The Urokinase Receptor Supports Tumorigenesis of Human Malignant Pleural Mesothelioma Cells

Abstract

Malignant pleural mesothelioma (MPM) is a lethal neoplasm for which current therapy is unsatisfactory. The urokinase plasminogen activator receptor (uPAR) is associated with increased virulence of many solid neoplasms, but its role in the pathogenesis of MPM is currently unclear. We found that REN human pleural MPM cells expressed 4- to 10-fold more uPAR than MS-1 or M9K MPM cells or MeT5A human pleural mesothelial cells. In a new orthotopic murine model of MPM, we found that the kinetics of REN cell tumorigenesis is accelerated versus MS-1 or M9K cells, and that REN instillates generated larger tumors expressing increased uPAR, were more invasive, and caused earlier mortality. While REN, MS-1, and M9K tumors were all associated with prominent extravascular fibrin deposition, excised REN tumor homogenates were characterized by markedly increased uPAR at both the mRNA and protein levels. REN cells exhibited increased thymidine incorporation, which was attenuated in uPAR-silenced cells (P < 0.01). REN cells traversed three-dimensional fibrin gels while MS-1, M9K, and MeT5A cells did not. uPAR siRNA or uPAR blocking antibodies decreased REN cell migration and invasion, while uPA and fetal bovine serum augmented the effects. Transfection of relatively low uPAR expressing MS-1 cells with uPAR cDNA increased proliferation and migration in vitro and tumor formation in vivo. These observations link overexpression of uPAR to the pathogenesis of MPM, demonstrate that this receptor contributes to accelerated tumor growth in part through interactions with uPA, and suggest that uPAR may be a promising target for therapeutic intervention.

Malignant pleural mesothelioma (MPM) is a relatively rare but aggressive form of thoracic cancer that is in most cases associated with previous exposure to asbestos. The incidence of MPM is 1 in 1,000,000 internationally, and approximately 3,000 new cases occur annually in the United States (1, 2). However, due to its protracted latency period (20–40 yr), the incidence of MPM is expected to increase. Current treatments for MPM include chemo- or radiation therapy and/or surgery; however, none of these treatments alone or in combination are curative. Because most patients die within 15 to 18 months of their initial diagnosis and more effective treatment is currently lacking, the search for new interventional targets is well-justified (3, 4).

Neoplastic cells are often characterized by increased expression of the urokinase plasminogen activator receptor (uPAR). Increased uPAR expression has been associated with poor prognosis in lung, prostate, and other cancers (5–7). We previously reported that uPAR regulates cell surface plasminogen activation by uPA and mitogenesis of cultured MPM cell lines (7). We also found that uPAR expression in MPM and MeT5A human pleural mesothelial cells is responsive to a number of stimuli otherwise implicated in the pathogenesis of MPM, including transforming growth factor-β (TGF-β1) and tumor necrosis factor-α (TNF-α) through regulation at the post-transcriptional level (7) and that uPAR was expressed in MPM tumor tissue harvested from patients (8). To extend these observations, we investigated the possible contribution of uPAR to virulence of MPM in a novel orthotopic mouse model in which three MPM cell lines that differentially express uPAR were administered by intrapleural injection. We then sought to determine the contribution of uPAR to specific effects that participate in tumorigenesis of MPM cells, including proliferation, migration in fibrin and other matrices, and invasiveness.

MATERIALS AND METHODS

Cell Culture

Cell lines used in these studies included MeT5A human pleural mesothelial cells (American Type Culture Collection, Manassas, VA) and three lines of MPM cells: REN (from Dr. S. Albelda, University of Pennsylvania, Philadelphia, PA), MS-1 (from Dr. S-M. Hsu, The University of Texas Health Science Center at Houston, Houston, TX), and M9K cells (from Dr. B. Gerwin, NIH, Bethesda, MD). All cells were cultured in Complete RPMI media, which contained 10% fetal bovine serum (FBS), 1–2% antibiotic-antimycotic, 1% L-Glutamine (Gibco/BRL, Grand Island, NY), and 5 μg/ml plasmocin (Invivogen, San Diego, CA) and were grown at 37°C in a humidified 5% CO₂ environment. TNF-α was used for treatment of the cells, and was obtained from R&D Systems (Minneapolis, MN). In the literature, REN cells are described as having an inflammatory epithelioid phenotype (9); MS-1 cells as exhibiting a spindle shaped or sarcomatous phenotype (8, 10), and M9K cells as exhibiting a mixed epithelioid and sarcomatous phenotype (11). However, in our initial characterizations, we found that all three MPM cell lines demonstrated an epithelioid phenotype in culture, possibly reflecting the effects of cell passage (data not shown).

A New Orthotopic Mouse Model of Thoracic Human MPM

All experiments involving the use of animals in this study were approved by the UTHSCT Animal Review Committee. REN, MS-1, and M9K cells were harvested and prepared for injection into the pleural cavity of nude mice (BALB/c Athymic NCr-nu/nu; National Cancer Institute at Frederick, Frederick, MD). The cells were detached using trypsin-EDTA, washed once with PBS, counted, pelleted, and then resuspended in equal amounts of PBS containing Matrigel (1:4 in PBS; BD Bioscience, San Jose, CA) in a final volume of 150 μl/mouse. A 1-cc syringe with a 25-gauge 5/8-inch needle was loaded with the mixture, which was then injected into the pleural cavity. Our approach is novel in that the cells are injected medially, perpendicular to the sternum at the left 4th or 5th interspace at the costal margin. This approach was chosen to prevent pneumothorax, which commonly occurred with more lateral injections in our preliminary experiments. We used injections with Coomasie Blue dye to determine that the instilled volume distributed uniformly within the right and left thoracic cavities using the central injection technique. Tumors were allowed to grow for defined intervals of up to 28 days, after which the mice were killed, the chest cavities photographed, and tumors counted (12). Tumor volumes were calculated from electronic digital vernier caliper measurements (cat. no. 14-648-17, lower limit of sensitivity for measurement of 0.1 mm; Thermo Fisher, Freemont, CA) of tumor length and width: volume (mm³) = length × (width²)/2. Tumor volumes were calculated in situ for each tumor and then combined for total tumor volume per animal. Detectable tumors that were below the caliper range of measurement were assigned length. The tumors were carefully resected and weighed as an index of tumor burden. Tumor homogenates were also prepared by flash freezing in Triton X-100 lysis buffer for protein extraction and TriZol for RNA/DNA extraction. Tumor homogenates and RNA preparations were stored at −80°C until used. Whole animal gated noncontrast CT imaging was done using a small animal GE eXplore Locus CT scanner (General Electric, Milwaukee, WI).

Histology and Immunohistochemistry

Tumor tissue was embedded in paraffin and 5-μm sections were cut and fixed to positively charged histology slides, which were incubated at 60°C for 30 minutes to melt the paraffin and further deparaffinized using Clear-Rite 3 (Thermo Fisher) and ethyl alcohol washes. Tumor sections were immunostained for the following MPM markers: cytokeratin-7, thrombomodulin, and Hector Battifora mesothelial-1 (HBME) (Cat. No. RB-10456-P1, 141C01, MS-1494-S1, respectively; Lab Vision, Thermo Fisher) and Calretinin (Cat. No. 18-0211; Zymed, San Francisco, CA). The antibodies were used at dilutions recommended by the manufacturer and allowed to incubate for 1 hour at room temperature. A Lab Vision DAB kit (Cat. No. TP-015-HD; Thermo Fisher) was used to stain the slides, antigen retrieval was performed, and the slides were finally counterstained and mounted. The tumor sections were also stained using Trichome (Cat. No. S1816; Polyscientific, Bay Shore, NY, and 2,576 [chromic acid]; Mallinckrodt, Hazelwood, MO) and factor VIII antigen detection (Cat. No. MS-722-P; Thermo Fisher).

Tumor sections were also stained for human uPA, uPAR, PAI-1, and human fibrin (Cat. No. 3689, 3936, 3785, 350, respectively; American Diagnostica, Stamford, CT). Slides were incubated in each of the antibodies at a concentration of 10 μg/ml overnight at 4°C. Super-sensitive multi-link (Cat. No. HK340; BioGenex, San Ramon, CA) and alkaline phosphatase–conjugated strepavidin label (Cat. No. HK331; BioGenex) were used to detect the antigens with the Fast Red (Cat. No. HK182; BioGenex) chromagen. Tissue sections were also incubated with a rabbit anti-mouse fibrin antibody (Cat. No. ASMFBGN-GF; Molecular Innovations, Novi, MI) at a concentration of 17 μg/ml overnight at 4°C using a rabbit anti-mouse AEC stain kit from Lab Vision (Cat. No. TP-015-HAX; Thermo Fisher). All slides were counterstained and mounted for examination. Immunohistochemical expression of the selected antigens or tumor vascularity was scored as 1, mild expression; 2, moderate expression; or 3, strong expression based on independent analyses of 50 high-power (×400) fields per mouse or alternatively all available tumor tissues (n = 3 REN, MS-1, or M9K mice/group).

Western Blot Analysis of Fibrinolytic Pathway Components

MPM cell lines were allowed to incubate at 37°C in serum-free RPMI media for 24 hours before collection of conditioned media and cell lysis in a 1% Triton X-100/PBS buffer. The lysates and conditioned media were then centrifuged at 13,000 rpm for 30 minutes. The lysates were cleared, and protein concentrations were determined, aliquoted, and frozen at −80°C. Tumor and cell line homogenates and conditioned media were resolved via SDS PAGE and uPA, PAI-1, and uPAR proteins were detected by Western blotting. Fifty micrograms of tumor lysate protein was resolved by SDS-PAGE and transferred to a PVDF membrane. Rabbit polyclonal primary antibodies were used at a dilution of 1:1,000 to detect uPA (American Diagnostica) 1:3,000 for uPAR (Attenuon, LLC, San Diego, CA), and 1:5,000 for PAI-1 (Abcam, Cambridge, MA) and incubated overnight at 4°C. Donkey anti-rabbit secondary antibody conjugate was used at a dilution of 1:15,000 for 30 minutes. Membranes were developed by enhanced chemiluminescence and images were scanned using an HP 3210 Scanner.

RNA Isolation and uPAR mRNA Detection

RNA was isolated and resolved as previously described (13). The membrane was then probed with [³²P]-labeled DNA against uPAR mRNA and exposed to Kodak BioMAX Xray film (Eastman Kodak Co., Rochester, NY). The stability of uPAR mRNA was assayed in transcription-chase experiments using Actinomycin D (10 μg/ml; Sigma Aldrich, St. Louis, MO), as we previously reported (14).

uPAR Gene Silencing

REN cells were transfected with uPAR siRNA as previously described (15, 16). Briefly, 50 to 80% confluent cells were transfected with 100 nM negative control oligonucleotide (#5; Dharmacon, Chicago, IL), an siRNA that does not specifically target any gene in the cell and controls for nonspecific effects, or uPAR siRNA composed of a previously described sequence (Dharmacon) (15) using LipofectAMINE 2000 (Invitrogen, Carlsbad, CA) for 6 hours. The cells were then incubated for 24 hours in RPMI media containing 10% FBS and 24 hours in serum-free RPMI before utilization.

Analysis of siRNA Induction of IFN-γ in REN Cells

REN cells were transfected with siRNA as described above. Briefly, 50 to 80% confluent cells were transfected using LipofectAMINE 2000 (Invitrogen) for 6 hours with the following: transfection control (PBS), 100 nM negative control oligonucleotide #5, or 100 nM uPAR siRNA. Transfection naïve cells were also used as controls. The cells were then incubated for 24 hours in RPMI media containing 10% FBS and 24 hours in serum-free RPMI before utilization. Conditioned media were collected and assayed using an enzyme-linked immunosorbent assay kit for IFN-γ production according to manufacturer's instruction (Cat. No. 88–7316; eBioscience, San Diego, CA).

Proliferation Assays

DNA synthesis was measured by thymidine incorporation, as previously described (7, 17). [³H]-thymidine was added to the media at a final concentration of 1 μCi/ml. Data describing the effects of uPA and FBS on thymidine incorporation was then normalized against the serum-free control for each respective cell line. The standard error of the mean was then calculated for each sample and normalized for percent error.

Migration Assays

Migration of MPM cells was first assayed by determining the rate at which cells crossed a porous membrane as previously described (18, 19). Briefly, 2.0 × 10⁴ cells in 200 μl of serum-free RPMI were seeded on the apical chamber of 6.5 mm, 8 μm pore Transwell filter inserts. Five hundred microliters of 2.5% FBS containing RPMI media was added to the basolateral chamber. The cells were then incubated at 37°C for 6 hours. The cells were washed three times in Mg²⁺/Ca²⁺ containing PBS, fixed in 100% methanol at 4°C and counted, taking an average of three optical fields at ×250 magnification for each stained filter. The effect of uPA on the migration of REN cells was tested in assays in which two chain “active” uPA (tcuPA) was used as a chemoattractant. Briefly, 2.0 × 10⁴ serum starved REN cells were added to the apical chamber of Transwell filter inserts. The basolateral chamber was filled with 500 μl of serum-free RPMI; serum-free RPMI with 1 μg/ml of tcuPA (124; American Diagnostica), or 2.5% FBS containing RPMI. The chambers were then incubated for 6 hours at 37°C. The cells were then fixed, stained, and counted as described above.

The effects of uPAR blockade on migration of REN cells were next determined in assays in which cells transfected with uPAR or control siRNA were tested. Twenty-four to 48 hours after siRNA transfection, siRNA treated cells were placed in serum-free RPMI media for 12 to 24 hours. A quantity of 2 × 10⁴ cells was then added to the apical chamber of Transwell filter inserts, after which the cells were then allowed to migrate for 6 hours, fixed, stained, and counted as previously described. In parallel assays, serum-starved REN cells were incubated with serum-free RPMI supplemented with PBS, 25 μg/ml of rabbit IgG, or a purified rabbit polyclonal uPAR neutralizing antibody (D3036-2; Attenuon), which prevents the binding of uPA to uPAR. We confirmed that D3036-2 prevents the binding of tcuPA to REN cells by uPAR immunoprecipitation and amidolytic analyses (data not shown).

Matrigel Invasion Assays

Invasion assays were performed as previously described with minor modifications (18, 19). Briefly, 6.5 mm, 8 μm pore, Matrigel-coated Boyden Chambers (BD Biosciences, Bedford, MA) were rehydrated according to manufacturer's instructions. A quantity of 5.0 × 10⁴ cells was then added to the filter supports in serum-free media. The bottom chamber was then filled with 500 μl of RPMI supplemented with 2.5% FBS. The cells were then allowed to invade the matrix for 12 to 15 hours at 37°C. Counts of invading cells were made using an average of three optical fields at ×250 for each of three stained filters.

Control and uPAR siRNA-treated REN cells were also tested for their ability to invade the Matrigel matrix. Twenty-four to 48 hours after siRNA transfection, control oligonucleotide and uPAR siRNA-treated cells were placed in serum-free RPMI media for 12 to 24 hours. A quantity of 5 × 10⁴ cells was then added to the apical chamber of Matrigel-Boyden chambers, after which the cells were then allowed to migrate for 12 hours at 37°C, fixed with 100% methanol at 4°C, and then stained with 0.1% Crystal Violet solution. Invading cells were then counted as described above.

Fibrin Gel-Overlay Invasion Analyses

The fibrin overlay experiments were conducted as previously described with minor modifications (19, 20). REN, MS-1, or M9K cells were grown to subconfluence on 24-well plates for 24 hours at 37°C in serum-containing RPMI followed by a 24-hour incubation in serum-free RPMI. The fibrin gel was then prepared by adding human thrombin (0.2 units/ml; Enzyme Research Laboratories, South Bend, IN) to sterile filtered human fibrinogen (3 mg/ml; Sigma) in a HEPES saline buffer (130 mM NaCl, 25 mM HEPES, 5 mM CaCl₂, pH 7.4). Five hundred microliters of this solution was then added to each well of cells, and a fibrin gel formed over the cells by 30 minutes at 37°C in a humidified CO₂ incubator. Cells were maintained with Complete RPMI media supplemented with 225 KIU/ml of aprotinin. After 72 hours, the fibrin gel was fixed in 4% formaldehyde at 4°C and incubated with a 1:5,000 dilution of Hoechst nuclear stain in PBS. Fibrin gels were then washed with PBS and imaged with a fluorescence microscope (Nikon Epifluorescent Microscope TE 100, Melville, NY). Three optical fields from each of three wells were used for cell counting at ×250 (Sensicam Camera; 3I Imaging Innovations, Denver CO).

Binding of Bovine uPA to REN Cells

REN cells were incubated in serum-free RPMI media for 24 hours before experimentation. In the first set of treatments, attached REN cells were incubated with RPMI media supplemented with 10% FBS for 24 hours at 37°C followed by treatment with serum-free RPMI media or RPMI media supplemented with 10% FBS for 60 minutes at 4°C. In the second set of treatments, cells were first acid-stripped to remove cell-surface uPA using a low pH Glycine buffer (50 mM Glycine, 100 mM NaCl, pH 3.0) for 3 minutes at 4°C and neutralized with a HEPES buffer rinse (500 mM HEPES, 100 mM NaCl), then incubated with serum-free RPMI media; serum-free RPMI supplemented with 10% FBS, 0.5 μg/ml of human tcuPA or 1.0 μg/ml of human tcuPA for 1 hour at 4°C. A quantity of 2.0 × 10⁵ cells from each treatment condition was then washed in a 0.2% BSA/PBS wash buffer, incubated with a 1:100 dilution of a rabbit anti-human uPA antibody (389, American Diagnostica) in wash buffer for 1 hour at 4°C, washed three times, then incubated with a 1:100 dilution of a goat anti-rabbit PE conjugated secondary antibody (Jackson ImmunoReseach Laboratories, West Grove, PA) for 1 hour at 4°C. The cells were then fixed in 4% PBS-buffered formalin and analyzed via flow cytometry (BD Coulter FACS Caliber; BD Bioscience, San Jose CA).

Binding Human and Murine uPA to REN Cells

REN cells were grown to confluence in Complete RPMI medium, incubated overnight in serum-free RPMI media, and washed with HBSS (pH 7.4; Gibco/BRL). Washed cells were incubated with tcuPA or purified single chain sc-human uPA (a generous gift from Abbott Laboratories, Chicago, IL), or recombinant mouse scuPA (0–10 nM; Attenuon) for 15 minutes (room temperature) in 0.5 to 1.0 ml of the same buffer. Cells were then transferred to 96-well white assay flat bottom Costar plate (Corning Inc., Corning, NY), and mixed with an equal volume of 0.1 mM fluorogenic uPA substrate (Pefafluor uPA; Centerchem, Norwalk, CT) in HEPES buffer (100 mM HEPES, 20 mM NaCl, pH 7.4). Amidolytic activity of uPA was determined from changes in fluorescence emission at 440 nm (excitation 344 nm) using a Varian Cary Eclipse fluorescence spectrophotometer equipped with 96-well plate reader accessory (Varian Inc, IL). Since the activity of scuPA is only 0.1% of tcuPA, human recombinant plasmin (1.6 ng) (421; American Diagnostica) was added to the aliquots of cells, and incubated with scuPA 20 minutes before addition of Pefafluor uPA. The amount of bound uPA was estimated by comparison of its activity with that of a standard known concentration of tcuPA (American Diagnostica). To estimate the affinity of uPA to REN cells, the amount of the active enzyme bound to the cells ([E_b]) was plotted against the concentration of the enzyme, which was used for incubation with cells ([E₀]). SigmaPlot 10.0 (SPSS, Inc.) was used for data analysis, employing the Levenberg-Marquardt algorithm. Finally, suPAR competition assays in which uPA-related amidolytic activity of REN cells incubated with 1 nM human or mouse uPA in the presence or absence of suPAR (60 nM; Attenuon LLC) were performed to determine if mouse scuPA binds uPAR on REN cells.

Effect of Anti-uPAR Antibodies on the Interaction Between uPA and suPAR

Precipitation of complexes of uPAR with antibodies by Protein A/G Plus beads (Santa Cruz Biotechnology, Santa Cruz, CA) was used to determine the effects of anti-uPAR and anti-uPA antibodies on the uPA/suPAR interaction. Soluble receptor (50–100 nM), as well as its complexes with 100- to 200-nM monoclonal uPAR blocking antibodies (ATN-658 [Attenuon], which does not interfere with uPA/uPAR interactions but blocks uPAR-mediated signaling; a monoclonal uPAR-binding antibody, which blocks the interaction between uPA and uPAR [3936; American Diagnostica], and a rabbit polyclonal antibody to human uPAR [D3036-2; Attenuon]) were tested for their ability to block the uPAR–uPA interaction. These antibodies and isotype control antibodies were incubated with human uPA (10 or 100 nM) for 15 minutes in a HEPES wash buffer, pH 7.4. Nonspecific mouse IgG, uPA, and suPAR were assayed alone or in combination as controls. Antibody complexes were precipitated with 100 μl of a 1:1 suspension of Protein A/G Sepharose in HEPES buffer. Active uPA bound to the uPAR/antibodies was detected via measurement of uPA's amidolytic activity as described above. Amidolytic uPA activity in FBS (Gibco/BRL) was measured in the samples of FBS diluted 4 to 40 times with HEPES buffer. Changes in fluorescence emission with time were compared with a tcuPA positive control.

Creation of REN Cells Stably Transfected with uPAR shRNA and Determination of Tumor Aggressiveness

A uPAR shRNA construct was engineered based on the siRNA sequence previously described (15, 21). This sequence was then cloned into the pSilencer 2.1 U6-Puro expression vector according to manufacturer's instructions (Applied Biosystems/Ambion, Austin, TX). REN cells were then transfected with 3 μg of the uPAR shRNA or the pControl shRNA vector using LipofectAMINE 2000. Two days after transfection, the cells were placed in Complete RPMI media containing puromycin (3 μg/ml; Sigma-Aldrich). Individual clones were then selected and expanded. Clones were then assayed for uPAR expression via Western blotting analysis as described above.

Only 5 of the initial 24 uPAR shRNA clones could be grown in culture, propagated, and characterized. A clone demonstrating 90% suppression of uPAR expression compared with naïve cells was selected and expanded. While these cells were viable, adherent, and morphologically indistinguishable from naïve REN cells, they grew slowly in culture. Despite repeated attempts to propagate these cells, only a limited number; enough to inject three animals, could be generated for testing in the mouse model. Cells were inoculated into nude mice as described above. Briefly, naïve REN cells and REN cells stably expressing the shRNA control vector, pControl, or uPAR shRNA were injected into the pleural space of nude mice (n = 3 per group) as described above. Tumors were allowed to grow for 28 days, at which time the mice were killed and the tumors were counted, assessed volumetrically by caliper measurement, excised, and weighed.

Effect of uPAR Overexpression on MS-1 Tumor Aggressiveness In Vitro

MS-1 MPM cells were engineered to stably express increased amounts of uPAR compared with parental MS-1 cells. MS-1 cells were stably transfected with pcDNA 3.1 empty vector (EV) or uPAR cDNA. Two days after transfection, cells were selected in Complete RPMI media containing G418 (400 μg/ml; Gibco/BRL). Individual clones were then expanded and assayed for increased uPAR expression. An MS-1 clone expressing uPAR at 7-fold higher levels than naïve or EV-transfected MS-1 cells was identified. Naïve, EV, and uPAR-expressing MS-1 cells were assayed in proliferation, migration, and invasion assays as described above.

Effect of uPAR Overexpression on MS-1 Tumor Progression In Vivo

The MS-1 clone expressing a 7-fold increment of uPAR, EV-transfected, and transfection naïve MS-1 cells were expanded and inoculated into nude mice as described above. Briefly, naïve MS-1 cells and MS-1 cells stably expressing EV and uPAR cDNA cells were injected into the pleural space of nude mice (n = 7 per group). Tumors were allowed to grow for 49 days (MS-1 tumors grow less aggressively than REN), at which time mice were killed, tumors counted, assessed volumetrically, excised, and weighed. Representative tumors from each group were then allocated for Western and Northern blotting analysis, as described above.

Statistical Analysis

For tumor progression analyses, Poisson regression analysis was used in which interaction between cell types was fitted. All other data are expressed as the mean ± SEM of replicate determination as indicated. For volumetric and tumor weight analyses, values of the sensitivity of the device used for the measurement (0.1 mm for determination of tumor volume by caliper measurements and 0.00001 g for weights) were assigned when the tumors were detected, but were too small to be volumetrically assessed or weighed. Statistical significance was determined by unpaired two-tailed Student's t test, where P < 0.05 was considered statistically significant.

RESULTS

Expression of uPAR Is Increased While uPA Is Not Expressed in REN Cells

REN MPM cells expressed increased amounts of uPAR protein relative to MS-1 and M9K MPM cells or MeT5A cells (by 5-, 4-, and 10-fold, respectively; Figure 1A). uPA protein was detectable in MS-1 and M9K cells, but not in the MeT5A and REN cell lysates. PAI-1 protein expression was comparable in REN, MS-1, M9K cells, and MeT5A cells (Figure 1A). The conditioned media of REN and M9K cells contained comparable amounts of PAI-1, with the MS-1 cells expressing relatively less. MS-1 and M9K conditioned medias were found to contain significantly more uPA than the MeT5A, but uPA was not detectable in the REN media (Figure 1B). uPAR mRNA was commensurately increased in REN cells. In transcription chase experiments, we found that REN uPAR mRNA was relatively stable, with a half life of approximately 13 hours (results not shown). Its decay was prolonged versus the MS-1 and M9K cell lines, which we previously reported to be 5 and 7 hours, respectively (14). MeT5A and REN cells expressed comparable levels of PAI-1 mRNA, with the MS-1 and M9K cells expressing lower amounts. MS-1 and M9K cell lines express large amounts of uPA mRNA; however, uPA mRNA was not detectable in the MeT5A or REN cells (Figure 1C). Treatment with 10 ng/ml TNF-α for 18 hours increased uPAR protein and message by Northern blotting in REN, MS-1, and M9K lysates, but uPA protein or mRNA remained undetectable in REN cells (data not shown).

Figure 1.

Expression of urokinase plasminogen activator receptor (uPAR), uPA, and PAI-1 by MeT5A mesothelial cells and REN, MS-1, or M9K MPM cells. (A) Fifty micrograms of cell lysates were resolved via SDS-PAGE and analyzed by Western blot analysis as described in Materials and Methods. β-actin expression was assessed as the loading control. (B) Serum-starved cells were incubated with serum-free RPMI media for 12 hours at 37°C. The conditioned media was then collected from each cell line and 40 μl was then resolved on an SDS-PAGE and analyzed by Western blot analysis as described in Materials and Methods. (C) uPAR, uPA, and PAI-1 mRNA expression in cultured MPM cells were analyzed via Northern blot analysis. Total RNA was isolated from serum-starved MeT5A and REN, MS-1, or M9K cells and resolved on an agarose/formaldehyde gel as detailed in Materials and Methods. The panels each illustrate findings representative of those of three independent experiments.

REN Cells Generate Larger Tumors More Rapidly and Are More Invasive in a New Orthotopic Model of MPM in Nude Mice

Since uPAR has been implicated in the pathogenesis of a number of solid neoplasms and in proliferation of MPM cells in vitro (22), we next tested the possibility that the uPAR-overexpressing REN cells grow more rapidly in vivo. We first established a model of MPM in BALB/c athymic nude mice injected intrapleurally with 1 × 10⁶ cells/ mouse in a 1:4 Matrigel/PBS mixture, as described in Materials and Methods. In preliminary experiments, excised REN, MS-1, and M9K MPM tumors were tested for retention of classical MPM markers. REN- and MS-1–derived tumors expressed cytokeratin 7, thrombomodulin, HBME, and calretinin, while the M9K-derived tumors expressed cytokeratin 7, thrombomodulin, and HBME but not calretinin, confirming their MPM origins and marker retention in vivo (data not shown).

We next assessed the ability of mice to tolerate intrapleural injection of REN, MS-1, or M9K cells. We found that REN cells rapidly formed exophytic tumors (n = 5 mice/group receiving either 2 or 4 × 10⁶ cells in Matrigel, range 16–22 intrapleural tumors found in each animal) by Days 19–24 after intrapleural administration, at which point the animals died or were killed because of the development of apparent distress. In contrast, tumors were relatively smaller and tumor burden was better tolerated, with no mortality or overt distress after intrapleural administration of 4 × 10⁶ cells/mouse in the MS-1 (n = 4, range 8–17 intrapleural tumors) and M9K-injected group (n = 4, range 6–27 tumors), which were both killed at Day 62. These initial experiments suggested that the REN cells were more virulent in vivo after intrapleural injection.

We lastly compared the ability of an equal number (1 × 10⁶ cells/mouse) of intrapleurally administered REN, MS-1, and M9K cells to generate tumors. The mice were killed at 2-, 3-, and 4-week intervals to compare the rate of tumor growth and response to tumor burden (n = 9 mice/interval, 3 mice/group). The animals were found to tolerate this level of tumor burden over the 4-week time course, and none required euthanasia due to distress. The numbers of tumors increased in each group with time, and there were more tumors at Week 4 than at Weeks 2 or 3 in all groups (Figure 2A, P < 0.01 in each case). Intrapleural administration of REN cells resulted in more tumors than M9K (P < 0.0001; risk ratio [RR], 3.62; 95% confidence interval [CI], 2.39–5.50) or MS-1 group (P < 0.0001; RR, 3.53; 95% CI, 2.36–5.30). By direct visualization and photography (Figure 2B), relatively larger, exophytic tumors occurred earlier in the REN group. At the 28-day interval, only REN tumors were invasive, which was evident in all three cases and included perineural, skeletal muscle, brown fat, perivascular, and lung parenchymal invasion. By whole animal respiratory-gated CT scanning and autopsy, tumors were limited to the thorax (data not shown). The tumors generated from all three MPM cell lines in the in vivo model displayed an epithelioid phenotype (data not shown).

Figure 2.

REN, MS-1, and M9K generate intrathoracic tumors after intrapleural injection in nude mice. (A) Intrapleural administration of REN, MS-1, and M9K cells all generated progressively more discrete tumor masses over time and REN cell-derived tumors produced significantly more tumors over the 4-week time course than MS-1 and M9K cells (P < 0.0001). (B) A representative photograph shows exophytic tumors generated by REN cells in the mouse model. The yellow arrows indicate the tumors. A broad arrow indicates the heart and a broken arrow indicates the contiguous lung tissue complicating with tumor. In this example, the heart has been displaced to the left to reveal a large retrocardiac tumor proximate to the injection site. Pleural tumors are also observed. (C) Excised exophytic tumors were fixed and stained for fibrin using an antibody to murine fibrin using FAST Red, which produces a red chromagenic signal. Tumors derived from each cell line consistently demonstrated fibrin at the periphery of the tumor masses and fibrin intercalated within the REN, MS-1, and M9K tumors. The figures are representative of the findings of the tumors harvested in each of the groups (n = 3 animals/group). Black arrows indicate the presence of fibrin deposition in tumor stroma. The magnification is at ×400 in each of the images and are representative of the findings in all animals (n = 3 mice/group at each of the 3 intervals). (D) Characterization of fibrinolytic pathway components, uPAR, uPA, and PAI-1 in excised exophytic REN, MS-1, and M9K tumors. At 28 days after injection, MPM tumors were excised and solubilized in Triton X-100 lysis buffer. Fifty micrograms of lysate was then resolved on an SDS-PAGE and then assayed for uPAR, PAI-1, and uPA antigen expression by Western blotting. β-actin expression was assessed as the loading control. This figure is representative of three independent experiments.

REN, MS-1, and M9K Tumors Are Characterized by Prominent Extravascular Fibrin Deposition and uPAR Expression

The transitional neomatrix of the REN, MS-1, and M9K tumors was fibrinous, and fibrin deposition was robust and most prominent at the periphery of REN tumors, while more focal fibrin positivity was detected at the periphery of the M9K and REN tumors (Figure 2C). Interestingly, prominent fibrin deposition was apparent by 2 weeks after intrapleural injection of the cells and was maintained at 4 weeks after injection with the same distribution. At 3 and 4 weeks after injection, tumor neovascularization was readily identified by Trichrome stain (23) in all REN tumors (n = 3 mice), while a single tumor-associated vessel was found in one of three MS-1–challenged mice and none of three tumors from M9K-injected mice. By immunohistochemical analyses at Week 3, moderate on average 2/3 immunostaining intensity of uPAR was detectable only in REN tumor cells and was at background levels within the MS-1 or M9K tumors (n = 3 mice/group). By Week 4, uPAR expression remained consistently increased in the REN tumor cells (n = 3 mice at each interval), again averaging a staining score of 2/3 within the REN cells versus an average of 1/3 or relatively mild staining in the MS-1 and M9K tumors (n = 3 mice/group). Moderate (2/3) uPA immunostaining was detectable in all of the REN tumors and was mild (1/3) intensity in the MS-1 or M9K groups. The same pattern of moderate immunostaining for PAI-1 was present in the REN tumors, with mild staining likewise detected within the MS-1 and M9K groups. All immunohistochemical analyses were performed by a pathologist (T. Allen) blinded to the origin of the samples.

uPAR Expression Is Increased in REN-Derived Tumors

Western blotting of REN tumor homogenates revealed relative overexpression of uPAR and PAI-1 compared with MS-1 and M9K but demonstrated uPA, as did the homogenates from MS-1– and M9K-derived thoracic MPM tumors (Figure 2D). Since neither uPA protein nor mRNA was detected in cultured REN cells (Figure 1), we speculated that cross-reactivity of the antibody with mouse uPA could account for its detection and found that this antibody detected purified recombinant murine uPA (data not shown), although we cannot rule out that uPA expression could be induced in the REN cells in vivo. We next addressed the further possibility that murine uPA could bind REN cells and tested the ability of murine versus human uPA to bind to REN cells and to soluble human uPAR. REN cells bound both human and mouse uPA, and soluble uPAR demonstrated effective competition only with human uPA at the surface of REN (data not shown).

REN Cells Increase Their Proliferation and Migration to a Relatively Greater Degree When Stimulated with Either FBS or uPA

Thymidine incorporation was used to measure the proliferation of MeT5A and REN, MS-1, and M9K cells. Serum-starved subconfluent cells were treated with serum-free RPMI media with or without uPA, or RPMI plus 10% FBS. Figure 3A illustrates that the REN, MS-1, and M9K cells exhibited more thymidine incorporation than MeT5A cells, indicating an increased rate of MPM proliferation. Among the MPM cells, the REN cells exhibited the greatest increment of thymidine incorporation in the presence of FBS. REN cells alone exhibited increased thymidine incorporation in response to treatment with tcuPA. REN cells were also found to migrate more rapidly than MS-1 or M9K or MeT5A cells when compared with serum-free treated controls (P < 0.01 and P < 0.001, respectively) (Figure 3B). Migration of REN cells also increased in the presence of tcuPA compared with serum-free controls, but less than FBS-treated cells (P < 0.05 and P < 0.001, respectively). REN cells were the only tested cell line that migrated in response to tcuPA treatment. Interestingly, invasion of Matrigel by REN and MS-1 cells was comparable and exceeded that of M9K and MeT5A cells (P < 0.01 and P < 0.001, respectively) (Figure 3C).

Figure 3.

MPM cells exhibit increased proliferation, migration, and invasiveness. (A) Proliferation was measured by measuring thymidine incorporation in serum-free, tcuPA-, and FBS-treated cells. REN mesothelioma cells exhibited greater thymidine incorporation than the other MPM cell lines in the presence of tcuPA and FBS (*P < 0.05 when compared with serum-free controls). The bar graphs illustrate data representative of experiments performed twice in quadruplicate in each of the duplicate experiments. (B) MPM cell migration was compared in serum-free RPMI, tcuPA containing serum-free RPMI, or FBS supplemented RPMI media. REN were the only line that exhibited increased migration in the presence of tcuPA versus the serum-free RPMI controls (*P < 0.01, ^†P < 0.001 when compared with the serum-free controls). (C) REN, MS-1, and M9K cells exhibit greater invasiveness than MeT5A normal mesothelial cells. Cells were seeded on Matrigel-coated Boyden chambers and allowed to invade the matrix for 12 hours. **P < 0.05, *P < 0.001 when compared with MeT5A cells; ^†P < 0.01 when compared with M9K cells. The bar graphs illustrate data representative of three independent experiments.

Inhibition of uPAR Protein Expression by uPAR siRNA

We next sought to assess the malignant potential of REN mesothelioma cells in which uPAR gene expression had been silenced. uPAR-specific siRNA reduced uPAR protein expression by 95% compared with cells transfected with a nonsense control siRNA oligonucleotide (Figure 4A), allowing for further analysis of the effects of uPAR expression on REN cell aggressiveness. uPAR silenced REN cells exhibited no morphologic derangements, and uPAR silencing was not associated with cytotoxicity. uPAR siRNA-transfected and control oligonucleotide-transfected cells both demonstrated greater than 98% cell viability by Trypan Blue exclusion analyses at 24, 48, and 72 hours in culture.

Figure 4.

uPAR siRNA-treated REN cells exhibit reduced proliferation, migration, and invasion potentials. (A) uPAR siRNA treatment significantly reduced the expression of uPAR in REN mesothelioma cells. Western blot analysis demonstrates a significant decrease in uPAR expression in uPAR siRNA–transfected cells. The bar graphs illustrate data representative of that of three independent experiments (*P < 0.01). (B) Serum-starved, semi-confluent, untransfected (naïve) and uPAR and control siRNA transiently transfected REN cells were treated with 1 μCi/ml of [³H]-thymidine for 12 hours in serum-free media. uPAR siRNA–treated cells proliferated at a lower rate than control oligonucleotide–treated cells (*P < 0.01). (C) uPAR siRNA–treated cells demonstrated a reduced migration potential versus naïve and control siRNA–treated cells. The bar graphs illustrate data representative of five independent experiments (*P < 0.05). (D) In Matrigel invasion assays, uPAR siRNA–treated REN cells demonstrated reduced invasiveness versus naïve and control siRNA–treated cells. The bar graphs portray data representative of three independent experiments (*P < 0.001). (E) uPAR blocking antibodies reduce the migration of REN, MS-1, and M9K cells. Serum-starved REN cells were incubated with rabbit IgG or the uPAR blocking antibody (D3036-2). The uPAR-neutralizing antibody significantly reduced the migration potential of MPM cells (*P < 0.01). The bar graphs illustrate data representative of three independent experiments.

Silencing of uPAR Decreases Proliferation of REN Cells

Since the REN tumors were more aggressive in vivo, the effect of uPAR on the rate of cell proliferation was first analyzed as a determinant of tumor growth potential. To determine the role of uPAR expression in the proliferation of REN cells, naïve, control oligonucleotide and uPAR siRNA-transfected cells were labeled with [³H]-thymidine and the rate of incorporation was measured in serum-free conditions. As illustrated in Figure 4B, silencing of uPAR gene expression significantly decreased REN cell proliferation. siRNA treatment has been shown to induce antiviral responses through the induction of IFN-γ expression, which results in the inability of a given siRNA to mediate gene silencing. This nonspecific effect is usually associated with siRNAs longer than 30 bp (24, 25). The uPAR siRNA used in this study is only 19 bp, gene silencing of uPAR expression is observed in the REN cells, and we found no change in levels of IFN-γ in siRNA-transfected REN cells versus those treated with control oligonucleotide #5 siRNA or naïve and transfection-reagent–treated control cells (data not shown).

uPAR Expression Contributes to Migration and Invasiveness of REN Cells

To investigate the role of uPAR in REN mesothelioma cell migration, REN cells transfected with control or uPAR siRNA were seeded on Transwell filter inserts using 2.5% FBS as the chemoattractant at 37°C. Cells treated with uPAR siRNA migrated at a significantly lower rate than the naïve and control oligonucleotide treated cells (Figure 4C). Matrigel-coated Boyden chambers were next used to evaluate the role of uPAR on REN cell invasion. Naïve and control or uPAR-specific siRNA-treated REN mesothelioma cells were seeded on Matrigel inserts and allowed to invade at 37°C for 12 hours. uPAR silencing reduced REN cell invasion of the Matrigel-coated inserts (Figure 4D). Finally, uPAR neutralization with a uPAR-binding rabbit polyclonal antibody (D3036-2) that prevents the binding of uPA to uPAR likewise reduced REN cell migration (Figure 4E), indicating that the binding of uPA to uPAR contributes to the migration response.

uPAR Mediates REN Mesothelioma Cell Invasion of Fibrin Matrices

Given the prominent extravascular fibrin we observed in our murine MPM model, we next tested the contribution of uPAR to invasiveness of MPM cells in a three-dimensional fibrin gel invasion system (19, 20). Three days after the application of the fibrin overlay, serum-starved semiconfluent REN, MS-1, M9K, and MeT5A cell invasion of the fibrin was measured. MeT5A and MS-1 or M9K cells did not invade in the three-dimensional fibrin gels (n = 3 independent experiments). Only the REN cells entered and then vertically traversed the gel. uPAR siRNA–treated REN cells were found to invade the fibrin matrix at a much lower rate than untransfected (naïve) and control oligonucleotide–treated cells (Figure 5A). We also found that uPA was detectable in FBS by Western blotting using a cross-reacting rabbit polyclonal antibody (#389; American Diagnostica) and that uPA-related amidolytic activity was detectable in FBS. By flow cytometry using an antibody that cross reacts with bovine uPA (#389; American Diagnostica), REN cells incubated with RPMI containing 10% FBS for 12 hours bound bovine uPA with a 30% increase in uPA labeling in serum-treated versus serum-free–treated cells (results not shown). These observations suggested that bovine uPA could associate with uPAR to enhance invasion in this system. We therefore used an alternative independent approach in which we determined the effect of uPAR-neutralizing antibodies on migration in the fibrin gel system. A rabbit polyclonal antibody (D3036-2) against human uPAR was used to block uPA binding to uPAR and other uPAR-interacting proteins. We found that this antibody blocked the association of uPA with soluble uPAR (data not shown), and at a concentration of 25 μg/ml, D3036-2 blocked the invasion of REN cells within fibrin matrices (Figure 5B).

Figure 5.

uPAR silencing and blockade significantly reduced the ability of REN cells to traverse three-dimensional fibrin gels. (A) uPAR siRNA–treated cells traversed less than untransfected (naïve) and control oligonucleotide–treated cells (*P < 0.05). (B) Cells treated with the uPAR-neutralizing antibodies were found to have a significantly reduced invasion potential of a fibrin matrix in comparison to the control IgG–treated samples (*P < 0.001). The panels illustrate representative findings from three to five independent experiments.

uPAR shRNA Expressing REN MPM Cells Are Difficult to Propagate, and Exhibit a Trend Toward Decreased Tumor Burden after Intrathoracic Injection in Nude Mice

Since uPAR siRNA treatment significantly reduced proliferation, migration, and invasion of REN cells in vitro, REN cells were next engineered to stably express uPAR shRNA. uPAR shRNA expressing REN cells with approximately 90% less uPAR than the naïve or pControl expressing REN cells were used in these experiments. We could only propagate enough uPAR shRNA cells to inject the thoraces of three nude mice. Naïve, pControl, and uPAR shRNA expressing clones were then injected into the pleural cavity of the nude mice (n = 3 mice/group). Tumors formed in mice injected with uPAR shRNA cells (range: 0–9 tumors) appeared to be smaller but did not significantly differ from those of naïve and pControl transfectants, in which large, exophytic tumors were uniformly observed at 28 days after injection. While there was no significant difference between the naïve and pControl-treated mice, there was a trend toward decreased tumor volume in the shRNA group (3.5 ± 1.8 mm³ versus 17.8 ± 4.9 mm³ in naïve controls; n = 3/group, P = 0.052). Tissue invasion was demonstrated in the tumors harvested from mice injected with uPAR shRNA-treated, vector-treated, and naïve control animals (n = 3 animals/group analyzed, data not shown).

uPAR-Overexpressing MS-1 MPM Cells Migrate and Proliferate to a Greater Extent than Naïve and EV-Expressing MS-1 cells

Since uPAR siRNA decreased proliferation, migration, and invasion in REN cells, a uPAR knock-in strategy was used on a MPM cell line that expresses a low level of uPAR antigen. MS-1 cells were engineered to stably overexpress uPAR protein. uPAR-overexpressing MS-1 cells were found to express 7-fold higher levels of uPAR compared with naive and empty vector–expressing MS-1 cells via Western blotting analysis (Figure 6A). Untransfected (naïve) MS-1 cells and MS-1 cells stably transfected with empty vector and uPAR were then studied in proliferation and migration assays. uPAR overexpression increased both proliferation (Figure 6B) and migration (Figure 6C) in the MS-1 cell line.

Figure 6.

uPAR overexpression in MS-1 MPM cells contributes to increased tumor aggresiveness in vitro and in vivo. (A) MS-1 cell lines were engineered to stably overexpress uPAR. By Western blotting analysis, uPAR-overexpressing MS-1 cells exhibit a 7-fold densitometric increase in uPAR expression in comparison to naïve and empty vector MS-1 constructs. (B) Serum-starved naïve (MS-1 naive), empty vector (MS-1 EV), and uPAR-overexpressing transfectants (MS-1 uPAR) were treated with 1 μCi/ml of [³H]-thymidine for 12 hours in the presence or absence of serum-free media. Enhanced uPAR-expressing MS-1 cells were found to proliferate to a greater extent than naive and empty vector–expressing MS-1 cells (*P < 0.05 in comparison to serum-free controls, ^#P < 0.05 in comparison to naïve FBS treated MS-1 cells). (C) uPAR-overexpressing MS-1 cells demonstrated increased migration compared with naïve and empty vector MS-1 transfectants. The bar graphs illustrate data representative of five independent experiments (*P < 0.05). (D) Intrapleural administration of uPAR-overexpressing MS-1 cells produced more exophytic tumors than naïve and empty vector–expressing MS-1 cells over the 49-day time course (*P < 0.05).

uPAR-Overexpressing MS-1 Cells Produce More Exophytic Tumors than Naïve and EV-Expressing MS-1 Cells

Since increased uPAR expression was found to increase proliferation and migration of MS-1 MPM cells, naïve, EV, and uPAR overexpressing MS-1 clones were inoculated intrapleurally in nude mice. Tumors were allowed to grow for 49 days, at which time the mice were killed and tumor growth was assessed as described above. uPAR-overexpressing MS-1 cells produced significantly more tumors (P = 0.043) compared with those generated by the naïve and EV-expressing MS-1 cells (Figure 6D). There was no significant difference of in vitro invasiveness of uPAR-overexpressing MS-1 cells versus naïve and empty vector–treated controls, nor was invasiveness of surrounding tissues demonstrated in the tumors generated in MS-1–overexpressing, naïve, or empty vector–treated mice (data not shown).

DISCUSSION

The literature strongly implicates the fibrinolytic system, transitional fibrin deposition, and its remodeling and uPAR in particular in the pathogenesis and outcomes of a broad range of malignant neoplasms (6, 26–29). These observations suggest that remodeling of extravascular fibrin could likewise promote the growth of MPM and provide a strong rationale to more precisely define the contribution of uPAR to the pathogenesis of MPM. uPAR is a heavily glycosylated, glycosylphosphatidylinositol (GPI)-anchored membrane protein that regulates stromal remodeling by localizing uPA-mediated proteolytic activity to the cell surface. uPAR-mediated signaling can alternately influence a range of processes relevant to neoplastic growth, including cellular proliferation, migration, or invasiveness (15, 27, 28, 30). Our data extend these observations and indicate that uPAR may contribute to accelerated tumor growth and invasiveness of MPM.

To assess the ability of MPM cells differentially expressing uPAR (and other components of the fibrinolytic system) to generate intrathoracic tumors, we developed a new orthotopic model that exploits murine anatomy in that there is a shallow, medially accessible compartment anterior to the heart that is amenable to injection. Using this approach, we reliably generated intrapleural MPM tumors in nude mice with less than 5% risk of pneumothorax, which in our hands commonly occur with more lateral approaches. Although orthotopic models of intrapleural MPM have previously been reported in rats and mice, they were not used to evaluate mechanisms that underlie tumor growth (12, 31). Our model affords a reproducible means to interrogate pathways that promote growth of MPM and represents a potentially powerful tool for the future testing of new therapeutics. We used nude mice as hosts given our study objective, which was to compare the ability of human MPM cells with differential expression of uPAR, to generate tumors in vivo. Because expression of other components of the fibrinolytic system and likely other effectors differed between the MPM lines, we used independent in vitro analyses to determine the contribution of uPAR to proliferation, migratory capacity, and invasiveness.

Tumor tissues derived from the REN, MS-1, and M9K cells retained expression of appropriate markers during the progression of tumor growth and the REN cells demonstrated accelerated tumor growth in the nude athymic mice, compared with MS-1 and M9K. Extravascular fibrin was also a prominent component of the transitional stroma of the tumors derived from the REN, MS-1, and M9K cells. This situation faithfully recapitulates the findings in tumors from patients with MPM, as we previously reported (8). Invasion of the surrounding tissues was observed in REN cell–derived tumors but not MS-1 or M9K tumors. By whole body CT scanning and postmortem analyses, the model also recapitulates the lack of distant metastases usually seen with clinical MPM.

By immunohistochemical analyses, the REN cells expressed discernibly increased uPAR versus the other cell lines. We found that uPAR expression was induced by TNF-α in these cells and that the stability of uPAR mRNA was increased in the REN cells in transcription chase experiments, extending our previously reported findings that uPAR mRNA is subject to post-transcriptional control in MS-1 and M9K cells (14, 22). Interestingly, uPA, PAI-1, and uPAR were all detectable by immunostaining of the excised tumors, even though the REN cells did not express uPA protein or message in culture. Consistent with these observations, increased uPAR expression was found in REN-derived tumor homogenates, while comparable levels of uPA were observed in all homogenates and PAI-1 was decreased in those derived from the MS-1 cells. We speculated that we most likely detected murine uPA within the REN stroma or associated with tumor cells tumor and confirmed that the antibody we used recognizes murine uPA. We also found that murine uPA binds REN cells but that there is little competition with soluble human uPAR for the murine enzyme. These findings raise the possibility that murine uPA binds receptors other than uPAR at the REN cell surface and that these interactions might likewise contribute to their ability to proliferate or migrate in vivo.

We chose REN cells for transfection with uPAR siRNA as they expressed increased levels of uPAR and were more aggressive in vivo. We found that proliferation and migration of uPAR-silenced REN cells was significantly impaired. Using an alternative approach, a polyclonal uPAR-neutralizing antibody that blocks the association of uPA with uPAR and likely binds other uPAR epitopes significantly decreased REN cell migration and invasion. In all, the data confirm that uPAR expression substantively contributes to the ability of REN cells to proliferate, migrate, and invade. Given that REN cells also increased proliferation and migration in response to exogenous human uPA, we conclude that the in vitro effects are at least in part related to the association of uPA with uPAR. As we found that uPA antigen and activity was detectable in FBS and that the bovine uPA binds REN cells by flow cytometry, we further infer that bovine uPA could associate with REN cell uPAR to contribute to these effects. The differential response of the MPM cell lines in response to antibody-mediated uPAR blockade could in part be due to differences in uPA expression. uPAR can mediate both uPA-dependent and -independent migration and invasion (32, 33). The MS-1 and M9K cells both make uPA, whereas REN cells do not. uPAR in REN cells may mediate migration through uPA-signaling effects, whereas the localization of uPA to the cell surface may be more important for MS-1 and M9K cell migration. Thus, the uPAR antibody we used could block uPAR-mediated migration via effects on uPA–uPAR binding or uPAR-mediated signaling that vary between the different cell lines.

Given the prominent representation of extravascular fibrin in the MPM tumors in our model, we also sought to determine the ability of the MPM cells to invade three-dimensional fibrin gel matrices. Interestingly, we found that REN cells were highly migratory in this system, while the MS-1 and M9K or MeT5A pleural mesothelial cells were not. REN and MS-1 cells invaded Matrigel matrices to a comparable extent, but the increased ability of REN cells to invade fibrin may contribute to their enhanced invasiveness in vivo, given the fibrinous neomatrices we found in our MPM model. The contribution of uPAR to this response was confirmed as uPAR-silenced cell migration in fibrin matrices was significantly decreased. While we found that addition of exogenous uPA lyses the three-dimensional fibrin gel, an antibody that blocks the association of uPA with uPAR (D3036-2) interferes with migration in the fibrin gel system. These observations link uPAR expression to the differential ability of REN cells to migrate in a three-dimensional fibrin matrix, either through association with uPA present in the media or alternatively in a uPA-independent manner.

We and others have reported that uPA binding to uPAR can drive proliferation, migration, and invasion in cancer cells (34–38). There are likely several mechanisms that drive these processes and that involve the localized proteolytic activity of uPA when it binds to uPAR. An alternative mechanism involves uPAR-mediated signaling after ligation with uPA that may be independent of the proteolytic activity of uPAR. The role of uPA bound to uPAR in proteolysis, invasion, and migration has been well documented in the literature (reviewed in Ref. 39). Along those lines, the down-regulation of uPAR in human carcinoma cell lines decreases uPA-induced ERK activation and leads to tumor dormancy through disruption of a uPAR/uPA/α5β1 complex (40). uPA has also been found to drive proliferation in cells expressing uPAR and EGFR through ERK and STAT5b activation (15, 41).

The effects of exogenous uPA were relatively muted versus that of serum, raising the possibility that uPAR mediated the effects via alternate pathways. In fact, uPAR has the ability to interact with other proteins and is known to mediate tumorogenic effects via these interactions. Resnati and colleagues reported that cleavage of uPAR by its ligand, uPA, causes a modification in uPAR which allows it to interact with the FPR-like receptor-1 receptor, FPRL1. This interaction activates the FPRL1 receptor, which then facilitates an increase in cell migration (42, 43). Work by Jo and colleagues further suggests that expression of uPAR by cancer cells could mediate the effects of EGF on EGFR interactions and facilitate cell proliferation. The phosphorylation of Tyr-845 and activation of STAT5b are dependent upon the presence of uPAR and EGFR and not necessarily uPA (15, 44, 45). uPAR expression can also activate ERK and Rac1 in an EGFR-independent manner and thereby facilitate cell migration (44, 46). uPAR is also known to interact with a number of integrins, including β1 forms (reviewed in Ref. 29), and may serve as a ligand to mediate integrin-dependent binding and signaling (45, 47–49). However, our data strongly suggest that the uPA–uPAR interaction could contribute, at least in part, to REN cell proliferation and migration, although other mechanisms are also likely invoked as observed when REN are stimulated with 2.5% FBS.

Due to the effects of uPAR down-regulation and blockade on classic in vitro readouts of tumor aggressiveness (e.g., decreased tumor cell proliferation, migration, and invasion), stably transfected uPAR shRNA–expressing REN cells were constructed. Only a fraction of the uPAR shRNA clones propagated to the extent that they could be assayed in subsequent analysis. The uPAR shRNA transfectant with the greatest inhibition of uPAR expression was chosen for further analysis. Unfortunately, this clone was very difficult to propagate and we were only able to obtain enough cells to inoculate three mice. A trend toward decreased volumetric tumor burden was observed in tumors generated by uPAR shRNA REN cells versus naïve or control shRNA-treated cells, but tissue invasion was observed in all tumor types. While differential patterns of invasion could emerge over more extended analyses, these experiments were not feasible. Nevertheless, the difficulty propagating uPAR shRNA–transfected REN cells themselves suggests that the receptor substantively contributes to tumor growth, and the in vivo findings are consistent with that notion. As the interpretation of these data is limited, we used an alternative approach to test the effects of stable transfection of uPAR on tumor growth of MS-1 cells.

Relatively low uPAR expressing MS-1 MPM cells were next engineered to stably overexpress uPAR. uPAR-overexpressing MS-1 cells were found to proliferate and migrate to a greater extent than naïve and EV-expressing MS-1 cells, indicating that increased uPAR expression in MS-1 cells increased MPM tumor progression in vitro. Naïve, EV-, and uPAR-overexpressing MS-1 cells were next inoculated into the pleural space of nude mice and the tumors were allowed to grow for 49 days. uPAR-overexpressing MS-1 cells were found to produce significantly more exophytic tumors and greater tumor burden than the naïve and EV-transfected cells, but evidence of tissue invasion could not be demonstrated. It is possible that the level of overexpression of uPAR was not sufficient to generate an invasive phenotype in vivo, that invasiveness might be demonstrable later in the progression of the tumor, or that uPAR is not critical to this response in MPM cells.

In summary, our data implicate uPAR in the growth of MPM cells in vitro and strongly suggest that it plays a role in accelerated tumor growth in vivo. We established a new orthotopic model of MPM which can be used to compare the growth of human MPM, evaluate underlying mechanisms responsible for tumor growth and invasiveness, or test the efficacy of new interventions. It is unclear that any interaction of murine uPA with human uPAR occurs in REN tumors in vivo or over a pathophysiologically relevant range of enzyme concentrations. uPAR-overexpressing REN cells fail to express uPA and formed tumors that grew more rapidly, were more invasive, and were more poorly tolerated than MS-1 and M9K cells that both express uPA but lesser amounts of uPAR. uPAR gene silencing and antibody blockade indicate that uPAR contributes to the ability of REN cells to proliferate, migrate, and invade. The ability of REN cells to migrate in fibrin matrices is in part related to uPAR expression, and is a potentially important property given the prominent extravascular fibrin we observed in all of the MPM tumors we studied. Further, we have established a direct link between uPAR expression and MPM tumor aggresiveness in vivo by demonstrating that increasing uPAR expression in a low uPAR expressing cell line (MS-1) increases tumor growth in nude mice. Our findings suggest that, as in other malignancies, uPAR represents a potential target for therapeutic intervention in MPM.