Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

During the course of mechanical ventilation, ~ 5-15% of patients present with enhanced pulmonary edema as a result of this treatment (1). This malady, known as ventilator-induced lung injury (VILI), has an associated mortality rate of 34-60% (2, 3). Extensive animal experimentation has shown that ventilation with high lung volumes or airway pressures induces air leaks, acute respiratory failure, pulmonary edema, and alveolar cell dysfunction (for review, see Ref. 4). In particular, ventilation with high tidal volumes has been demonstrated to cause increased edema (4) and increased mortality in clinical studies (8). Alveolar epithelial injury or dysfunction alone can potentiate alveolar flooding even in the presence of normal pulmonary microvascular pressure, plasma oncotic pressure, and endothelial permeability (9). Thus, the integrity of the alveolar epithelium is believed to play a critical role in the development of VILI.

The epithelial lining of the alveoli of the healthy lung provides the dominant barrier to passage of proteins and more than 90% of the total resistance to the passage of hydrophilic solutes (10). If this barrier were to be compromised, solute diffusion would occur unchecked into the alveolar airspace. Water would then passively follow into the airspace along the osmotic gradient, causing the existence or worsening of pulmonary edema as seen in VILI. In animals, increased overall or regional lung inflation resulted in increased effective epithelial pore sizethat is, the lungs became more permeable to macromolecules (11). These changes were irreversible at high inflation volumes, consistent with behavior observed clinically in VILI cases.

The low permeability of the healthy pulmonary epithelium to paracellular solute transport is due to the presence of tight junctions (TJ), belt-like interconnected junctional strands of protein connecting alveolar cells and separating the apical and basolateral surfaces of the epithelium (Figure 1). These structures are thought to provide the primary rate-limiting barrier to paracellular transport (10, 14). Two of the best characterized proteins of the TJ are occludin, a transmembrane protein believed to provide the majority of the barrier function of the tight junction (18, 19), and zonula occludens (ZO)-1, an intracellular protein putatively located between occludin and cytoskeletal proteins, also thought to affect paracellular permeability (20). Recently, the claudin family of TJ proteins has been found to contribute to paracellular barrier function in kidney cell lines, but no claudins have been found in significant quantity in the pulmonary alveolar epithelium (23).

View larger version (10K): [in this window] [in a new window]   Figure 1.   Tight junction (TJ) structure. (A) Schematic of TJ in cell showing its apical location and proximity to the actin cytoskeleton. (B) Closeup of TJ. The TJ is thought to consist of several strands of connected proteins in parallel. (C) Putative protein structure of TJ. Notice that occludin is a transmembrane protein known to associate with ZO-1. ZO-1 is itself believed to associate with F-actin, either directly or indirectly. Many other TJ proteins have been identified; however, only proteins relevant to this study are shown. Modified from Mitic and Anderson (58).

TJ physiology has been investigated in great detail, primarily in the epithelia of the gastrointestinal tract and kidney (for recent review articles, see Refs. 26-28). Two of the chemical stimuli known to result in TJ dissociation are low extracellular calcium or low intracellular ATP (ATP_i) concentrations (29, 30). In addition, disruption of the actin cytoskeleton has been shown to increase paracellular permeability in cultured cells (31). Unfortunately, there is a paucity of data in the literature regarding the effects of mechanical forces on TJ structure and function. Only Pitelka and Taggart investigated the effects of physical loads on TJ structure, noting alterations in strand arrangement and orientation when mammary cells were under tension (32). Despite the importance that the alveolar epithelium is believed to play in the maintenance of fluid balance between the alveolar and interstitial spaces, the role of mechanical injury on epithelial TJ integrity has never been studied in cultured pneumocytes experiencing well-controlled physical stimuli.

Previously, we have demonstrated that 1 h of cyclic stretch at 25% and 37% change in surface area (SA) resulted in 3.9 ± 3.4% and 8.9 ± 3.9% cytotoxicity in rat alveolar epithelial cells maintained in primary culture for 5 d in 10% fetal bovine serum. These stretch magnitudes correspond to the strains experienced by the alveolar epithelium at tidal volumes of 80% and 100% total lung capacity (TLC), respectively (33). However, 1 h of cyclic 50% SA resulted in 49.3 ± 10.4% cell death (34). Thus, in these cells there exists a viability threshold between 37% and 50% SA, which corresponds to a point above 100% TLC. Epithelial barrier function was not examined in these cells, however.

The goal of this study was to evaluate the effect of physiologically relevant magnitudes of equibiaxial strain on TJ morphology and oxidative metabolism in the alveolar epithelium as a means of determining the mechanism(s) by which mechanical ventilation can induce or enhance VILI. Our hypothesis is that alveolar strain associated with large tidal volumes causes disruption of the TJ structure, possibly resulting in decreased degree of cell-cell attachment (CCA), and causes a decrease in ATP_i levels. We show that the tight junctions of cultured alveolar epithelial cells respond to applied strain in a dose-dependent manner, decreasing in degree of attachment and immunofluorescent intensity at mechanical strains above a certain deformation threshold. In addition, the cultured cells demonstrate a decrease in ATP_i with stretch. Disruption of glycolytic metabolism and the actin cytoskeleton in the absence of stretch also resulted in atypical TJ structure. In cells whose TJ morphology was altered by strain, total expression of occludin was significantly decreased as well.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture Protocol

Alveolar type II cells were isolated from healthy male Sprague-Dawley rats (180-200 g) according to a modification of the method of Dobbs and coworkers (35). This protocol was approved by the University of Pennsylvania Institutional Animal Care and Use Committee. The rats were anesthetized with pentobarbital sodium (50 mg/kg body wt intraperitoneally). The trachea was cannulated, the lungs were mechanically ventilated, an abdominal aortotomy was performed to exsanguinate the animal, and excess blood was removed via pulmonary arterial perfusion. The lungs were then excised and type II cells were isolated using an elastase digestion technique, in which the lungs are instilled and incubated with an elastase solution (3 U/ml; Worthington Biochemical, Lakewood, NJ) and then minced in the presence of DNase (Sigma Chemical, St. Louis, MO) with a tissue chopper (Sorvall, Newtown, CT). The elastase reaction was stopped with fetal bovine serum (Life Technologies, Rockville, MD). Cells were filtered through progressively finer Nitex mesh (Crosswire Cloth, Bellmawr, NJ), and plated on an IgG-coated culture dish (3 mg · 5 ml Tris-HCl¹). After a 1-h incubation at 37°C, gentle panning isolated type II cells from the macrophages, and contaminating cells preferentially adhered to the culture dish. Ultimately, cells were spun down and resuspended in minimum essential medium (MEM) with Earle's salts and supplemented with 10% fetal bovine serum, 25 µg/ml Gentamicin, and 0.25 µg/ml Amphotericin B (Life Technologies).

All cells were seeded at a density of 1 × 10⁶ cells/cm² onto fibronectin-coated (10 µg/cm²; Boehringer Mannheim Biochemicals, Indianapolis, IN) flexible Silastic membranes (Specialty Manufacturing, Saginaw, MI) mounted in custom-made wells. The cells were cultured in MEM supplemented as above for 5 d. It has been previously reported that rat alveolar type II cells differentiate into type I cells (which compose over 95% of the alveolar epithelium in vivo) after 5 d under similar culture conditions (36). The media was replaced daily. By the fifth day of culture, the cells had formed a confluent monolayer (Figure 2) and displayed a phenotype consistent with that observed for cultured type I cells. At this time, the cells were washed with dye-free Dulbecco's modified Eagle's medium (DMEM; Life Technologies) and subjected to protocols relating to either TJ morphology, ATP_i determination, or Western analysis, each of which is described below. All experimentation was performed at 37°C in room air. Unstretched, untreated wells served as controls in all experiments.

View larger version (169K): [in this window] [in a new window]   Figure 2.   Phase microscopy picture of Day 5 cultured alveolar epithelial cells. This photograph was taken at 20× magnification. Bar = 100 µm. The confluence of the cells in this image is representative of all wells used in experimentation.

Four isolations were performed for each type of experiment. For the TJ morphology and ATP_i analysis experiments, the control and each treatment group consisted of 2-3 wells from each isolation (8-12 wells total for each group). For Western analysis, control and treatment groups consisted of 7 wells from each isolation (28 wells total for each group).

TJ Morphology Studies

Cells were either used as no-treatment controls or treated as follows. Two populations of wells were mounted onto a custom-built cell-stretching device capable of applying equibiaxial strain to the samples at a precise user-defined magnitude and frequency (34). The cells were stretched at 15 cycles/min for 1 h, one group of wells at 25% SA and another at 37% SA, which corresponds to strains experienced by the epithelium at 80% and 100% TLC, respectively (33). Other groups of wells were incubated for 1 h in Hanks' balanced salt solution (HBSS) containing the F-actin altering agent latrunculin A (70 nM, Sigma), which sequesters monomeric G-actin and thus interrupts F-actin polymerization, or the glycolytic metabolic inhibitors 2-deoxy-D-glucose (2 mM; Sigma) and antimycin A (10 µM; Sigma) which have been shown to interrupt cellular glucose metabolism, thus depleting the cells of ATP (29). This method of ATP depletion has been shown to reduce intracellular ATP levels by 45-90% in similarly cultured alveolar epithelial cells (37). Incubation of cells in HBSS alone did not result in any significant cytological changes for over 1 h (data not shown).

After 1 h of treatment or stretch, the cells were washed 2× with phosphate-buffered saline (PBS) and fixed for 20 min in 4% paraformaldehyde. The cells were then washed again 3× in PBS, and treated for 5 min with 0.1% Triton X-100 in PBS to permeabilize the cells. After washing the cells 3× in PBS, the cells were incubated for 30 min in 5% normal goat serum (NGS; Jackson ImmunoResearch, West Grove, PA) in PBS to block nonspecific binding sites. The cells were then incubated overnight with 2.5 µg/mL monoclonal mouse anti-occludin and 1.0 µg/mL polyclonal rabbit anti-ZO-1 (Zymed Laboratories, South San Francisco, CA) in 1% NGS in PBS. After washing the cells 3× in PBS, the cells were incubated for 2 h with fluorescein isothiocyanate- goat antimouse and Rhodamine Red X-goat antirabbit (Jackson ImmunoResearch) in 1% NGS in PBS, followed by another 3× PBS wash. The Silastic membrane on which the cells were grown was then mounted onto a standard microscope slide and the cells were analyzed as described below. Two additional wells from the control and latrunculin-treated groups were fixed as described above and stained with Oregon Green 488-phalloidin (Molecular Probes, Eugene, OR) to visualize F-actin.

Slides were observed under 60× magnification using a Nikon TE-300 inverted microscope. Two fields from each slide were selected randomly, and two images of each field viewed using a red (ZO-1) or green (occludin) emission filter were captured and stored using Metamorph imaging software (Universal Imaging, West Chester, PA) and a microscope-mounted Hamamatsu camera and controller. Identical image acquisition times were used for all images acquired. The intensity and distribution of the fluorescent band at the periphery of the cells, the fluorescence of the cell interior, and the degree of CCA were scored for each image. Scoring consisted of rating each of these attributes higher, lower, or equal to that of the corresponding controls. The determination of the degree of CCA was a binary decision, based on the number of visible spaces between adjacent cells in each field. If a field containing stretched or treated cells displayed any visible intercellular spaces, it was considered to have less CCA. The observed parameters of stretched or chemically treated cells were compared with those of unstretched, untreated control samples from the same experiment using the Wilcoxon signed-rank test for statistical significance (40). Significance was defined as P  0.05.

Stretch-Induced Intracellular ATP Determination

Cells were either used as no-stretch controls or mounted onto the custom-built cell-stretching device described above. The cells were stretched at 15 cycles/min for 1 h at 25% or 50% SA, corresponding to epithelial strains at tidal volumes of up to 80% and >100% TLC, respectively (33). After stretch, the cells were lysed using 0.1% Triton X-100 in buffered solution. The cells were scraped from the Silastic membrane and vortexed. A 50-µL sample of cell lysate from each well was then added to an equal volume of firefly luciferase reagent and an additional 400 µL of ATP buffer. The luminescence was measured immediately (Chronolog LumiAggregometer, Chronolog Corp., Havertown, PA) and ATP content determined by comparing with samples of known ATP concentration. The remainder of the lysate was used for total cellular protein quantification using the colorimetric method of Bradford (41).

ATP_i was normalized with respect to total cellular protein for each sample. Average ATP_i per protein values for each magnitude of stretch were then normalized with respect to those of unstretched samples from the same experiment. The normalized ATP values were analyzed using an F-test to determine if the mean of each group was significantly different from that of the control group. Data from the two stretch groups were also compared using the Tukey-Kramer multiple comparison test.

Western Analysis

Cells were either used as no-stretch controls or mounted onto the custom-built cell-stretching device described above and stretched at 15 cycles/min for 1 h at 25% SA or 37% SA, corresponding to epithelial strains at tidal volumes of 80% and 100% TLC, respectively (33). The cells were then scraped from the Silastic membrane in the presence of 2% sodium dodecyl sulfate (SDS) solution supplemented with the Complete protease inhibitor cocktail (1:24; Boehringer Mannheim Biochemicals). This suspension was then homogenized by passing it 5× through a high-gauge pipet, and the homogenate was then boiled for 10 min. A portion of this homogenate was treated with the DetergentOut detergent removal kit (Geno Technologies, St. Louis, MO) to remove SDS, and the detergent-free portion of the cell lysate was used for total protein quantification as described above.

The remainder of the cell homogenates was analyzed by 10% SDS-polyacrylamide gel electrophoresis (PAGE) (200 V for 60 min), and the resolved proteins were transferred electrophoretically onto polyvinylidene difluoride (PVDF) membranes (70 V for 90 min). The membranes were incubated for 1 h in PBS containing 5% powdered milk and 0.1% Tween-20 (PBS-MT) to block nonspecific binding. The PVDF membranes were then incubated overnight at 4°C in the presence of the mouse antioccludin primary antibody mentioned above (1:1500 in PBS-MT). After washing the membranes 3× in 0.1% Tween-20 in PBS (PBS-T), the membranes were incubated for 1 h with HRP-conjugated donkey antimouse secondary antibody (1:1500; Jackson ImmunoResearch). The PVDF membranes were again washed 3× with PBS-T, and developed using enhanced chemiluminescence (Amersham Life Sciences, Arlington Heights, IL).

The developed film from each experiment was digitized (Personal Densitometer SI; Molecular Dynamics, Sunnyvale, CA), and average intensity of each band was determined using the image analysis software package MCID (Imaging Research, Inc., St. Catharine's, ON, Canada). To normalize for variations in Western band densities due to differences in total protein in each lane, the intensities were divided by the total lysate protein content for each sample. The normalized occludin levels of the stretched cells were compared in a pairwise fashion with those of unstretched cells from the same isolation using Student's t test for statistical significance.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Determination of Morphologic Changes in TJ Proteins after Stretch or Treatment

Immunofluorescent staining of the TJ proteins occludin and ZO-1 of unstretched, untreated cells which had been in culture for 5 d resulted in a strong signal for both proteins (Figures 3A and 3B). The staining was located primarily at the periphery of each cell in all fields examined. This is consistent with previous findings in other cell types (18, 42), and with the reputed role of these proteins as important regulators of TJ permeability.

View larger version (53K): [in this window] [in a new window]   Figure 3.   Occludin (A, C, E, G, and I) and ZO-1 (B, D, F, H, and J) distribution after stretch or treatment. (A and B) Unstretched, untreated wells. (C and D) Wells stretched at 25% SA. (E and F) Wells stretched at 37% SA. (G and H) Unstretched wells after disruption of F-actin. (I and J) Unstretched wells after depletion of ATP. Bar = 10 µm. White arrows point to regions of cell-cell detachment. All images are representative of their particular experimental group.

Comparison between control samples and those that had been stretched for 1 h at 25% SA revealed no significant cytologic differences (Figures 3C and 3D). Stretch at 37% SA, however, produced a peripheral occludin band that was significantly fainter than that of the unstretched group (Figures 3E and 3F; Table 1). Additionally, wells stretched at this magnitude exhibited a decrease in the degree of CCA. ZO-1 distribution was not affected by either magnitude of stretch. There was no evidence of cell death in fields examined at either stretch magnitude as measured by ethidium homodimer exclusion, consistent with previously published results (34) (data not shown).

Treatment for 1 h with 70 nM latrunculin A resulted in a peripherally concentrated F-actin distribution, similar to that seen after 1 h of cyclic deformation at 25% SA (Figure 4). Actin disruption did not affect the intensity of staining of either TJ protein examined, but CCA decreased significantly in the treated wells (Figures 3G and 3H; Table 1). The occludin distribution in the latrunculin A-treated wells was very punctate and intense in regions of the cell where more than two cells remained in close proximity.

View larger version (70K): [in this window] [in a new window]   Figure 4.   Actin distribution in cultured alveolar epithelial cells after stretch or perturbation of F-actin polymerization. (A) Untreated cells. (B) Cells after 1 h incubation in 70 nM latrunculin A. (C) Cells after 1 h cyclic stretch at 25% SA (51). Bar = 20 µm.

One hour of ATP depletion with 2 mM 2-deoxy-D-glucose and 10 µM antimycin A resulted in a weaker peripheral occludin band in the treated wells (Figures 3I and 3J). CCA was also significantly reduced in these wells. Peripheral occludin distribution was punctate in ATP-depleted wells, and this group also demonstrated a higher occludin staining in the cell interior (i.e., away from the periphery of the cell).

Intracellular ATP Determination

After 5 d in culture, cells were stretched at 25% SA or 50% SA for 1 h or served as unstretched controls, and the intracellular ATP per mass total cellular protein after strain was determined. The results after stretch were normalized with respect to paired, unstretched controls from the same isolation. One hour of cyclic strain at 25% SA caused an average decrease in intracellular ATP of over 40%, compared with unstretched controls (Figure 5). A change in surface area of 50% resulted in an average ATP_i decrease of over 75% compared with controls. This decrease at 50% SA is also significantly different from that observed at 25% SA, indicating that the decrease in ATP_i is dependent on the magnitude of applied strain.

View larger version (22K): [in this window] [in a new window]   Figure 5.   Total moles ATPi per mass total cellular protein (normalized by unstretched value) after cyclic strain. Two strain magnitudes, resulting in 25% and 50% SA, were examined. Results are given as mean ± standard error. *Significantly different from controls (P < 0.01). **Significantly different from controls (P < 0.0001), significantly different from 25% stretch (P < 0.05).

Results of Western Analysis

SDS-PAGE of cell lysate from stretched and unstretched cell populations followed by incubation with a monoclonal occludin antimouse antibody resulted in a band of ~ 65 kD molecular weight (Figure 6). This corresponds to the published molecular weight of occludin studied in other cell types (43). An additional weaker band of ~ 50 kD molecular weight was also expressed. The average intensity of these bands were determined for the control, 25% SA, and 37% SA populations from each experiment and normalized with respect to total cellular protein content for each sample. Application of 25% SA strain did not result in a statistically significant change in cellular occludin content. However, an applied cyclic biaxial strain of 37% SA did produce a statistically significant (P < 0.01) decrease in normalized intracellular occludin levels, to 49% of unstretched control levels (Figure 7).

View larger version (43K): [in this window] [in a new window]   Figure 6.   Western analysis of stretched and unstretched cell lysates. (A) Unstretched cells. (B) Cells stretched at 25% SA. (C) Cells stretched at 37% SA. Each population displayed an intense protein band ~ 65 kD in size. A fainter band ~ 50 kD in size was also detected in all cells.

View larger version (20K): [in this window] [in a new window]   Figure 7.   Total expression of occludin per mass total cellular protein (normalized by unstretched value) in cells after 1 h. (A) Unstretched cells. (B) Cells stretched at 25% SA. (C) Cells stretched at 37% SA. Results are given as mean ± standard error. NS, no significant difference from controls. *Significantly different from controls and from 25% SA (both P < 0.01).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Our immunocytologic data indicate that healthy rat alveolar epithelial cells displaying a type I phenotype (36) express an intense peripheral distribution of the TJ proteins occludin and ZO-1 after 5 d in culture. This result agrees with the work of other investigators, who have found strong peripheral occludin staining in pulmonary epithelial cells cultured for a shorter period of time (3 d) (44). It has also been shown that alveolar epithelial cells cultured for a similar length of time possess functional TJs, as determined by high (> 1000  · cm²) transepithelial resistance (TER) measurements, which are commonly considered as inverse measures of paracellular permeability (45). Finally, our observation that occludin and ZO-1 appear in close proximity in cultured alveolar epithelial cells is consistent with the proposal that the protein ZO-1 acts as a linkage between the TJ and the cytoskeleton (46).

Large inflation volumes have long been known to increase the permeability of the otherwise healthy alveolar epithelium. Kim and Crandall found that quasi-static inflations within the physiologic range did not affect the transport properties of isolated bullfrog lungs, while overinflation of the lung resulted in a significantly increased equivalent pore radius, indicative of an impairment in epithelial barrier function (47). In other experiments, Egan and coworkers discovered that quasi-static regional overdistension of the rabbit lung resulted in an increased epithelial protein permeability (11). Another finding of this group was that the equivalent pore radius of adult sheep increased with increasing quasi-static inflation volume, up to 100% TLC, at which point barrier function was completely eliminated (13). In dogs, the epithelial pore radius was found to be significantly different at 47% and 82% TLC, and again there was a failure of barrier function at 100% TLC (12). Similarly, large-scale clinical trials have suggested that low-volume mechanical ventilation reduces mortality in patients with acute or chronic lung damage (8, 48). All of these findings indicate that high-volume ventilation is detrimental to healthy solute and fluid balance. However, at the present time, the exact mechanisms by which large lung inflations are transduced into a worsening of barrier function are unknown.

Previously, we demonstrated that, on average, the alveolar epithelium stretched 58% between functional residual capacity and TLC (33). In this communication, we hypothesize that sustained cyclic epithelial stretch at large amplitudes adversely affects TJ structure. Our results show that 1 h of cyclic applied epithelial strain at 25% SA did not produce significant alterations in the distribution of occludin or ZO-1, the degree of CCA, or in total intracellular occludin content, compared with unstretched cells. This is the magnitude of strain experienced by the alveolar epithelium in the whole lung at 80% TLC (33). However, stretch at 37% SA did cause a decrease in peripheral occludin staining, as well as a decrease in the degree of CCA. This stretch corresponds to that experienced by the alveolar epithelium at 100% TLC (33). The TJ structural results at 37% SA stretch suggest that TJ function may also decrease after strain. In fact, Egan and colleagues found that quasi-static inflation to this same percent of lung capacity eliminated alveolar epithelial barrier function in dogs and sheep (12, 13).

Previous studies have been performed to determine the relationship between TJ permeability and TJ structure, specifically occludin expression. In Madin-Darby canine kidney (MDCK) cells, a cell line commonly used for TJ and permeability studies, overexpression of occludin led to increased TER, while introduction of mutant occludin truncated at its COOH terminus resulted in increased paracellular permeability in these cells (19). COOH-terminated occludin also led to decreased TJ "fence" function, resulting in increased diffusion between the apical and basolateral domains of the plasma membrane (49). Because separation of these domains is required for proper solute transport, disruption of the membrane-bound occludin band can affect not only paracellular but also transcellular permeability as well.

In addition to the previously mentioned immunocytologic alterations, overall cellular expression of occludin decreased significantly at 37% SA stretch, as determined by Western analysis. The mean decrease in stretched cells was 51% of unstretched values. This result agrees with the observed reduction in perijunctional occludin immunofluorescent intensity after stretch. Together, these data indicate that occludin is not merely internalized after stretch, but that the protein itself is altered. Mechanical stretch may possibly cause cleavage of the occludin molecule, or at least alter its conformation enough so that its ability to tightly bind cells together is diminished.

These findings indicate a possible strain magnitude threshold below which the TJ remains intact, but above which the TJ can be damaged, resulting in less CCA and, therefore, a leakier epithelium. Previously, we have reported that 1 h of cyclic stretch at 25%, 37%, and 50% SA resulted in 3.9 ± 3.4%, 8.9 ± 3.9%, and 49.3 ± 10.4% cell death in similarly cultured cells (34). The percentage of dead cells at 37% SA is not sufficient to account for the 51% decrease in occludin observed at this stretch magnitude. Thus, these data indicate that alveolar epithelial cells tolerate mechanical stretch up to a certain point, above which physiologic dysfunction results (37% SA), followed by cell death at even higher magnitudes (> 37% SA).

One pathway by which TJ structure appears to be regulated involves the actin cytoskeleton. The cytoskeleton is one of the principal pathways by which external mechanical stimuli are transduced into a physiologic response by the cell (50, 51). We have shown that 25% SA stretch alters the actin distribution in these cells (52) (Figure 4). Previous investigations established that the cytoskeleton associates with the cytoplasmic domain of the TJ (20). Addition of cytochalasin D (believed to fragment existent actin microfilaments) has been shown to rapidly decrease TER and the number of visible TJ strands and increase transepithelial mannitol transport in kidney, gallbladder, and intestine epithelia (31). Thus, we hypothesized that cytoskeletal perturbation associated with stretch may affect the ability of the TJ of one cell to remain bound to those of its neighbors.

We perturbed the actin cytoskeleton using latrunculin A, which is somewhat more specific in its action than the cytochalasin family. Latrunculin A does not dismantle the actin network directly; it instead sequesters monomeric G-actin. Thus, further formation and repair of polymeric F-actin fibers is inhibited. Disruption of the actin cytoskeleton decreased CCA and resulted in a more heterogeneous occludin distribution. Occludin was especially concentrated in regions of contact of three or more cells (see Figure 3G). It is worth noting that, although we have shown that 25% SA affects actin distribution (Figure 4) but does not affect TJ morphology, the latrunculin A treatment may perturb the actin distribution in a more severe or different manner than does 25% SA stretch, thus altering TJ morphology.

Proper TJ formation, maintenance, and physiology may also rely on sufficient ATP_i (53, 54). It is well documented that removal of cytoplasmic ATP from epithelial cells (typically through blockage of glycolytic metabolic pathways) quickly results in dissociation of TJ proteins and a rapid and dramatic decrease in TER (29). Obviously, many other cellular processes are also energy-dependent and therefore require ATP.

Depletion of ATP in our cultured epithelia through the disruption of cellular glucose metabolism resulted in drastic changes in TJ structure. The degree of CCA was reduced, while peripheral occludin distribution was faint and often punctate, indicating a discontinuous TJ band. These results are consistent with previously reported effects of ATP depletion. A particularly interesting result was that most of the fields viewed displayed intense staining for occludin in the cytoplasm, away from the plasma membrane. This agrees with the results of Tsukamoto and Nigam, who showed similar staining in identically treated MDCK cells (53, 55). These investigators postulated that ATP depletion causes an internalization of occludin from the periphery to an intracellular site. Despite the major changes in occludin distribution, ZO-1 distribution was not significantly altered in the fields studied. Previously, ATP depletion had been shown not to significantly affect ZO-1 localization in MDCK cells (55).

Chemical modulation is not the only way to reduce intracellular ATP. Our results show that 1 h of mechanical stretch at two different magnitudes reduces ATP_i in cultured cells. This decrease was magnitude-dependent, ranging from a 40% decrease at 25% SA to an over 75% decrease at 50% SA. In comparison, 1 h of treatment with 2-deoxy-D-glucose and antimycin A (as performed in this experiment) has been shown to reduce ATP_i levels to 15% of untreated levels in MDCK cells (29) and to 10-55% in alveolar epithelial cells (37). Given that ATP is required for nearly every cellular process, the more modest decreases occurring with stretch may contribute to the alveolar cell dysfunction seen by other investigators at high inflation volumes in vivo, such as decreased Na⁺,K⁺-ATPase activity (56). This stretch-induced ATP_i decrease may be a reason for the perturbed occludin distribution and decreased CCA seen at 37% SA stretch; but interestingly, despite the 40% decrease in ATP_i seen at 25% SA, there were no immunocytochemical changes observed at this stretch level. It is possible that the cells are able to maintain normal TJ structure even with reduced ATP_i, but that an ATP_i threshold exists below which TJ structure and other critical cell functions are disrupted. According to our data, this threshold would exist somewhere between 40% and 75% of normal unstretched ATP_i concentration for cultured alveolar epithelial cells.

Because of the relative lack of available information regarding the relationship between mechanical forces and TJ structure, it is difficult to compare these findings with the works of others. The only previous data on the effect of mechanical stretch on TJ structure come from Pitelka and Taggart, who demonstrated that applied uniaxial stretch of cultured mammary epithelia resulted in altered TJ strand orientation, as visualized using electron microscopy (32). Strands that demonstrated prolific undulation before stretch became straighter and more orderly after strain application. These observations are only structural in nature, however, with no functional measurements.

At present, no functional tests correlating mechanical strain and TJ permeability have been performed with cultured epithelial cells. Although the findings presented here, together with the previous findings of other investigators as described above, may indicate that perturbation of occludin distribution after stretch results in increased paracellular permeability, this is not implicit. Saitou and coworkers have found that disruption of both alleles for the occludin gene in embryonic stem cells did not affect paracellular permeability, as determined by surface biotinylation assay (57). Therefore, the necessary next step in this direction of research is to measure what, if any, changes in TJ barrier function arise after mechanical stretch of the alveolar epithelium. Such information is necessary to fully interpret the significance of our results.

Our data show that applied mechanical strain can deplete ATP_i and, at sufficient magnitude, can disrupt TJ structure. We have also shown that ATP depletion alone can cause morphologic changes in the TJ. These results indicate that one relevant mechanotransduction pathway involves ATP, yet ATP is required for so many vital cellular processes. It is currently unclear whether TJ structure is directly affected by loss of ATP_i, is affected by some ATP-dependent secondary pathway or pathways, or whether a combination of both primary and secondary pathways is responsible for TJ disruption. Similarly, the actin cytoskeleton is obviously very important in mechanotransduction in the healthy epithelium, and the effects of its disruption are most likely widespread. There are undoubtedly other key molecules and processes by which mechanical stretch is able to regulate TJ structure, and so further experimentation must be performed to elucidate the specific mechanisms by which strain alters TJ structure and possibly permeability as well.

We have shown that tight junction structure and intracellular ATP content in cultured alveolar epithelial cells are adversely affected by cyclic biaxial epithelial strain. Strain at 25% change in surface area did not affect TJ structure, but a magnitude of 37% SA was able to perturb occludin distribution and overall expression. ATP depletion and actin perturbation also resulted in altered TJ structure. ATP_i concentration decreased at 25% SA and demonstrated a further decrease at 50% SA. These findings are an additional step toward understanding the etiology of ventilator-induced lung injury, and may prove influential in the direction of future VILI research and in the clinical development of less injurious ventilation strategies.