|Home | About | Journals | Submit | Contact Us | Français|
Transforming Growth Factor Beta (TGF-β) is involved in regulating many biological processes and disease states. Cells secrete the cytokine as a latent complex that must be activated for it to exert its biological functions. We previously discovered that the epithelial-restricted integrin αvβ6 activates TGF-β and that this process is important in a number of in vivo models of disease. Here, we show agonists of G-protein coupled receptors (Sphingosine-1-Phosphate and Lysophosphatidic Acid) that are ligated under conditions of epithelial injury directly stimulate primary airway epithelial cells to activate latent TGF-β through a pathway that involves Rho Kinase, non-muscle myosin, the αvβ6 integrin, and the generation of mechanical tension. Interestingly, lung epithelial cells appear to exert force on latent TGF-β using sub-cortical actin/myosin rather than the stress fibers utilized by fibroblasts and other traditionally “contractile” cells. These findings extend recent evidence suggesting TGF-β can be activated by integrin-mediated mechanical force and suggest that this mechanism is important for an integrin (αvβ6) and a cell type (epithelial cells) that have important roles in biologically relevant TGF-β activation in vivo.
Transforming Growth Factor Beta (TGF-β) is a central mediator in multiple in vivo disease models, including fibrosis in the lungs, kidney, and biliary tract [1–5], acute lung injury [6, 7], and pulmonary emphysema . Cells secrete the pleiotropic cytokine as a large latent complex, which must be activated to exert its biological functions. There are many ways to activate latent TGF-β in vitro, for example, via heat, acidic pH, and matrix-metalloproteinases. However, the relevance of these processes in vivo is poorly understood . Several studies have demonstrated that a subset of integrins that interact with an arginine-glycine-aspartic acid (RGD)-binding site motif also directly bind and activate latent TGF-β [1, 10].
Integrins are a widely expressed family of cell surface receptors that mediate cell adhesion and bidirectional signaling to regulate cellular processes. The importance of integrins in TGF-β activation in vivo is highlighted by transgenic mice that harbor a knock-in mutation of Tgfb1, so that the RGD binding site is mutated to RGE. These mice exhibit multi-organ inflammation and autoimmunity, a phenotype that completely phenocopies that of Tgfb1−/− mice, presumably due to a lack of binding and activation of TGF-β by RGD-binding integrins . However, the only RGD integrins that have been definitively shown to regulate TGF-β activation in vivo are αvβ6 and αvβ8 [1, 12]. In particular, Itgb8−/− mice treated with blocking antibodies against αvβ6 from birth demonstrate all of the developmental phenotypes of mice lacking both TGF-β isoforms 1 and 3 .
We previously demonstrated that the epithelial-restricted integrin αvβ6 activates TGF-β and that Itgb6−/− mice exhibit a phenotype mildly similar to Tgfb1−/− mice [1, 14]. These mice are also dramatically protected in models of diseases that are mediated by TGF-β, supporting a role for αvβ6 in regulating TGF-β bioactivity in vivo [1, 4, 5, 9, 14]. Because of the importance of αvβ6-mediated TGF-β activation in the pathogenesis of various disease states, we wanted to determine the molecular signals and mechanisms that regulate activation of this pathway.
Binding of αvβ6 to latent TGF-β1 and 3 in vitro is insufficient for activation of the cytokine, and cytoplasmic interactions between the integrin and the actin cytoskeleton are required for this process . However, the mechanism and significance of these interactions in modulating αvβ6-mediated TGF-β activation is unknown. Furthermore, tethering of the latent complex to the extracellular matrix (ECM) by latent TGF-β binding protein-1 (LTBP-1) is required . These findings suggest that TGF-β activation by αvβ6 potentially involves a mechanical mechanism. In vitro evidence supporting a role for mechanical force in integrin-dependent TGF-β activation has been provided by studies of αvβ3 and αvβ5-mediated TGF-β activation in fibroblasts , but thus far, there are no convincing data demonstrating roles for either of these integrins in activating TGF-β in vivo. The recently solved crystal structure of the small latent complex of TGF-β1 provides a model for how an RGD-binding integrin could exert mechanical force on a tethered latent complex and activate it. Electron microscopic analysis of purified integrin αvβ6 and the small latent complex provided further evidence that this integrin binds to the latent complex in an appropriate fashion to exert activating mechanical force . However, the relevance of mechanical force in TGF-β activation in vivo remains unknown.
The purpose of this study was to identify signals that regulate αvβ6-mediated TGF-β activation and determine what role, if any, mechanical force plays in the activation process. Here, we describe a novel activator, Sphingosine 1-Phosphate (S1P), that regulates αvβ6-mediated TGF-β activation in primary lung epithelial cells. We also demonstrate that both S1P and Lysophosphatidic Acid (LPA) induced αvβ6-mediated TGF-β activation require cell contraction and the generation of cellular tension. Finally, we show that cell tension generated by these primary epithelial cells is associated with organization of sub-cortical actin rings without evidence of actin stress fibers. Our findings highlight the potential for the development of additional therapeutic strategies for diseases that involve aberrant αvβ6-mediated TGF-β activation signaling and demonstrate a role for mechanical force in mediating TGF-β activation in a cell type and by an integrin that has been clearly shown to be relevant in vivo.
NHBE cells were cultured at 37°C, 5% CO2 (Lonza, Walkersville, MD, USA). The β6-blocking antibody, 3G9, was provided by Paul Weinreb and Shelia Violette (Biogen Idec and Stromedix, Cambridge, MA, USA), and the pan-TGF-β blocking antibody, 1D11, was purchased (R&D Systems, Minneapolis, MN, USA). Agonists and inhibitors used were S1P, LPA, Y-27632, and Blebbistatin (Sigma-Aldrich, St. Louis, MO, USA), and active human recombinant TGF-β1 (R&D Systems, Minneapolis, MN, USA).
NHBE cells were cultured as a monolayer in a 48-well plate (Corning, El Sobrante, CA, USA). Cells were pre-treated with blocking antibodies or inhibitors and S1P or LPA added. Cell lysates were harvested after two hours using RIPA buffer (50 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% NP-40, 1% Sodium Deoxycholate, 0.1% SDS) containing a cocktail of protease inhibitors and phosphatase inhibitors (50 mM NaF, 1 mM Na3VO4, 1 mM Beta-glycerophosphate, 2.5 mM Sodium Pyrophosphate) (Sigma-Aldrich, St. Louis, MO, USA). Protein was measured using a Bradford Lowry assay (Bio-Rad Laboratories, Hercules, CA, USA), and 20–50 μg of protein was separated on an SDS-PAGE gel and transferred onto a PVDF membrane (Amersham Biosciences, Chicago, IL, USA). The membrane was incubated with 3% nonfat dry milk in Tris-buffered saline (50 mM Tris, 150 mM NaCl, pH=7.4) with 0.05% Tween-20 before probing with a primary rabbit polyclonal anti-pSmad2 or a mouse monoclonal anti-Total Smad2 (Cell Signaling Technology, Danvers, MA, USA) at 4°C overnight. An HRP-conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was used. Proteins were detected using ECL Plus (Amersham Biosciences, Chicago, IL, USA). Blots were stripped using Western Restore stripping buffer (Pierce Biotechnology, Rockford, IL, USA) prior to examination of Total Smad2.
The ratio of pSmad2 density to that of Total Smad2 for at least three independent Western blot experiments was analyzed using Image J software. Data are expressed as means ± standard error of the mean (s.e.m.). We performed an analysis of variance with a post hoc Dunnett's or Fisher PLSD's test to make comparisons within the data. Values of P < 0.05 were considered statistically significant.
NHBE cells were seeded onto collagen I coated transwells (Corning, El Sobrante, CA, USA) in bronchial epithelial differentiation medium (BEDM), consisting of 50% Bronchial Epithelial Basal Medium (BEBM) and 50% Dulbecco's Modified Eagle Medium (DMEM) (Mediatech Inc., Manassas, VA, USA) supplemented with all of the included SingleQuots (Lonza, Walkersville, MD, USA) except: retinoic acid, gentamycin/amphotericin, and triiodothyronine. Retinoic acid solution was added to a final concentration of 50 nM. Cultures were maintained for two weeks containing media on the apical and basal side of the transwells. Afterwards, only the basal side contained media for an additional two weeks. Agonists or inhibitors were added to the basal side, and cell lysates were harvested after two hours.
Polyacrylamide flexible substrates were created according to the protocol described [18, 19]. Briefly, 12 mm round coverslips were treated with 0.1 N NaOH, 3-aminopropyltrimethoxy silane and 0.5% glutaraldehyde (Sigma-Aldrich, St. Louis, MO, USA). After “activation”, 12 μL of polyacrylamide mixture (bis-acrylamide, acrylamide, 50 mM HEPES, pH=8.5, Ammonium Persulfate, Temed) (Bio-Rad Laboratories, Hercules, CA, USA) was pipetted onto the coverslips. Another 12 mm coverslip was placed directly on top. After polymerization, the top coverslip was removed and substrates were rinsed with 50 mM HEPES. The substrates were UV irradiated with 0.5 mg/mL Sulfo-SANPAH (Pierce Biotechnology, Rockford, IL, USA) and coated with collagen I (Sigma-Aldrich, St. Louis, MO, USA) at 4°C overnight. Cell lysates were harvested after two hours.
NHBE cells were seeded onto collagen I (Sigma-Aldrich, St. Louis, MO, USA) coated 12 mm coverslips (Fisher Scientific, Santa Clara, CA, USA). Cells were pre-treated with inhibitors and S1P or LPA was added. After two hours, cells were fixed with 1% paraformaldehyde, permeabilized with 0.1% Triton-X, and stained with rhodamine-phalloidin (Invitrogen, Carlsbad, CA, USA). Coverslips were mounted onto slides using ProLong Gold Anti-Fade mounting medium (Invitrogen, Carlsbad, CA, USA). Images were captured using a Leica DMI6000B inverted microscope (Leica Microsystems Inc., Buffalo, IL, USA).
S1P is a G-protein coupled receptor (GPCR) agonist that is abundantly stored in platelets and released at sites of increased TGF-β signaling, such as inflamed or injured tissues. The phospholipid induces contraction in airway smooth muscle cells and epithelial cells and has been implicated to play a role in tissue fibrosis [20, 21]. To determine whether S1P could induce αvβ6-mediated TGF-β activation in epithelial cells, primary normal human bronchial epithelial (NHBE) cells were treated with increasing concentrations of S1P in the presence or absence of a β6-blocking antibody, and pSmad2 was examined as an early readout of TGF-β activity. We found that S1P increased pSmad2 levels in a dose-dependent manner, which was reduced in the presence of a β6-blocking antibody (Fig. 1A). We then examined whether the Rho/Rho Kinase (ROCK) pathway was involved in mediating this process, since this pathway plays an important role in actin remodeling and the actin cytoskeleton is required for β6-mediated TGF-β activation . NHBE cells treated with Y-27632, an inhibitor of ROCK1 and 2, no longer showed induction of αvβ6-mediated TGF-β activation in the presence of S1P (Fig. 1B). Finally, to obtain the best in vivo representation of the airway epithelium, we differentiated NHBE cells by culturing them at an air-liquid interface, so that the apical side of the cell is exposed to air, and the basal side to growth media. S1P also induced αvβ6-mediated TGF-β activation under this more physiologic condition, which was blocked by treatment with Y-27632 (Figure 1C).
Our data demonstrate that S1P induces αvβ6-mediated TGF-β activation by acting through ROCK and suggests the potential importance of this pathway in vivo. However, while the primary tissue culture system we used does allow airway epithelial cells to differentiate to resemble their counterparts in vivo, we cannot be certain that epithelial cells in intact organs would utilize an identical pathway. Multiple cell types secrete TGF-β; for example, platelets are a major source for this cytokine in vivo. The pathway we describe could activate TGF-β secreted by platelets that have exited the vasculature in epithelial organs. However, platelets themselves do not express the αvβ6 integrin, so it is likely that other mechanisms are involved in activation of TGF-β secreted by platelets at a distance from αvβ6-expressing epithelial cells.
We previously reported in β6-transfected mouse embryonic fibroblasts and NHBE cells that thrombin and LPA, agonists of GPCRs, also induce αvβ6-mediated TGF-β activation by acting through ROCK [7, 22]. As diverse molecules that converge on the Rho/ROCK pathway cause activation of TGF-β by αvβ6, this pathway may represent a potential therapeutic target to modulate αvβ6-mediated TGF-β activation.
We previously determined that the actin cytoskeleton plays an important role in αvβ6-mediated TGF-β activation. In addition, the finding that this process requires physical tethering of latent TGF-β to the ECM by LTBP-1 , suggests that integrin-mediated application of physical force may be required to activate the tethered latent cytokine. We hypothesized that this process is the result of increased actin/myosin contraction within the epithelial cells. To test this possibility, we treated NHBE cells with Blebbistatin, a specific inhibitor of non-muscle myosin II, and examined its effects on TGF-β activation in response to S1P. We found that Blebbistatin inhibited both S1P and LPA-induced αvβ6-mediated TGF-β activation (Fig. 2A,B). To assess whether actin/myosin-mediated induction of TGF-β activation by αvβ6 required the generation of cellular tension, we cultured NHBE on flexible polyacrylamide gel substrates with varying degrees of stiffness. If generation of tension is required for αvβ6-mediated TGF-β activation, we would not expect the latent cytokine to be activated on flexible substrates, which would be deformed by cell contraction and thus not allow transmission of force to the bound latent complex. In contrast, contraction of cells cultured on rigid substrates should allow transmission of force to the bound latent complex and facilitate its activation. We observed that NHBE cultured on a low rigidity substrate did not activate TGF-β upon stimulation with S1P, whereas cells cultured on a stiff substrate activated the cytokine (Fig. 2C). These results show that induction of αvβ6-mediated TGF-β activation by diverse lipid GPCR agonists requires the generation of actin/myosin-mediated cell contractility.
Epithelial cells are not generally thought of as contractile cells. However, contraction of these cells has been demonstrated to be important in maintaining the integrity of epithelial protective barriers . In endothelial cells and fibroblasts, the actin/myosin cytoskeleton is organized into stress fibers that mediate cellular contraction and exert force on the extracellular environment. In contrast, cortical actin mediates contraction in epithelial cells . Consistent with this finding, we found that treatment with S1P and LPA did not induce stress fiber formation in NHBE cells, but reorganized actin and myosin into cortical structures. We also found that Y-27632 and Blebbistatin treatment inhibited cortical actin formation, suggesting that reorganization of actin and cellular contraction are required for TGF-β activation in epithelial cells by the αvβ6 integrin (Fig. 3A,B). Finally, we observed cortical actin structures in S1P and LPA treated NHBE cells cultured at an air-liquid interface. Treatment with Y-27632 and Blebbistatin also inhibited cortical actin formation under this condition, suggesting that contraction of epithelial cells in vivo may result in αvβ6-mediated TGF-β activation (M.M.G., unpublished results).
Previously published work demonstrating an important role for linkage of the αvβ6 integrin to the actin cytoskeleton  and a requirement for LTBP-1 to tether the latent complex to the ECM , provided indirect evidence that mechanical force might regulate activation of latent TGF-β by the αvβ6 integrin. The recently solved crystal structure of the small latent complex and parallel electron microscopic results confirming the predicted structural interaction between the αvβ6 integrin and the small latent complex provided further evidence that activation by mechanical force was plausible . We now demonstrate that non-muscle myosin II is also required and that αvβ6-mediated TGF-β activation only occurs on substrates that are rigid enough to allow the cell to generate retractile force. Our findings together suggest that contractile force and mechanical deformation of that latent complex is responsible for activation of TGF-β by the αvβ6 integrin on epithelial cells.
Increased stiffness in the ECM may thus constitute a positive feedback loop to promote the progression of fibrotic disorders by stimulating αvβ6-mediated TGF-β activation. Cytoskeletal tension itself is influenced by the stiffness of the ECM substrate, which may further amplify this positive feedback loop. Overall, such a positive feedback loop likely contributes to progression of tissue fibrosis, since excessive scar tissue and rigidity of the matrix may lower the threshold of cell contractile stimuli necessary for subsequent TGF-β activation.
This work was supported by grant R37HL53949 from the NHLBI to D.S., U19AI077439 from the NIAID to D.S., and by a Pre-doctoral Fellowship in Pharmacology/Toxicology from the PhRMA Foundation to M.M.G.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of interest: The authors declare that no conflict of interest exists.