Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cells of the respiratory tract are submitted to normal cyclic stretch during tidal respiration. In some situations, such as positive pressure mechanical ventilation or during severe asthma attacks, bronchial and/or alveolar cells can be elongated to supranormal values, a condition referred to as overstretching (1). This has been proposed as a possible mechanism for cell activation (2), and may participate in the pathogenesis of ventilator-induced lung injury (VILI) (2) or in the bronchial inflammation during acute severe asthma. It was recently demonstrated that cyclic stretch induced several proinflammatory genes both in animals (5- 7) and in vitro experiments (8, 9). Among these, the chemokine interleukin (IL)-8 is particularly relevant, because in humans it represents the major chemoattractant for neutrophils, and neutrophil recruitment to the airways is a critical step in the pathogenesis of VILI (10). The likely cellular sources of IL-8 in the lung are the respiratory epithelium (8, 9) and the alveolar macrophage (8). Fibroblasts, smooth muscle cells, and capillary endothelial cells are also candidates for the stretched-induced IL-8 secretion, although this could not be confirmed in in vitro experiments (11). In response to proinflammatory stimuli, the IL-8 production is dependent on at least three signaling pathways: mitogen-activated protein (MAP) kinases (MAPK), nuclear factor (NF)-B, and NF-IL-6 pathways (12), but the precise molecular mechanisms by which IL-8 is activated in response to cell stretching remain not completely understood.

In this study, we set up an in vitro model of cyclic stretch- induced IL-8 production by human bronchial epithelial BEAS-2B cells. Using this model, we investigated various signaling pathways and found that IL-8 secretion induced by cyclic stretch was a transcriptional effect and that MAPK played a pivotal role.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cells

Several human epithelial cell lines were initially screened for IL-8 production in response to cell stretching. Human bronchial epithelial BEAS-2B cells (ATCC number CRL-9609 [Rockville, MD]) were chosen, because they consistently produced IL-8 in response to cell stretch, and because they were derived from primary bronchial cells and immortalized using SV40, rather than originating from undifferentiated lung carcinoma. These cells have kept many features of the primary bronchial epithelial cells (13). BEAS-2B cells were cultured in a 50/50 mixture of Dulbecco's modified Eagle's medium, and F-12 nutrient mixture supplemented with 10% FCS, 2 mM L-glutamine, 10 mM HEPES (Life Technologies, Paisley, Scotland, UK), 50 U/ml penicillin, and 2 µg/ml gentamycin (Garamycin; Essex Chemie AG, Lucerne, Switzerland). Stably transfected BEAS-2B cells were cultured in the same medium supplemented with 150 µg/ml Hygromycin B (Clontech, Palo Alto, CA).

Cell Transfection

Several cell lines derived from BEAS-2B cells were established by stable transfection of artificial promoter constructs of consensus sequences of responding elements of classic transcription factors controlling inflammatory genes. BEAS-2B cells were transfected with tandem repeats of NF-B, activator-protein (AP)-1, glucocorticoid receptor-responding element (GRE), cyclic AMP (cAMP)-responding element (CRE), and heat-shock-responding element (HSE) (Mercury Pathway Profiling Luciferase System; Clontech) driving a firefly luciferase gene. The sequences of the synthetic promoters are shown in Table 1.

Cells were cotransfected with a pHygEGFP vector (Clontech) for selection. One stable cell line was established for each promoter construct. Cell sorting by flow cytometry using the green fluorescent protein (GFP) was used to obtain similar expression levels in all cell lines. Stably transfected cells homogeneous for their GFP expression were submitted to cell stretching as described below. Positive controls consisted of cells in static conditions incubated with specific agonists: tumor necrosis factor (TNF)- for NF-B, phorbol ester (PMA) for AP-1, dexamethasone for GRE, isoprenalin for CRE, and heat shock (42°C for 30 min) for HSE. Stretched and control cells were lysed and cell lysates were assayed for luciferase activity using Promega's Luciferase Assay System protocol (Madison, WI). Luciferase activity was normalized for the protein concentration of the lysate.

Mechanical Stretch

Native or transfected BEAS-2B cells were seeded on collagen I-coated 6-well BioFlex silastic-bottom culture plate (Flexcell International Corp., Hillsborough, NC) at the concentration of 1.5 × 10⁵ cells/well. Cells were grown in a 5% CO₂ incubator at 37°C for 48-72 h in the BioFlex plates until a single-cell monolayer confluence was achieved. Plates were then transferred to the baseplate of the cell stretching device FX-3000 Flexercell strain unit (Flexcell International), and placed in a 37°C, 5% CO₂ incubator. Before stretching, the culture medium was changed with fresh medium. Cells were stretched using the following protocol: stretching rate of 20 cycles/min with a square signal, a 1:1 stretch:relaxation ratio, and a 20% maximal equibiaxial elongation, in most of the experiments. In some experiments, cells were submitted to 5, 10, and 20% maximal equibiaxial elongation. Control cells were cultured in BioFlex plates but not submitted to cell stretch (static condition). Cells and/or supernatants were harvested after various times, and processed immediately. Cell viability was assessed using a classic DNA-binding propidium iodide (PI; BD Pharmingen, San Jose, CA) staining, with detection of PI-negative living/PI-positive dead cells using flow cytometry. Cell viability was not found to be influenced by cell stretching up to 20% of elongation.

In some experiments, protein kinase inhibitors or their diluent (DMSO) were added to the cells 30 min before cell stretching. These included the nonspecific tyrosine kinase inhibitor genistein used at 50 µM (Sigma, St. Louis, MO), a specific p38 inhibitor (SB 203580, 1 µM, Calbiochem, San Diego, CA), and a specific MEK 1/2 inhibitor (PD 98059, 25 µM, upstream of p44/ 42 MAPK; Calbiochem, San Diego, CA). In other experiments, the inhibitor of transcription actinomycin D (5 µg/ml; Sigma) was added to the cells either 30 min before (measurement of the IL-8 protein) or 2 h after (measurement of IL-8 mRNA by RNase protection assay [RPA]) the beginning of cell stretching.

IL-8 Protein Levels in Supernatants

Cell-free supernatants from control and stretched BEAS-2B were harvested and stored at 20°C until assayed. In some experiments, cells were lysed in a Tris-buffered saline containing 1% Triton-X100, glycerol, antiproteases, and EDTA. Levels of IL-8 in supernatants and in cell lysates were determined by a sandwich ELISA using paired monoclonal antibodies (Endogen, Woburn, MA) as described by the manufacturer. Maximal levels of cyclic stretch-induced IL-8 may vary between experiments. However, the ratio of induced IL-8/baseline IL-8 remained very similar from one experiment to the other, which allowed comparison between experiments.

RNA Extraction and RNase Protection Assay

Control or stretch BEAS-2B cells were lysed after various times in 500 µl Trizol/well (Life Technologies). Total RNA was extracted and resuspended in DEPC-H₂O. Five micrograms of total RNA per condition were used for RPA analyses. The IL-8 probe protected the complete ORF of the IL-8 mRNA (300 nucleotides) and was kindly provided by C. Power (Serono International SA, Plan-les-Ouates, Geneva, Switzerland). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe protected 100 nucleotides (positions 67-166 of the mRNA) and was obtained from USB (Cleveland, OH). Radiolabeled riboprobes were obtained by the transcription of the DNA templates using [-³²P] UTP and the T7 RNA polymerase according to the manufacturer's protocol (Promega). Total RNA/riboprobe mixtures were digested with A and T1 RNases at 30°C for 30 min. Digested RNA mixtures were run on urea gels as previously described (14). Dried gels were submitted to autoradiography and PhosphorImager (Amersham Bioscience, Sunnyvale, CA). The intensity of the protected bands was quantified using the ImageQuant software (Amersham Bioscience). The level of the IL-8 mRNA was defined as the intensity of the IL-8-protected band divided by the intensity of the GAPDH-protected band.

Western Blot

BEAS-2B cells were submitted to cell stretching for 5, 10, 20, 30, and 60 min. Stretched BEAS-2B and control cells were lysed directly into 100 µl/well of SDS sample buffer containing 62.5 mM pH 6.8 Tris-HCl, 2% SDS, 25% glycerol, and 0.01% bromophenol blue. Twenty microliters of cell extract were loaded onto 10% SDS-PAGE, and proteins were electrotransferred to polyvinylidene difluoride membranes (Bio-Rad, Hercules, CA). Membranes were immunoblotted with 1/1,000 dilution of anti-phospho-MAPK antibodies (anti-phospho-p38, anti-phospho-p44/42 MAPK, and anti-phospho-SAPK/JNK were from Cell Signaling Technology, Beverly, MA) and a 1/2,000 dilution of a horseradish peroxidase-conjugated secondary antibody (Cell Signaling Technology) in nonfat dry milk- or BSA-based incubation buffers, according to the manufacturer's recommendations. Detection was performed using enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Sunnyvale, CA) and autoradiography. Membranes were then submitted to a stripping protocol by incubation for 30 min at 52°C in a buffer containing 62.5 mM pH 6.8 TRIS-HCL, SDS 2%, and 100 mM -mercaptoethanol, and reprobed using anti-"total" MAPK antibodies (anti-SAPK/JNK and p44/ 42 MAPK from Cell Signaling Technology, and anti-p38 from Santa Cruz Biotechnology, Santa Cruz, CA). Second steps and detection were the same as described above.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Stretching of BEAS-2B Cells Induces the Production of IL-8

BEAS-2B cells submitted to cyclic stretch produced and secreted IL-8 protein in a dose-and time-dependent manner (Figure 1A and 1B). This was accompanied by an increase in IL-8 cellular mRNA levels as determined by RNase protection assay (Figure 1C). The levels of GAPDH mRNA were not modified by cell stretching (results not shown). After 8 h of cyclic stretch, IL-8 levels were not found to be different in cell lysates from cells submitted to stretch compared with static controls (0.8 ± 0.1 versus 0.7 ± 0.1 ng/µg protein of cell lysate, respectively), whereas IL-8 significantly increased in the supernatants of the same cells submitted to stretch (1.4 ± 0.1 versus 0.7 ± 0.1 ng/ ml). Both the IL-8 protein and the IL-8 mRNA induced by cell stretching were completely inhibited by the presence of Actinomycin D (Figure 2), suggesting an activation of IL-8 gene transcription by cyclic stretch. Importantly, the concentration of Actinomycin D (5 µg/ml) used in these experiments was not toxic for the cells as determined using a cell viability MTT assay (data not shown) (15), and was comparable to that used by others (16). These experiments highly suggest a transcriptional control of the IL-8 gene by cyclic stretch.

View larger version (15K): [in this window] [in a new window]   Figure 1.   IL-8 production by BEAS-2B cells submitted to cyclic stretch. (A) IL-8 protein concentration in supernatants from cells submitted to 5, 10, and 20% stretch for 8 and 24 h. (B) Time course of IL-8 protein concentration in supernatants from cells submitted to 20% stretch (squares) and cells in static conditions (circles). (C) Time course of IL-8 mRNA levels in cells submitted to stretch (squares) and cells in static conditions (circles). The results presented in each panel are representative of three separate experiments.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Effect of transcription inhibition on stretch-induced IL-8 production by BEAS-2B cells. (A) IL-8 protein concentrations were measured in conditioned supernatants after 6 h of cyclic cell stretching in the presence or absence of actinomycin D (Act. D). (B) Time course of IL-8 mRNA levels in BEAS-2B cells submitted to cell stretching, in the presence (circles) or absence (squares) of actinomycin D, added 2 h after the initiation of cell stretching. The results presented in each panel are representative of three separate experiments.

Signaling Pathways Activated by Cell Stretching

To screen for various signaling pathways that may be induced by cell stretching, different BEAS-2B cell lines were established by stable transfection of inducible luciferase under the control of consensus promoter constructs. As shown in Figure 3, all these synthetic promoters were functional as determined by their response to specific agonists, but only AP-1- and cAMP-responding elements were activated by cyclic stretch. A time course (2, 4, 8, and 24 h) was performed with cells transfected with the NF-B tandem repeats. Whereas TNF- induced a significant luciferase activity already after 2 h, stretch did not activate NF-B with this reporter assay throughout the experiment.

View larger version (36K): [in this window] [in a new window]   Figure 3.   Activation of various responding elements by cyclic stretch. BEAS-2B cell lines were established by transfection of various tandem repeats of responding elements: NF-B, AP-1, cyclic AMP (CRE), heat shock (HSE), and glucocorticoid (GRE) responding elements driving a luciferase reporter gene. Upper panel: cells were submitted to cyclic stretch (striped bars) or kept in static conditions (open bars) for 4 h, and then lysed. Luciferase activity was then measured and reported as fold induction compared with static conditions for each cell line. Lower panel: each cell line was incubated for 4 h with specific agonists (striped bars) to demonstrate the inducibility of the promoter constructs. The following agonists were used: recombinant human TNF- (10 ng/ml) for NF-B; PMA (200 ng/ml) for AP-1; isoprenalin (106 M) for CRE; 42°C heat shock during 30 min for HSE; and dexamethasone (1 µg/ml) for GRE. Open bars, no agonist. *P < 0.05 by unpaired Student's t test. The results presented here correspond to a representative experiment out of four similar experiments.

Inhibition of IL-8 Secretion by Inhibitors of Kinases

To investigate potential pathways activated by cyclic stretch, we first used a nonspecific tyrosine kinase inhibitor, genistein, which abolished the IL-8 secretion induced by cyclic stretch (Figure 4).

View larger version (14K): [in this window] [in a new window]   Figure 4.   Effect of genistein (A) (a tyrosine kinase inhibitor), of SB 203580 (B) (a p38 kinase inhibitor), and of PD 98059 (C), an inhibitor of the p44/42 MAPK pathway on the IL-8 production in BEAS-2B cells submitted to cell stretching for 8 h. Results are expressed as mean ± SEM of three different experiments.

Because AP-1 is controlled by stress MAPK (17), and seemed to be activated by cyclic stretch, and because MAPK are known to be downstream of tyrosine kinases, we next tested the potential role of MAPK in stretch-induced IL-8 production. We initially tested the effect of two specific MAPK inhibitors: SB 203580 (p38 inhibitor) and PD 98059 (inhibitor of MEK1/2 upstream of p44/42 MAPK) on stretch-induced increase of IL-8 protein secretion and on IL-8 mRNA cellular contents. Cells were harvested after 2, 5, and 8 h, and supernatants were collected after 8 h of cell stretching. SB 203580, but not the DMSO diluent, completely prevented the increase of IL-8 secretion induced by cyclic stretch (Figure 4). PD 98059 consistently decreased the secretion of IL-8 induced by cyclic stretch by 70% (Figure 4). Neither SB 203580 nor PD 98059 had any effect on cell viability, as measured by an MTT assay (data not shown). To further evaluate the possible role of MAP kinases on IL-8 transcription, we tested the effects of these two inhibitors on stretch-induced IL-8 mRNA levels determined by RPA. We found that SB 203580 almost totally prevented the increase in IL-8 mRNA induced by cyclic stretch (Figure 5). PD 98059 did not have a significant effect on IL-8 mRNA levels induced by stretch as shown in Figure 5.

View larger version (42K): [in this window] [in a new window]   Figure 5.   IL-8 mRNA expression by BEAS-2B cells submitted to cyclic stretch in the presence or absence of MAPK inhibitors SB 203580 (upper panels) and PD 98059 (lower panels). Left panels: autoradiographs (one representative experiment) of gels from RNase protection assays. RNAs were protected with 32P-radiolabeled IL-8 and GAPDH riboprobes. Right panels: quantification of bands (PhosphorImager, ImageQuant), presented as mean ± SD: squares, stretch; circles, static; and triangles, stretch in the presence of the MAPK inhibitor.

MAPK Phosphorylation Induced by Cell Stretching

We further investigated whether stretch did induce MAPK phosphorylation, a step necessary for MAPK activation. This was done by the detection of phosphorylated forms of MAPK by Western blot using specific phospho-MAPK antibodies. Cyclic stretch of BEAS-2B cells resulted in a rapid and specific phosphorylation of all MAPK tested (p44/42 MAPK, p38, and SAPK/JNK) (Figure 6). Phospho-MAPK were detected after 5 min of cell stretching (p38) and 10 min for p44/42 MAPK and SAPK/JNK. Phospho-p38 peaked at 10 min and then decreased and was nearly undetectable at 60 min. Phospho-p44/42 MAPK and phospho-SAPK/JNK peaked at 20 min, and remained detectable at 60 min. Unphosphorylated MAPK bands were similar in stretched and unstretched cells (Figure 6).

View larger version (61K): [in this window] [in a new window]   Figure 6.   Early p44/42 MAPK (A), p38 (B), and SAPK/JNK (C) phosphorylation in BEAS-2B cells submitted to cell stretching. Phosphorylation of the MAPK was detected by immunoblot on cell lysates using antibodies specific to the phosphorylated forms of the kinases (phospho-MAPK, upper panels). Equal loading of cell lysates was determined by immunoblotting with antibodies recognizing phosphorylated and unphosphorylated forms of MAPK (lower panels). The results presented here correspond to a representative experiment out of three similar experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Lung neutrophilic inflammation is a hallmark of the acute respiratory distress syndrome (ARDS), VILI, and acute severe asthma (ASA) (18). Overstretching of airways is believed to play an important pathogenic role in these conditions, and the degree of airway distention conditions the recruitment of neutrophils. During VILI, the chemokine IL-8 is responsible in great part for the recruitment of blood neutrophils to the airspace (2, 3, 20). IL-8 secretion by lung macrophages, human alveolar type II-like A549 cells, and endothelial cell submitted in vitro to cyclic cell stretching has been reported by several groups (8, 9, 11, 22, 23). However, the mechanisms of the IL-8 production remained to be determined. Herein, we show that cell stretching of human bronchial epithelial cells results in an increase in IL-8 mRNA expression, and IL-8 protein secretion, and that this process is dependent on MAPK activation.

The degree of lung epithelial cell stretching in patients submitted to mechanical ventilation is largely unknown, particularly in patients with diseased lungs, or with hyperinflation due to bronchospasm. Using in vitro models, several investigators have attempted to quantify the elongation factor of cells from alveoli, alveolar ducts, or bronchioles in response to variation of lung volumes (24, 25). The alveolar surface area was quantified in animal isolated lungs by both light and electron microscopy techniques (24). Changes in lung volumes from 40% of the total lung capacity (TLC) to 80% of the TLC were associated with variations of 8-28% of the alveolar surface area, which corresponded to alveolar cell elongation factors of 3.6- 12.5%, respectively, assuming that the alveolus is perfectly spherical. Inflating isolated rat lungs from residual capacity to 80 and 100% of the TLC resulted in increased alveolar surface areas of 25-37%, which translated into clinically relevant transpulmonary pressures (12 and 25 cm H₂O, respectively) (24, 27). The alveolar stretching factors studied in isolated lungs were somewhat smaller than the cell elongation factor most frequently used in our study (20%). However, a 10% elongation factor, more compatible with these morphometric studies, also induced bronchial cells to release IL-8 in our in vitro model.

The cell type used in our studies is derived from human bronchial cells. Interestingly, Goldstein and coworkers found bronchiolar dilatation in diseased areas of infected pig lungs submitted to mechanical ventilation (26). The transverse section area of bronchioles contained in consolidated parenchyma and measured by light microscopy was increased by 100% when compared with unaffected regions (26). Assuming that bronchioles are cylinders, such a degree of distention translates into a 41% elongation factor. However, this represents a static measurement, and the degree of stretch of these structures during tidal ventilation is unknown. Yager and colleagues measured the elongation of various parenchymal structures in lungs from dogs and humans submitted to isotropic biaxial stretch in vitro corresponding roughly to tidal breathing. They found that alveolar ducts stretched more than the rest of the parenchyma by a factor of 3.8% in humans and 10.3% in dogs, suggesting that small tubular airway structures may well be submitted to overstretching (28).

In our model of cyclic bronchial cell stretching, IL-8 was released in a dose-dependent manner by cells submitted to 5, 10, and 20% stretching. The elongation factor used in most of our cyclic cell stretching experiments was 20%. This level of cell elongation was chosen because it induced a maximal IL-8 production and did not affect cell viability. Tschumperlin and Margulies had previously shown that stretch-induced loss of plasma membrane integrity in alveolar type II cells and secondary cell death were dependent on the elongation factor, the time during which the cells were kept adherent on the silicone membrane, and the seeding density (27). The authors concluded that type II cells in culture acquired stretch resistance due to type I- like phenotypic changes (27). Elongation factors of 20- 30% were also commonly used by many investigators in in vitro models of lung cell stretching (9, 29). Nevertheless, extrapolation of these in vitro studies to human physiology and pathology requires great care.

IL-8 production in response to proinflammatory stimuli is generally regulated both at transcriptional and post-transcriptional levels (33). Our data, essentially based on actinomycin D experiments, strongly suggest that cell stretching induces IL-8 transcription. However, one cannot exclude an additional stabilizing effect of stretch on IL-8 mRNA (12, 34). IL-8 transcription has been shown to be under the control of various signaling pathways. These include NF-B, AP-1, and NF-IL-6 (12, 33). Using artificial promoter constructs, we could show that at least two classic responding elements were activated by cyclic stretch: AP-1 and CRE. These results should, however, be taken with caution, because the levels of induction of the reporter gene were low, as compared with natural agonists. In addition, the use of synthetic tandem repeats of responding elements also has limitations in that they do not allow cooperation of different transcription factors, frequently required for efficient gene transcription.

Both AP-1- and cAMP-responding elements bind dimers of basic leucin zipper family proteins, downstream targets of MAPK, such as c-jun, c-fos, and transcription factors of the ATF/CREB family (17, 35). This is consistent with the report of Du and coworkers. These authors reported activation of AP-1, CRE, and NF-B in endothelial cells submitted to cyclic strain. Whereas AP-1 and NF-B sites have been identified in the 5' region of the human IL-8 gene, CRE is not found in this promoter. Glucocorticoid- and heat shock-responding elements, which have not been described in the promoter region of IL-8, nor linked with mechanical stress-induced cell activation, were not turned on in our study. NF-B had been previously shown to be activated by cyclic stretch in an in vitro model of stretched macrophages (8), as well as in animals submitted to injurious ventilatory regimen (5). Using a synthetic promoter construct of tandem repeats of a consensus sequence for NF-B, we did not observe increased luciferase activity after cell stretching. This could be explained by different pathways of cell activation in macrophages versus epithelial cells, as nicely demonstrated for endothelial cells, which show divergent transcription factor activation to cyclic strain depending on the vascular bed origin (38). Another possibility is that nuclear translocation and DNA binding of NF-B was not sufficient for gene transcription. Indeed, a critical phosphoinositide-3 kinase (PI3K)/Akt-dependent nuclear phosphorylation of p65 is required to confer its transcriptional activity (39, 40). Therefore, cyclic stretch may induce the initial NF-B activation and nuclear translocation, but not the secondary p65 phosphorylation. Alternatively, NF-B may be activated but require other transcription factors to cooperate with to turn on inflammatory genes (41).

The finding that AP-1 was activated by cyclic stretch prompted us to investigate the role of MAPK in the signal transduction pathway induced by cyclic stretch. MAPK of the three subclasses, p44/42, JNK, and p38, were strongly activated as shown by the rapid and sustained phosphorylation of these proteins upon cell stretching. Similar findings were recently reported by Sawada and colleagues. These authors showed that MAPK were activated by cyclic stretch in human embryonic kidney (293) cells and in mouse L929 fibroblasts (42). More importantly, specific inhibitors of MAPK or of upstream kinases were able to block the cell stretch-IL-8 production, both at the mRNA and at the protein level. The p38 inhibitor SB203580 abrogated the IL-8 signal, whereas an inhibitor of MEK1/2 a MAPK kinase upstream of p44/p42 MAPK only partially inhibited IL-8 production. We were not able to test JNK pharmacologic inhibitors because such compounds are not commercially available. This strongly suggested a prominent role of MAPK, and particularly of p38, in controlling IL-8 expression in response to mechanical stress in bronchial epithelial cells. Interestingly, p38 activation was recently implicated in IL-8 mRNA stabilization (12, 34). This stabilizing effect may well participate in the observed increase in IL-8 production by BEAS-2B cells submitted to cell stretching. Importantly, p38 was already implicated in the IL-8 production by lung cells with cell stimulation other than cell stretching (proinflammatory cytokines, rhinoviruses, and hyperosmolarity) (43). The phosphorylation and the activation of p38 by cyclic stretch could be the consequence of the activation of Rap-1, a small G-protein upstream of MAPKKK and MEK1/6 (42). Whether this protein participates directly to the mechanosensing apparatus remains to be determined. Activation of focal adhesion kinase (FAK) has been reported by several groups in response to shear stress and cyclic stretch in various cell types (46, 47). This may represent one of the very proximal steps of cellular mechanosensing as well as be responsible for the activation of the Grb2/Sos/ Ras pathway. Whereas it seems that Ras is activated by shear stress (48), Ras activation in response to cyclic stretch has been found inconsistently (42, 49). In endothelial cells, both PI3K-dependent and -independent pathways of Ras activation have been described (48). Cell stretch is a mechanical strain which resembles that of shear stress, and certainly shares similar activation and signaling pathways (4, 46). The role of calcium and inositol phosphates which was not studied here may also play a role in the mechanosensing and in the activation process of cells submitted to both types of mechanical strains (50, 51). A scheme (Figure 7) illustrates our current hypotheses of the molecular mechanisms responsible for IL-8 production in bronchial cells submitted to cyclic stretch.

View larger version (25K): [in this window] [in a new window]   Figure 7.   Putative sequence of signaling events, and IL-8 production in human BEAS-2B bronchial epithelial cells submitted to cell stretching. MAPK are activated in response to cyclic stretch. The blockade of p38 by the pharmacologic compound SB 203580 abolished IL-8 production. IL-8 production was only partially inhibited by PD 98059, an inhibitor of MEK1, upstream of p44/42 MAPK. The IL-8 gene is activated in an AP-1-dependent manner. p38 may play a role in stabilizing IL-8 mRNA.

In conclusion, we show that cyclic stretch of human bronchial epithelial cells induces IL-8 secretion in a MAPK-dependent manner. The p38 kinase is rapidly activated after initiation of cyclic stretch, and seems to play a crucial role, because its pharmacologic blockade abrogates IL-8 production. p38 inhibitors may therefore represent a promising therapeutic strategy in lung inflammatory diseases where airway overdistention occurs.