Abstract
The migratory and invasive potential of the epithelial-derived tumor cells depends on epithelial-to-mesenchymal transition (EMT) as well as the reorganization of the cell cytoskeleton. Here, we show that the tricyclic compound acetylenic tricyclic bis(cyano enone), TBE-31, directly binds to actin and inhibits linear and branched actin polymerization in vitro. Furthermore, we observed that TBE-31 inhibits stress fiber formation in fibroblasts as well as in non–small cell lung cancer cells during TGFβ-dependent EMT. Interestingly, TBE-31 does not interfere with TGFβ-dependent signaling or changes in E-cadherin and N-cadherin protein levels during EMT. Finally, we observed that TBE-31 inhibits fibroblast and non–small cell lung tumor cell migration with an IC50 of 1.0 and 2.5 μmol/L, respectively. Taken together, our results suggest that TBE-31 targets linear actin polymerization to alter cell morphology and inhibit cell migration. Cancer Prev Res; 7(7); 727–37. ©2014 AACR.
Introduction
Cell migration is essential in numerous physiologic processes such as embryogenesis, cell differentiation, and cell renewal. It is also key in the later stages of cancer metastasis, where cell migration and invasion allow cancer cells to establish secondary tumor sites, which accounts for more than 90% of cancer-related deaths (1).
Epithelial cells are usually organized in an apical-basal polarity, and multiple cell-to-cell adhesions help assemble epithelial cells into a sheet-like formation (2–5). For epithelial-based tumors to detach from the primary tumor and migrate, the cells will have to dissociate these cell-to-cell contacts in a process known as the epithelial-to-mesenchymal transition (EMT; refs. 2–5). These biochemical changes, among others, liberate the epithelial cells and enable them to assume a much more mesenchymal-like phenotype and prime them for cell migration and invasion (2–5).
Cell migration begins with an initial protrusion of the cell membrane. Newly forming cell adhesions at the front attach the protrusions to the substratum and provide traction, whereas the focal adhesions at the rear disassemble and contractions help retract the tail (6–8). This process relies on the localization of polarity proteins and the reorganization of different components of the cytoskeleton, which consists of the microtubule network, intermediate filaments, and the actin cytoskeleton. Although all the cytoskeletal components act in concert, it is the actin cytoskeleton that provides the force necessary for translocation. The localization of Rac1 or Cdc42 toward the leading edge with other polarity proteins is a key regulatory event that stimulates actin polymerization and results in membrane protrusion toward the direction of migration. G-actin subunits polymerize to form actin filaments creating filopodia and lamellipodia. Cdc42 and formins regulate actin polymerization to form long unbranched bundles of actin to form filopodia, which are thought to act as sensors that can probe for external cues. Rac1 activates actin-related proteins 2 and 3 (Arp2/3), causing it to bind to preexisting bundles of actin and promote branched actin polymerization to form lamellipodia, which are broad sheet-like protrusions that drive the cell forward (6–8). Bundles of actin also form stress fibers to give a cell its shape and work with focal adhesions and myosin to produce contractile forces in cell migration (9, 10). Because EMT and cell migration are important processes in tumor cell metastasis, chemotherapeutic drugs that target various aspects of these processes are continuously sought.
Triterpenoids are a family of naturally occurring compounds synthesized in plants by the cyclization of squalene. They are a chemically diverse family with more than 20,000 known compounds (11, 12). Oleanolic acid is of particular interest because of its anti-inflammatory and antitumorigenic properties (11–14). Continuous synthetic modifications on oleanolic acid led to 2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid (CDDO), which has much higher anti-inflammatory and cytoprotective potency than oleanolic acid, as well as its imidazolide (CDDO-Im) and its methyl ester (CDDO-Me; refs. 11–16). They have been demonstrated to effectively inhibit cytokines from inducing nitric oxide synthase in macrophage cells and inducing phase II cytoprotective enzymes, both of which help prevent a cancer-causing environment. In addition, they were found to inhibit proliferation, and induce differentiation and apoptosis in cancer cells in vitro. They have also been shown to inhibit carcinogen-induced primary tumor growth, orthotopic and ectopic tumor formation, as well as lung metastasis and in multiple experimental animal models (11, 12, 17–22).
Recently, we have observed that CDDO-Im and CDDO-Me have affect the cytoskeleton, leading to the interference of physiological processes (23, 24): CDDO-Im was observed to disrupt the microtubule network, displacing the polarity proteins IQGAP1 and RAC1 from the leading edge of migrating cells to abrogate cell polarity (23). Branched actin polymerization depends on the binding of actin-related protein 2/3 (Arp2/3) to preexisting bundles of actin to provide a nucleation site and allow for branches of actin to polymerize, and CDDO-Im and CDDO-Me were both found to inhibit the actions of Arp2/3 (24). Together, these actions potently inhibit cell migration (23, 24).
Structure-activity studies on the parental pentacyclic compounds have revealed that the 2-cyano-1-en-3-one in ring A and a 9(11)-en-12-one in ring C are responsible for the potency of these compounds (Fig. 1A). Therefore, new tricyclic compounds, termed tricyclic bis(enone) compounds (TBE), were generated, which have great potential for structural diversity, robust pharmacokinetic, and pharmacodynamic profiles (25–27). In the pool of semisynthetic triterpenoids and synthetic TBEs, acetylenic tricyclic bis(cyano enone) (TBE-31; Fig. 1B) is one of the most potent compounds (25–27). The oral administration of TBE-31 resulted in a profound and dose-dependent induction of the cytoprotective enzymes, NAD(P)H:quinone oxidoreductase 1 (NQO1) and glutathione S-transferases in the stomach, skin, and liver of mice, and indicated excellent oral bioavailability (28). Also, long-term topical daily applications of TBE-31 caused a robust systemic induction of the Keap1/Nrf2/ARE pathway and decreased 6-thioguanine incorporation in DNA of skin, blood, and liver of azathioprine–treated mice, indicating extraordinary bioavailability and efficacy (29). Furthermore, TBE-31 is orally highly protective against aflatoxin–induced liver cancer in rats (30).
Chemical structures of pentacyclic triterpenoids and TBE-31. A, the basic pentacyclic triterpenoid structure. CDDO-Im, which contains an imidazolide group, is used in this study. B, TBE-31, and below is shown the biotinylated TBE-31, TBE-56.
Chemical structures of pentacyclic triterpenoids and TBE-31. A, the basic pentacyclic triterpenoid structure. CDDO-Im, which contains an imidazolide group, is used in this study. B, TBE-31, and below is shown the biotinylated TBE-31, TBE-56.
To examine protein targets of TBE-31, a biotin conjugate of TBE-31, TBE-56, obtained from TBE-55, was designed and synthesized (31). In the present study, we investigated the effect(s) of TBE-31 on the actin cytoskeleton, and have determined that unlike the CDDO analogs, TBE-31 binds directly to actin to inhibit polymerization, and, by doing so, TBE-31 also inhibits stress fiber formation during EMT as well as non–small cell lung cancer (NSCLC) tumor cell migration. Therefore, TBE-31 is an exciting prospect for multi-targeted inhibition of metastatic potential.
Materials and Methods
Cell culture, antibodies, and reagents
All cell lines were purchased from ATCC. Rat2 (ATCC-CRL-1764) and NIH 3T3 (ATCC-CRL-1658) fibroblasts were cultured in Dulbecco's modified Eagle Medium (DMEM). A549 (ATCC-CCL-185) lung adenocarcinomas were cultured in Kaighn's Modification of Ham's F-12 Medium (F-12K) and H1299 (ATCC-CRL-5803) human NSCLC cells were cultured in Roswell Park Memorial Institute 1640 medium (RPMI-1640). All cells were cultured in a 37°C incubator with 5% CO2, and all media were supplemented with 10% fetal bovine serum (FBS) unless otherwise stated.
AlexaFluor 555–conjugated phalloidin (A34055) was purchased from Invitrogen. Anti-Smad2/3 (BD610843), anti-E-Cadherin (BD610182), anti-N-Cadherin (BD610921), and anti-Paxillin (BD610051) were purchased from BD transduction Laboratories. Anti-Phospho-Smad 2 (AB3849) was purchased from Millipore. Cytochalasin D (C8273) and Anti-Arp3 (A5979) were purchased from Sigma-Aldrich. NeutrAvidin Agarose beads (29200) were purchased from Thermo Scientific. Biotin Azide (B10184) was purchased from Life Technologies. CDDO-Im and TBE-31 compounds were provided by Dr. Michael B. Sporn (Dartmouth, NH). The purified actin (AKL99), Arp2/3 (RP01), VCA domain of WASP (VCG03), and actin polymerization kits (BK003) were purchased from Cytoskeleton Inc.
Immunofluorescence microscopy
In all immunofluorescence microscopy studies, cells were fixed in 4% paraformaldehyde for 10 minutes, permeabilized in 0.25% Triton X-100 for 5 minutes, blocked in 10% FBS, and immunostained overnight with the appropriate primary antibodies. All images were taken with an Olympus IX81 inverted epifluorescence microscope.
Scratch assays and immunofluorescence microscopy
H1299 or NIH 3T3 cells were grown to confluence and scratched with a pipette tip. Bright-field images were obtained at 0 hours with an Olympus IX81 microscope and the same coordinates were reimaged after 18 hours (H1299 cells) or 12 hours (NIH 3T3 cells). The extent of cell migration was measured using the difference in width of the scratch between 0 hours and final time point. To generate videos, an Olympus IX81 inverted microscope was fitted with an atmosphere chamber (37°C and 5% CO2). Bright-field images were collected at 15-minute intervals over the duration of the experiment. For immunofluorescence studies, Rat2 cells were scratched and given 4 hours to polarize, before incubating in media containing DMSO, 1.0 μmol/L of TBE-31, or 1.0 μmol/L of CDDO-Im for 2 hours. The cells were then fixed, permeabilized, and immunostained as described.
Subconfluent cell migration
Subconfluent NIH 3T3 or H1299 cells were treated with DMSO, 1 μmol/L TBE-31 or 0.5 μmol/L CDDO-Im. Time-lapse observations were made using an Olympus IX81 inverted microscope equipped an atmosphere chamber (37°C and 5% CO2). Bright-field images were collected at 10-minute intervals over 18 hours. Distance of migration was determined by tracking the positions of cell nuclei over 18 hours using the MtrackJ plugin (32) for ImageJ software (33).
In vitro actin polymerization
In vitro actin polymerization was carried out using a modified version of the protocol supplied by Cytoskeleton Inc. Purified pyrene-labeled actin was incubated for 1 hour on ice to depolymerize any actin oligomers followed by microcentrifugation at 4°C for 30 minutes. Two μmol/L actin (Actin) or actin in the presence of 100 nmol/L VCA domain of n-WASp protein and 13 μmol/L Arp2/3 complex (AVA) were incubated with DMSO, or 100 μmol/L of TBE-31 for 15 minutes on ice, before warming the mixtures to 37°C to initiate polymerization. The change in fluorescence was measured with a Wallac Victor3 V plate reader. As a control, actin was fully polymerized in the absence of treatment (30 minutes) and the fluorescence intensity was measured. DMSO, 100 μmol/L TBE-31, or 100 μmol/L CDDO-Im was then added and the fluorescence intensity was immediately remeasured to ensure that these molecules do not interfere or quench the fluorescence intensity.
Actin stress fiber repolymerization
Subconfluent NIH 3T3, H1299, or Rat2 cells were incubated in media containing 5 μmol/L of cytochalasin D for 30 minutes at 37°C. Afterward, the cells were rinsed with clean media to remove cytochalasin D, before reincubating in media containing either DMSO, 1 μmol/L TBE-31, or 1 μmol/L CDDO-Im. At various time points after reintroducing the new media, the cells were fixed, permeabilized, and labeled for actin using AlexaFluor 555–conjugated phalloidin.
Affinity pull-downs
Rat2 lysates were incubated with DMSO, biotin, TBE-55, or TBE-56 at 4°C for 2 hours, followed by incubating with NeutrAvidin beads for 1 hour at 4°C to precipitate the proteins interacting with the biotin conjugate. The beads were thoroughly washed, and 2× Laemmli sample prep buffer was added. These samples were then subjected to SDS-PAGE, and silver staining was performed. Proteins that were uniquely stained in the TBE-56 samples were excised from the gel, trypsinized, and analyzed by electrospray mass spectrometry. To confirm our results, the same pull-down experiments from above were performed with cell lysates or purified actin proteins; immunoblotting with anti-actin was performed after subjecting the samples to SDS-PAGE. For competitive binding studies, additional samples were added where TBE-31 or TBE-55 were incubated with lysates 1 hour before incubating with TBE-56, followed by the same steps as above.
EMT
A549 cells were incubated with 0.2% FBS F-12k media for 4 hours, before the addition of 200 pmol/L TGFβ was added for 48 hours at 37°C to initiate EMT. For immunofluorescence studies, the cells were grown on coverslips and fixed, permeabilized, and immunostained with monoclonal paxillin antibody, and AlexaFluor 555–conjugated phalloidin for filamentous actin after the 48-hour treatment. For Western blot analysis, the cells were lysed, subjected to SDS-PAGE, and immunoblotted for E-Cadherin, N-Cadherin, Smad2/3, Phospho-Smad2, or actin.
Results and Discussion
Hallmarks of epithelial-derived tumors include unregulated cell growth, and in later stages of tumor progression, cells undergo metastasis and migrate away from the primary tumor site and invade distant organs to establish secondary tumors (2, 34–37). Therefore, the development of chemotherapeutic agents that can inhibit different aspects of metastasis, including EMT and cell migration, will play a critical role in reducing in the rate of cancer mortality. EMT and cell migration are dependent on the reorganization of the cytoskeleton, and the actin cytoskeleton has been proposed to be a feasible target in tumor metastasis (38).
TBE-31 does not alter Arp3 localization in polarized cells
Cell migration is dependent on the proper localization of proteins involved in maintaining cell polarity (for e.g., Cdc42/Rac1 and IQGAP1) as well as proteins involved in the formation of the lamellipodia via Arp2/3-dependent branched actin polymerization at the leading edge of migrating cells (6–8, 39–41). We previously demonstrated that CDDO-Im displaces Arp2/3 from the leading edge, reduces branched actin polymerization, and inhibits cell migration (23, 24). TBE-31 contains a similar structural core as the pentacyclic triterpenoids (Fig. 1) and has been shown to induce the Keap1/Nrf2/ARE pathway in a similar fashion as the pentacyclic compounds (28, 29). However, the effect(s) of TBE-31 on the cytoskeleton remain unknown. We, therefore, used immunofluorescence microscopy to examine whether TBE-31 would target the cytoskeleton by first assessing if it targets the Arp2/3 complex and displaces it from the leading edge of migrating cells as efficiently as CDDO-Im. Briefly, Rat2 fibroblasts were grown to confluence, scratched, and allowed to polarize before being treated with DMSO, 1 μmol/L TBE-31, or 1 μmol/L CDDO-Im. The cells were then immunostained for Arp3 and filamentous actin, followed by processing for immunofluorescence microscopy (Fig. 2). We observed that 88% ± 3% of the cells treated with DMSO exhibited Arp3 localization at the leading edge, whereas CDDO-Im reduced leading edge staining of Arp3 to 53% ± 18% (Fig. 2), consistent with our previous observations (24). Interestingly, TBE-31 did not alter leading edge staining of Arp3 compared with the cells treated with vehicle (Fig. 2). This was intriguing because the core of the CDDO-Im molecule (similar to the TBE-31 core) was modeled to fit into the Arp3 inhibitory groove (24), similar to the bona fide Arp3 inhibitor, CK-869 (42). The lack of displacement of Arp3 from the leading edge in cells treated with TBE-31 suggests that the compound is not associating and interfering with the subcellular localization of Arp3. Therefore, to assess the binding of TBE-31 with Arp3 and/or other cytoskeletal proteins, we next carried out a pull-down approach.
TBE-31 does not alter Arp3 localization to the cell periphery. A, confluent Rat2 fibroblasts were scratched and incubated for 4 hours to allow them to polarize. Cells were then incubated for 2 hours in media containing DMSO, 1 μmol/L CDDO-Im, or 1 μmol/L TBE-31. The cells were then fixed, permeabilized, and immunostained with Arp3 (green), AlexaFluor 555 phalloidin (red) for filamentous actin, and DAPI (blue) for the nuclei. Shown are representative immunofluorescence images taken with an Olympus IX81 microscope at ×40 magnification. Bar, 10 μm. Arrows, staining of Arp3 at the cell periphery. B, experiments were carried out as described in A, quantitated, and graphed for the number of cells containing leading edge staining for Arp3 (n = 3 ± SD; *, P < 0.05).
TBE-31 does not alter Arp3 localization to the cell periphery. A, confluent Rat2 fibroblasts were scratched and incubated for 4 hours to allow them to polarize. Cells were then incubated for 2 hours in media containing DMSO, 1 μmol/L CDDO-Im, or 1 μmol/L TBE-31. The cells were then fixed, permeabilized, and immunostained with Arp3 (green), AlexaFluor 555 phalloidin (red) for filamentous actin, and DAPI (blue) for the nuclei. Shown are representative immunofluorescence images taken with an Olympus IX81 microscope at ×40 magnification. Bar, 10 μm. Arrows, staining of Arp3 at the cell periphery. B, experiments were carried out as described in A, quantitated, and graphed for the number of cells containing leading edge staining for Arp3 (n = 3 ± SD; *, P < 0.05).
TBE-31 binds to actin directly
To determine the binding partners for TBE-31, cell lysates were incubated with DMSO or TBE-56, followed by precipitation with NeutrAvidin beads. Proteins associating with TBE-56 were separated on SDS-PAGE, processed, and analyzed by mass spectrometry (Fig. 3A). Although we previously used this approach to identify that CDDO-Me associated with Arp3 (24), we were unable to detect an association between TBE-56 and Arp3. The main cytoskeletal proteins that were identified to associate with TBE-56 were actin and tubulin. Interestingly, we previously identified actin and tubulin to be triterpenoid-binding partners (24), and characterized the effects of CDDO-Im on microtubule organization (23). On the basis of the robustness of interaction between TBE-56 and actin, we further analyzed this interaction by immunoblot analysis (Fig. 3B). Briefly, Rat2 lysates were incubated with DMSO, TBE-55 or TBE-56, precipitated with NeutrAvidin beads and processed for immunoblotting with actin antibodies, and we observed that TBE-56 associated with actin in cell lysates (Fig. 3B, left). To assess whether this interaction was direct, we carried out the pull-down assay using purified actin and observed similar results (Fig. 3B, right), suggesting that TBE-56 binds directly to actin. Furthermore, we performed competitive binding assays where Rat2 lysates were pretreated with DMSO, or increasing concentrations of TBE-31 or TBE-55, before incubating with TBE-56. We found that by increasing the concentrations of TBE-31 or TBE-55 in the pretreatment, actin was unable to be precipitated by the biotin conjugate, TBE-56 (Fig. 3C).
TBE-31 binds to actin. A, Rat2 fibroblasts were incubated with 10 μmol/L biotin, TBE-55, or biotinylated-TBE-31 (TBE-56), lysed, and incubated with NeutrAvidin beads. SDS-PAGE and silver staining were performed and proteins that were uniquely stained in TBE-56–treated samples were excised from the gel, trypsinized, and analyzed by electrospray mass spectrometry. Protein identifications are indicated on the right of the gel (n = 3). Tpi1, triosephosphate isomerase 1; Hsp, heat shock protein; NS, nonspecific. B, Rat2 lysates or purified actin were incubated with DMSO, TBE-55, or biotinylated-TBE-31 (TBE-56) at 4°C for 2 hours, followed by incubating with NeutrAvidin beads for 1 hour at 4°C to precipitate the proteins interacting with TBE-56. The beads were then washed, subjected to SDS-PAGE, and immunoblotted for actin (n = 3). C, competitive binding studies, similar to the experiments described in B, were performed. Additional samples were added that had been pretreated with increasing concentrations of TBE-31 or TBE-55, before incubating with TBE-56 (n = 3).
TBE-31 binds to actin. A, Rat2 fibroblasts were incubated with 10 μmol/L biotin, TBE-55, or biotinylated-TBE-31 (TBE-56), lysed, and incubated with NeutrAvidin beads. SDS-PAGE and silver staining were performed and proteins that were uniquely stained in TBE-56–treated samples were excised from the gel, trypsinized, and analyzed by electrospray mass spectrometry. Protein identifications are indicated on the right of the gel (n = 3). Tpi1, triosephosphate isomerase 1; Hsp, heat shock protein; NS, nonspecific. B, Rat2 lysates or purified actin were incubated with DMSO, TBE-55, or biotinylated-TBE-31 (TBE-56) at 4°C for 2 hours, followed by incubating with NeutrAvidin beads for 1 hour at 4°C to precipitate the proteins interacting with TBE-56. The beads were then washed, subjected to SDS-PAGE, and immunoblotted for actin (n = 3). C, competitive binding studies, similar to the experiments described in B, were performed. Additional samples were added that had been pretreated with increasing concentrations of TBE-31 or TBE-55, before incubating with TBE-56 (n = 3).
These results suggested that biotinylated TBE-31 is able to associate with actin but not Arp3. We next assessed whether the tricyclic compound would have an effect on linear versus branched actin polymerization in vitro.
TBE-31 inhibits actin polymerization
Actin is a globular protein that binds end-to-end to form filaments, which can provide a basis for contractile forces, cell protrusions, or structure (6, 7, 9, 10, 43). We examined the effects of TBE-31 on actin dynamics by assessing in vitro polymerization of actin (Fig. 4A). Briefly, pyrene-labeled actin was incubated in the presence of DMSO (vehicle), TBE-31, CDDO-Im, or cytochalasin D (CD) for 15 minutes on ice, before adding the actin polymerization buffer and warming the mixtures to 37°C to initiate polymerization. To induce branched actin polymerization, Arp2/3, and the VCA domain of n-WASp were added to actin (AVA). By recording the change in fluorescence intensity, we measured the rate of polymerization. Cytochalasin D is a well-established mycotoxin that reversibly binds with globular actin monomers to prevent polymerization (44, 45). As expected, cytochalasin D inhibited the rate of linear and branched actin polymerization (Fig. 4A, top). CDDO-Im inhibited the Arp2/3-mediated branched actin polymerization and had little effect on the polymerization rate of linear actin (Fig. 4A, middle). This was expected, as we have previously demonstrated that CDDO-Im only acts on Arp2/3 to inhibit Arp2/3-mediated branched actin polymerization (24). Interestingly, TBE-31 was able to inhibit both branched and linear actin polymerization (Fig. 4A, bottom). Furthermore, TBE-31 also reduced the total overall amount of polymerized actin by approximately 40%, suggesting that unlike cytochalasin D, which is capable of depolymerizing filamentous actin, TBE-31 may be sequestering monomeric actin and preventing it from being incorporated into growing filamentous actin fibers. Finally, to ascertain that the addition of the tri- or pentacyclic compounds was not interfering with the fluorescence of the pyrene actin in the assay, we allowed actin to fully polymerize and then incubated the polymerized actin in the presence or absence of DMSO (control), CDDO-Im, or TBE-31 (Supplementary Fig. S1). We observed that the addition of the tri- or pentacyclic compounds did not quench the fluorescent signal.
TBE-31 inhibits actin polymerization and stress fiber formation. A, purified pyrene-labeled actin was polymerized in the presence of DMSO (vehicle), 5 μmol/L cytochalasin D (top), 100 μmol/L CDDO-Im (middle), or 100 μmol/L TBE-31 (bottom). For branched actin polymerization, 100 nmol/L VCA domain of n-WASp protein and 13 μmol/L Arp2/3 complex were added to actin (AVA). Actin polymerization was measured by pyrene fluorescence intensity and graphed (arbitrary units) versus time (minutes) (n = 3 ± SEM). Note that the data in the graphs were acquired together but separated into three graphs for the sake of clarity. The common control conditions (gray lines) are shown in each graph to indicate the differences in vehicle versus compound-treated samples. B, subconfluent NIH 3T3 cells were incubated in medium containing 5 μmol/L cytochalasin D at 37°C for 30 minutes to depolymerize the actin cytoskeleton. Afterward, the cells were washed with PBS. Media supplemented with DMSO, 1 μmol/L TBE-31, or 1 μmol/L CDDO-Im was then reintroduced to the cells while the actin stress fibers repolymerized in the absence of cytochalasin D. The cells were then fixed, permeabilized, and immunostained with AlexaFluor 555 phalloidin to visualize actin filaments (red), and DAPI (blue) for the nuclei. Shown here are representative immunofluorescence images taken with an Olympus IX81 microscope at ×40 magnification (n = 3). Bar, 20 μm. Quantitation of cells with actin stress fibers (inset; white arrows) was carried and graphed (bottom; n = 3 ± SEM; *, P < 0.05). Quantified cells were differentiated from cells containing cortical actin staining (inset; yellow arrowhead).
TBE-31 inhibits actin polymerization and stress fiber formation. A, purified pyrene-labeled actin was polymerized in the presence of DMSO (vehicle), 5 μmol/L cytochalasin D (top), 100 μmol/L CDDO-Im (middle), or 100 μmol/L TBE-31 (bottom). For branched actin polymerization, 100 nmol/L VCA domain of n-WASp protein and 13 μmol/L Arp2/3 complex were added to actin (AVA). Actin polymerization was measured by pyrene fluorescence intensity and graphed (arbitrary units) versus time (minutes) (n = 3 ± SEM). Note that the data in the graphs were acquired together but separated into three graphs for the sake of clarity. The common control conditions (gray lines) are shown in each graph to indicate the differences in vehicle versus compound-treated samples. B, subconfluent NIH 3T3 cells were incubated in medium containing 5 μmol/L cytochalasin D at 37°C for 30 minutes to depolymerize the actin cytoskeleton. Afterward, the cells were washed with PBS. Media supplemented with DMSO, 1 μmol/L TBE-31, or 1 μmol/L CDDO-Im was then reintroduced to the cells while the actin stress fibers repolymerized in the absence of cytochalasin D. The cells were then fixed, permeabilized, and immunostained with AlexaFluor 555 phalloidin to visualize actin filaments (red), and DAPI (blue) for the nuclei. Shown here are representative immunofluorescence images taken with an Olympus IX81 microscope at ×40 magnification (n = 3). Bar, 20 μm. Quantitation of cells with actin stress fibers (inset; white arrows) was carried and graphed (bottom; n = 3 ± SEM; *, P < 0.05). Quantified cells were differentiated from cells containing cortical actin staining (inset; yellow arrowhead).
Taken together, these results are intriguing because CDDO-Im and CDDO-Me were found to inhibit cell migration by blocking branched actin polymerization (24); however, TBE-31 may affect cell migration by targeting general actin polymerization. We, therefore, next assessed the effect TBE-31 would have an effect to actin polymerization in cells.
Actin polymerization in cell culture was also examined in parallel with our in vitro experiments by examining actin stress fiber formations in NIH 3T3 fibroblasts. Briefly, NIH 3T3 fibroblasts were incubated for 30 minutes in media containing cytochalasin D to completely depolymerize the actin stress fibers. Afterward, cytochalasin D was removed and the cells were incubated in media containing DMSO, TBE-31, or CDDO-Im while the actin cytoskeleton repolymerized. At various time points, the cells were fixed, permeabilized, and labeled for filamentous actin using phalloidin. Immediately after cytochalasin D treatment, the cells were round and did not have a polarized morphology (Fig. 4B, top). However, after removing the cytochalasin D and incubating cells in the presence of DMSO or CDDO-Im, the actin repolymerized and formed stress fibers. Interestingly, TBE-31–treated cells did not reestablish elongated stress fibers (Fig. 4B, top). Closer examination showed that only short, cortical stress fiber formation was occurring (Fig. 4B, inset). Quantitation of these cells for cells with actin stress fibers versus cells without stress fibers supported our observations. By 2 hours DMSO, CDDO-Im, and TBE-31–treated cells had 47% ± 1%, 40% ± 3%, and 39% ± 5% of cells with stress fibers formed, respectively (Fig. 4B, bottom). By 6 hours, DMSO and CDDO-Im–treated cells had 69% ± 3% and 73% ± 2% recovery of stress fibers, whereas TBE-31–treated cells had lagged behind with only 43% ± 4%. Similar results were obtained using Rat2 fibroblasts (Supplementary Fig. S2). These results further support the idea that TBE-31 inhibits actin polymerization and predict that stress fiber formation would be inhibited in tumor cells undergoing EMT.
TBE-31 inhibits TGFβ-dependent actin stress fiber formation during EMTs
During EMT, the actin cytoskeleton rearranges from a cortical alignment associated with cell–cell junctions into actin stress fibers, which are associated with cell migration (3, 5). We, therefore, investigated whether TBE-31 would have an effect on the actin-dependent changes that occur during EMT. Briefly, NSCLC A549 cells were incubated in serum-free medium or medium containing TGFβ for 48 hours, and either DMSO or TBE-31. Afterward, the cells were fixed, permeabilized, and immunostained with paxillin antibody and phalloidin to visualize filamentous actin. Under bright-field microscopy, we observed a cobblestone phenotype associated with epithelial cells in the cells that were not stimulated with TGFβ, regardless of the addition of DMSO or TBE-31 (data not shown). These cells only demonstrated cortical actin staining with little to no stress fibers, which is another characteristic of epithelial cells (Fig. 5A). Conversely, the cells stimulated with TGFβ in the presence of DMSO demonstrated an elongated, and spindle-like phenotype and under immunofluorescence microscopy these cells also demonstrated proper stress fiber formation (Fig. 5A), all of which are associated with a more mesenchymal phenotype. TGFβ stimulation in the presence of TBE-31 resulted in less stress fiber formation and elongation; instead, they displayed the cobblestone phenotype associated with epithelial cells (Fig. 5A). These results suggest that TBE-31 inhibits the actin-related changes in EMT induced by TGFβ.
TBE-31 inhibits TGFβ-dependent actin stress fiber formation during NSCLC tumor cell EMT. A, A549 cells were incubated in the presence or absence of TGFβ and with DMSO or 1 μmol/L TBE-31 for 48 hours at 37°C. The cells were then fixed, permeabilized, and immunostained with monoclonal paxillin antibody (green), and AlexaFluor 555–conjugated phalloidin (red) for filamentous actin. Shown here are representative immunofluorescence images at ×40 magnification (n = 3). Bar, 10 μm. Insets highlight area of interest and indicate actin stress fibers (white arrows) versus cortical actin staining (yellow arrowhead). B, A549 cells were incubated with 200 pmol/L TGFβ and either DMSO or 1 μmol/L TBE-31 for 30 minutes at 37°C. Cells were lysed before TGFβ stimulation, 0.5, 1.5, or 4.5 hours after stimulation. Lysates were subjected to SDS-PAGE and immunoblotted for p-Smad2, and Smad2 (n = 4). C, A549 cells were incubated in media containing 200 pmol/L TGFβ and either DMSO, 1 μmol/L TBE-31, or varying concentrations of cytochalasin D for 48 hours. Cells were subsequently lysed, subjected to SDS-PAGE, and immunoblotted for E-cadherin, N-cadherin, or actin (n = 3).
TBE-31 inhibits TGFβ-dependent actin stress fiber formation during NSCLC tumor cell EMT. A, A549 cells were incubated in the presence or absence of TGFβ and with DMSO or 1 μmol/L TBE-31 for 48 hours at 37°C. The cells were then fixed, permeabilized, and immunostained with monoclonal paxillin antibody (green), and AlexaFluor 555–conjugated phalloidin (red) for filamentous actin. Shown here are representative immunofluorescence images at ×40 magnification (n = 3). Bar, 10 μm. Insets highlight area of interest and indicate actin stress fibers (white arrows) versus cortical actin staining (yellow arrowhead). B, A549 cells were incubated with 200 pmol/L TGFβ and either DMSO or 1 μmol/L TBE-31 for 30 minutes at 37°C. Cells were lysed before TGFβ stimulation, 0.5, 1.5, or 4.5 hours after stimulation. Lysates were subjected to SDS-PAGE and immunoblotted for p-Smad2, and Smad2 (n = 4). C, A549 cells were incubated in media containing 200 pmol/L TGFβ and either DMSO, 1 μmol/L TBE-31, or varying concentrations of cytochalasin D for 48 hours. Cells were subsequently lysed, subjected to SDS-PAGE, and immunoblotted for E-cadherin, N-cadherin, or actin (n = 3).
To assess whether the inhibition of actin stress fiber formation in the A549 cells was due to an inhibition of the EMT process, we examined TGFβ signal transduction by examining Smad2 phosphorylation to ensure TBE-31 was not interfering with TGFβ signaling. Briefly, A549 cells were serum starved for 4 hours before treatment with DMSO, TBE-31, or varying concentrations of cytochalasin D for 30 minutes. Afterward, TGFβ was added to the media and was allowed to incubate for another 30 minutes before being lysed, subjected to SDS-PAGE, and immunoblotted for P-Smad2, Smad2, and actin. We observed that regardless of the pharmacologic treatment, TGFβ was able to induce the phosphorylation of Smad2, suggesting that none of our treatments affected the TGFβ signaling pathways (Fig. 5B).
To further verify proper TGFβ signaling, we examined other biochemical changes that mark EMT. E-cadherin is the major component of epithelial adherens junctions, which form a belt-like structure around the cells to tether adjacent cells into an immobile sheet-like formation. During EMT, the expression of E-cadherin becomes downregulated and the expression of N-cadherin, an adhesion molecule associated with highly migratory mesenchymal cells, becomes upregulated. This is known as the cadherin switch and is a major hallmark of EMT (2–5). These changes were examined by treating A549 cells with TGFβ for 48 hours in the presence or absence of DMSO, TBE-31, or cytochalasin D, subjected to SDS-PAGE and immunoblotted for E-Cadherin and N-Cadherin (Fig. 6). Regardless of treatment, TGFβ stimulation led to a decrease in E-cadherin expression, whereas N-cadherin expression increased. This suggested that a proper cadherin switch was occurring in the presence of TGFβ in vehicle, TBE-31, or cytochalasin D–treated cells and further reconfirms that TBE-31 does not have an effect on TGFβ signaling, or gene transcription associated with EMT.
TBE-31 inhibits NSCLC tumor cell migration. A, top, confluent H1299 lung tumor cells were scratched and incubated at 37°C in media containing either DMSO, 2 μmol/L TBE-31, or 1 μmol/L CDDO-Im for 18 hours. Bright-field microscopy images were taken at ×10 magnification at 0 and 18 hours postincubation. Bottom, scratch assays were carried out as described above using varying concentrations of TBE-31 or CDDO-Im and the relative cell migration was quantitated using ImagePro software (n = 4 ± SD). *, P < 0.05. B, top, confluent NIH 3T3 fibroblasts were scratched and incubated as described in A. Bright-field microscopy images were taken at ×10 magnification at 0 and 12 hours postincubation (n = 3). Bottom, scratch assays were carried out as described above using varying concentrations of TBE-31 or CDDO-Im and the relative cell migration was quantitated using ImagePro software (n = 3 ± SD). *, P < 0.05. C, subconfluent H1299 or NIH 3T3 cells were incubated with vehicle (DMSO), 1 μmol/L TBE-31, or 0.5 μmol/L TCDDO-Im and imaged over 18 hours. Using ImageJ software, the tracking patterns of individual migrating cells (multicolor plots) were measured and graphed. The average distance migrated by the H1299 or NIH 3T3 cells are shown on the right (n = 3 ± SD). *, P < 0.05.
TBE-31 inhibits NSCLC tumor cell migration. A, top, confluent H1299 lung tumor cells were scratched and incubated at 37°C in media containing either DMSO, 2 μmol/L TBE-31, or 1 μmol/L CDDO-Im for 18 hours. Bright-field microscopy images were taken at ×10 magnification at 0 and 18 hours postincubation. Bottom, scratch assays were carried out as described above using varying concentrations of TBE-31 or CDDO-Im and the relative cell migration was quantitated using ImagePro software (n = 4 ± SD). *, P < 0.05. B, top, confluent NIH 3T3 fibroblasts were scratched and incubated as described in A. Bright-field microscopy images were taken at ×10 magnification at 0 and 12 hours postincubation (n = 3). Bottom, scratch assays were carried out as described above using varying concentrations of TBE-31 or CDDO-Im and the relative cell migration was quantitated using ImagePro software (n = 3 ± SD). *, P < 0.05. C, subconfluent H1299 or NIH 3T3 cells were incubated with vehicle (DMSO), 1 μmol/L TBE-31, or 0.5 μmol/L TCDDO-Im and imaged over 18 hours. Using ImageJ software, the tracking patterns of individual migrating cells (multicolor plots) were measured and graphed. The average distance migrated by the H1299 or NIH 3T3 cells are shown on the right (n = 3 ± SD). *, P < 0.05.
TBE-31 inhibits cell migration
We next assessed cell migration, another actin-dependent process that follows EMT in metastasis. Actin polymerization provides the main driving force in cell migration, and we have previously shown that the triterpenoids CDDO-Im and CDDO-Me inhibit the Arp2/3-mediated branched actin polymerization by binding to Arp2/3 and decrease the rate of cell migration (24). Because we have demonstrated that TBE-31 is able to bind to actin directly and modify its polymerization into stress fibers, we next assessed whether it could inhibit cell migration as well. We used H1299 human NSCLC cells as well as migrating fibroblasts (NIH 3T3 cells), which exhibited robust migration in our scratch assays (Fig. 6). Briefly, H1299 or NIH 3T3 cells were grown to confluence and scratched to stimulate migration. The cells were then incubated with media containing DMSO or varying concentrations of TBE-31 or CDDO-Im. We observed that after 18 hours, DMSO-treated H1299 cells migrated on average 432 ± 25 μm, whereas the NIH 3T3 cells migrated 545 ± 30 μm after 12 hours. Overall, TBE-31 or CDDO-Im–treated cells demonstrated a dose-dependent decrease in cell migration (Fig. 6A, graphs; Supplementary Online Video 1). More specifically, the IC50 of TBE-31 for H1299 cells was found to be 2.5 μmol/L and 1 μmol/L for the NIH 3T3 cells. The IC50 for CDDO-Im was found to be approximately 1 μmol/L, for both cell lines, similar to our previous results using Rat2 fibroblasts (24).
To confirm that TBE-31 and CDDO-Im were indeed inhibiting the migration of both tumor and fibroblast cell lines, we carried out single-cell tracking studies of H1299 or NIH 3T3 cells (Fig. 6B; Supplementary Online Video 2). Briefly, subconfluent populations of cells were incubated with media containing vehicle (DMSO), CDDO-Im, or TBE-31 and imaged for 18 hours. The migration tracks as well as the average distance travelled were calculated and graphed (Fig. 6B). Consistent with the scratch assays, we observed that both TBE-31 and CDDO inhibited NIH 3T3 and H1299 cell migration at low micromolar concentrations.
Overall, our results show that TBE-31 associates and inhibits actin polymerization, resulting in reduced tumor cell stress fiber formation and cell migration.
Conclusions
Although the overall effect of TBE-31 on cell migration are not dissimilar from CDDO-Im and CDDO-Me (the compounds it was derived from), the effects on actin polymerization are strikingly different: CDDO-Im and CDDO-Me were shown to only inhibit Arp2/3-dependent branched actin polymerization and did not have an appreciable effect on linear actin polymerization (24). This CDDO-Im–dependent or CDDO-Me–dependent inhibition of Arp2/3 activity or perhaps branched actin formation was also associated with an absence of polarity proteins at the leading edge of migrating cell. The differences in TBE-31 and CDDO compound in their abilities to target linear versus branched actin polymerization may also effect polarity protein localization and activity and will be investigated in future studies. In addition, CDDO-Im demonstrated the ability to alter microtubule dynamics and organization (23). Future studies will also examine the effects of TBE-31 on the microtubule network and the possible impact it may have on microtubule-dependent processes such as protein trafficking, cell signaling, cell division, mitosis, and cell migration.
The enormous biologic diversity of cancer has led to limited promise for therapies that target single signaling molecules. It has been suggested that strategic combinations of agents targeting against the most critical alterations in cancer will be needed, or simply the use of more unspecific agents that modulate several relevant targets simultaneously (46, 47). Not surprisingly, drug discovery has moved toward investigating multitarget drugs in the last decade, in a large part due to the development of cancer therapeutics (48–50). TBE-31 is an attractive compound for cancer prevention due to its antioxidative, anti-inflammatory response capabilities by inducing the Keap1/Nrf2/ARE pathway (25–29), and now inhibiting cell migration by targeting actin. The parallels and contrasts between TBE-31 and the CDDO analogs may lend well to combinational therapies with cytotoxic drugs. The possibility for these and other synergistic effects between TBE-31 and drugs already used in the clinic will provide us with future avenues of research.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: E. Chan, T. Honda, G.M. Di Guglielmo
Development of methodology: E. Chan, G.M. Di Guglielmo
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Chan, G.M. Di Guglielmo
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Chan, G.M. Di Guglielmo
Writing, review, and/or revision of the manuscript: E. Chan, T. Honda, G.M. Di Guglielmo
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Chan, A. Saito, T. Honda, G.M. Di Guglielmo
Study supervision: G.M. Di Guglielmo
Grant Support
G.M. Di Guglielmo is supported by funding from the Canadian Institutes of Health Research (CIHR; MOP-93625) and T. Honda is supported by funds from Stony Brook Foundation and Reata Pharmaceuticals. A. Saito is grateful to the Institute of Chemical Biology & Drug Discovery Postdoctoral Scholarships.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.