Although patients with acute respiratory distress syndrome require mechanical ventilation, these ventilators often exacerbate the existing lung injury. For example, the cyclic closure and reopening of fluid-filled airways during ventilation can cause epithelial cell (EpC) necrosis and barrier disruption. Although much work has focused on minimizing the injurious mechanical forces generated during ventilation, an alternative approach is to make the EpC less susceptible to injury by altering the cell's intrinsic biomechanical/biostructural properties. In this study, we hypothesized that alterations in cytoskeletal structure and mechanics can be used to reduce the cell's susceptibility to injury during airway reopening. EpC were treated with jasplakinolide to stabilize actin filaments or latrunculin A to depolymerize actin and then exposed to cyclic airway reopening conditions at room temperature using a previously developed in vitro cell culture model. Actin stabilization did not affect cell viability but significantly improved cell adhesion primarily due to the development of more numerous focal adhesions. Surprisingly, actin depolymerization significantly improved both cell viability and cell adhesion but weakened focal adhesions. Optical tweezer based measurements of the EpC's micromechanical properties indicate that although latrunculin-treated cells are softer, they also have increased viscous damping properties. To further investigate the effect of “fluidization” on cell injury, experiments were also conducted at 37°C. Although cells held at 37°C exhibited no changes in cytoskeletal structure, they did exhibit increased viscous damping properties and improved cell viability. We conclude that fluidization of the actin cytoskeleton makes the EpC less susceptible to the injurious mechanical forces generated during cyclic airway reopening.
- microbubble flows
- surface-tension forces
- ventilation-induced lung injury
- power-law rheology
although mechanical ventilation is required to maintain adequate gas exchange in patients experiencing acute lung injury and/or the acute respiratory distress syndrome (ARDS), the overdistension of lung tissue (3, 11, 25) and/or the cyclic closure and reopening of fluid-filled airways (7, 21, 26, 28) during ventilation can exacerbate the existing lung injury by exposing alveolar epithelial cells (EpC) to pathological mechanical forces. Lung injury due to overdistension of alveolar EpC can be minimized by utilizing a low tidal volume ventilation strategy (1). However, at low lung volumes, dependent regions of an edematous lung become derecruited (i.e., nonventilated) and must be reopened during ventilation. The specific mechanisms of derecruitment at low lung volumes may include compliant airway collapse as well as the flooding of noncollapsed airways (17). In either case, the cyclic closure and reopening of collapsed or fluid-filled airways during ventilation involve the displacement of air-liquid interfaces and the propagation of microbubbles over the EpC lining airway/alveolar walls (4, 22, 27, 40). These microbubble flows exert large hydrodynamic and surface tension forces to the epithelium and may therefore cause necrosis of EpC due to plasma membrane rupture (4, 40), barrier disruption due to EpC detachment (40), and exacerbation of the existing lung injury (7).
Several investigators have tried to minimize cell injury during low lung volume ventilation by either preventing airway closure/collapse or reducing the magnitude of the hydrodynamic forces generated during airway reopening. For example, previous investigations (32) indicate that raising the positive end-expiratory pressure (PEEP) may prevent airway closure. However, the appropriate selection of PEEP is often difficult in a clinical setting (9) and the difficulty in setting the appropriate PEEP value has been demonstrated in the recently completed ARDS Network ALVEOLI Trial (6) where the use of a higher PEEP value did not significantly reduce the mortality rate. In vitro studies (4) have suggested that the use of exogenous surfactants, which decreases surface tension forces, may result in decreased cell necrosis. However, surfactant replacement therapy for ARDS has had limited clinical success (24, 31) potentially due to the deactivation of surfactant by plasma proteins (5). As a result, there is a pressing need for new therapies to address ventilator-induced lung injury during low volume ventilation.
While much research has focused on either changing ventilation parameters or developing therapies that minimize the magnitude of the injurious mechanical forces that occur during ventilation, an alternative approach is to alter the physical characteristics of the EpC themselves to make them less susceptible to injurious mechanical forces. For example, it is well established that the amount of cell necrosis due to plasma membrane rupture is directly related to the amount of cell deformation (35). Although reducing the magnitude of the hydrodynamic forces exerted on the EpC during cyclic airway reopening may result in less cell deformation, an alternative approach is to alter the cell's resistance to deformation or microrheological properties. Previous studies (13, 23) have demonstrated that biological cells exhibit complex rheological properties that are dependent on the cell's cytoskeletal structure. Specifically, cells exhibit a combination of elastic, solid-like properties as well as viscous, fluid-like properties (10, 13, 14, 18, 23). For a constant mechanical load, a compliant or “soft” cell will undergo more deformation than a rigid cell and cells that have more fluid-like properties will deform slower than highly elastic cells.
We hypothesize that the amount of cell necrosis and detachment that occurs during cyclic airway reopening depends on the rheological and biostructural properties of the EpC. To test this hypothesis, we first altered the cytoskeletal structure of the cell by treating EpC with drugs that either stabilize or depolymerize actin filaments. A previously developed in vitro cell culture model of cyclic airway reopening (40) was used to expose treated and untreated EpC to a propagating air-liquid interface, and changes in cell necrosis and cell detachment were monitored. We also used a previously developed optical tweezer microrheometer (38) to determine how changes in cytoskeletal structure influence the rheological properties of EpC. The effect of rheological properties on cell injury patterns was also investigated by conducting experiments at 37°C. These studies indicate that EpC with more fluid-like properties are less susceptible to microbubble-induced injury during airway reopening.
Human A549 alveolar EpC (CCL-185, American Type Culture Collection, Manassas, VA) were cultured at passage number 10–30 and were maintained in Ham's F-12K medium with 10% FBS, 1% penicillin/streptomycin, and 1% fungizone (Invitrogen, Carlsbad, CA). The cell culture medium was changed every 2–3 days. Cells were seeded onto 30-mm-diameter coverslips placed in 6-well plates by adding 2 ml of an 8 × 104 cells/ml solution to each well. Cells were grown under standard culture conditions (37°C, 95% O2-5% CO2) for 4 days to obtain a confluent monolayer.
Generation of airway reopening conditions.
Cells were exposed to airway reopening conditions using a parallel plate microfluidic flow chamber system described previously (40). Briefly, a POC-miniperfusion system (Hemogenix, Colorado Springs, CO) was modified to create an adjustable-height, parallel-plate flow chamber (Fig. 1, A and B). The flow chamber consisted of a silicon flow channel gasket sandwiched between an upper glass coverslip and a lower coverslip seeded with EpC. For this study, flow channels had a width of 1 cm and a thickness of 500 μm. This thickness, which specifies the flow channel height, was selected based on the diameters of terminal and respiratory bronchioles (39). The flow chamber was filled with PBS using a high accuracy, PHD2000 syringe pump (Harvard Apparatus, Holliston, MA) at a rate of 3 mm/s, and control studies demonstrated that this initial filling of the chamber does not result in cell necrosis (40). Airway reopening was then simulated by withdrawing the fluid from the chamber at a constant speed of 3 mm/s. This resulted in the propagation of a single, long air bubble over the surface of the EpC. The reopening velocity of 3 mm/s was selected based on the expected range of reopening velocities during normal breathing in terminal/respiratory bronchioles (40). The procedure of filling the chamber with PBS and withdrawing the fluid at 3 mm/s was repeated five times to mimic multiple airway reopening events. We note that a larger number of reopening events can produce nearly 100% detachment which makes assessing differences between treatment conditions difficult. Therefore, five reopening events were used, since it produced ∼50% detachment and facilitated comparisons between treatment groups. PBS was used as the occlusion fluid because it represents a surfactant deficient, high surface tension fluid characteristic of the fluid found in the lungs of ARDS patients (4). In this study, cell injury experiments were performed at both room temperature (23°C) and body temperature (37°C). To maintain cells at body temperature during the cell injury experiments, the flow chamber was placed on a heating plate connected to a water circulation heater that maintains a constant 37°C within the chamber. The occlusion fluid (PBS) used during the experiment was also maintained at the corresponding experimental temperature.
Quantification of cell viability and cell adhesion.
Before exposure to microbubble flow, cells were incubated in 1 μM CellTracker Green (Invitrogen) for 15 min to fluorescently label live cells. After exposure to microbubble flow, cells were incubated with 1.2 μM ethidium homodimer for 5 min to fluorescently label cells with ruptured plasma membranes (dead cells). Both before and after microbubble exposure, six different fluorescent images were taken with a ×10 objective lens at random along the center line of the flow channel. All fluorescent live/dead images were obtained using an IX-71 inverted microscope (Olympus, Melville, NY) and a cooled SPOT RT-KI 7–33 Shot Color CCD camera (Diagnostic Instruments, Sterling Heights, MI). We note that our modified POC-miniperfusion system was fully compatible with the IX-71 platform and that all images were obtained in situ, i.e., disassembly of the system was not required to obtain live/dead fluorescent images.
The number of live and dead cells in each image was counted using the Metamorph image analysis software (Molecular Devices, Downington, PA). The percentage of cell death was determined as the number of dead cells present in each image after exposure, normalized by the total number of cells (live and dead) in each image after exposure. The percentage of cell adhesion was determined as the total number of cells present in each image after bubble exposure, normalized by the average number of cells present before exposure. One-way ANOVA, followed by post hoc t-tests, was used to determine whether statistically significant differences in cell death or cell adhesion existed among treatment groups.
Fluorescent labeling of actin and vinculin.
To visualize actin filaments and focal adhesions, cells were washed in PBS, fixed in 10% neutral buffered formalin (3.7% formaldehyde in PBS), and permeabilized with 0.2% Triton X-100. Nonspecific staining was blocked with 1% BSA. To label actin filaments, cells were incubated with Alexa-488-labeled phalloidin (Invitrogen) for 20 min at room temperature. To label focal adhesions, cells were incubated with monoclonal mouse anti-vinculin antibody (1:200, clone hVIN-1, Sigma-Aldrich, St. Louis, MO) for 1 h at room temperature, followed by 1 h incubation with rhodamine-labeled goat anti-mouse IgG secondary antibody (1:40, Jackson Immunologicals, West Grove PA) at room temperature. Following labeling for actin or vinculin, cells were washed in PBS, flooded with 0.002 mg/ml DAPI (Sigma-Aldrich) to label nuclei, and mounted on glass slides using Gelmount mounting media (Biomeda, Foster City, CA). Actin filaments were visualized by epifluorescence using an IX-71 inverted microscope (Olympus, Melville, NY) with fluorescence capabilities and a ×40 objective lens. To visualize focal adhesions, images were taken at the base of the cell (×40 objective) using a LSM 510 Meta confocal microscope (Zeiss, Oberkochen, Germany).
Optical tweezer microrheology.
In this study, we used an oscillating optical tweezer technique developed by our laboratory (38) to measure changes in the microrheological properties of alveolar EpC. For these measurements, A549 cells were cultured on 30 mm glass coverslips and placed into the modified POC-miniperfusion system. The advantage of using our modified POC-miniperfusion system for these measurements is that it could be directly incorporated onto the optical tweezer platform as shown in Fig. 1C. Specifically, the optical tweezer platform has a very small working distance between the microscope objective and a large condenser used to collect the tracking laser beam (<1 mm) and the POC-miniperfusion system was ideally suited for this application. Note that all optical tweezer measurements were performed under no-flow conditions.
The details regarding system calibration and optical techniques have been previously published (38). Briefly, 1.5-μm silica beads coated with an anti-integrin antibody were attached to the apical surface of EpC and a polarized laser beam was used to sinusoidally oscillate the bead at different frequencies. A second laser beam was used to measure the bead's displacement (D) and phase shift (δ) as a function of oscillation frequency (ω). These measurements were used to calculate the elastic modulus, G′(ω), the loss modulus G′′(ω), and the complex shear modulus |G*|(ω) = sqrt[G′(ω)2 + G′′(ω)2] [see Wei et. al. (38) for details]. In this study, raw data are presented for G′(ω) and |G*|(ω). For ω > 63 rad/s, all data followed a power-law relationship with frequency [i.e., G′ = G0′(ω/ω0)α or |G*| = G0*(ω/ω0)β] and nonlinear least squared regressions were used to determine the power-law exponents (α and β) and prefactors G0′ and G0*). In addition, a one-sample Kolmogorov-Smirnov test was used to demonstrate that α and β followed a normal distribution while G0 and G0 followed a log-normal distribution. As a result, standard ANOVA was used to document statistical differences in α and β while a log-transformed ANOVA was used to document statistical differences in G0′ and G0*.
Effect of cytoskeletal structure on cell death.
The goal of this study was to investigate how changes in cytoskeletal structure influence the susceptibility of lung EpC to injury during a single airway reopening event and during cyclic airway reopening. We first investigated how changes in cytoskeletal structure affect cell death during a single airway reopening event. Cells were pretreated with 0.5 μM jasplakinolide for 1 h to stabilize actin filaments or with 0.5 μM latrunculin A for 1 h to depolymerize actin filaments. Changes in actin cytoskeleton were verified by immunofluorescence as shown in Fig. 2, B and C. Specifically, actin filaments in cells treated with jasplakinolide (Fig. 2B) appear intact with a slightly increased actin staining intensity relative to nontreated cells. In contrast, actin filaments in cells treated with latrunculin A (Fig. 2C) are disrupted with greatly decreased F-actin staining intensity. Note that all images were captured using identical exposure settings.
Following treatment with either jasplakinolide or latrunculin A, cells were exposed to single or multiple bubble propagations at room temperature. Figure 3 shows representative fluorescence images of live and dead cells for each treatment condition before and after microbubble exposure. As shown in Fig. 4A, cells treated with latrunculin A experienced significantly less cell death than nontreated cells (7.17 ± 1.3 vs. 21.4 ± 3.2%, P = 0.001), while cells treated with jasplakinolide showed only a slight reduction in cell death relative to untreated cells, which was not statistically significant (19.0 ± 3.2 vs. 21.4 ± 3.2%, P = 0.62). Under control conditions, i.e., filling of the flow chamber with PBS only, untreated cells and cells treated with jasplakinolide or latrunculin A experienced negligible cell death. Although these experiments were conducted using confluent cells, we (40) have previously shown that stress distributions in confluent cells differ from stress distributions exerted on subconfluent cells during microbubble flows. However, additional experiments conducted using subconfluent monolayers showed similar trends in cell death with drug treatment, i.e., lower cell death with latrunculin A and no change with jasplakinoide (data not shown).
Effect of temperature on cell death.
In addition to altering cytoskeletal structure and mechanics with drug compounds, we also investigated the effects of temperature-induced changes in cell mechanics on cell necrosis during a single airway reopening event. For these experiments, cells were not treated with any cytoskeletal agents and were maintained at either room temperature (23°C) or body temperature (37°C). Immunostaining of actin fibers (Fig. 2, A and D) shows that cells at 37°C do not have significant changes in cytoskeletal structure relative to cells at 23°C.
Cells at each temperature were exposed to single or multiple bubble propagations. Figure 3 shows representative fluorescent images for each temperature condition before and after microbubble exposure. As shown in Fig. 4B, cells maintained at 37°C experienced significantly less cell death following microbubble exposure than cells maintained at 23°C (2.21 ± 0.07 vs. 21.7 ± 1.53%, P = 0.0004). At both temperatures, cells that were not exposed to a forward-propagating microbubble experienced no death.
Effect of cytoskeletal structure on cell adhesion.
In addition to cell death, another important mechanism of lung injury during cyclic airway reopening is disruption of the alveolar-capillary barrier due to the detachment of EpC. In this study, we sought to understand how drug-induced changes in cytoskeletal structure or temperature-induced changes in cell mechanics affect cell adhesion during cyclic airway reopening.
First, we investigated how depolymerizing the actin cytoskeleton with latrunculin A or stabilizing actin filaments with jasplakinolide influenced cell adhesion following both one airway reopening event and five airway reopening events. As shown in Fig. 5A, after one bubble passage, untreated cells and cells treated with jasplakinolide or latrunculin A showed no significant difference in cell adhesion after one reopening event (ANOVA, P > 0.5) and overall cell adhesion levels were very high (>90% for all treatments). Note that adhesion is reported as a percentage of the average number of cells present before bubble propagation. However, after five bubble passages, cell adhesion was substantially reduced in all groups. The percentage of cells that remained adhered was significantly greater in cells treated with either jasplakinolide (51.0 ± 2.3%, P = 0.03) or latrunculin A (64.2 ± 3.6%, P = 0.001) relative to untreated cells (37.1 ± 5.0%). In addition, cells treated with latrunculin A showed significantly greater cell adhesion relative to cells treated with jasplakinolide (P = 0.014). Of the cells that remained adhered after five bubble passages, there was also a trend toward less cell death in latrunculin A-treated cells relative to untreated cells (Fig. 5B, 29.5 ± 6.3 vs. 51.9 ± 9.7%, P = 0.07). Although there was less cell death in jasplakinolide-treated cells relative to untreated cells (Fig. 5B, 38.2 ± 6.1 vs. 51.9% ± 9.7), this difference was not significant (P = 0.26). Because it is not possible to determine whether detached cells are live or dead, the data shown in Fig. 5B do not account for necrotic cells that have become detached. However, if we assume that detached cells are necrotic, this would shift the cell death values for each treatment upward in proportion to the amount of cell detachment under that treatment. Since detachment after five bubble passages is greater in untreated cells than in jasplakinolide-treated cells and greater in jasplakinolide-treated cells than in latrunculin treated cells, including the necrosis of detached cells would only accentuate the differences in cell death between treatments shown in Fig. 5B.
While both jasplakinolide and latrunculin A treatment were found to have beneficial effects on cell adhesion, immunofluorescent detection of vinculin, a focal adhesion protein, demonstrated that these drugs have the opposite effects on focal adhesion patterns. As shown in Fig. 6, confluent and subconfluent cells treated with jasplakinolide appear to contain stronger and more numerous focal adhesions compared with untreated cells, while focal adhesions are more diffuse in latrunculin A-treated cells. Note that all images were obtained at ×40 under identical exposure settings. Thus the jasplakinolide-associated improvements in cell adhesion may be accounted for by the strengthening of cell attachments to the substrate, while latrunculin-associated increases in cell adhesion do not appear to be due to strengthening of focal adhesions.
Effect of temperature on cell adhesion.
We also investigated the effects of temperature-induced changes in cell mechanics on cell adhesion during cyclic airway reopening. Figure 7A shows the effect of multiple bubble passages on cell adhesion for cells held at 23°C or 37°C. For a single bubble passage, almost all cells remained adhered at both temperatures. However, after five bubble passages, significantly more cells remained adhered at 37°C than at 23°C (58.9 ± 5.5 vs. 40.7 ± 2.8%, P = 0.02). In addition, as shown in Fig. 7B, of the cells that remained adhered, there were significantly fewer dead cells at 37°C than at 23°C following both one bubble passage (2.21 ± 0.07 vs. 21.7 ± 1.53%, P < 0.001) and five bubble passages (8.46 ± 2.28 vs. 53.26 ± 3.36%, P < 0.001).
Optical tweezer measurements of cell mechanical properties.
As described above, altering the cell's cytoskeletal structure via latrunculin or jasplakinolide treatment and holding cells at 37°C had a significant effect on cell viability and adhesion during cyclic airway reopening. To investigate the potential biomechanical and/or microrheological mechanisms responsible for improved cell viability and adhesion, we utilized a previously developed oscillating optical tweezer microrheometer (38) to measure changes in cell mechanics after pretreating with latrunculin A or jasplakinolide or holding cells at 37°C. A representative sample of the raw data obtained from the optical tweezer technique is shown in Fig. 8, where data have been normalized with respect to the mean values of G′ and G* measured in untreated cells at ω0 = 63 rad/s, i.e., G0,untreated′ and G0,untreated*. Figure 8, A and B, shows how the elastic modulus, G′, varies as a function of the oscillation frequency, ω, while Fig. 8, C and D, shows how the magnitude of the complex shear modulus |G*| varies as a function of frequency. For ω > 63 rad/s, both G′ and |G*| consistently followed a power law relationship with frequency as shown by the solid, dashed, and dash-dotted lines in Fig. 8. These lines represent a nonlinear least-squared regression of the data to the following relationships: G′ = G0′(ω/ω0)α and |G*| = G0*(ω/ω0)β, where ω0 is a reference frequency selected to be 63 rad/s (i.e., f = 10 Hz) and G0′ and G0* represent the elastic and shear moduli at ω0. As a result, we were able to calculate the elastic and shear moduli, G0′ and G0*, and the power-law exponents, α and β, on a cell-by-cell basis. For this study, G0′, G0*, α, and β were obtained in n = 20 untreated cells, n = 11 latrunculin A-treated cells, n = 7 jasplakinolide-treated cells, and n = 9 cells held at 37°C.
Figure 9A shows how latrunculin A, jasplakinolide, and holding cells at 37°C influence the power-law exponents α and β. The mean power-law exponents ± SE were as follows: α = 0.19 ± 0.01 and β = 0.27 ± 0.01 for untreated cells at 23°C, α = 0.25 ± 0.02 and β = 0.39 ± 0.03 for latrunculin A-treated cells, α = 0.23 ± 0.02 and β = 0.27 ± 0.02 for jasplakinolide-treated cells, and α = 0.36 ± 0.04 and β = 0.47 ± 0.03 for cells held at 37°C. A one-sample Kolmogorov-Smirnov test was used to demonstrate that α- and β-values under each treatment condition followed a normal distribution (P > 0.6 for all cases). Both cells held at 37°C and treated with latrunculin A exhibited higher power-law exponents than the untreated cells held at 23°C (ANOVA, P < 0.05), while cells treated with jasplakinolide did not exhibit a statistically significant difference in power-law exponents compared with untreated cells (P = 0.22 and 0.91 for α and β, respectively). In addition, cells held at 37°C exhibited a statistically significant difference in power law exponents compared with latrunculin A-treated cells (P < 0.05).
Figure 9B shows how latrunculin A, jasplakinolide, and holding cells at 37°C influences elastic/shear moduli at ω0, G0′, and G0*. Note that data in Fig. 9B are normalized with respect to the mean values measured in untreated cells, i.e., G0,untreated′ and G0,untreated*. Unlike the power-law exponents, G0′ and G0* were found to follow a log-normal distribution. Specifically, a one-sample Kolmogorov-Smirnov test on the log′ed G0′ and G0* values indicate that for each treatment condition (i.e., untreated, latrunculin A, jasplakinolide, and 37°C) the data are log-normally distributed (P > 0.85 for all cases). As a result, “log-transformed” statistics were used to calculate the mean and 95% confidence intervals. As shown in Fig. 9B, the mean value and 95% confidence interval for the normalized elastic modulus were G0′/G0,untreated′ = 1.0 (1.21, 0.82) in untreated cells, G0′/G0,untreated′ = 0.46 (0.56, 0.37) in latrunculin A-treated cells, G0′/G0,untreated′ = 1.38 (1.69, 1.13) in jasplakinolide-treated cells, and G0′/G0,untreated′ = 0.41 (0.60, 0.28) in cells held at 37°C. The mean value and 95% confidence interval for the shear modulus were G0*/G0,untreated* = 1.0 (1.21, 0.83) in untreated cells, G0*/G0,untreated* = 0.45 (0.55, 0.37) in latrunculin A-treated cells, G0*/G0,untreated* = 1.31 (1.60, 1.07) in jasplakinolide-treated cells, and G0*/G0,untreated* = 0.41 (0.63, 0.27) in cells held at 37°C. ANOVA and post hoc tests on the log-transformed data indicate that both latrunculin A treatment and holding cells at 37°C result in a lower elastic modulus compared with untreated cells (P < 0.05). However, the elastic modulus in jasplakinolide-treated cells was not statistically different than the modulus measured in untreated cells (P = 0.39). There were also no statistically significant differences between latrunculin A and 37°C treatment conditions (P = 0.77). Similarly, latrunculin A treatment and holding cells at 37°C resulted in a lower shear modulus compared with untreated cells (P < 0.05) and the shear modulus in jasplakinolide-treated cells was not statistically different than the modulus measured in untreated cells (P = 0.48).
It is well established that ventilation of injured lungs at larger than normal volumes imparts large stretching deformations to alveolar EpC and that these large deformations can result in rupture of the plasma membrane, cell necrosis (15, 36, 37), and disruption of the alveolar-capillary barrier (20). In addition, clinical trials (1) have demonstrated a significant reduction in mortality when patients are ventilated at lower volumes. Although these low volume ventilation strategies minimize the amount of stretching deformations experienced by EpC, several studies (8, 12, 26) have demonstrated that even these low volume strategies result in significant lung damage. At low volumes, airways and alveoli become derecruited (i.e., nonventilated) due to either airway collapse (16) and/or the fluid occlusion of noncollapsed airways and alveoli (19). The cyclic closure and reopening of these collapsed and/or fluid-filled airways will involve the flow of microbubbles over the epithelium and the application of complex fluid mechanical forces to the EpC (17). Unlike high volume ventilation where substrate stretching directly imparts a known amount of deformation to EpC, the amount of cellular deformation and associated injury that will occur during the cyclic airway reopening is not known a priori and will depend on both the magnitude of the imposed fluid mechanical forces and the rheological/biostructural properties of the EpC. Although several previous studies (4, 22, 40) have investigated how different fluid mechanical forces influence cell injury during cyclic airway reopening, the goal of this study was to investigate how changes in the cell's biostructural and rheological properties influence cell injury in a model of airway reopening. We specifically hypothesized that changes in the cell's cytoskeletal structure would result in changes in both cell rheology and the cell's susceptibility to injury during airway reopening. To test this hypothesis, we used a previously developed in vitro model of airway reopening to expose EpC to microbubble flows that mimic cyclic airway reopening conditions and quantified changes in injury patterns after pretreating the cells with cytoskeletal agents that either depolymerize or stabilize the actin cytoskeleton. We also investigated how changes in the ambient temperature influence cell injury. Finally, we used a previously developed optical tweezer-based microrheometer to quantify how actin depolymerization or stabilization and changes in temperature influence the rheological properties of EpC.
Stabilization of actin filaments improves cell adhesion but not cell viability.
Jasplakinolide is a common cytoskeletal agent that can stabilize actin filaments, and in this study we demonstrated that treatment of alveolar EpC with 0.5 μM jasplakinolide for 1 h had no statistically significant effect on the amount of cell necrosis that occurred after either one or five reopening events (see Fig. 4A and 5B) but did result in an improvement in cell adhesion after five reopening events (Fig. 5A). Although jasplakinolide-treated cells exhibited an increase in actin fiber intensity (see Fig. 2B), they only exhibited a small increase in elastic modulus that was not statistically significant (Fig. 9B). To investigate the potential mechanisms for improved adhesion, we visualized changes in focal adhesion patterns by fluorescently labeling a focal adhesion protein, vinculin, and obtaining laser scanning confocal images at the base of the cell. As shown in Fig. 6A, focal adhesions appear in punctuated regions and are peripherally distributed in subconfluent cells. Stabilization of actin filaments with jasplakinolide resulted in more numerous and potentially stronger focal adhesion sites (Fig. 6B). Note that the images in Fig. 6 were obtained under identical exposure settings and that we associate higher fluorescent intensities with stronger focal adhesions. The increase in focal adhesion number and strength is consistent with the decrease in cell detachment in the jasplakinolide-treated cells. Note that the cells used to obtain the images shown in Fig. 6 were not exposed to microbubble flow conditions and that these changes in focal adhesion distribution/density are only due to treatment with the cytoskeletal agents.
Depolymerization of actin filaments improves cell viability due to fluidization.
Several previous studies (23, 30, 42) have demonstrated that depolymerizing the actin cytoskeleton results in a significant reduction in the cell's elastic modulus, i.e., a softer cell. In addition, disruption of the actin cytoskeleton impairs the cell's ability to repair the plasma membrane breaks that develop during defined stretching deformations (37). We therefore originally hypothesized that treatment with latrunculin A would result in a softer cell that would undergo more deformation during microbubble flows and thus develop more permanent membrane breaks (i.e., necrosis). However, as shown in Figs. 4A and 5B, cells treated with latrunculin A actually showed a significant improvement in cell viability after either one or five reopening events. We note that although actin disruption was previously shown to reduce cell viability during controlled stretching deformation (37), the current result of improved viability in latrunculin A-treated cells was obtained in a different system that applies a known force to the cells and where the deformation is unknown a priori. The current measurements indicate that actin depolymerization induces a change in cell rheology, which minimizes microbubble-induced cell deformation and which, in turn, prevents the initial formation of plasma membrane breaks.
To further investigate this hypothesis, we used a previously developed optical tweezer microrheometer (38) to measure the rheological properties of EpC before and after treatment with latrunculin A. Previous investigators have used a wide variety of techniques, including atomic force microscopy (2), magnetic twisting cytometry (29), laser tracking microrheometry (41), and optical magnetic twisting cytometry (33), to measure the rheological properties of biological cells. These techniques quantify cell rheology by measuring the frequency dependence of the cell's elastic (G′) and complex shear (|G*|) moduli. Measurements in several different cell types have consistently demonstrated that G′ and |G*| follow a weak power-law relationship over several decades of frequency. These data can be used to quantify both the stiffness and viscoelasticity of the cells. First, the magnitude of G′ and |G*| is a measure of the cell's stiffness where smaller G′ and |G*| values represent softer cells that would undergo more deformation under a constant load. Second, the frequency dependence of G′ and |G*| is a measure of the cell's viscoelastic properties. For example, previous investigators (13) have used structural damping and soft glassy rheology models to analyze the power-law dependence of G′. In these models, changes in the power-law exponent (α) represent a transition from solid-like behavior (α = 0) to fluid-like behavior (α = 1). Specifically, materials with α = 0 are purely elastic and deform instantaneously to an applied load. By contrast, a fluid-like material does not deform instantaneously and a more fluid-like material (i.e., larger α) will deform slower than a less fluid-like material (smaller α).
As shown in Fig. 8, our optical tweezer technique measured a weak power-law relationship with frequency for both G′ and |G*| for all treatment conditions and is therefore consistent with previous studies (13, 18). To quantify cell stiffness and viscoelasticity, we correlated several sets of data similar to those shown in Fig. 8 with power-law relationships: G′ = G0′(ω/ω0)α and |G*| = G0*(ω/ω0)β, where G0′ and G*0 are measures of cell stiffness and α and β are measures of the cell's viscoelasticity (i.e., solid-like or fluid-like behavior). As shown in Fig. 9B, latrunculin A-treated cells exhibit smaller G0′ and G0* values and are therefore “softer” than the untreated cells. However, Fig. 9A indicates that the latrunculin A-treated cells also exhibit larger power-law exponents (α and β) than the untreated cells and are therefore more fluid-like (i.e., fluidized). We note that previous investigators (23) have also measured a similar decrease in elastic modulus and increase in power-law exponents after treatment with latrunculin A in airway smooth muscle cells. In addition, the slight increase in elastic/shear modulus and no change in power-law exponents for jasplakinolide-treated cells shown in Fig. 9 are also consistent with the previous measurements in airway smooth muscle cells (23).
Although the “softer” latrunculin A-treated cells would be expected to deform more than the untreated cells under constant loading conditions, the hydrodynamic forces exerted by the microbubble on the EpC are not “constant” since they are rapidly applied and removed (22), i.e., the forces are time dependent. Under these conditions, the more fluid-like latrunculin A-treated cells, which deform slower than the less fluid-like untreated cells, may not have time to respond and deform to the rapidly applied hydrodynamic forces exerted by the microbubble. Since the amount of cell necrosis due to plasma membrane rupture is a direct function of the amount of cell deformation (34, 36), the data in Figs. 4, 5, and 9 support the hypothesis that the more fluid-like latrunculin A-treated cells are able to “damp out” the rapidly applied hydrodynamic forces, undergo less deformation than untreated cells, and therefore experience less plasma membrane rupture and necrosis during microbubble flows.
Depolymerization of actin filaments improves cell adhesion due to fluidization.
In addition to altering the amount of cell necrosis after cyclic airway reopening, Fig. 5A demonstrates that treatment with latrunculin A also results in a statistically significant improvement in cell adhesion. Unlike treatment with jasplakinolide, latrunculin A-treated cells exhibited more diffuse focal adhesions (Fig. 6C). As a result, we initially hypothesized that latrunculin A-treated cells would experience more cell detachment during microbubble flows. However, as discussed above, latrunculin A-treated cells are significantly more fluid-like than normal cells. As a result, latrunculin A-treated cells, which have increased viscous damping properties, may not effectively transmit the hydrodynamic forces applied at the apical surface of EpC to the basally located focal adhesion sites. As a result, even though latrunculin A-treated cells may have weaker physical attachments with the substrate, their increased viscous damping properties may prevent detachment by interfering with force transmission from the apical to the basal surface.
Fluidization of actin cytoskeleton at 37°C improves cell viability.
The results described above clearly indicate that changes in cytoskeletal structure due to actin depolymerization prevent EpC injury and detachment during airway reopening. In addition, our rheological measurements indicate that these improvements in cell viability and adhesion were due to the more fluid-like mechanical properties of the latrunculin A-treated cells. To further investigate the effect of fluidization on cell injury and detachment during airway reopening, we compared experiments conducted at room temperature (23°C) and body temperature (37°C). Although holding cells at 37°C had no effect on cytoskeletal structure (Fig. 2), cells held at 37°C exhibited significant changes in both elastic and shear moduli, G′0 and G*0, and power-law exponents, α and β (Fig. 9). First, cells at 37°C were “softer” than untreated cells but exhibited similar values of G′0 and G*0 compared with latrunculin A-treated cells (Fig. 9B). We note that this decrease in cell stiffness with increasing temperature is consistent with previous studies (36). In addition, cells at 37°C also exhibited larger power-law exponents and are therefore more fluid-like than both the untreated and latrunculin A-treated cells (Fig. 9A). This fluidization at 37°C resulted in significantly less cell necrosis compared with both untreated and latrunculin A-treated cells (Fig. 5) as well as less cell detachment during cyclic airway reopening (Fig. 7). These results at 37°C therefore strongly support the protective mechanism of “fluidization,” i.e., cells with more fluid-like properties undergo less deformation, injury, and detachment during microbubble flows.
The current study utilized the cytoskeletal agents jasplakinolide and latrunculin A to investigate the hypothesis that the amount of cell necrosis and detachment that occurs during cyclic airway reopening depends on the EpC rheological and biostructural properties. However, jasplakinolide and latrunculin A are not clinically relevant compounds and future studies should therefore investigative how clinically relevant pharmaceutical agents that have similar effects on the cytoskeleton influence cell injury/detachment during airway reopening. Another limitation is that in the current study EpC were cultured on a rigid glass substrate. To better mimic the collapsibility of pulmonary airways, future studies should expose EpC to microbubble flows in a flexible-walled microfluidic system and should investigate how altering substrate elasticity influences both cell injury/detachment during airway reopening and the microrheological properties of the EpC. Finally, we note that computational techniques may be required to further investigate the hypothesis that “fluidization” of the cytoskeleton results in less cell deformation and detachment during airway reopening. Specifically, it is difficult to measure the rapid (and small) cell deformations that occur during microbubble flows. Computational models of this system could provide quantitative predictions of cell deformation and could therefore be used to further investigate the effect of cytoskeletal fluidization.
In conclusion, in this study we utilized an in vitro cell culture system to investigate how changes in cytoskeletal structure and microrheological properties influence the necrosis and detachment of EpC during cyclic airway reopening. Results indicate that the stabilization of actin filaments does not improve cell viability but does improve cell adhesion due to an increase in focal adhesion size/strength. Disruption of the actin cytoskeleton resulted in significantly less cell necrosis and detachment during cyclic airway reopening and also resulted in a “softer” and more fluid-like cell. Furthermore, cells held at body temperature also exhibited significantly less cell necrosis and detachment during cyclic airway reopening as well as “softer” and more fluid-like rheological properties. These experimental data strongly indicate that cells with more fluid-like properties dissipate the rapidly applied hydrodynamic loads generated during airway reopening and therefore undergo less deformation, necrosis, and detachment. The implication of this study is that fluidization of the cytoskeleton may be a novel way to reduce the susceptibility of EpC to microbubble-induced injury during cyclic airway reopening.
This research was supported by National Science Foundation CAREER Grant No. 0852417, a Beginning Grant-In-Aid from the American Heart Association, and a Parker B. Francis Fellowship in Pulmonary Research to S. N. Ghadiali.
- Copyright © 2009 American Physiological Society