Peroxisome Proliferator-Activated Receptor-γ Ligands Alter Breast Cancer Cell Motility through Modulation of the Plasminogen Activator System

We investigated peroxisome proliferator-activated receptor-γ (PPAR-γ) ligands effect on cell motility and the plasminogen activator system using normal MCF-10A and malignant MCF-10CA1 cell lines. Ciglitazone reduced both wound-induced migration and chemotaxis. However, the effect was not reversed with pretreatment of cells with the PPAR-γ-specific antagonist GW9662. Immunoblot analysis of conditioned media showed ciglitazone decreased plasminogen activator inhibitor-1 (PAI-1) in both cell lines; this effect was also unaltered by PPAR-γ antagonism. Alternatively, treatment with the ω-6 fatty acid arachidonic acid (ArA), but not the ω-3 fatty acid docosahexanoic acid, increased both MCF-10A cell migration and cell surface uPA activity. Pretreatment with a PPAR-γ antagonist reversed these effects, suggesting that ArA mediates its effect on cell motility and uPA activity through PPAR-γ activation. Collectively, the data suggest PPAR-γ ligands have a differential effect on normal and malignant cell migration and the plasminogen activation system, resulting from PPAR-γ-dependent and PPAR-γ-independent effects.


Introduction
A function of any tumor cell that allows for propagation of diseased cells is the ability for that tumor cell to invade the surrounding tissue. One family of proteins involved in this pathological process is the plasminogen activator (PA) system [1,2]. The PA system includes the urokinase-type plasminogen activator (uPA). uPA is most active when bound to its cell surface urokinase receptor, uPAR. In addition to the role of the uPA/uPAR complex in the degradation of the ECM, this complex plays a role in cell adhesion. uPAR is able to engage cell surface integrins, allowing for attachment of cells expressing the uPA/uPAR complex to other surrounding cells. Another key component is plasminogen activator inhibitor-1 (PAI-1), the physiological inhibitor of uPA activity [1,2]. PAI-1 binds uPA bound to the cell surface, forming a PAI-1/uPA/uPAR complex that is then recognized by the scavenger protein low-density lipoprotein receptor-related protein (LRP), which internalizes the tertiary complex [3,4]. Paradoxically, elevated levels of PAI-1 in breast cancer patients are associated with decreased patient survival [5].
Peroxisome proliferator-activated receptor-gamma (PPAR-γ) is a transcription factor that is considered the master regulator of adipogenesis [6,7]. However, PPAR-γ has been found in numerous cell lines, including endothelial cells [8,9], normal and malignant prostate epithelium [10,11], and normal and malignant breast epithelium [12]. PPAR-γ is a ligand-activated nuclear transcription factor and the target of the thiazolidinedione (TZD) class of insulin sensitizing drugs [6,13]. Drugs in this family bind PPAR-γ, resulting in the activation of the PPAR-γ/retinoid X receptor (RXR) heterodimer. PPAR-γ then binds the PPAR response element (PPRE) in the promoter of target genes, recruits coactivators, and then the gene is transcribed. In addition to TZD drugs, PPAR-γ has been shown to be activated by the naturally occurring 15-deoxy-Δ 12,14 -prostaglandin J2 (15d-PGJ2) [14]. Although 15d-PGJ2 is a potent agonist for PPAR-γ in vitro, there is data suggesting 15d-PGJ2 is not found at a high enough concentration to act as an in vivo ligand for PPAR-γ [15].
In addition to the TZD class of drugs and 15d-PGJ2, PPAR-γ has also been shown to be activated by a number of dietary fatty acids, specifically omega-3 (ω-3) and omega-6 (ω-6) fatty acids. A diet high in fat is associated with the development of a number of diseases, including cardiovascular disease, type 2 diabetes mellitus, and a variety of cancers. Dietary fat intake has been linked to prostate cancer risk [16], colon cancer [17][18][19], and breast cancer [20]. Thoennes, et al., showed differential transcriptional activity by PPAR-γ following treatment of MCF-7 cells with ω-3 and ω-6 fatty acids [21]. Treatment with ω-3 fatty acids inhibited levels of PPAR-γ activation, while ω-6 fatty acids increase PPAR-γ activity over control [21].
The goal of this study was to investigate the effect of PPAR-γ ligands on breast cancer cell motility and the plasminogen activator system. The TZD ciglitazone decreased cell motility, independent of PPAR-γ. PAI-1 levels were lower following ciglitazone treatment. The naturally occurring PPARγ ligand 15d-PGJ2 also reduced wound-induced cell migration. Interestingly, treatment with the ω-6 fatty acid arachidonic acid (ArA) increased cell motility, while the ω-3 fatty acid docosahexanoic acid (DhA) had no significant effect. Our collective results suggest that the PPAR-γ ligand ciglitazone decreases cell motility, in a PPAR-γ independent manner, potentially though the down-regulation of PAI-1; alternatively, the PPAR-γ ligand ArA promotes migration in a PPAR-γ dependent manner that increases uPA.

In Vitro Wound Healing
Assay. Cells were plated at 1.0 × 10 5 cells per well in a 12-well tissue culture treated plate as detailed previously [24,25]. At confluence, cells were serumstarved overnight. Cells were then scratched with the tip of a sterile yellow pipet tip and serum-free media containing various concentration of 15d-PGJ2 (Calbiochem, San Diego, Calif, USA) or ciglitazone ranging to 10 μM (Cayman Chemical, Ann Arbor, Mich, USA) from ethanol stocks were added to each well. Migration was monitored at 0, 6, and 12-hours using a Kodak MDS290 camera. Wound closure was quantified by measuring distance as pixels between each leading edge of the wound (10 lines/wound) at each time point using the measuring tool in Adobe Photoshop, with a grid superimposed on image to guide measurements.

Modified-Boyden Chamber Assay.
Following serum starvation, cells were treated with PPAR-γ ligands ranging to 10 μM of ciglitazone, ArA (ArA-sodium salt, Sigma, St. Louis, Mo, USA), or DhA (DhA-sodium salt, Sigma, St. Louis, Mo. USA) in serum-free media for 24 hours. Lower wells of chamber contained DMEM:F12 plus 1 mg/mL fatty-acidfree bovine serum albumin (BSA, Sigma, St. Louis, Mo, USA) with or without 5 ng/mL EGF (Invitrogen, Carlsbad, Calif, USA). Cells (1 × 10 5 ) were plated in upper wells in DMEM:F12 containing 1 mg/ml fatty-acid free BSA, above a collagen IV coated, 10 mm porated membrane. Chambers were incubated at 37 • C for 6-hours in a humidified atmosphere. Cells were fixed and stained with Diff-Quick (Dade-Behring, Newark, DE). Cells that migrated to the undersurface of the membrane were examined microscopically at 200x magnification. Each condition was done in triplicate, with 4 fields counted per well [1]. In experiments with GW9662, serum-starved cells were pretreated for 30 minutes with GW9662 (5 μM) (Calbiochem, San Diego, Calif, USA), then ciglitazone or fatty acid treatment was added to cells for 24-hours. GW9662 is an irreversible PPAR-γ antagonist and it was used at a concentration where it is selective for PPAR-γ in cells [26,27].

Immunoblot Analysis.
Conditioned media from treated cells was collected and concentrated with centrifugal concentrators (Amicon Ultracel 30 kD, Millipore, Billerica, Mass, USA). Protein concentration was determined using BioRad Protein DC assay (BioRad, Hercules, Calif, USA). Proteins were separated by SDS-PAGE in 10% polyacrylamide and electrotransferred to PVDF membrane. Phosphate buffered saline/0.1% Tween-20 (PBS/Tween) buffer was used in all steps of immunoblot analysis. Each step was preceded by three 9-minute washes at room temperature. Nonspecific binding was blocked by 5% nonfat dry milk for 30 minutes at room temperature. Membrane was incubated at 4 • C overnight with primary antibody diluted 1 : 1000 (unless otherwise noted) in 1% nonfat dry milk. Membrane was exposed for 1-hour at room temperature to horseradish peroxidase conjugated secondary antibody diluted 1 : 5000 in 1% nonfat dry milk in PBS/Tween. Membrane was exposed to luminal substrate for 1 minute, covered in plastic wrap then exposed to X-ray film. Primary antibodies were: rabbit antihuman PAI-1 (1 : 2000 dilution) (Molecular innovations, Novi, MI) and rabbit anti-human uPA (no. 389, American Diagnostica, Stamford, Conn, USA) as described previously [1].

Indirect Cell-Surface Associated UPA Activity Assay.
MCF-10A and MCF-10CA1 cells (1 × 10 5 ) were plated in a 96-well plate [1]. Following 24-hour serum starvation, cells were pretreated with PPAR-γ antagonist GW9662 or vehicle control for 30 minutes at 37 • C. Cells were then treated with various concentrations (up to 10 μM) of ciglitazone or arachidonic acid for 24 hours at 37 • C. After treatment, cells were washed with PBS and plasminogen was added to cells and incubated at room temperature. The supernatant was removed and added to another 96well plate containing buffer amiloride to inhibit any residual uPA activity. Chromogenic substrate is then added to the well and hydrolyzed by plasmin generated by plasminogen cleaved by uPA on the cell surface. Rate of chromogenic substrate cleavage by plasmin was measured at 405 nm for 90 minutes.

Plasminogen Activator, PPAR-γ, and RXR in MCF-10A
and MCF-10CA1 Cells. As previously reported, MCF-10A cells express less uPA and uPAR but more PAI-1 than MCF-1CA1 breast cancer cells [1]. Both cell lines express PPAR-γ and RXRs (data not included). Based on these findings, we performed a study with some PPAR-γ ligands on uPA/PAI-1-mediated cell migration processes comparing near normal MCF-10A cells to oncogenic Ras-transformed metastatic MCF-10CA1 cells.

PPAR-γ Ligands Decrease In Vitro Wound Closure.
Ciglitazone decreased wound closure dose dependently (Figure 1(a)), with 5 μM ciglitazone reducing cell closure by 39% compared to no ciglitazone. 15d-PGJ2 also decreased cell closure dose dependently, with 10 μM 15d-PGJ2 reduced cell closure by 50% compared to no 15d-PGJ2 (Figure 1(b)). These results show that PPAR-γ ligands decrease wound closure of MCF-10A cells, and they further support the literature that PPAR-γ activation inhibits migration of cancer cells in vitro.

Ciglitazone Treatment Decreases Chemotaxis, Decreases
PAI-1 Expression, but Increases uPA Activity. Ciglitazone decreased cell chemotaxis to EGF in a dose-dependent manner ( Figure 2). To determine if these effects were mediated by PPAR-γ, we pretreated the cells with the PPAR-γ specific antagonist GW9662. Interestingly, blocking PPARγ activation with GW9662 (5 μM) pretreatment did not reverse the effect of ciglitazone (5 μM) in either cell line. Control experiments with 5 μM GW9662 showed neither detrimental effect on cell viability nor changes in cell mo- tility. The data suggest ciglitazone is working in a PPAR-γ independent manner to reduce cell migration. The effect of ciglitazone on cell viability was then determined by MTT assay. Treatment of MCF-10A and MCF-10CA1 cells with 5 μM ciglitazone partially reduced cell viability (Abs 595 nm of cells with no and 5 μM ciglitazone was 0.331 ± .014 and 0.292 ± .003 for MCF-10A cells and 0.304 ± .006 and 0.279 ± .002 for MCF-10CA1 cells, resp.). There was a substantial loss of cell viability at 10 μM ciglitazone for both cell lines; thus, all further experiments used 5 μM ciglitazone. In additional control experiments, there was no loss of cell viability with the PPAR-γ ligands 15d-PGJ2 or ArA when tested up to 10 μM (data not included). These results imply that the effect of ciglitazone in MCF-10A and MCF-10CA1 cell motility is not due to a substantial reduction in cell viability.
In both MCF-10A and MCF-10CA1 cell lines, ciglitazone treatment resulted in decreased PAI-1 protein expression (Figure 3). To determine if this decrease in PAI-1 expression was mediated by PPAR-γ, we pretreated with GW9662 prior to ciglitazone treatment. We did not see a reversal  of ciglitazone-mediated reduction in PAI-1 expression (Figure 3) suggesting ciglitazone is affecting PAI-1 levels independently of PPAR-γ.
In MCF-10A cells, ciglitazone treatment alone or in conjunction with GW9662 pretreatment increases uPA activity on the cells surface (Figure 4(a)). Ciglitazone treatment in MCF-10CA1 cells did not significantly alter uPA activity although it seems GW9662 treatment in these cells results in more plasmin generation (Figure 4(b)).

Arachidonic Acid Increases Both MCF-10A Cell
Motility and uPA Activity. With 10 μM ArA, we see a significant increase in cell motility compared to control cells ( Figure 5(a)). When we pretreated the MCF-10A cells with the PPAR-γ antagonist GW9662, we see a reduction in cell motility ( Figure 5(a)). ArA treatment increases uPA activity on the cell surface though GW9662 did not seem to fully reduce uPA activity ( Figure 5(b)). Additionally, treatment with the ω-3 fatty acid DhA, up to 10 μM, had no effect on either cell motility or cell viability (data not shown). These results suggest ArA is able to activate PPAR-γ, resulting in increased cell motility and uPA activity.

Discussion
PAI-1 and uPA protein expression have been used as strong independent prognostic indicators for breast cancer [5,[28][29][30]. In addition to cancer, PAI-1 overexpression is linked to a variety of disease states. Morbidly obese individuals have elevated circulating PAI-1 levels, likely due to an increase in PAI-1 expression from adipose tissue [31]. In rats with streptozocin-induced diabetes, PAI-1 levels are increased 60-80% over control [32]. In humans, elevated PAI-1 levels have been reported in patients with T2DM [33] and is related to cardiovascular dysfunction [33,34]. While the literature on PPAR-γ activation and PAI-1 alterations is conflicting, it has been shown in a number of cell types and in vivo that PPAR-γ does modulate PAI-1 expression [34][35][36][37]. We treated cells with ciglitazone, 15d-PGJ2, and ArA acid to investigate effects of PPAR-γ activation on migration and PAI-1 expression following treatment. Based on previous literature, we expected to see differential effects of PPAR-γ activation, specifically with ArA treatment [21].
In vitro, treatment of tumor cells with TZDs results in a number of antitumor effects. In prostate cancer cells, PPAR-γ ligands reduced proliferation, induced terminal differentiation, and downregulated E-cadherin and c-myc expression [38]. Pioglitazone, in combination with valproic Journal of Oncology acid, upregulates E-cadherin and reduced invasion and migration in prostate cancer cells [39]. We found that treatment with either ciglitazone or 15d-PGJ2 resulted in a significant decrease in wound closure of MCF-10A cells. Ciglitazone treatment decreased chemotaxis toward EGF in both MCF-10A and MCF-10CA1 cells. GW9662 is a specific PPAR-γ antagonist, which binds PPAR-γ and blocks ligand binding and subsequent activation of the receptor [40]. Surprisingly, pretreatment with GW9662 did not reverse the effects of ciglitazone, which suggests that ciglitazone mediates this reduction in migration through a PPAR-γindependent mechanism. Emery et al. showed rosiglitazone and pioglitazone inhibited proliferation of pituitary tumors; however, PPAR-γ antagonists did not reverse these effects, suggesting the antiproliferative effect was independent of PPAR-γ activation [41]. Another study found ciglitazone and 15d-PGJ2 induced apoptosis in normal and malignant B Journal of Oncology cell, independent of PPAR-γ [42]. Finally, ciglitazone and 15d-PGJ2 have been shown to activate p38 MAPK signaling, which were reported to be independent of PPAR-γ activation [43,44]. Interestingly, ArA treatment of MCF-10A cells enhanced cell migration. These effects were reversed in cells pretreated with GW9662, suggesting ArA is acting in a PPAR-γdependent manner. Since ω-3 and ω-6 fatty acids have been shown to have differential effects on PPAR-γ activation [21], we also investigated if ω-3 fatty acids had an effect on cell migration in our system. We saw no change in migration in MCF-10A cells treated with DhA, which agrees with past studies that ω-6, but not ω-3, fatty acids promote cell motility [45]. GW9662 pretreatment did not fully reverse ArA-induced uPA activity; one possibility for this is ArA also signals through PI3K [46] to upregulate uPA expression [47]. It is also possible ArA is engaging PPAR-γ intracellularly, resulting in increased cell migration, while independently initiating the PI3K signaling cascade and then upregulating uPA activity. One limitation of our study was the exclusive use of GW9662 for its irreversible PPAR-γ antagonist effect [26,27]. Future studies with MCF-10A and MCF-10CA1 cells would benefit from either silencing PPAR-γ expression or expressing a dominant negative PPAR-γ to investigate any possible differences in cell motility or proliferation following treatment with ciglitazone or other PPAR-γ agonists. Another limitation to our study was the absence of reporter studies for PPAR-γ gene regulation.
TZDs may be useful adjuvant therapies in cancer treatment. One clinical trial in phasetwo investigated the effect of pioglitazone in conjunction with a COX-2 inhibitor in glioma patients and saw moderate results in patients with high-grade glioma, suggesting pioglitazone treatment may be beneficial to a subset of patients [48]. A phase-I trial of a non-TZD PPAR-γ agonist LY29311 studied maximum tolerated dose in a combination regimen in patients with advanced solid tumors and determined there was no limiting toxicity and no disease progression [49]. To date, these advances have not been realized with PPAR-γ agonists in contrast to their preventative benefits in diabetic patients.

Conclusions
This study shows ciglitazone treatment reduces both normal and malignant epithelial cell migration in vitro, independently of PPAR-γ activation. Additionally, we found ciglitazone treatment reduces PAI-1 protein levels, and this effect was not reversed by antagonism of PPAR-γ. We hypothesize that the antimigratory effects of ciglitazone are mediated by the alteration of the PA system in these cells. We know PAI-1 inhibits apoptosis, can promote cell motility, and plays a role in intracellular signaling [1,2]. Given the role of PAI-1 in these tumor processes, the in vivo data showing FDA-approved TZDs decrease PAI-1 in diabetic patients, and our results and those of others, one could draw the conclusion that TZD therapies may eventually prove to be a valid adjuvant therapy for some breast cancer patients.