the upregulation of a variety of inflammatory mediators and the presence of thrombin or histamine in the alveolar fluid due to disruption of the alveolar-capillary barrier. These chemical agents have been shown to alter the mechanical properties of EpC (28), and future studies should therefore investigate whether these biomechanical changes influence the amount of bubble-induced cell injury.
In this study, we utilized an in vitro cell culture model to investigate how bubble velocity, airway diameter, repeated reopening events, and degree of cell confluence influence cell necrosis during the reopening of fluid-filled airways. Results indicate that slow reopening velocities and small airway diameters both result in a significant increase in cell necrosis. The implication of this result is that airways in the deep lung, which are smaller, would experience slower reopening velocities and are most susceptible to injury. These results also indicate that ventilation techniques that seek to reopen the lung rapidly may prevent cellular injury in fluid-filled airways. An analysis of the data obtained in 100% confluent monolayers with PBS as the occlusion fluid indicates that the large spatial gradients in pressure that develop near the bubble tip are responsible for rupturing the cell's plasma membrane. The repeated-reopening experiments indicate that the major implication of cyclic airway reopening is detachment of EpC and a disruption of the epithelium. A comparison of data obtained with PBS and cell culture medium as the occlusion fluid indicates that in addition to the magnitude of the hydrodynamic stress applied to the EpC, cellular responses to changes in the fluid environment may also alter EpC susceptibility to bubble-induced injury. This study also demonstrated that EpC in subconfluent monolayers are more susceptible to the hydrodynamic stresses generated during reopening. Although there are significant differences in the morphology of confluent and subconfluent cells, analysis of this data suggests that these changes would not alter the magnitude of the hydrodynamic stresses applied to the EpC. Therefore, changes in other cell-based properties might be responsible for the different injury levels observed in confluent and subconfluent EpC. For example, visualization of the EpC cytoskeletal structure indicates that subconfluent cells may have more heterogeneous biomechanical properties than confluent cells and that confluent EpC contain significant cell-cell attachments that might increase EpC resistance to bubble-induced injury. These results therefore indicate that in addition to the magnitude of the applied hydrodynamic forces, the degree of cell injury during airway reopening may also depend on cellular responses to environmental conditions and the biomechanical/biostructural properties of the EpC. Therefore, future studies should be designed to specifically investigate how changes in different environmental and/or cellular properties influence EpC injury during airway reopening. The results of these studies may be useful in the future development of ventilation strategies that seek to reduce the magnitude of the injurious stress components or pharmaceutical therapies that seek to reduce EpC susceptibility to bubble-induced cell injury.
This work was supported by a Beginning Grant-In-Aid from the Pennsylvania-Delaware affiliate of the American Heart Association. S. Ghadiali is a Parker B. Francis Fellow of Pulmonary Research.
We thank E. David Bell for his assistance with the cell adhesion and repeated-reopening experiments.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society