Mechanical strain of alveolar type II cells in culture: changes in the transcellular cytokeratin network and adaptations

Abstract

Mechanical forces exert multiple effects in cells, ranging from altered protein expression patterns to cell damage and death. Despite undisputable biological importance, little is known about structural changes in cells subjected to strain ex vivo. Here, we undertake the first transmission electron microscopy investigation combined with fluorescence imaging on pulmonary alveolar type II cells that are subjected to equibiaxial strain. When cells are investigated immediately after stretch, we demonstrate that curved cytokeratin (CK) fibers are straightened out at 10% increase in cell surface area (CSA) and that this is accompanied by a widened extracellular gap of desmosomes–the insertion points of CK fibers. Surprisingly, a CSA increase by 20% led to higher fiber curvatures of CK fibers and a concurrent return of the desmosomal gap to normal values. Since 20% CSA increase also induced a significant phosphorylation of CK8-ser431, we suggest CK phosphorylation might lower the tensile force of the transcellular CK network, which could explain the morphological observations. Stretch durations of 5 min caused membrane injury in up to 24% of the cells stretched by 30%, but the CK network remained surprisingly intact even in dead cells. We conclude that CK and desmosomes constitute a strong transcellular scaffold that survives cell death and hypothesize that phosphorylation of CK fibers is a mechano-induced adaptive mechanism to maintain epithelial overall integrity.

many cell types are subjected to deformation or other mechanical forces. An increasing number of studies has addressed cellular and molecular mechanisms that lead to coordinated responses of cells to such stimuli (for review, see Refs. 1, 23). Besides physiological responses occurring within a tolerable range of mechanical stress, a certain level of deformation inevitably threatens the structural integrity of the cell and leads to cell death. Little is known about structural cell changes in this life-threatening state, and even less is known about possible defense mechanisms a cell might employ to resist potentially fatal amplitudes of tensile stress. The goal of this study was to gather basic information regarding these questions on a cellular level by application of controlled strain to cells in culture, allowing direct comparison of stretched cells with unstretched controls.

We used primary cultures of alveolar type II (ATII) cells, the most abundant cell type in the lung, because in vivo this cell type is exposed to life-long mechanical strain during breathing. Although different lung volumes (as % of total lung capacity) were suggested (2, 36, and for a review, see Ref. 5) to induce single cell stretch on a cellular level, it has been shown that within a certain range, these cells respond to mechanical distension with release of surfactant into the alveolus in vivo (25, 27) as well as in vitro (13, 43). Repetitive stretch induces a variety of cellular responses associated with gene regulation and protein expression (for review, see Refs. 8, 11). Beyond this physiological range, overdistension of alveoli is considered a main cause of ventilation-induced lung injury, resulting in respiratory distress syndrome (for review, see Refs. 9, 39). Despite this obvious clinical importance, only little is known about structural changes of the cell caused by stretch.

When cell biological aspects on tissue level are considered, epithelial ATII cells face two most obvious threats when subjected to high mechanical strain: 1) a rupture of the virtually inelastic membrane that leads to a breakdown of osmotic and ionic gradients (for review, see Ref. 3) and 2) a disintegration of the transcellular cytokeratin (CK) network including the respective intercellular cell junctions, the desmosomes. The latter reduces the mechanic integrity of the whole alveolus and is apparently of pathophysiological importance as CKs were recently detected in exhaled breath condensate of mechanically ventilated patients (14).

Focusing on these two points, we investigated stretch-induced, morphological changes on a light and electron microscopical level and determined phosphorylation of CK8-ser431. We demonstrate here that the transcellular CK network is much more resistant to strain than the cell itself and that it remains intact even after the harboring cells have died. We also show that tensile stress alters CK curvature and desmosomal structure when observed immediately after stretch and suggest a mechano-protective mechanism that might involve phosphorylation of CK fibers.

MATERIALS AND METHODS

ATII cell isolation and cell culture.

Cells were isolated from male Sprague-Dawley rats (150–200 g) as previously described (6, 7). Briefly, rats were anesthetized, injected with heparin, and then lungs were perfused in situ to remove blood. After removing them from the body, lungs were lavaged several times to remove most alveolar leukocytes. Lung digestion and release of ATII cells were achieved by incubating them at 37°C with combination of 0.25 mg/ml elastase (EPC) and 0.05 mg/ml trypsin (Sigma) solution. Lungs were thoroughly minced together with 1.3 mg/ml DNAse I (Sigma) to minimize cell clumping. The tissue was filtered six times through gauze and nylon meshes (mesh width: 150, 20, and 10 μm) and the filtrate was centrifuged. After resuspension in DMEM (GIBCO), the cells were plated on IgG (Sigma, 0.5 μm/ml in 50 mM Tris)-coated Petri dishes and incubated at 37°C for 15 min to let the remaining macrophages and leukocytes adhere to the dish. Nonadherent cells were centrifuged and the cell pellet was resuspended in DMEM with 10% FBS (PAA A15-041) and 1% penicillin-streptomycin (GIBCO). Two milliliters of cell suspension were seeded in central portion of an elastic silicone membrane. The membrane had been autoclaved and coated previously with fibronectin. After an additional 10–15 min, 15 ml of medium were added. Cells were then incubated at 5% CO₂, at 37°C for 1–2 days (Fig. 1E) after which the majority of cells still show the typical morphology of cultivated ATII cells.

Fig. 1.

Preparation of silastic membranes and use on the cell stretcher (AdvancedLab). The silastic membrane was clamped between 2 metal rings (A), autoclaved (B), and coated (C). After seeding (D), the alveolar type II (ATII) cells were cultivated for 2 days (E). Then, the membrane was mounted on the stretch device (F; and as cross section in G), where a hollow, cylindrical shaped piston can be lowered into a ring-shaped indentation (H). Thereby, the membrane is stretched equibiaxially. For immunostaining or EM protocols, the inner part of the membrane was clamped after stretch between another metal ring and a circular magnet (I) and removed from the device (J). Cell stretching can be observed with an inverted microscope (H).

Preparation of silastic membranes, handling of stretch device, and stretch protocol.

An elastic and transparent polydimethylsiloxane membrane (“silicon rubber”) was clamped between two metal rings (Fig. 1A; both AdvancedLab), excess membrane was removed, and the membrane was autoclaved and coated with fibronectin (5 μg/ml; Sigma) overnight (Fig. 1, B and C). After seeding of cells and cultivation (Fig. 1, D and E), the clamped membrane was mounted on the equibiaxial strain device (Cell Stretcher 2.0, AdvancedLab) and the membranes were stretched at a constant, computer-controlled rate of 1%/s by 10, 20, and 30% of the initial membrane area (Fig. 1, F–I). Control experiments in the absence of strain were performed on the same device. In a first experimental set, the cells were used immediately after the three different amounts of stretch. Images and Western blots of the different stretches therefore reflect snapshots during a 30-s lasting, continuous stretch from 0 to 30% increase in cell surface area (CSA). In another experimental set the cells were fixed 5 min after the different stretch amplitudes had been applied. In all experiments, the cells remained stretched in the device until fixation (or lysis) was complete. For immunolabeling or electron microscopy, the membrane was clamped between two smaller rings (Fig. 1I) and removed from the stretcher. This allowed to maintain the stretch while the cells on the membranes were subjected to immunolabeling or electron microscopy protocols. For Western blotting, the cells were lysed immediately after stretch (as in Fig. 1H) and the whole stretcher was rapidly transferred on a cooling device where an ice-cold metal piston in direct contact with the bottom of the silicone membrane instantaneously reduced the temperature of the lysate. Then, the cells were removed with a cell scraper (Greiner) and the lysate was frozen at −20°C for later Western blotting. For quantification of dead cells, the stretcher was mounted on an inverted microscope (Axiovert 100, Zeiss) and images of the stretched cells were continuously acquired for 5 min with a 5× objective (Fig. 1H).

Transmission electron microscopy and immunofluorescence.

For electron microscopy, the cells were fixed with 2.5 glutaraldehyde (Plano) and 1% sucrose (Sigma) in PBS followed by a postfixation with 2% OsO₄ (Chempur) for 1 h. After dehydration in ethanol, the cells were enblock stained with saturated uranyl acetate in ethanol for 30 min at 37°C and embedded in EPON (Fluka) and hardened 48 h at 60°C. Ultra thin sections (50 nm) and 400-nm sections were cut on a Reichert ultramicrotome. The former were observed in a Zeiss EM10, whereas a Zeiss 902 under zero loss energy conditions was used for 400-nm sections. Negatives were then scanned with a UMAX scanner for further proceeding.

For CK immunostainings, the cells were fixed in 4% paraformaldehyde for 3 min at unstretched control conditions or after applying 10, 20, or 30% increase in CSA. Cells were then permeabilized with 0.1% Triton X-100 (Sigma) for 3 min and blocked with 5% goat serum (GIBCO) for 2 h. After overnight incubation at 36°C with the pan-CK antibody (dilution 1/150; ab17155, Biozol), the secondary fluorescent Alexa 488 antibody (dilution: 1/2,000; Molecular Probes) was applied for 1 h. All steps were performed at room temperature and in PBS except where indicated otherwise. Images were acquired on a Zeiss cell observer with a cool snap EZ CCD camera and Metamorph software.

Quantification of stretch-induced cell injury and morphologic parameters.

Stretch-induced injuries of the cell membrane have been determined with the stretching device mounted on a Zeiss Axiovert 100 (Fig. 1H) by monitoring the uptake of the nuclear dye ethidium homodimer-1 (final concentration: 4 μM; Molecular Probes) via the injured membrane during 5 min of persistent stretch. The number of injured (labeled) cells was compared with the total number of unlabeled cells before the stretch.

For determining the curvature index of the basal cell membrane, the actual length of the membrane profile in electron micrographs (as in Fig. 2, E and F) was measured between two points and divided by shortest distance between these two points. The CK curvature index was determined similarly by measuring the fiber length of a CK fiber in immunofluorescence images and dividing by the shortest distance between these two points. Pictures from the cell periphery were chosen randomly and pictures were taken with a 100× objective (Zeiss apochromat) and CK fibers were traced from the cell border toward the center of the cell as long as no clear branching point occurred or where a superimposition of two fibers impeded the clear identification of an individual fiber. This limited the measurement to the very outermost portions of the cell (∼3 μm from the cell border, see also arrows in Fig. 4, E–H). For determining the width of the extracellular, desmosomal gap, the distance between the outermost borders (closest to the cell membrane) of the dense plaques of one desmosome was measured in digitized electron micrographs (see also white arrowheads in Fig. 6, C–F). Only desmosomes with a clearly delineated outline of the plaque have been considered and several EM tilting series of desmosomes in 400-nm-thick sections of different stretch degrees have been performed to evaluate the effect of the observing angle on the measurement. To test whether the actual stretch of cells corresponds to that of the membrane, the average increase of the CSA, delineated by the outline of a cell, was determined on EPON-embedded cells and compared with the unstretched control. Average values were obtained by measuring the area of a cell cluster with DIC optics on a Zeiss Axiovert 200 and dividing it by the number of nuclei in this area.

Fig. 2.

General ultrastructure of stretched cells and stretch-induced membrane rupture. Unstretched control (A) and cells stretched by 30% cell surface area (CSA) increase (B) show no difference in the overall morphology. Some of the characteristic lamellar bodies (LBs) in the cell are marked with arrows. The apparent increase in size of the LBs at 30% CSA increase is not representative. The vertical section of cell stretched by 30% in C shows the detail of a fusion pore and the release of several LBs. The same cell is shown as an overview in D. Arrowheads point at released LBs. The basal cell membrane (arrowheads) of control cells (E) is wavy. Note the electron-dense deposits that cover the surface of the silastic membrane and resemble a basal lamina (arrows). Cells stretched by 30% (F) show a flat basal cell membrane, occasionally with ruptures, which are enlarged in H and I. In contrast to E, the silastic membrane in F tore very close to the basal cell membrane during the embedding procedure explaining the much brighter appearance of the space below the cell. The curvature index of the basal cell membrane in G demonstrates that already 10% ΔCSA leads to an unfolding of the initially wavy cell membrane. Statistical significance is indicated with asterisks (see materials and methods for explanation).

Image analysis for all measurements was performed with ImageJ software and data were processed in Excel software.

Western blots and extraction of keratins.

ATII cells were lysed immediately after stretch in 20 mM Tris·HCl at pH 7.4, 0.6 M potassium chloride, 1% Triton X-100, protease inhibitor cocktail (Roche), and 1 mM PMSF (triton high-salt buffer). Lysates were incubated for 20 min at 4°C and clarified by centrifugation at 10,000 g for 20 min at 4°C. For keratin extraction, the pellet was resuspended in the same buffer, incubated for further 20 min on ice, and again subjected to centrifugation at 10,000 g for 20 min at 4°C. The pellet of insoluble proteins was resuspended in 5 volumes of 8 M urea and the same volume of 5× SDS-PAGE sample buffer was added to the solution. Samples were then separated by SDS-PAGE. Supernatant was resuspended in 5× SDS sample buffer and separated by SDS-PAGE.

Gels were electrophoretically transferred onto a PVDF membrane and incubated with Pan-keratin (Clontech) and phosphospecific antibodies against K8-Ser431 (5B3) (20). After incubation with the secondary horseradish peroxidase-labeled antibody, the signals were measured in Chemismart (Paclab) and quantified with Bio1D software (Vilber Lourmat).

Statistic analysis.

Data are presented as means ± SD in text and figures. For statistical analysis, the data were tested for normal distribution (Kolmogorov-Smirnov test) and P values were determined either with Mann-Whitney U-test (not normally distributed) or with the Student's t-test (normally distributed). Data of the time course in Fig. 4 were also compared with a paired t-test for correlated samples. The statistical significance was indicated in the figures with *P ≤ 0.05, with **P ≤ 0.01, and with ***P ≤ 0.001.

RESULTS

Actual stretch of cells.

To verify that the stretch of the silicon membrane also reliably results in a stretch of the ATII cells, we measured the average CSA at different amounts of stretch in one experiment. We found that a nominal stretch of 10, 20, and 30% increase in membrane area resulted in 10.2% (n = 189 cells), 14.6% (n = 356), and 27.7% (n = 121) increase of the average CSA compared with unstretched control cells (n = 135), demonstrating that cells do not detach from the membrane during stretch.

General ultrastructure and membrane damage immediately after stretch.

When comparing the ultrastructure of unstretched controls (Fig. 2A) with cells stretched to the maximum of 30% CSA increase (Fig. 2B), the overall morphology and the appearance of the cells were unaltered. The characteristic surfactant containing lamellar bodies (some indicated with arrows) and other organelles were clearly visible under both conditions. As shown in earlier studies, mechanical stretch stimulates LB fusion with the plasma membrane (12, 41), and also we found fusion events in the transmission electron microscopy (TEM) only in stretched cells (see vertical section in Fig. 2, C and D). In accordance with cell injury measurements, most cells did not show any obvious sign of cell damage, like emptied cytoplasm or damaged organelles at this point of time. Nevertheless, occasional membrane ruptures could be observed on the apical (not shown) as well as in the basal cell membrane (Fig. 2, F–I) when stretch exceeded 20% increase of CSA. Figure 2E shows that the basal cell membrane was wavy in control cells but, as demonstrated by the curvature index in Fig. 2G, was straightened out entirely at a stretch of 10%. Consistent with this observation, higher stretches to 20 and 30% increase of CSA lead to ruptures in the basal cell membrane (Fig. 2, F–I), where flattening of the membrane could not compensate the increased tensile stress. Apparently, size and/or incidence of the cellular lesion did not suffice to cause severe, immediate damage which is consistent with the low number of injured cells immediately after stretch as shown in Fig. 3.

Fig. 3.

Time course of stretch-induced cell injury under conditions of permanently increased CSA by 10, 20, and 30%. (For numeric values of mean and SD, see Table 1.) The number of injured cells increased with higher extent and duration of stretch. Statistical significance between different stretches at the same time point is indicated with * beside dotted lines (see materials and methods for explanation), whereas the endpoints of the lines in the bottom indicate significance between the respective time points within the same amount of stretch (0.05 ≥ P ≥ 0.01). One experiment at 30% stretch did not show an increase in cell damage after 3 min resulting in no significance.

Cell injury.

Table 1 and Fig. 3 show the stretch-induced cell injury in three independent experiments for each stretch amplitude. The number of injured cells depended on the percentage and duration of stretch. Consistent with the observation that membrane ruptures occurred in the basal cell membrane at a stretch higher than 10% increase in CSA, cell injury clearly increased at 20 and 30% and showed a maximum (avg. = 23.76%) when 30% stretch was applied for 5 min.

Table 1.

Percentage and duration of stretch

	Control		10%		20%		30%
	Mean	Std.	Mean	Std.	Mean	Std.	Mean	Std.
0 min	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%
1 min	0.13%	0.22%	4.22%	3.23%	6.08%	6.15%	4.37%	1.44%
2 min	0.92%	1.17%	5.13%	4.03%	10.13%	7.19%	8.65%	1.91%
3 min	1.70%	1.92%	6.71%	4.17%	13.76%	7.33%	15.35%	9.17%
4 min	1.79%	2.08%	7.69%	5.01%	17.42%	10.35%	18.17%	9.77%
5 min	2.35%	3.05%	7.92%	5.10%	20.18%	13.79%	23.76%	14.30%

CK and desmosomes immediately after stretch.

The immunofluorescence images in Fig. 4, A–D, show that ATII cells at different stretch amplitudes displayed a well-developed CK network that spanned the entire cytoplasm except the nucleus. Most notably, none of the applied stretch amplitudes resulted in an apparent breakage of the CK fibers. Instead, a straightening out of slack fibers could be observed when individual CK bundles at the periphery of control cells were compared with cells stretched to 10% or more. Slack CK fibers of control cells are shown in Fig. 4E and tense fibers (Fig. 4, F–H) at stretch of 10, 20, and 30% increase in CSA. The straightening out has been quantified as the curvature index (for description, see materials and methods), in which n_f = number of individual fibers, n_cell = number of cells, and n_isol = number of independent primary cell isolations: 1.0105 ± 0.0176 at control conditions (n_f = 648, n_cell = 159, and n_isol = 3), 1.0045 ± 0.0086 at 10% (n_f = 459, n_cell = 143, and n_isol = 3), 1.0077 ± 0.0140 at 20% (n_f = 395, n_cell = 131, and n_isol = 3), and 1.0073 ± 0.0105 at 30% (n_f = 267, n_cell = 73, and n_isol = 2). One or two elastic membranes with the same stretch amplitude were examined per cell isolation. These data are also presented in the histogram in Fig. 4I including standard deviation and statistical significance. Examples for fibers as used in the measurements are indicated with white arrows in Fig. 4, E–H.

Fig. 4.

Cytokeratin (CK) immediately after stretch. A–D: pan-CK immunolabelings as overviews and E–H show details from the cell periphery of different amounts of stretch. The staining pattern in the center portions of the cells in B and in particular in C and D is less distinct caused by a lower number of prominent fiber bundles. Representative areas are marked with white squares, enlarged, and shown at the right bottom corner of each overview in A–D. For better fiber discrimination, the gray values have been digitally inverted. E: CK fibers (arrows) in the cell periphery of control cells are bent but had straightened out at 10% stretch (F) or higher (G, H). This is confirmed by the mean fiber curvature index in I. Statistical significance is indicated with asterisks (see materials and methods for explanation).

Although a statistically reliable quantification of the fiber curvature or other morphometric parameters in the central portion of the cell was impossible because of a generally high fiber density, we also observed a change in the general appearance of the CK network in this part of the cell. Stretch, especially above 10% CSA increase, led to a less well-defined CK staining pattern. The number of prominent CK fibers was lower and the CK network apparently consisted of more but thinner fibers. This phenomenon is depicted in Fig. 4, A–D, where the central portion of the cell is shown. In the insets, the gray value had been inverted for better visibility. The micrograph in Fig. 6A shows prominent CK bundles (arrows) but also an example how splitting of CK fibers (white double arrowhead) appears in the TEM.

To determine whether the observed morphological changes of the CK network were accompanied by phosphorylation of relevant proteins, we determined the phosphorylation of CK8-ser431 site as an example of CK phosphorylation. The quantitative analysis of 7 independent experiments in Fig. 5 showed that a stretch of more than 20% led to a statistically significant increase of CK8-ser431 phosphorylation.

Fig. 5.

CK8-s431 phosphorylation relative to control at different amounts of stretch. Analysis of Western blots (average ± SD) shows a statistical significance increase at 20% increase of CSA, which is indicated with an * (see materials and methods for explanation).

Besides CK fibers, we also focused on the functionally and structurally related desmosomes that link CK bundles of adjacent cells and represent the only anchor point for CK fibers in ATII cells, which lack classic hemidesmosomes (16). The overview in Fig. 6A shows cells stretched to 30% where desmosomes (arrowheads) and the inserting CK bundles (arrows) are clearly visible. TEM revealed that the width of the extracellular desmosomal gap varied at different degrees of stretch (see Fig. 6B and distance between arrowheads in Fig. 6, C–F). We observed an increase as shown in the list below in which n = number of desmosomes, n_sec = number of sections (with 15–40 cells per section), and n_isol = number of independent primary cell isolations: from a normal value of avg. = 26.7 ± 6.5 nm (n = 52, n_sec = 66, and n_isol = 4) under unstretched control conditions, the gap width increased to avg. = 35.0 ± 7.5 nm (n = 30, n_sec = 54, and n_isol = 2) when stretched by 10%. A stretch of 20% resulted in an almost normal width of avg. = 28.2 ± 9.8 nm (n = 48, n_sec = 48, and n_isol = 3) and a stretch of 30% led to a narrowing of the gap to avg. = 22.5 ± 8.6 nm (n = 35, n_sec = 48, and n_isol = 3). The results of these measurements are also presented in Fig. 6 including information about the statistical significance. Considering that we used a stretch protocol, where a constant stretch rate in all stretch experiments was applied, these measurements at 10, 20, and 30% display single time points during a continuous stretch from 0 to 30%. Accordingly, the gap first widens after 10 s and 10% stretch and then returns to normal values at increased stretch of 20 and 30%–20 and 30 s after starting the stretch, respectively. To assess the effect of the viewing angle of the desmosome, we performed several tilting series of 400-nm-thick sections. These sections comprised a major part of a desmosome and as shown in Fig. 6, G–J, an observation angle of 6° starts to affect visibility of the desmosome by obscuring the edge of dense plaque. At this section thickness, the measured values of the desmosomal width under optimal viewing angle were 22.06 nm (n = 3) under control conditions, 34.05 nm (n = 5) at 10% stretch, and 17.3 nm (n = 2) at 30% stretch. As the section thickness used for the measurements (Fig. 6) was only 50 nm, the critical viewing angle is substantially higher and has less effect on the measurement. However, only desmosomes with a clearly visible edge were considered for measurement.

Fig. 6.

Desmosomes immediately after stretch. The electron micrograph in A shows the overview of a cell-cell contact. Desmosomes are labeled with arrowheads and one of the inserting CK bundles is indicated with arrows. An example of longitudinal fiber splitting is indicated with a white double arrowhead. B: mean width (±SD) of the extracellular gap (see white arrows in C–F) at different amounts of stretch shows an increase at 10% but not at higher stretch amplitudes. Statistical significance between columns is indicated with asterisks (see materials and methods for explanation). C–F: examples of desmosomes at different amounts of stretch and G–J show 4 different tilting angles of a 400-nm section to illustrate the effect of the viewing angle on the visibility of the desmosomal gap. The scale bar in C applies for C–J.

CK and desmosomes 5 min after stretch.

As shown in Fig. 3, the number of injured ATII cells subjected to stretch depended on the extent of stretch and the stretch duration. Stretch of 10% did not show obvious signs of cell death after 5 min in most of the cells as observed in the TEM: a prominent CK network could be detected and the gap between the dense plaques (avg. = 29.5 nm, std. = 13.84, n = 11, n_isol = 1) did not differ practically from normal values, although few values exceeded the average value by far. The situation changed dramatically at 20% increase in CSA where many cells were fatally injured, some of which are shown in Fig. 7. Organelles were damaged and the cytoplasm was less dense, indicative for a leaky cell membrane. Figure 7, A and B, shows the interface of two cells. Both cells are dead which is most strikingly demonstrated by the blemished mitochondria in Fig. 7A (double arrowheads). Despite cell death, CK fibers (arrows in Fig. 7, A and B) often spanned the cell and were still attached to the desmosomes. The ultrastructure of desmosomes at 20% stretch (Fig. 7) and above differed in many cases dramatically from the normal situation. The gap between the dense plaques widened to a multiple of the normal value (avg. = 116.2 nm, std. = 58.3, n = 44, n_isol = 1), intestingly without being torn apart. Desmosomes with values beyond 400 nm were not considered for evaluation, because under these conditions, their structural entity was apparently lost.

Fig. 7.

Ultrastructure of desmosomes and CK of lethally injured cells after 5 min of 20% persistent stretch. A and B: 2 adjacent cells with several desmosomes (black frames). Some of the CK fibers are marked with arrows. The cytoplasm is depleted and the remainders of former cell organelles are indicated with double arrowheads in A. B: cell contact often remains only where groups of desmosomes function as mechanical strongholds. As clearly shown in the enlarged C–F, the distance of the extracellular gap dramatically exceeds normal values reflecting the tensile stress of the inserting CK fibers. Note that fine molecular structures (best visible in D) span the gap between the dense plaques.

DISCUSSION

The major focus of this study was to elucidate stretch-induced morphological changes of primary ATII cells, that could be relevant for cell damage and partly also whether CK phosphorylation might function as a mechano-protective mechanism. Using an experimental set up that allowed us to investigate cells in culture in the TEM and with immunofluorescence microscopy, we could directly compare stretched cells with unstretched controls. We found that acute stretch led to changes of the CK network and the desmosome structure. Surprisingly, neither was simply dictated by the stretch and they were partly accompanied by phosphorylation of CK. Long-lasting stretch of 5 min caused more pronounced and strain-related changes of these structures. Importantly, the CK network remained largely intact even in cases of severe and lethal damage of the entire cell.

The range of the stretch with a maximum of 30% CSA increase was based on the similar increases in vivo of the basal cell membrane at 100% total lung capacity (2, 36) and on the range used in other studies (12, 28, 35). However, ATII cells may experience considerably less deformation than alveolar type I cells in vivo (29). This should especially be considered when studying the effects of high stretch amplitudes on ATII cells in culture with the situation in vivo. Some parameters in our study show an unexpected course when cells were stretched more than 10%, which could be explained by exceeding the maximal amplitude an ATII cell experiences during normal breathing. However, this interpretation is challenged by the fact that cultured ATII cells differ markedly in their overall morphology compared with the situation in the alveolus which might affect their response to mechanical strain. In addition, the response of a cultured cell to mechanical strain might differ from its native counterpart by an altered overall morphology and geometry.

Our data demonstrate that the moderate stretch to 10% CSA increase was sufficient to significantly straighten out CK fibers in the cell periphery. Combined with the finding that the extracellular gap widens at 10% stretch, this strongly indicates that fiber curvature no longer compensates the stretch, and the actual tensile stress on individual CK fibers and the anchoring desmosomes is increased. Although neither the straightening out of CK fibers nor the significant change in the desmosomal gap has been reported previously, these observations were not surprising. The phenomena can plausibly be interpreted as a simple passive response of the initially slack CK network and their insertion points, the desmosomes, which could not provide enough mechanical resilience to maintain their original width between the dense plaques.

In contrast, the finding that the extracellular gap of the desmosomes did not widen further, but instead decreased, when the stretch was acutely increased by 20 and 30% is counterintuitive. As not one incidence of torn CK fibers could be observed in TEM or in CK immunostainings, we can exclude that the abolished pulling force of a destroyed CK network allowed the desmosomes to return to their original gap width. Another mechanism must account for the effect and we believe that the significant increase of the CK fiber curvature in the periphery of the cells at 20 and 30% stretch compared with 10% (Fig. 4, E–I) plausibly explains the phenomenon. The CK fibers obviously relax at higher amounts of stretch which reduces the tensile force of the CK fibers and also lowers the pulling force on the desmosome. Thereby, desmosomes would be protected from being torn apart, which is consistent with the idea of the transcellular CK network being the most stress resilient cellular component in the alveolus as previous studies suggest (13).

Although intermediate fibers in general and CKs in particular are considerably flexible when investigated in vitro (for a review, see Ref. 19), only little is known about the behavior of mechanically stretched CK fibers in a living cell. We were interested whether the suggested CK fiber relaxation is purely a passive response to increased tensile stress or whether it may be actively facilitated by a mechano-induced change of the CK fiber properties. As the observed changes appeared on a very fast time scale, we investigated CK phosphorylation as the most promising candidate. CKs can adapt to numerous cellular conditions (10, 26, 40, 42 and for review, see Refs. 2, 15, 18) and it is known that changes in the phosphorylation status affect filament organization and solubility (30, 32). It has been shown previously that two main CK types in lung CK18 and in particular CK8 are phosphorylated upon cell stress (17, 20–22) and Ridge et al. (30) observed that persisting shear stress leads to a change in the distribution of CK fibers in lung epithelial cells accompanied by a phosphorylation of CK8-ser73. We also found a statistically significant phosphorylation of CK8, although we could not investigate CK8-ser73 because this phosphorylation site does not exist in rats. Instead, we quantified the phosphorylation of CK8-ser431, which might compensate for the lack of CK8-ser73. It is conspicuous that the strongest (and significant) increase in phosphorylation could be oberserved at 20% stretch where CK fiber curvature unexpectedly increases (fibers are becoming more slack) and where the width of the desmosomal gap narrows. This suggests that the CK fiber phosphorylation might indeed be involved in lowering the tensile force.

When comparing the morphological changes of the CK network in our experiments (Fig. 4) with the changes observed by Ridge et al. (30), an obvious difference can be noticed as we could not observe thicker bundles of CK fibers. We explain this by the substantially different stimulation protocol on a completely different time scale that was used.

Cell injury (Fig. 3) is presented as the relative number of cells that show a leaky plasma membrane as determined by the uptake of a nuclear dye. Although it is difficult to compare these values with other studies as experiments differ in rate and duration of stretch (35, 37) and the use of cell lines (33, 38), we consider the range of our values within those of previous observations. When focusing on the effects of persistent strain after 5 min, we observed that 7.9% of the ATII cells at 10% stretch were injured. Higher amounts of stretch showed clearly elevated injury rates with a maximum of 23.8% at a stretch of 30% CSA increase. Although this value is higher compared with Tschumperlin and Margulies (35) (16% injury at 25% stretch after one stretch cycle of 4 s with 1-day-old ATII cells), our study employs a longer stretch duration and a slightly higher stretch amplitude, which might explain the discrepancy. The fact that exceeding the stretch beyond 10 to 30% CSA increase leads to a threefold increase after 5 min is consistent with our observations in the TEM where the basal cell membrane on the silastic membrane was wavy under unstretched control conditions but flattened out already at 10% stretch (Fig. 2, H and I). Further stretching of the cells must lead to rupture of the membrane and is confirmed with our observation in the TEM. It is likely that these small injuries of the cell membrane are the initial membrane injuries that eventually lead to cell death after 5 min of persistent strain, probably by continued leakage of ions and other compounds from the cytosol into the bath solution and vice versa. However, we can certainly not exclude that the observed injuries were already subject to, or would be compensated later by, plasma membrane wound-healing processes. The cell membrane injury measurements (Fig. 3) fail to detect such a mechanism as the DNA dye was able to pass the leaky plasma membrane as soon as the membrane injury occurred and remained there even in case of a later plasma membrane repair.

Cells that were dead after 5 min of stretch still showed a well-developed network of CK fibers, although the cell had ceased to exist as a biological functional entity. This is consistent with previous findings showing that breaks in the epithelium do not occur in the vicinity of an intercellular junction (13) which also lead to the suggestion that “cell viability may be killed in preserving junctional integrity” (35). Moreover, even those desmosomes that displayed gap widths far beyond normal values still showed electron-dense, fiber-like structures between the dense plaques, which indicate physical contact between adjacent cells and possibly some resilience against tensile stress. This unexpected stability of the transcellular CK network might be of utmost importance for maintaining the structural integrity of an alveolus during alveolar injury as it could remain as a scaffold in those cells that were not able to survive tensile stress. Thereby, it could still hold the neighboring cells (or another adjacent network of remaining CK) in place and act as a guide rail for outgrowing cells. Also, adjacent injured endothelial cells unable to prevent leakage of blood cells into lumen of the alveolus would benefit from such a CK network, as it would provide a certain barrier and structural support function, at least for a certain time span (for a review, see Ref. 31). From that point of view, the suggested mechano-protective mechanism that protects the transcellular CK desmosome network becomes even more important as it might contribute to epithelial repair in case of lung injury. Therapeutical as well as pharmacological treatments of lung injury might benefit from a better understanding of the underlying principles.

GRANTS

This work was supported by the Deutsche Forschungsgemeinschaft, Grant D1402, the Fonds zur Förderung der wissenschaftlichen Forschung, Grant P15743, and the 6th framework of the EU, Pulmo-Net.