| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Physiology Program, School of Public Health, Harvard University, Boston, Massachusetts; and 2Unitat de Biofísica i Bioenginyeria, Facultat de Medicina, Universitat de Barcelona - IDIBAPS, Barcelona, Spain
Submitted 12 October 2005 ; accepted in final form 3 January 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
alveolar epithelium; cell mechanics; magnetic twisting cytometry; prestress; protease-activated receptors
The serin-protease thrombin is found in bronchoalveolar fluids in a variety of lung inflammatory conditions (14, 20). However, its role in the pathogenesis and resolution of lung injury remains poorly understood. Recent studies have revealed that, in addition to its role in the coagulation cascade, thrombin causes a number of proinflammatory and profibrotic effects in the lung through activation of protease-activated receptors (PARs) (14, 22). Work from our laboratory (24, 34) and others (18) suggests that thrombin alters the balance of forces at the alveolar epithelial cell level by means of at least three competing mechanisms. First, thrombin has been shown to increase centripetal contractile forces via phosphorylation of the myosin light chain (MLC) (18, 24). Second, thrombin has also been reported to increase levels of tight junction-associated proteins and circumferential reorganization of the cytoskeleton, potentially leading to enhanced cell-cell adhesion as suggested by increased transepithelial resistance in confluent alveolar epithelial monolayers (18). Finally, thrombin rapidly increases cell stiffness by threefold, leading to a similar increase in the centripetal tension that cell-cell and cell-matrix adhesions would need to withstand during lung expansion (34).
The aim of this work was to study the effect of stretch on the structural integrity and micromechanics of human alveolar epithelial cell monolayers exposed to thrombin. We used a novel technique to subject thrombin-treated cell monolayers to stepwise stretch and to measure simultaneously cell strain and viscoelasticity (35). The technique is based on binding ferrimagnetic beads to cell surface receptors. Because the beads are tightly connected to cell surface and cytoskeleton, they can be used as markers to measure cell strain during substrate stretching. The structural integrity of the cell monolayer is assessed by comparing the intracellular and intercellular strains within the monolayer. In addition, the beads are twisted by an oscillatory magnetic field (11) that allows us to measure cell viscoelasticity during stretch.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
For experiments, cells were harvested with a brief exposure to trypsin EDTA (Sigma) and plated (3 x 105 cells/well) on collagen I-coated flexible-bottomed culture wells (35-mm diameter; Bioflex, Flexcell International). Measurements were performed in confluent cells (7–14 days after plating) and in subconfluent cells (2–4 days after plating).
On the day of experiments, 400 µg of coated ferrimagnetic beads (Fe3O4) of 4.5-µm diameter with magnetic moment 9 x 10–13 A·m2 suspended in HEPES-buffered RPMI 1640 were added to each individual well. After a 20-min incubation, the well was washed with HEPES-buffered RPMI 1640 to remove unbound beads and mounted on the experimental setup.
The beads were coated with a synthetic RGD (Arg-Gly-Asp)-containing peptide (Peptide 2000; Integra life Sciences, San Diego, CA) at 50 µg of peptide/mg bead in 1 ml of carbonate buffer (pH 9.4).
Stretching device. Experiments were performed with a stretching device mounted on an inverted optical microscope (Axiovert S100; Zeiss, Göttingen, Germany). As described in detail elsewhere (35), the cell-stretching device is based on deforming the cell substrate by applying vacuum to its underside. A cylindrical loading post is located underneath the central region of the well, coaxial to the microscope objective. When vacuum is applied underneath the outer annular region of the flexible cell substrate, the central region is stretched equibiaxially and homogeneously while roughly remaining on the microscope focal plane. The rising and falling times of the applied negative pressure step are set to 4 s with a first-order pneumatic filter.
Optical magnetic twisting cytometry. An optical magnetic twisting cytometer (OMTC) (11) was coupled to the stretching device to measure cell mechanics (35). In OMTC, ligand-coated ferrimagnetic beads are specifically bound to cell surface receptors. The beads are permanently magnetized with a brief (20 ms) and strong (120 mT) pulse of magnetic field in the horizontal direction of the cell monolayer and subsequently twisted in a weak sinusoidal magnetic field applied in the vertical direction. The resulting lateral bead displacement is measured with nanometer resolution using videomicroscopy (33). Images were obtained with a progressive scan black-and-white charge-coupled device camera (CV-M10 BX; JAI, Glostrup, Denmark) and digitized by an eight-bit resolution frame grabber (PC Eye4; Eltec, Mainz, Germany). In this study, microscope magnification was x10, resulting in a field of view of 640 x 480 µm. The camera trigger and the current fed to the coils were controlled with an analog-digital/digital-analog PCI board (PCI-MIO-16XE-10; National Instruments, Austin, TX) driven by LabVIEW software (National Instruments).
Protocol. Confluent and subconfluent cell monolayers were subjected to the protocol illustrated in Fig. 1A. The first OMTC and cell strain measurement was performed by twisting the beads for 15 s in an oscillatory magnetic field of 5 mT amplitude and 1 Hz frequency (baseline). After 1 min, thrombin (0.5 U/ml) or vehicle (HEPES-buffered RPMI 1640) was added to the well. Another measurement was carried out 5 min later as previously described. A stretch of 0% (control) or 20% of the substrate was then produced, and the beads were twisted again 1.5 min after stretching. Stretch was held, and the bead oscillation was repeated 5 min later. Finally, the substrate was relaxed to its initial unstretched conformation, and the beads were twisted again 1.5 min after relaxation. The beads were magnetized before each measurement. For each experimental condition, n = 7 wells were measured.
|
100–150 beads/well) throughout the experiments using a centroid algorithm.
We computed two indexes of strain to assess the extent to which cells followed the imposed substrate deformation (Fig. 1B) (35). These indexes were obtained from the bead position in the first image of each OMTC measurement (no twisting field applied). The first index accounted for the intracellular deformation and was termed cell strain (
c). This index was defined as the fractional change of the distance (d) between beads bound to the same cell relative to the distance (dII) before stretch application (time point II in Fig. 1A)
 | (1) |
c was computed as the median of at least 15 pairs of beads in each well. The second index accounted for the intercellular deformation of the cell monolayer and was termed layer strain (L). L was calculated as the fractional change of the distance (D) between beads that were at least 200 pixels apart from each other (D > 164 µm) relative to the distance before stretch application (DII)
 | (2) |
In a previous study, we showed that L matched the applied substrate strain (35). To assess differences between bulk strain of the cell monolayer and strain of individual cells, we defined the loss of strain transmission (
) as
 | (3) |
indicates the extent to which strain is transmitted from the substrate to overlying cells and was taken as an index of paracellular monolayer disruption (see DISCUSSION).
OMTC measurements were processed as follows. The specific torque (T) applied to a bead was computed as
 | (4) |
*)
 | (5) |
 | (6) |
= G"/G' is the hysteresivity or loss tangent that reflects the balance between elastic and frictional stresses in the cell. When < 1, cell mechanical behavior is predominantly elastic or solidlike. This mechanical property confers the cell with the ability to rapidly recover its shape in response to deformation. By contrast, when > 1, the cell behavior is predominantly dissipative or liquid like, which enables the cell to alter its shape and flow in functions such as crawling, spreading, division, or contraction (10). G* was computed as the median of 100–150 beads in each well. Statistics. Data are reported as means ± SE. Statistical comparisons were performed by unpaired Student’s t-test. Statistical significance was assumed at P < 0.05.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
c was slightly smaller than L, resulting in a small but significant (inset). No significant differences were observed between cells treated with thrombin or with vehicle. When the substrate was returned to its unstretched conformation, c of vehicle-treated cells recovered its baseline values, whereas thrombin-treated cells exhibited a tendency to maintain a small positive . The deformation of nonconfluent cells in response to substrate stretch differed from that of confluent cells (Fig. 2, bottom). When stretch was applied, c was considerably smaller than L in both vehicle-treated and thrombin-treated cells, resulting in large , which was significantly larger in cells treated with thrombin than in cells treated with vehicle. When the substrate was unstretched, both groups exhibited negative c, resulting in > 0. Holding the stretch for 5 min induced only minor changes in of both confluent and nonconfluent cells (not shown).
|
= 0.344 ± 0.007 for confluent cells and G' = 1.03 ± 0.06 Pa/nm and = 0.336 ± 0.006 for subconfluent cells. Thrombin markedly increased G' (192 ± 25% in confluent cells and 263 ± 22% in nonconfluent cells) and decreased (33 ± 3% in confluent cells and 36 ± 2% in nonconfluent cells). Stretch application induced an increase in G' and a drop in in vehicle-treated cells (Fig. 3). By contrast, both G' and of thrombin-treated cells remained roughly unchanged when the step of stretch was applied. Only minor changes were observed in G' and after holding strain for 5 min (not shown). When the substrate was relaxed to its initial unstretched conformation, G' of thrombin-treated cells dropped to values close to the original baseline (Fig. 4), whereas G' of thrombin-treated cells that had not been stretched remained elevated. However, of both stretched and unstretched thrombin-treated cells remained >25% lower than the baseline value.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
A major limitation of this study is that it does not include several essential features of the extracellular environment of the injured epithelium, such as the presence of inflammatory cells and proinflammatory mediators, the deposition of fibrin, the inactivation of surfactant, and the secretion of extracellular matrix (40). In the absence of this pathophysiological environment, the simple extrapolation of our findings to in vivo conditions may be misleading. However, this simple model captures a central feature of the acutely injured epithelium, i.e., mechanical determinants of cell-cell junctions.
We used a recently developed technique that allowed us to subject adherent cells to biaxial stretch of their substrate and to simultaneously measure their strain and viscoelasticity (35). Cell strain was computed from changes in the distance between the ligand-coated ferrimagnetic beads tightly bound to cell surface via focal adhesions. If the cell was a mechanically homogenous body firmly attached to its substrate, the imposed substrate deformation would be wholly transmitted to the apical surface of the cell. In such case, strain of the cell monolayer L and strain of individual cell c would be equal and = 0. However, in the present study, we found significant differences between L and c (Fig. 2). We reason that this loss of strain transmission might be attributable to two main factors: mechanical inhomogeneities in the cell body and partial detachment of the cell from the substrate. Mechanical inhomogeneities in the cell body could result in regional differences in cell stiffness (1). In response to substrate strain, stiffer regions would deform less than the substrate, leading to < 0. Conversely, softer regions would deform more than the substrate and yield > 0. Given that c and L were computed over at least 15 pairs of beads in each well studied and that the beads were randomly distributed over the cell surface, mechanical inhomogeneities probably do not account for the discrepancies between L and c. Underestimation of c could also arise from basal-to-apical strain gradients. Such gradients could differ between confluent vs. subconfluent cells and between thrombin-contracted vs. noncontracted cells, thus partially explaining the different values of between these different experimental conditions. However, if the values of during stretching arise from basal-to-apical strain gradients, they would vanish once the substrate is unstretched to its initial conformation. This was not the case in our experiments (Fig. 2, inset). Indeed, upon substrate unstretching, remained significantly positive with values similar to those obtained during stretching. Therefore, we conclude that the contribution of basal-to-apical inhomogeneities to the average during stretching, if any, is small. Instead, discrepancies between L and c are more likely to be due to the breakdown of cell-cell and cell-substrate adhesions during cell distention. As a result of this loss in attachments, cells retracted from their substrate and paracellular gaps formed or increased in size. This interpretation is consistent with the fact that remained elevated after stretch cessation. Given the aforementioned considerations, the index was assumed as an index of formation or change in size of paracellular gaps.
The study was carried out in human alveolar epithelial cells from the cell line A549. Although this cell line is extensively used (18, 31, 39), its appropriateness as a cultured model of alveolar epithelial cells remains controversial. Specifically, the ability of A549 cells to establish functional tight junctions and to form nonpermeable monolayers has been called into question (41). A recent study, however, has provided evidence that A549 cells have the ability to form adherens junctions and tight junctions when grown to confluence (18). In addition, the authors showed that A549 monolayers exhibit significant transepithelial resistance and that this resistance can be modulated by thrombin. A549 cells have also been reported to express functional PARs and to respond to thrombin by releasing inflammatory mediators including IL-6, IL-8, and PGE2 (2). Therefore, despite the limitations inherent in transformed cell lines, we considered that A549 cells were a suitable model to study the micromechanics and structural integrity of alveolar epithelial cell monolayers subjected to thrombin and stretch.
Thrombin is well known for its key role in the coagulation cascade, in which it cleaves circulating fibrinogen to fibrin. In this function, thrombin is involved in coagulation abnormalities that lead to alveolar fibrin deposition and lung dysfunction in ALI (15, 19). In addition, thrombin also mediates cellular responses through proteolytic activation of the PAR family (6, 13, 22). PAR activation by thrombin has been comprehensively studied at the endothelial level (4, 9, 23, 38). By contrast, the role of thrombin in the alveolar epithelium is only starting to be elucidated. In this connection, increasing evidence supports the notion that activation of PARs also mediates relevant cellular responses in the alveolar epithelium such as release of inflammatory mediators, MLC phosphorylation, increased levels and translocation of tight junction proteins, and cytoskeletal remodeling (2, 18, 34). However, the potential role of thrombin in the pathogenesis and resolution of ALI and ARDS under dynamic mechanical conditions remains elusive.
The structural integrity of the alveolar epithelium is determined by a force balance between centripetal cytoskeletal tension and centrifugal cell-cell and cell-matrix tethering forces. Given that the epithelium undergoes stretch during breathing and mechanical ventilation, cell stiffness is a central determinant of this dynamic balance: the stiffer the cell, the larger centripetal tension that cell-cell and cell-matrix adhesions need to withstand during lung expansion. Prompted by the findings that thrombin induced a prolonged threefold increase in cell stiffness (34), we studied whether exposure with thrombin followed by a single mechanical stretch with an amplitude characteristic of mechanical ventilation (20%)-induced changes in epithelial barrier integrity (36). We found that strain experienced by thrombin- and vehicle-treated confluent cells differed from imposed substrate strain (Fig. 2). These results suggest that cell-cell and cell-matrix adhesions were not able to completely withstand the increased centripetal tension induced by stretching, leading to gap formation as reflected by > 0. Interestingly, we did not find significant differences in between thrombin- and vehicle-treated confluent cells, indicating that the increase in centripetal tension induced by thrombin did not significantly contribute to gap formation. This apparent paradox could be explained by the fact that, in addition to increased cell stiffness, thrombin also increases the levels of tight junction proteins (ZO-1 and occludin) and induces peripheral actin remodeling in A549 cells (18, 34). Our findings suggest that this peripheral remodeling of the cortical cytoskeleton and of cell junctions induced an increase in tethering forces. As a result, the confluent cell monolayer was able to balance the increase in centripetal force caused by cell stiffening and contraction, thus preventing paracellular gaps to form or increase in size.
In contrast to confluent cells, nonconfluent cells exhibited substantial retraction from the substrate as shown by the large we found. This reveals a critical mechanical role of cell-cell junctions in counterbalancing increased centripetal forces during stretching. was dramatically increased in subconfluent cells that had been exposed to thrombin, indicating a substantial rise in the size of paracellular gaps. The differential behavior we found between confluent and subconfluent cells can be explained in terms of a simple force balance. In confluent cells, the increase in centripetal force caused by stretch was balanced by the reinforcement of cell-cell junctions induced by thrombin. However, in subconfluent cells where anchorages to adjacent cells are reduced or absent, cell-substrate adhesions failed to withstand centripetal tension, leading to partial cell retraction from the substrate. Given that ALI and ARDS are characterized by widespread epithelial disruption, impairment of cell-cell junctions is a hallmark of disease progression (3, 40). In this context, our results suggest that the combined effect of cell-cell junction impairment and cell stretching can result in further paracellular gap formation and increased epithelial permeability. This effect might be dramatically exacerbated by the presence of thrombin.
When the cell substrate was relaxed to its initial unstretched conformation, subconfluent thrombin-treated cells exhibited significant compressive strain. This negative strain could be a consequence of cell detachment and retraction during stretching that turned into negative strain after stretch removal. To the best of our knowledge, the response of alveolar epithelial cells to compressive stress has not been studied. However, compressive stress has been shown to slow wound healing and to trigger profibrotic responses in cultured bronchial epithelial cells (27, 37). Similar responses could occur in the alveolar epithelium as a result of the combined effect of thrombin and stretch and could contribute to the profibrotic environment of acutely injured lungs (14, 15).
Simultaneously with global cell deformation, we measured cell stiffness G' and hysteresivity with OMTC. In line with a previous study (35), vehicle cells experienced significant stiffening and a drop in in response to stretch (Fig. 3). By contrast, stretch had little effect in G' and in thrombin-treated cells. This differential behavior could be attributed to stretch-induced changes in cytoskeletal tension (prestress). Cytoskeletal prestress is thought to be carried by actin microfilaments and intermediate filaments and mainly modulated by cell adhesions to adjacent cells and to the extracellular matrix and by the cell contractile apparatus (16, 30). A number of studies in different cell types have demonstrated that increased prestress is paralleled by increased G' and decreased (26, 29, 35). Given that we observed similar viscoelastic response to stretch in confluent and nonconfluent cells, the differential behavior of vehicle and thrombin-treated cells does not seem to lie in cell-cell and cell-matrix adhesion forces. Indeed, G' and of confluent and nonconfluent cells exhibited parallel behavior in response to stretch. Instead, the absence of stretch-induced stiffening in thrombin-treated cells probably arises from structural changes at the cytoskeleton level. Thrombin has been shown to enhance MLC phosphorylation, which is indicative of an increased number of actin-myosin interactions (18). Disruption of these interactions by cell distention could explain the differential response to stretch of thrombin- and vehicle-treated cells. Indeed, in thrombin-treated cells, the increase in prestress resulting from cytoskeletal distention could be counterbalanced by a loss of cell contractile stress due to detachment of myosin from actin, ultimately leading to unchanged stiffness and hysteresivity after stretch. By contrast, in vehicle-treated cells where levels of MLC phosphorylation are lower (18), the fall in prestress due to myosin disruption would be smaller than the increase in prestress caused by cytoskeletal distention, resulting in the stretch-induced cell stiffening observed. Disruption of actin-myosin interactions has been postulated as a cause of stretch-induced softening in airway smooth muscle tissues (12) but has not yet been observed at the single cell level. Our data suggest that this phenomenon could be inherent not only in smooth muscle cells but also to other cell types. In addition to detachment of actin-myosin interactions, stretch could also provide enough mechanical energy to disrupt or unfold cytoskeletal crosslinks, which might be more abundant in the cortical cytoskeleton of thrombin-treated cells than in untreated cells (18, 34). The structural rearrangements discussed above are consistent with the observed drop of G' on release of the substrate stretch (Fig. 4). Indeed, G' of thrombin-treated cells recovered the untreated baseline levels after stretch release, suggesting that stretch disrupted the cytoskeletal structures that determined thrombin-induced stiffening.
The potential implications of the disruption and rearrangement of cytoskeletal structures induced by a single stretch in cells exposed to thrombin are at least twofold. First, it is well known that the cytoskeleton is the scaffold that interconnects organelles in the cell and provides a large surface area for proteins to bind. Mechanical breakdown of this scaffold has been shown to influence relevant cellular functions, including apoptosis, which is believed to be a major cause of epithelial cell death in lung injury (17, 21). Second, the observation that application of stretch reversed thrombin-induced stiffening is a new factor to take into account in the balance of forces at the alveolar-capillary barrier. Our results suggest that the force balance paradigm (9, 23) cannot be interpreted in terms of static forces because some of its mechanical determinants, i.e., cell stiffness and contraction, are substantially altered by stretch.
Although the stiffness of thrombin-treated cells recovered a value close to baseline after substrate relaxation, cell hysteresivity remained 20–25% lower than its original untreated baseline (Fig. 4). Cell hysteresivity reflects the degree of solid-like vs. liquid-like behavior of the cell. In addition, recent findings suggest that is a robust measure of cytoskeleton dynamics and remodeling (5, 10, 28). In light of these studies, our findings that thrombin induces a sustained decrease in suggest that thrombin could impair essential functions of alveolar epithelial cells such as migration and proliferation, in which cells require high to change their shape and flow.
In summary, we showed that thrombin impairs the ability of alveolar epithelial cells to follow imposed substrate deformations when cell-cell junctions are reduced or absent, which could result in increased epithelial permeability. This effect was absent in confluent cell monolayers, which highlights the central role of cell-cell junctions in withstanding increased tension in the presence of thrombin and mechanical stretch. Stretching thrombin-treated cells resulted in profound changes in cell mechanics possibly due to disruption and rearrangement of cytoskeletal structures. Overall, our findings suggest that thrombin has a potential role in the pathogenesis and resolution of ALI and ARDS in patients subjected to mechanical ventilation. Our novel experimental approach can be used to assess the contribution of inflammatory mediators to the structural integrity of the alveolar capillary barrier in mechanically stimulated cell monolayers.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
* X. Trepat and F. Puig contributed equally to this work. 
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. C. Geiger, C. D. Kaufman, A. P. Lam, G. R. S. Budinger, and D. A. Dean Tubulin Acetylation and Histone Deacetylase 6 Activity in the Lung under Cyclic Load Am. J. Respir. Cell Mol. Biol., January 1, 2009; 40(1): 76 - 82. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Trepat and R. Farre Alveolar permeability and stretch: too far, too fast Eur. Respir. J., October 1, 2008; 32(4): 826 - 828. [Full Text] [PDF]
|
|
|
|
 |
|
 E. U. Azeloglu, J. Bhattacharya, and K. D. Costa Atomic force microscope elastography reveals phenotypic differences in alveolar cell stiffness J Appl Physiol, August 1, 2008; 105(2): 652 - 661. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. R. Rana, C. Zobel, E. Saygili, K. Brixius, F. Gramley, T. Schimpf, K. Mischke, D. Frechen, C. Knackstedt, R. H. G. Schwinger, et al. A simple device to apply equibiaxial strain to cells cultured on flexible membranes Am J Physiol Heart Circ Physiol, January 1, 2008; 294(1): H532 - H540. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Chimenti, G. Morici, A. Paterno, A. Bonanno, L. Siena, A. Licciardi, M. Veca, W. Guccione, F. Macaluso, G. Bonsignore, et al. Endurance Training Damages Small Airway Epithelium in Mice Am. J. Respir. Crit. Care Med., March 1, 2007; 175(5): 442 - 449. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Gavara, R. Sunyer, P. Roca-Cusachs, R. Farre, M. Rotger, and D. Navajas Thrombin-induced contraction in alveolar epithelial cells probed by traction microscopy J Appl Physiol, August 1, 2006; 101(2): 512 - 520. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
