Recently, a number of steps in the progression of metastatic disease have been shown to be regulated by redox signalling. Electrophilic lipids affect redox signalling through the post-translational modification of critical cysteine residues in proteins. However, the therapeutic potential as well as the precise mechanisms of action of electrophilic lipids in cancer cells is poorly understood. In the present study, we investigate the effect of the electrophilic prostaglandin 15d-PGJ2 (15-deoxy-Δ12,14-prostaglandin J2) on metastatic properties of breast cancer cells. 15d-PGJ2 was shown to decrease migration, stimulate focal-adhesion disassembly and cause extensive F-actin (filamentous actin) reorganization at low concentrations (0.03–0.3 μM). Importantly, these effects seem to be independent of PPARγ (peroxisome-proliferator-activated receptor γ) and modification of actin or Keap1 (Kelch-like ECH-associated protein 1), which are known protein targets of 15d-PGJ2 at higher concentrations. Interestingly, the p38 inhibitor SB203580 was able to prevent both 15d-PGJ2-induced F-actin reorganization and focal-adhesion disassembly. Taken together, the results of the present study suggest that electrophiles such as 15d-PGJ2 are potential anti-metastatic agents which exhibit specificity for migration and adhesion pathways at low concentrations where there are no observed effects on Keap1 or cytotoxicity.
Approximately 90% of all cancer-related deaths are the result of metastasis; thus understanding the regulation of this complex process is important in developing new anti-metastatic treatment strategies. It is becoming clear that a number of steps in the metastatic cascade are regulated by redox signalling. The primary mechanism by which redox signalling occurs is through the post-translational modification of critical cysteine residues (thiols) in redox-sensitive proteins. The modification of thiols in proteins such as PPARγ (peroxisome-proliferator-activated receptor γ), actin, Keap1 (Kelch-like ECH-associated protein 1) and H-Ras (Harvey rat sarcoma viral oncogene homologue) can change the protein structure and/or function of these target proteins and thereby alter signalling pathways [1–5]. Known downstream effects of modification of redox-sensitive signalling pathways include modulation of MMP (matrix metalloproteinase) expression , NF-κB (nuclear factor κB)-regulated gene expression and mitochondrial ROS (reactive oxygen species) generation [7–10], and activity of these pathways has been shown to be directly linked to metastatic potential in multiple cancer types [7,11–13]. Taken together, these studies suggest there are redox-sensitive signalling pathways controlling basic processes required for metastasis.
Species capable of modifying redox signalling pathways can be derived from several sources such as the diet, environment or endogenously through enzymatic or non-enzymatic processes [14,15]. One such redox signalling molecule is the electrophilic cyclopentenone prostaglandin 15d-PGJ2 (15-deoxy-Δ12,14-prostaglandin J2), which primarily modifies cysteine residues through a Michael-type addition . In the context of cancer, 15d-PGJ2 has garnered much interest because of its ability to inhibit angiogenesis, cause growth arrest and induce cell death in several cancer cells lines [17–20]. Interestingly, although 15d-PGJ2 has been shown to be cytotoxic in cancer cells, little is known about its effects on metastasis.
There are two basic mechanisms that have been described to explain the biological actions of 15d-PGJ2. First, 15d-PGJ2 has been proposed as the endogenous ligand for PPARγ. PPARs are ligand-inducible transcription factors which belong to the nuclear hormone receptor superfamily [21,22]. New evidence suggests they may also play a role in oncogenesis, as they modulate proliferation and apoptosis and are expressed in many human tumours including breast . The second mechanism of action by which 15d-PGJ2 alters cellular signalling pathways is through the post-translational modification of redox-sensitive signalling molecules as mentioned above. There are multiple protein targets of 15d-PGJ2 which can mediate diverse biological responses. We have termed this group of proteins the electrophile-responsive proteome . This latter mechanism probably underlies the pleiotropic effects of 15d-PGJ2 reported in the literature .
Cellular migration plays an important role in metastasis, and 15d-PGJ2 has been shown to inhibit migration [26,27]. There is also evidence demonstrating that 15d-PGJ2 alters cytoskeletal structure in multiple cell types including neuroblastoma and mesangial cells; however, these studies reported cytotoxicity associated with cytoskeletal alterations [2,3]. The cytoskeletal effects of 15d-PGJ2 have been largely attributed to the direct modification of proteins such as actin, vimentin and tubulin [2,3]. In the present study, we investigated the effects of 15d-PGJ2 on the F-actin (filamentous actin) cytoskeleton at lower concentrations which do not cause cytotoxicity. The effect of 15d-PGJ2 on the cytoskeleton and migration might have important implications in the inhibition of metastatic processes such as invasion, intravasation and extravasation.
The goals of the present study were to determine the effects of non-toxic low concentrations of 15d-PGJ2 on the regulation of cytoskeletal organization and its influence on cell migration, and to determine the mechanism of action of 15d-PGJ2 at these low concentrations. We first investigated the effect of 15d-PGJ2 on cell viability, migration and focal-adhesion disassembly. In addition, we determined the effects of 15d-PGJ2 on F-actin cytoskeletal structure and examined the roles of direct actin adduction, PPARγ activation and redox signalling pathways in 15d-PGJ2-mediated cytoskeletal regulation. Our present study is the first to demonstrate that 15d-PGJ2 can alter actin organization with minimal direct adduct formation with actin, and that this effect coincides with decreased migration and increased focal-adhesion disassembly. These results suggest a role for redox signalling pathways, rather than direct cytoskeletal disruption, in the mechanism of 15d-PGJ2 in cancer cells.
MATERIALS AND METHODS
BODIPY FL EDA (4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl ethylenediamine) was purchased from Molecular Probes. Alexa Fluor® 633 Phalloidin was purchased from Invitrogen. 15d-PGJ2, PGE2 (prostaglandin E2), 15(R)-PGD2 (prostaglandin D2) and ROSI (rosiglitazone) were purchased from Cayman Chemicals. The p38 inhibitor SB203580 was purchased from Calbiochem. BODIPY FL EDA-tagged 15d-PGJ2 (BD-15d-PGJ2) was synthesized using the method previously described by Higdon et al. . EZ-link 5-(biotinamido)pentylamine was purchased from Pierce for the synthesis of biotinylated 15d-PGJ2 (bt-15d-PGJ2) as described previously . Structures of the parent compound 15d-PGJ2 and tagged derivatives are shown in Figure 1. It was determined that tagging 15d-PGJ2 does not affect its biological activity by comparing the effect of tagged and untagged analogues on migration, colony formation and the F-actin cytoskeleton (Supplementary Figure S1 at http://www.BiochemJ.org/bj/430/bj4300069add.htm). Additionally, the BODIPY fluorophore itself does not alter any of the biological effects examined (Supplementary Figure S1). All other reagents used were of analytical grade.
Structures of 15d-PGJ2, bt-15d-PGJ2 and BD-15d-PGJ2
JC mouse mammary adenocarcinoma cells (A.T.C.C.) were cultured in RPMI 1640 medium (Mediatech) supplemented with 10% FBS (fetal bovine serum; Atlanta Biologicals). Cultures were maintained in 5% CO2 and humidified in a 37 °C incubator. All experiments were performed in 0.5% FBS in RPMI 1640 medium at ~50% confluence.
Assessment of cell viability
Cell viability was assessed using two different methods. Apoptosis and necrosis were measured after treatment with the indicated concentrations of 15d-PGJ2 for 16 h by flow cytometric analysis using an Annexin V–FITC Apoptosis Detection Kit (Calbiochem). Briefly, treated cells were trypsinized and then incubated with Annexin V–FITC and PI (propidium iodide). Gating parameters were set using PI only, Annexin V only and no staining controls, and 10000 events were collected for each experimental sample. Cells staining positive for both PI and Annexin V were considered late apoptotic. Cells staining positive for PI only or Annexin V only were scored as necrotic or early apoptotic respectively. Cells which stained negative for both PI and Annexin V were scored as viable cells. Fluorescence was measured using a BD LSR II flow cytometer. Colony formation was measured after treatment with indicated concentrations of 15d-PGJ2 for 16 h. Cells from experimental dishes were trypsinized and collected. All cells from each dish were then centrifuged (600 g for 10 min at 20 °C), resuspended in fresh medium and counted. Cells were then plated in six-well plates at a low density (100–200 cells per well), and clones were allowed to grow for 14 days in complete medium in the presence of 0.1% gentamycin. Cells were then fixed with 70% ethanol and stained with Coomassie Blue for analysis of colony formation as described previously .
Measurement of GSH
JC cells were treated with 15d-PGJ2 for 16 h at the concentrations indicated. Total GSH (glutathione+glutathione disulfide) was determined in lysates as described previously . Briefly, after treatment, cells were lysed in 10 μM DTPA (diethylenetriaminepenta-acetic acid) containing 0.1% Triton X-100 in PBS (pH 7.4). Total GSH was determined by monitoring the reduction of 5,5′-dithio-bis(2-nitro-benzoic acid) spectrophotometrically at 412 nm. The protein content was assayed using the Bradford method (Bio-Rad).
Fluorescence microscopy was used to visualize p-FAK [phosphorylated FAK (focal-adhesion kinase)] and F-actin. JC cells were plated on to glass coverslips, treated and fixed using paraformaldehyde (3.7%) for 10 min. The cells were then rinsed twice with PBS and permeabilized with 0.1% Triton X-100 in PBS. To determine the relative levels of p-FAK, samples were blocked with 5% normal goat serum (Vector Laboratories) and 0.3% Triton X-100 in PBS for 60 min at room temperature (20 °C). Cells were then incubated in p-FAK (Tyr397) antibody (Cell Signaling) at a 1:100 dilution in antibody dilution buffer [1% (w/v) BSA and 0.3% Triton X-100 in PBS] overnight at 4 °C. Alexa Fluor® 488-conjugated goat anti-rabbit secondary antibody (1:500 dilution for 1 h; Invitrogen) was applied, and then coverslips were then mounted on to glass slides using Vectashield Hard Set Mounting Medium containing DAPI (4′,6-diamidino-2-phenylindole; Vector Laboratories). p-FAK foci were visualized using confocal fluorescence microscopy on a Leica DMIRBE laser-scanning confocal microscope with excitation from a 488 nm laser line and emission detection suitable for fluorescein. Nuclei were imaged using a UV laser line and emission detection suitable for DAPI.
To determine the extent of actin polymerization, JC cells were prepared as described above Next, cells were incubated with blocking solution [PBS containing 1% (w/v) BSA] for 30 min followed by application of 2 units of Alexa Fluor® 633 Phalloidin (Invitrogen) for 30 min at room temperature in blocking solution. F-actin imaging was performed with a 633 nm laser line for excitation and emission detection suitable for Cy5 (indodicarbocyanine). In p-FAK and F-actin co-staining experiments, Alexa Fluor® 633 Phalloidin was co-incubated with secondary antibody. All confocal images represent single sections and were manipulated using linear histogram correction in Adobe Photoshop CS3 (Adobe Systems). Fluorescence quantification was performed using SimplePCI software (Hamamatsu Corporation).
Determination of the modification of Keap1 and β-actin by 15d-PGJ2
To determine the extent of modification of Keap1 by 15d-PGJ2, JC cells were treated with increasing concentrations (0.3–20 μM) of bt-15d-PGJ2 for 4 h. After the treatment, cell lysates were prepared in 1% Triton X-100 in 10 mM Tris/HCl lysis buffer. Biotinylated proteins were affinity-precipitated using 100 μl of a 50% slurry of neutravidin beads (Pierce) which were pre-washed with 20 mM Tris/HCl (pH 7.4, six times). Cell lysates (6 mg of protein) containing protease inhibitor cocktail were added to the beads and incubated for 3 h at room temperature with rotation. Beads were then washed with 600 μl of 0.1 M glycine (pH 2.8, six times) followed by 600 μl of 20 mM Tris base (pH 10, six times) and then 600 μl of 20 mM Tris/HCl (pH 7.4) to neutralize the beads. Samples were then prepared for analysis by heating the beads to 80 °C for 10 min in 80 μl of 5× sample buffer (0.5 M Tris, 20% SDS, 50% glycerol and 1% Bromophenol Blue, pH 6.8) containing 100 mM 2-mercaptoethanol to release the biotin-labelled proteins. Samples were centrifuged at 600 g for 10 min at 4 °C, and supernatants were used for analysis. Proteins were separated by SDS/PAGE (10% gels), and transferred on to nitrocellulose membranes at 100 V for 2 h and then membranes were blocked with 5% (w/v) dried skimmed milk in TBS-T [Tris-buffered saline (25 mM Tris, pH 7.4, 137 mM NaCl and 2.6 mM KCl) with 0.05% Tween 20]. Membranes were incubated with a polyclonal primary antibody against Keap1 [E-20; 1:1000 dilution in 5% (w/v) dried dried skimmed milk in TBS-T; Santa Cruz Biotechnology], followed by an HRP (horseradish peroxidase)-conjugated donkey anti-goat secondary antibody [1:1000 dilution in 5% (w/v) dried skimmed milk in TBS-T; Santa Cruz Biotechnology]. Membranes were developed by chemiluminescence using SuperSignal West Dura substrate (Pierce) and sequential images were taken with quantification only performed on bands that had not reached saturation.
To determine the modification of β-actin by 15d-PGJ2, JC cells were treated as described above. After treatment, cell lysates were prepared in NEM (N-ethylmaleimide; Pierce) containing lysis buffer (10 mM NEM and 1% Triton X-100 in PBS) for 1 h in order to alkylate sulfhydryls and prevent auto-oxidation during the sample processing. 2-Mercaptoethanol (20 mM) was added to quench the excess NEM. Cell lysates were incubated with sodium borohydride (10 mM NaBH4 in 5 mM NaOH) overnight to reduce the carbonyl group on the pentene ring in order to stabilize the lipid adducts on proteins. Biotinylated proteins were affinity-precipitated using 100 μl of a 50% slurry of neutravidin beads (Pierce) which were pre-washed with 20 mM Tris/HCl (pH 7.4, six times). Cell lysates (3 mg of protein) containing protease inhibitor cocktail were added to the beads and incubated for 3 h at room temperature with rotation. Proteins were eluted from the beads as described above for Keap 1, separated by SDS/PAGE (10 % gels), and transferred on to nitrocellulose membrane at 100 V for 2 h and membranes were blocked with 5% (w/v) dried skimmed milk in TBS-T. Membranes were incubated with a polyclonal primary antibody against β-actin [Cell Signaling Technology; 1:1000 dilution in 5% (w/v) dried skimmed milk in TBS-T], followed by a HRP-conjugated donkey anti-rabbit secondary antibody [1:1000 dilution in 5% (w/v) dried skimmed milk in TBS-T; GE Healthcare]. Membranes were developed as described above.
Measurement of cellular migration
JC cells were grown to confluence in six-well plates, and then scratched with the narrow end of a sterile pipette tip. Medium was immediately changed to remove floating cells and was replaced with medium containing increasing concentrations of 15d-PGJ2 or the ethanol vehicle control. The width of the scratch was measured at four points in each well after initial wounding, and cells were incubated for 8 h at 37 °C in a CO2 incubator. After 8 h, the scratch width was measured again, and the ability of the cells to migrate into the cell-free zone (relative motility) was expressed as the normalized percentage change in the width of the scratch after 8 h compared with the ethanol control.
Focal-adhesion disassembly assay
Focal adhesions were assessed using interference reflection microscopy. JC cells were plated on to glass coverslips, allowed to attach and then grown for 24 h before being serum-starved (0.5% FBS in RPMI 1640) for 30 min prior to the treatments indicated. Cells were fixed in 3% glutaraldehyde (Sigma) for 30 min at 37 °C and then rinsed and mounted on to glass slides. Slides were imaged using interference reflection microscopy using a modified inverted Zeiss microscope as described previously . Cells containing >6 focal adhesions were scored as positive by an observer without prior knowledge of sample conditions. In total, 300 cells/coverslip were scored for each treatment group in triplicate.
Western blot analysis
Cell lysate proteins were resolved using SDS/PAGE and transferred on to nitrocellulose membranes (Bio-Rad). Protein levels were quantified using the method of Bradford (Bio-Rad), and equivalent amounts of protein were loaded. Uniform protein loading was confirmed using Ponceau S staining of membranes and showed no significant differences in protein levels on blots among samples. Membranes were blocked in 5% (w/v) BSA in TBS-T, and then incubated with primary antibodies overnight at 4 °C. Antibody conditions were as follows: anti-FAK (1:1000 dilution; Cell Signaling Technology), anti-phosphorylated-p38 (Cell Signaling; 1:1000 dilution), anti-p38 (Cell Signaling Technology; 1:1000 dilution) and β-actin (1:1000 dilution; Cell Signaling Technology). After washing with TBS-T, membranes were incubated with HRP-conjugated secondary antibody. Membranes were developed using SuperSignal West Dura chemiluminescence substrate (Pierce) and imaged using a CCD (charge-coupled device) camera imaging system.
Data were expressed as means±S.E.M. at a minimum in triplicate, and subjected to a Student's t test or one-way ANOVA followed by Bonferroni's multiple comparison tests. P values less than 0.05 were considered statistically significant.
15d-PGJ2 toxicity in JC mouse mammary adenocarcinoma cells
15d-PGJ2 has been shown to induce apoptotic cell death in a number of cancer cell lines at concentrations ranging from 5 to 50 μM [32–35]. In order to determine non-toxic concentrations of 15d-PGJ2, cytotoxicity was assessed using PI and Annexin V–FITC co-staining measured by flow cytometry. JC cells were treated with 15d-PGJ2 (0.01–3 μM) for 16 h. Using this method, we are able to distinguish apoptotic and necrotic cell death. As seen in Figure 2(A), at concentrations ranging from 0.01 to 3 μM, there are no significant changes in viability. Furthermore, there was no indication of apoptotic or necrotic cell death in response to 15d-PGJ2 treatment as determined by the lack of cells staining positive for PI or Annexin V–FITC (results not shown). In contrast and consistent with reports in the literature [32–35], when JC cells were treated with higher concentrations (20 μM) of 15d-PGJ2, there was a significant decrease in cell viability as well as an increase in both late apoptotic and necrotic cell populations (Supplementary Figure S2 at http://www.BiochemJ.org/bj/430/bj4300069add.htm).
Effect of 15d-PGJ2 on cell death and colony formation
15d-PGJ2 toxicity was also assessed using a colony formation assay. This assay measures the replicative ability of cells to form colonies after treatment, an important characteristic of cancer cells. Interestingly, when JC cells were treated with 15d-PGJ2, a marked decrease in colony formation was measured at concentrations as low as 0.01 μM (Figure 2B). Taken together, these data suggest that treatment with low concentrations of 15d-PGJ2 (0.01–3 μM) attenuates the clonogenic capacity of JC cells, but does not cause apoptosis or necrosis. Since re-adherence to tissue culture plates and proliferation are two critical steps for successful colony formation in this assay, these data also suggest that 15d-PGJ2 may attenuate proliferation and/or adhesion pathways. Further investigation demonstrated that 15d-PGJ2 does not cause alterations in cell-cycle progression in this model; however, re-adherence of cells to tissue culture plastic after treatment with 15d-PGJ2 and subsequent trypsinization was impaired (59.02±9.79% of vehicle-treated cells, P<0.05; results not shown). Therefore we conclude that 15d-PGJ2 does not cause cell death at concentrations at or below 3 μM, and the decreased colony formation caused by 15d-PGJ2 appears to be due to a decreased ability of cells to re-adhere after plating, rather than decreased cell viability itself.
15d-PGJ2 attenuates migration
Having established sub-lethal concentrations of 15d-PGJ2, we next examined the effect of this electrophile on cell motility using a scratch assay. Treatment with 15d-PGJ2 caused a concentration-dependent decrease in cell migration over 8 h with significant changes seen at concentrations of 15d-PGJ2 equal or greater than 0.03 μM (Figure 3). These results demonstrate that low non-toxic concentrations of 15d-PGJ2 attenuate cancer cell migration.
Effect of 15d-PGJ2 on cell migration
Effects of 15d-PGJ2 on FAK signalling
FAK is a cytoplasmic protein tyrosine kinase whose expression has been shown to be frequently deregulated in cancer (reviewed in ). To investigate the potential role of FAK signalling in 15d-PGJ2-induced attenuation of migration, we treated JC cells with 0.3 μM 15d-PGJ2 for 30 min and 4 h, and determined total FAK protein levels. Treatment with this sub-lethal concentration of 15d-PGJ2 did not alter total FAK protein levels (Figure 4A).
Effect of 15d-PGJ2 on focal-adhesion disassembly and migration
We also investigated the activity of the FAK signalling pathway by examining levels of p-FAK. It is well-established that activation of FAK results in the autophosphorylation of Tyr397 which reveals a binding site for Src family kinases and mediates many of the downstream signalling events . After treatment with 15d-PGJ2 (0.3 μM for 30 min), p-FAK distribution was markedly altered, whereby p-FAK was localized to the terminal ends of F-actin filaments in untreated cells, and exposure to 15d-PGJ2 resulted in a more diffuse and perinuclear pattern of p-FAK protein (Figure 4B). However, FAK phosphorylation assessed by Western blot analysis after treatment with 15d-PGJ2 (0.3 μM, 15–30 min) showed no significant difference in p-FAK levels in whole-cell lysates from 15d-PGJ2-treated cells compared with the vehicle control (results not shown). The total number of cells scoring positive for focal adhesions (>6 focal adhesions/cell) was also quantified after treatment with 15d-PGJ2. There was no difference in the number of cells scoring positive for focal adhesions after treatment with 15d-PGJ2 (0.3 μM) for 30 min (results not shown); however, decreases in cells scoring positive for focal adhesions was evident after treatment for 4 h (Figure 4C), suggesting that treatment with 0.3 μM 15d-PGJ2 induces focal-adhesion disassembly.
It is clear that 15d-PGJ2 can act through multiple mechanisms including redox signalling, PPARγ-dependent pathways and the G-protein-coupled PGD2 receptor DP2 (reviewed in ). Interestingly, treatment with the DP2 agonist 15(R)-PGD2 or the PPARγ agonist ROSI also resulted in a decrease in the number of cells scoring positive for focal adhesions (Figure 4C), whereas neither of these compounds altered cell motility (Supplementary Figures S3Anad S3B at http://www.BiochemJ.org/bj/430/bj4300069add.htm). Given that 15d-PGJ2 changes focal-adhesion disassembly and p-FAK localization, with no significant change in total FAK, our results suggest that sub-lethal concentrations of 15d-PGJ2 may alter FAK-mediated signalling or localization. However, the fact that the focal-adhesion disassembly caused by 15d-PGJ2 can be recapitulated using PPARγ and DP2 agonists suggests that this effect is probably mediated through different mechanisms than those which mediate migration.
15d-PGJ2 changes F-actin morphology
Since focal adhesions are the site at which the actin cytoskeleton is linked to the extracellular matrix , we investigated the effect of 15d-PGJ2 on the F-actin cytoskeletal structure using Phalloidin. In the same samples which demonstrate p-FAK changes in response to 15d-PGJ2 (Figure 4B), vehicle-treated cells exhibited a filamentous elongated morphology of the F-actin cytoskeleton (Figure 4B, ‘EtOH’ panel). However, treatment with 15d-PGJ2 for 30 min caused extensive reorganization of the F-actin cytoskeleton resulting in rounding of the F-actin cytoskeleton (Figure 4B, ‘15d-PGJ2’ panel). The effects of PPARγ and DP2 agonists on the F-actin cytoskeleton were also examined. Neither ROSI nor 15(R)-PGD2 had any gross effect on the F-actin cytoskeletal morphology (Supplementary Figure S3C), suggesting that the reorganization of the F-actin cytoskeleton in response to 15d-PGJ2 does not occur through PPARγ- or DP2-dependent pathways.
Concentration-dependent effect of bt-15d-PGJ2 modification of actin and Keap1
It is well accepted that the actin cytoskeleton plays an important role in cellular migration . It was shown previously that 15d-PGJ2 can form covalent adducts with a number of important cytoskeletal components including actin, tubulin and vimentin [2,3]. Moreover, 15d-PGJ2 can affect cytoskeletal organization in neuroblastoma and mesangial cells [2,3]. We therefore sought to characterize further the effect of 15d-PGJ2 on the F-actin cytoskeleton. In order to determine whether 15d-PGJ2 forms a direct adduct with actin at low concentrations, JC cells were treated with 0.3, 3 and 20 μM bt-15d-PGJ2 for 4 h. bt-15d-PGJ2-adducted proteins were then enriched from cell lysate protein using a neutravidin column. Total and bt-15d-PGJ2-modified actin was detected by Western blotting. In Figure 5(A), actin modification was detected in cell lysates treated with 3 and 20 μM bt-15d-PGJ2, consistent with reports in the literature of a direct modification of actin by 15d-PGJ2 [3,41]. Importantly, minimal modification of actin was detected after 4 h with 0.3 μM bt-15d-PGJ2.
Dose-dependent adduct formation of bt-15d-PGJ2 with β-actin and Keap1
As a positive control, we compared actin adduct formation with another protein which is known to be adducted by 15d-PGJ2, Keap1. bt-15d-PGJ2 directly adducted Keap1 when cells were treated with 3 μM or 20 μM bt-15d-PGJ2; however, minimal adduct formation on Keap1 was detected at 0.3 μM bt-15d-PGJ2 (Figure 5A). The amount (percentage) of Keap1 and actin which were pulled-down from the total cell lysate was assessed by calculating the quantity of each protein from the densitometry of the respective Western blot and adjusting for the amount of total protein loaded per lane. The amount of bt-15d-PGJ2-modified Keap1 or actin was also measured in the eluate from the Western blot by densitometry, and the percentage of each protein which was recovered by neutravidin pull-down from the total cell lysate protein is shown in Figure 5B (percentage modification). Cell lysate and pulled-down samples were analysed and quantified from the same Western blot membrane in order to minimize variability from blotting development or exposure. Although these experiments do not rule out the possibility that actin modification may occur in response to 15d-PGJ2, they do suggest that 15d-PGJ2 does not cause extensive actin modification or damage at very low concentrations which alter F-actin organization, focal adhesions and migration.
Activation of EpRE (electrophile-response element)-dependent intracellular antioxidants
It is known that modification of critical thiols in Keap1 by 15d-PGJ2 results in an increase in the activity of the transcription factor Nrf2 (nuclear factor erythroid 2-related factor 2), and subsequent transcription of genes under the control of the EpRE. Such genes include subunits of glutamyl cysteine ligase which controls the production of GSH [4,42,43]. To demonstrate that adduct formation with Keap1 correlates with a biological response in JC cells, we monitored Keap1 modification by bt-15d-PGJ2 and the subsequent increase in GSH levels after 16 h.
GSH levels were significantly increased after exposure to 3 μM 15d-PGJ2, but not at 0.3 μM (Figure 5C). Thus these data demonstrate that adduct formation with Keap1 correlates with activation of EpRE-dependent gene expression in JC cells, but that this effect occurs at concentrations of 15d-PGJ2 higher than those required to elicit focal-adhesion turnover, attenuation of migration or reorganization of the F-actin cytoskeleton.
Effects of p38 inhibition on 15d-PGJ2-induced focal-adhesion disassembly and F-actin cytoskeletal changes
Since direct modification of actin, PPARγ activation and DP2-dependent pathways cannot adequately account for the effects of 15d-PGJ2 described herein, we investigated the role of a known redox-active MAPK (mitogen-activated protein kinase) signalling pathway. The p38 pathway has been implicated in the regulation of actin dynamics and can be activated downstream of focal-adhesion signalling [44–46]. Therefore we assessed activation of p38 in response to low levels of 15d-PGJ2 using Western blot analysis. Treatment with 15d-PGJ2 (0.3 μM) for 15 and 30 min resulted in a significant increase in p-p38 (phosphorylated p38) compared with the vehicle control (EtOH)-treated cells (Figure 6A).
Role of p38 in focal-adhesion disassembly and cytoskeletal arrangement regulation
The effect of p38 inhibition on 15d-PGJ2-induced alterations in focal adhesions and the F-actin cytoskeleton were examined next. There was a significant decrease (18%, P<0.01) in the percentage of cells positive for focal adhesions upon treatment with 0.3 μM 15d-PGJ2 (Figure 6B). The p38 inhibitor SB203580 alone had no effect on focal adhesions, but pretreatment with SB203580 for 30 min prevented the decrease in focal-adhesion-positive cells in response to 15d-PGJ2 (Figure 6B).
JC cells were also pretreated with the p38 inhibitor SB203580, and then the effect of 15d-PGJ2 on F-actin morphology was assessed. Figure 6(C) shows the F-actin cytoskeletal structure in JC cells that were pretreated with the p38 inhibitor prior to exposure to 0.24 μM BD-15d-PGJ2. Cells treated with the electrophile exhibited significant F-actin alterations. The p38 inhibitor itself had no apparent effect on the F-actin structure. Interestingly, pretreatment with SB203580 was able to prevent 15d-PGJ2-induced F-actin cytoskeletal rounding. Taken together, these data suggest a role for the p38 pathway in the F-actin cytoskeletal reorganization and focal-adhesion disassembly in response to 15d-PGJ2. The effect of SB203580 on 15d-PGJ2-induced attenuation of migration was also assessed; however, treatment with the p38 inhibitor itself resulted in the inhibition of migration (results not shown). Therefore it is unclear what role p38 plays in mediating the effect of 15d-PGJ2 on cellular migration.
Breast cancer metastasis is a major cause of mortality and morbidity in patients, and therefore agents that can inhibit this process, particularly with minimal toxicity to the patient, are desirable therapeutic options. In the present study, we demonstrate that 15d-PGJ2 at low concentrations (<1 μM), which do not cause cytotoxicity, stimulate focal-adhesion disassembly, cause F-actin reorganization and attenuate migration of JC mouse mammary adenocarcinoma cells. Since these processes are required for successful metastasis, our data point to a potential anti-metastatic activity of 15d-PGJ2.
This lipid electrophile can work through multiple mechanisms of action including post-translational modification of thiols, PPARγ and DP2 receptors. It is for this reason that we examined the effects of PPARγ and DP2 agonists on the end points described herein. ROSI, a PPARγ agonist, has previously been shown to alter focal-adhesion signalling and impair migration . Additionally, Powell  and Monneret et al. [49,50] demonstrated that 15d-PGJ2 can bind to and activate the DP2 receptor on eosinophils. Although DP2 receptor expression seems to be limited to Th2 T-cells, cytotoxic T-cells, eosinophils and basophils in humans , whereas PPARγ is expressed more ubiquitously , we examined the effect of both DP2 and PPARγ agonists [15(R)-PGD2 and ROSI respectively] on focal-adhesion disassembly. Both agonists were used at concentrations 20-fold higher than the reported EC50 for each compound [53,54], and both agonists had an effect on focal adhesions which was comparable with 15d-PGJ2 (Figure 4C). Importantly, the effects on F-actin and migration appear to be independent of activation of PPARγ or the PGD2 receptor and direct modification of actin (Figure 5 and Supplementary Figure S3), but instead are likely to be modulated by one or more redox signalling pathways. It is expected that events upstream of these signalling pathways include the covalent modification of specific protein targets of 15d-PGJ2, which have yet to be elucidated in this model.
There have been a number of previous studies demonstrating the ability of 15d-PGJ2 to cause cancer cell death, and this is thought to occur primarily through PPARγ-mediated activation of cell-death pathways [23,47]. However, studies by our group and others have also shown that 15d-PGJ2 causes apoptosis in a number of cell types, including endothelial cells, through the direct modification of protein thiols in mitochondrial proteins . The modification of these proteins leads to permeability transition and activation of apoptotic cell death . This has raised the concern that 15d-PGJ2 might have toxic side effects when used therapeutically at doses which kill cancer cells . For this reason, we chose to investigate the possibility of targeting metastatic properties of cancer cells at concentrations of 15d-PGJ2 that are not lethal. Our results demonstrate that cell processes which promote metastasis including migration, focal-adhesion disassembly and F-actin reorganization can be effectively modulated by low sub-lethal concentrations of 15d-PGJ2. Additionally, this is consistent with the finding that, at low micromolar concentrations, 15d-PGJ2 attenuates neutrophil migration after an inflammatory stimulus in a mouse model of peritonitis .
Our observation that 15d-PGJ2 causes profound reorganization of the F-actin cytoskeleton (Figure 4B) is consistent with previous reports in neuroblastoma cells . In their study, Aldini et al.  attributed the F-actin changes to direct modification of actin by 15d-PGJ2 through formation of covalent adducts. In the present study, we were able to recapitulate this result insomuch as direct protein adduct formation of 15d-PGJ2 with actin was observed at 3 and 20 μM bt-15d-PGJ2 (Figures 5A and 5B). Interestingly, bt-15d-PGJ2 did not appreciably form protein adducts with actin at 0.3 μM, although there was still a profound effect on the F-actin cytoskeleton at this concentration. These results indicate that, although 15d-PGJ2 forms protein adducts with actin at higher concentrations, this adduct formation does not adequately explain the extensive effect on F-actin reorganization observed at low concentrations of 15d-PGJ2 (<1 μM).
Instead, we have focused on the p38 signalling pathway which is known to be redox-regulated and has been implicated in actin structural dynamics in a number of cancer model systems . The p38 signalling pathway was also shown previously to be activated by 15d-PGJ2 at low micromolar concentrations in two human endothelial cell models [58,59]. Multiple stimuli that regulate the actin cytoskeleton, focal-adhesion disassembly, cell motility and invasion converge on the p38 signalling pathway. For example, in neuroblastoma cells, the WASP (Wiskott–Aldrich Syndrome protein)/WAVE (WASP verprolin homologous protein) family member WAVE3 has been shown to regulate actin polymerization and cytoskeletal organization through p38-dependent signalling . Orr et al.  also showed that focal-adhesion disassembly in response to thrombospondin is regulated by p38 . Furthermore, activation of Hsp27 (heat-shock protein 27) by p38 is well-established as an important regulator of actin polymerization and depolymerization . PMA-induced migration of glioblastoma cells has been shown to occur through the p38/Hsp27 signalling axis . More recently, it was shown that the motility of glioma cells is inhibited by flavonid silibinin by a mechanism involving ROS generation and p38 activation . Taken together, these reports demonstrate the integral role p38 plays in modulating cytoskeleton organization, focal-adhesion disassembly, motility, and invasion initiated by diverse stimuli. Although our experiments implicate p38 in the mechanism of 15d-PGJ2-mediated actin reorganization and focal-adhesion disassembly (Figure 6), further studies are necessary to determine the role of other potentially important redox signalling pathways on this effect.
The mechanism by which 15d-PGJ2 causes focal-adhesion disassembly appears to be distinct from those responsible for migration and F-actin reorganization. FAK activation occurs primarily through integrin-mediated signal transduction. Downstream signalling events regulate multiple biological process including cell survival, proliferation, angiogenesis, and of particular interest in this context, migration and invasion (reviewed in ). Chen et al.  demonstrated previously that 15d-PGJ2 treatment of thyroid carcinoma cells caused decreased levels of the focal-adhesion proteins vinculin, integrin β1, FAK and paxillin; however, these effects were observed at concentrations of 15d-PGJ2 which also caused cell death. Focal adhesions have been shown to be modulated by a number of pathways including FAK, extracellular matrix components and integrin signalling (reviewed in ). FAK signalling is altered in response to 15d-PGJ2, not through changes in total FAK levels as described previously , but through changes in FAK signalling (Figure 4). Since PPARγ and DP2 agonists were able to decrease the number of cells which stain positive for focal adhesions to a similar extent as 15d-PGJ2, it is likely that focal-adhesion signalling is regulated by multiple mechanisms. Future studies will examine the role of signalling downstream of PPARγ and DP2 in the regulation of focal-adhesion disassembly to determine whether 15d-PGJ2 activates common pathways.
In summary, we have shown that 15d-PGJ2 attenuates mammary cancer cell motility at sub-lethal concentrations. This effect is preceded by extensive alterations in the F-actin cytoskeletal organization resulting in the rounding of the F-actin cytoskeleton and significant focal-adhesion disassembly. Moreover, these effects appear to be independent of PPARγ activation or the direct modification of actin by the electrophile. Our results indicate that the p38 signalling pathway plays an integral role in mediating the 15d-PGJ2-induced alterations in the aforementioned parameters. Although further studies are required to identify the redox-sensitive protein target or targets of 15d-PGJ2 responsible for changes in F-actin, focal adhesions and ultimately migration, it is clear that modulation of redox signalling pathways by electrophiles may constitute important anti-metastatic therapeutic avenues in the future.
prostaglandin D2 receptor
15-deoxy-Δ12,14-prostaglandin J2, BD-15d-PGJ2, BODIPY FL EDA (4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl ethylenediamine)-tagged 15d-PGJ2
fetal bovine serum
heat-shock protein 27
Kelch-like ECH-associated protein 1
nuclear factor erythroid 2-related factor 2
peroxisome-proliferator-activated receptor γ
reactive oxygen species
Anne Diers, Brian Dranka, Karina Ricart, Joo Yeun Oh, Michelle Johnson, Fen Zhou, Manuel Pallero and Thomas Bodenstine performed experiments. Anne Diers, Fen Zhou, Joo Yeun Oh, Joanne Murphy-Ullrich, Danny Welch and Aimee Landar designed experiments and analysed results. All authors contributed to the preparation of the manuscript.
This work was supported by a Junior Faculty Development Grant from the Comprehensive Cancer Center at the University of Alabama at Birmingham (to A.L.); the National Institutes of Health [grant numbers NIH R01 HL079644 (to J.E.M.-U.), NIH R01 CA87728 (to D.R.W.)]; and the National Foundation for Cancer Research (D.R.W.). A.R.D. is supported by T32 HL007918 (Medical Scientist Training Program Grants, University of Alabama at Birmingham).