Enhanced Peritoneal Ovarian Tumor Dissemination by Tissue Transglutaminase

Epithelial ovarian cancer (EOC) arises from the epithelial layer covering the surface of the ovaries, and i.p. metastasis is commonly observed at diagnosis. Ovarian tumor spread in the i.p. space leads to the characteristic symptoms and complications of the disease, ascites, and small bowel obstruction. Several features set apart ovarian cancer spread from the metastatic model of other epithelial tumors. First, EOC cells are in direct contact with the overlying peritoneal fluid and this allows exfoliated cells to disseminate freely in the i.p. space. Second, ovarian cancer cells derived from the mullerian epithelium have dual epithelial and mesenchymal characteristics and can convert to either phenotype in response to factors in the microenvironment (1, 2). Adopting a mesenchymal phenotype favors dislodgement from the primary tumor, as cells are more motile and not bound by tight cellular junctions. Thus, EOC cells can spread passively to distant sites by exfoliating from the primary tumor, floating in the peritoneal fluid, and nesting along the i.p. space, where they adhere and grow as metastatic implants. This type of spread, which is uniquely characteristic to EOC, is accompanied by specific changes at the interface between tumor and the peritoneal “oncomatrix” that allow cancer cells to move, attach, and grow.

Such changes include increased expression of integrins (3–6) and of the hyaluran receptor CD44 (7) that promote adhesion of EOC cells to the peritoneum and overexpression of the chemokine receptor CXCR4 and secretion of its ligand CXCL12 that regulate cell motility in the i.p. milieu (8). Cancer and mesothelial cells secrete lysophosphatidic acid (9) and other proteins [fibronectin, periostin, osteopontin, and laminin (10–13)], which stabilize the extracellular matrix (ECM) and promote establishment of metastases. These interactions with the mesothelium and the peritoneal stroma activate “outside-in” signaling (14), which stimulates cancer cell proliferation and survival. In this context, neovascularization is facilitated and peritoneal metastases form and grow.

Tissue transglutaminase (TG2) is a ubiquitously expressed enzyme involved in protein cross-linking via acyl transfer between glutamine and lysine residues. TG2 promotes Ca²⁺-dependent post-translational protein modification effected by the insertion of isopeptide bonds and incorporation of polyamines into peptide chains. We previously reported that TG2 mRNA expression is up-regulated in transformed ovarian epithelial cells and tumors compared with normal ovarian surface epithelial cells (15). Other reports link TG2 to epithelial cancers, particularly breast (16), pancreatic (17), and non–small cell lung cancer (18). In breast cancer cells, membrane-bound TG2 has kinase activity and phosphorylates the insulin-like growth factor-binding protein-3 (19) and the enzyme is overexpressed in drug- and radiation-resistant breast cancer cells (20, 21). The goal of this study was to analyze the role of TG2 as a mediator of ovarian tumor dissemination in the peritoneal space.

We show here that TG2 is expressed in a cancer-specific manner in human ovarian tumors and is secreted in ascites fluid. We show that TG2 facilitates ovarian cancer cell adhesion to fibronectin and directional cell migration and promotes i.p. tumor seeding. TG2 exerts its effects by interacting with β₁ integrin, modulating its expression and function. These data suggest a novel role for TG2 as a regulator of i.p. metastasis.

Immunohistochemistry. Twenty-seven paraffin-embedded epithelial ovarian tumor specimens from the Cooperative Human Tissue Collection and 6 normal ovarian specimens from six patients undergoing oophorectomy for benign disorders from the Indiana University Tissue Bank Collection were immunostained using a TG2 monoclonal antibody (CUB 7402, NeoMarkers) at a dilution of 1:200 after antigen retrieval using sodium citrate. Secondary labeling was based on the avidin/biotin system (LSAB2 kit, DAKO). Slides were stained with 3,3′-diaminobenzidine and counterstained with hematoxylin. Negative controls were run in parallel, with omission of the primary antibody. Staining was graded from 0 (no staining) to 3+ (strong staining) by a board-certified pathologist. Immunoreactivity was recorded only if noted in more than 15% to 20% of tumor cells. The Indiana University Institutional Review Board approved the use of human tissue specimens (protocol 0412-54).

Ascites fluid. Thirty samples of ascites fluid cytologically positive from patients with ovarian cancer and 8 samples of ascites fluid from patients with noncancerous conditions (inflammatory pleural or ascitic fluid) were included in this analysis (University of California at Los Angeles and Indiana University Cancer Center Tissue Bank, protocol collection approved by the Institutional Review Board, protocol 0409-02). After collection, samples were centrifuged to remove cellular debris, aliquoted, and stored at −80°C until use.

Cell lines. Human SKOV3 and OV90 ovarian cancer cell lines were obtained from the American Type Culture Collection and cultured in growth medium containing 1:1 MCDB 105 (Sigma) and M199 (Cellgro) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Cellgro) and 1% antibiotics (100 units/mL penicillin and 100 μg/mL streptomycin). All cells were grown at 37°C in a humidified 5% CO₂ atmosphere.

Transfection. To overexpress TG2, OV90 cells in the logarithmic phase of growth were transfected with wild-type TG2 cloned into the pcDNA3.1 vector using Fugene (Roche Applied Science). To knock down TG2, an antisense construct cloned into pcDNA3.1 vector was transfected in SKOV3 cells. As a control, cells were transfected with the pcDNA3.1 vector carrying the G418 resistance gene. Transfection efficiency in these conditions is typically 5% to 10% in OV90 cells and 30% to 40% in SKOV3 cells as determined by estimation of green fluorescent protein expression. Stable transfected clones were established by selection with G418 (Sigma) at concentrations of 600 μg/mL for SKOV3 cells and 150 μg/mL for OV90 cells. Plasmids were generous gift from Professor Janusz Tucholski (University of Alabama, Birmingham, AL). Overexpression and knockdown of TG2 in selected clones were shown by Western blot analysis.

Conditioned medium. Serum-free conditioned medium was collected from SKOV3 cells stably transfected with vector (pcDNA3.1) or the TG2 antisense construct (AS-TG2) and centrifuged at 3,000 rpm for 5 min to sediment cellular debris. Equal volumes of conditioned medium (30 μL) were used for immunoblotting.

Western blot analysis. Cells were lysed in radioimmunoprecipitation assay buffer containing protease inhibitors leupeptin (1 μg/mL), aprotinin (1 μg/mL), phenylmethylsulfonyl fluoride (PMSF; 400 μmol/L), and sodium orthovanadate (Na₃VO₄; 1 mmol/L). Cell lysates were sonicated briefly and subjected to centrifugation at 14,000 rpm for 15 min at 4°C to sediment particulate material. Equal amounts of protein (50 μg) were separated by SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (Millipore). After blocking, membranes were probed with primary antibody overnight at 4°C with gentle rocking. Antibodies used are β₁ integrin antibody (MAB2251, 1:1,000 dilution; Chemicon), TG2 (CUB 7402, 1:1,000 dilution), glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 1:5,000 dilution; Biodesign International), and epidermal growth factor receptor (EGFR; 1:1,000 dilution; Cell Signaling). After incubation with specific horseradish peroxidase–conjugated secondary antibody, antigen-antibody complexes were visualized using the enhanced chemiluminescence detection system (Amersham Biosciences). Images were captured by a luminescent image analyzer with a CCD camera (LAS 3000, Fuji Film).

Separation of membrane and cytosolic proteins. SKOV3 cells were collected in a hypotonic lysis buffer containing 10 mmol/L KCl, 1.5 mmol/L MgCl₂, 10 mmol/L Tris-HCl (pH 7.4), and 2 mmol/L PMSF. The cell lysate was centrifuged at 4,000 × g for 15 min to remove cell debris and nuclei. The supernatant was then centrifuged at 100,000 × g for 60 min to separate the membrane fraction. The final crude membrane pellet was resuspended in a buffer containing 0.25 mol/L sucrose, 10 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, and 2 mmol/L PMSF.

Coimmunoprecipitation. To detect the interaction between TG2 and β₁ integrin, SKOV3 cells were plated on fibronectin-coated (5 μg/mL) dishes, allowed to adhere for 2 h, and then lysed in a buffer containing 20 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1 mmol/L β-glycerolphosphate, 1 mmol/L EDTA, 1 mmol/L EGTA, 1 mmol/L Na₃VO₄, 2.5 mmol/L sodium pyrophosphate, 1% Triton X-100, 10% glycerol, 1 μg/mL leupeptin, 1 μg/mL aprotinin, 400 μmol/L PMSF, and 1 mmol/L DTT. After 30 min of incubation on ice, cells were scraped from the plates, sonicated, and centrifuged at 14,000 rpm for 15 min to pellet cellular debris. Protein (500 μg) from the supernatant was incubated overnight at 4°C with β₁ integrin monoclonal antibody or mouse IgG (Santa Cruz Biotechnology). Immune complexes were recovered by adding 60 μL of a slurry of protein G plus/protein A agarose (Calbiochem) and shaking for 2 h at 4°C. After washing with a buffer containing 0.2% Triton X-100, the immune complexes were eluted with 2× sample buffer and boiled for 10 min at 100°C.

Reverse transcription-PCR. RNA was extracted from SKOV3 cells stably transfected with AS-TG2 or empty vector using RNA STAT-60 Reagent (Tel-Test, Inc.). RNA (2.5 μg) was used for first-strand cDNA synthesis using the SuperScript II System (Invitrogen). The following primers were used: β₁ integrin, ATCTGCGAGTGTGGTGTCTG (forward) and ACAACATGAACCATGACCTC (reverse); GAPDH, GATTCCACCCATGGCAAATTCC (forward) and CACGTTGGCAGTGGGGAC (reverse; Integrated DNA Technologies). The reverse transcriptase product (1 μL) and primers were heated at 94°C for 90 s followed by 28 rounds of amplification for GAPDH and 34 cycles for β₁ integrin (30-s denaturing at 94°C, 30-s annealing at 60°C, and 30-s extension at 72°C) followed by a final extension of 10 min at 72°C. The reverse transcription-PCR (RT-PCR) product was visualized under UV light after fractionation on a 1.5% agarose gel.

Immunofluorescence. SKOV3 cells were plated on fibronectin-coated chamber slides (BD Biosciences) and allowed to adhere. After fixation in 4% paraformaldehyde, cells were permeabilized using Triton X-100 (0.2% in PBS; 15 min) and blocked for 1 h with 3% goat serum in PBS. Then, cells were incubated for 2 h with primary antibody diluted in blocking buffer at room temperature [TG2 polyclonal antibody RB-060 (1:100 dilution; NeoMarkers) and β₁ integrin monoclonal antibody (MAB2251, 1:100 dilution)] followed by a 30-min incubation with AlexaFluor⁴⁸⁸ anti-mouse secondary antibody (1:1,000; Molecular Probes) or Cy5-conjugated anti-rabbit antibody (1:500; Zymed). Staining to visualize the cytoskeleton was done with rhodamine-phalloidin (Molecular Probes). Isotype-specific IgG served as a negative control. Nuclei were visualized by 4′,6-diamidino-2-phenylindole (DAPI) staining (Vectashield, Vector Laboratories). Analysis was done using a Zeiss LSM 510 META confocal multiphoton microscope system under UV excitation at 488 nm (for AlexaFluor⁴⁸⁸), 630 nm (for Cy5), 540 nm (for rhodamine), and 340 nm (for DAPI). Protein colocalization was estimated by calculating the area of color overlap in a Z-stack of images using MetaMorph software.

Solid-phase adhesion assays. Exponentially growing cells were detached from culture plates by trypsinization and labeled with calcein acetoxymethylester (calcein AM; 2 μmol/L; Molecular Probes) for 20 min. After washing, cells were resuspended in serum-free medium. Equal numbers of cells (4 × 10⁴ per well) were seeded into 96-well plates precoated with fibronectin (Sigma) at different concentrations (1–10 μg/mL) or bovine serum albumin (1% w/v). After 1 h of incubation at 37°C, the plate was immersed into PBS containing 1 mmol/L MgCl₂ to remove nonadherent cells. The number of adherent cells was measured in a fluorescence plate reader (Applied Biosystems) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm. All experiments were done in quadruplicate and repeated twice.

Cell migration assays. A migration assay was done in a modified Boyden chamber method using 6.5-mm-diameter, 8.0-μm pore size polycarbonate membrane transwell inserts in a 24-well plate (Corning). To assess directional migration, the lower surfaces of the transwell membrane were coated with 50 μg/mL fibronectin or with 0.01% type I collagen (Sigma). SKOV3 cells stably transfected with AS-TG2 or vector were serum starved for 18 h and then plated in the upper well at a concentration of 2 × 10⁵ in 100 μL of serum-free medium. After 4 h of incubation at 37°C in a CO₂ incubator, the cells on the upper surface of the membrane were wiped off with a cotton swab. The cells on the lower surface of the transwell were stained with HEMA3 stain (Fisher) and counted at ×200 magnification. Cells were counted in five high-power fields (HPF) in duplicate experiments. Results are expressed as mean number of migrating cells ± SE. A similar experiment was done using serum-free conditioned medium from cells stably transfected with AS-TG2 or pcDNA3.1 as cell attractant in the lower chamber of the transwell.

In vivo growth of SKOV3 cells in nude mice. The human ovarian cancer cell line SKOV3 stably transfected with AS-TG2 or vector was injected i.p. into 7- to 8-week-old female nude mice (nu/nu BALB/c) from Harlan. Eight weeks after the injection, the mice were euthanized and a necropsy was done. Tumor formation was estimated by two methods. First, we measured bidimensionally tumors >0.4 cm with calipers and calculated tumor volume according to the formula L × W² / 2, where L is length and W is width. A cumulative tumor volume was calculated by adding the volumes of dominant tumors for each animal. Second, we estimated peritoneal seeding with unmeasurable 1- to 3-mm tumors by counting the number of implants on the mesentery, omentum, and peritoneum. Harvested tumors were preserved in formalin for future histologic and immunohistochemical studies. When possible, tumors were minced and cultured in medium supplemented with G418 in the presence of hyaluronidase at a concentration of 1 mg/mL. The xenograft-derived cultures were characterized by immunoblotting and solid-phase assay. Animal experiments were approved by the Indiana University School of Medicine Animal Care and Use Committee (protocol 2680) and were in accordance with federal regulations. Three independent experiments were done.

Flow cytometry. Quantification of cell surface β₁ integrin was done using the FACScan/CellQuest system (Becton Dickinson). Trypsinized cells were incubated with β₁ integrin monoclonal antibody (1:100 dilution) or mouse IgG for 1 h on ice. After incubation with secondary AlexaFluor⁴⁸⁸-labeled anti-mouse IgG (1:500 dilution), immunofluorescent staining was quantified using the FACScan/CellQuest system. Ten thousand events were accumulated for each analysis. Three independent readings were obtained from separate experiments, and data were averaged for statistical analysis.

Statistical analysis. For the analysis of the immunohistochemical and immunoblotting data in cancer and noncancer specimens, the χ² test was used. Likewise, the χ² was used for the comparison between animals developing peritoneal studding in the groups injected with AS-TG2–transfected or control cells. For the solid-phase adhesion and migration assays, flow cytometry analysis, and comparison of volumes and number of peritoneal implants between the two animal groups, we used the Student's t test.

TG2 is expressed in EOC. We used immunohistochemical assay to determine expression of TG2 in ovarian tumors. Among 27 tumors, we found intense cytoplasmic and membrane staining (2 to 3+) in 21 specimens (79% of tumors; Fig. 1A). Two specimens stained weakly (1+) and four tumors did not stain. All histologic subtypes were immunoreactive: 9 of 9 clear cell carcinoma were intensely positive, 3 of 3 endometrioid tumors displayed 2+ staining, and 9 of 14 papillary serous tumors were 2 to 3+ positive. One poorly differentiated carcinoma did not immunoreact. Details of tumor characteristics, including stage, histologic type, and grade, are included in Supplementary Table S1. TG2 expression was noted in advanced tumors (10 of 13 stage III and IV tumors) as well as in early-stage disease (13 of 14 stage I and II tumors). In normal ovary, TG2 immunoreactivity was weak in normal stroma and absent in the surface epithelial layer. None of six normal ovarian specimens immunostained for TG2 in the epithelium, suggesting that TG2 expression is specific to transformed ovarian epithelial cells (P = 0.005), but weak staining was observed in surface epithelial inclusions (Fig. 1A). Control staining (without primary antibody) was consistently negative.

TG2 is secreted in EOC malignant ascites. Because ovarian cancer disseminates in the peritoneal cavity, large volumes of ascites that contain proteins secreted by tumor and mesothelial cells are generated (22). Such secreted proteins modulate the growth and spread of carcinoma cells in the peritoneal milieu (9, 23, 24). Knowing that TG2 is a secreted protein, we tested whether it is detectable in ascites. Immunoblot analysis revealed the presence of TG2 in 25 of 30 ascites specimens from patients with EOC (representative immunoblots illustrated in Fig. 1B). There was variable level of TG2 secretion among the malignant ascites specimens. Eight specimens of nonmalignant, inflammatory effusions contained none or negligible TG2 (one of eight samples), suggesting that TG2 secretion is specific to cancer cells (P = 0.005). Immunoblot analysis revealed a higher molecular weight band, migrating at ∼170 kDa in several ascites specimens. This was also observed in conditioned medium from some of the ovarian cancer cell lines (data not shown) and was partially disrupted by stringent denaturing conditions (SDS or 2-mercapthoethanol), suggesting that it represents a covalently linked dimer. These data suggest that TG2 is up-regulated in EOC cells and secreted in malignant ascites.

TG2 facilitates ovarian cancer cell adhesion to fibronectin and directional cell migration. To understand the function of TG2 in ovarian cancer cells, we generated stable human cell lines, in which TG2 was overexpressed or knocked down. To knock down TG2, we used an antisense construct (AS-TG2) cloned in pcDNA3.1 (25) in SKOV3 ovarian cancer cells, which expresses abundant TG2. Two stable clones (G and M) were selected based on G418 resistance and screening by immunoblotting. Decreased TG2 was noted in whole-cell lysates and conditioned medium from these clones compared with vector-transfected cells (Fig. 2A).

Next, we wished to determine whether TG2 mediated ovarian cancer cell dissemination in the peritoneal space and examined its involvement in cell adhesion and directional migration. Using a solid-phase assay, we found that stable knockdown of TG2 in SKOV3 cells decreased ovarian cancer cell adhesion to fibronectin by >50% at all concentrations of fibronectin tested compared with control cells (Fig. 2B). Cell spread on fibronectin was visualized by staining the cytoskeleton with rhodamine-phalloidin. AS-TG2–transfected cells were round and failed to extend lamellipodia, whereas cells transfected with vector spread readily on fibronectin (Fig. 2C).

The effects of TG2 on directional cell migration were measured by a transwell assay. Fibronectin- and collagen-stimulated chemotaxis was decreased in SKOV3 cells stably transfected with AS-TG2 compared with cells transfected with vector (Fig. 2D). In addition, secreted TG2 acted as an attractant, stimulating ovarian cancer cell migration across the transwell. Directional cell migration of SKOV3 cells induced by conditioned medium was significantly lower when the assay was done with conditioned medium lacking TG2 (from AS-TG2 cells) compared with conditioned medium from control cells (Fig. 2D).

Next, we tested whether stable overexpression of TG2 increases adhesion to fibronectin. For this, an ovarian cancer cell line with low endogenous TG2 (OV90) was transfected with TG2. A stably transfected clone was identified by immunoblotting after selection with G418 (Fig. 3A). Stable expression of TG2 enhanced adhesion to fibronectin compared with cells transfected with empty vector (Fig. 3B). Likewise, directional cell migration stimulated by fibronectin or collagen was enhanced by stable expression of TG2 (Fig. 3C). Coupled with the effects of TG2 knockdown, these experiments show that TG2 is critical to ovarian cancer cell adhesion and directional migration, essential steps in metastasis.

TG2 knockdown inhibits tumor development and spread on the peritoneal surface in vivo. We injected nude mice i.p. with SKOV3 cells stably transfected with AS-TG2 or empty vector, and after 8 weeks, we observed a significant difference in pattern of tumor development. Mice injected with pcDNA3.1-transfected (empty vector) cells developed tumors in the omentum and the retroperitoneal space and numerous 1- to 3-mm tumors studding the mesentery adjacent to the bowel and on the peritoneal surface of abdominal flanks (Fig. 4A and B). The average number of implants was 76 ± 8. Mice injected with AS-TG2–transfected cells developed one large tumor in the omentum, invading into the retroperitoneal space, and a tumor nodule at the injection site, but significantly fewer mesenteric implants (17 ± 6; Table 1). The tumor volume of dominant masses was not significantly different for AS-TG2–derived xenografts compared with controls (Table 1). Twelve of 14 mice injected with control cells developed milliary peritoneal metastases, whereas only 2 of 17 mice injected with AS-TG2 cells developed milliary studding of the mesentery (P < 0.0001; Supplementary Table S2). Tumors were of similar histologic appearance, with high nuclear grade for both groups, and a serous papillary pattern was discernible in mesenteric implants.

TG2 knockdown was retained in vivo as shown by immunohistochemistry in xenografts and by Western blot analysis of cell cultures established from explanted xenografts (Fig. 4C and D). Occasional small and isolated islands of TG2-positive cells were observed in tumors from AS-TG2 cells, consistent with the emergence of some TG2-positive subpopulations in the absence of G418 selection in vivo. However, general preservation of the TG2 knockdown in vivo is shown by the observation that cell cultures from explanted xenografts conserved their original phenotype (Fig. 4D). In addition, decreased adhesion to fibronectin was noted between cells cultured from AS-TG2 and control tumors. The pattern of tumor formation in the peritoneal space was consistent with the phenotype observed in vitro, suggesting an important role for TG2 in peritoneal studding.

TG2 interacts with β₁ integrin and modulates its expression. As alteration in the level of TG2 expression modulates EOC cell adhesion in vitro and in vivo, we examined whether TG2 interacts with integrins. Knowing that β₁ integrin has been linked to invasion and metastasis in EOC (26), we evaluated interaction of TG2 with this subunit. Immunoprecipitation with β₁ integrin antibody followed by immunoblotting for TG2 shows endogenous interaction between TG2 and β₁ integrin in EOC cells. TG2 and β₁ integrin also colocalized in cytoplasmic organelles (Fig. 5A). We also examined the level of β₁ integrin in cells with diminished TG2 expression and found that the β₁ subunit is expressed at decreased levels in AS-TG2–transfected cells (Fig. 5B). However, mRNA levels were not different between AS-TG2 and control cells, suggesting that TG2 affects integrin processing posttranscriptionally. Decreased β₁ subunit in the plasma membrane fraction was confirmed by Western blotting in SKOV3 transfected with AS-TG2 compared with controls. Expression of β₁ integrin on the cell surface was estimated by fluorescence-activated cell sorting (FACS) and immunofluorescent staining. FACS analysis revealed reduced levels of β₁ integrin on the surface of AS-TG2 cells compared with controls (Fig. 5C). In cells expressing AS-TG2, immunofluorescence also showed reduced distribution of β₁ integrin to the cell membrane (Fig. 5D).

To measure whether changes in rates of protein degradation contribute to the difference in β₁ integrin expression dependent on the TG2 status of ovarian cancer cells, we treated AS-TG2 and control cells with cycloheximide and measured β₁ integrin levels by immunoblotting. In AS-TG2 cells, β₁ integrin was degraded more rapidly [half-life (t_1/2) = 3–6 h] compared with control cells (t_1/2 = 12–24 h; Supplementary Fig. S1A). To assess lysosomal-, ubiquitin-, or calpain-mediated β₁ integrin proteolysis in AS-TG2 or pcDNA3.1-transfected (vector) ovarian cancer cells, we measured integrin levels by immunoblotting after incubating cells with lysosomal inhibitor (leupeptin), proteasomal inhibitors (MG132 and lactacystin), calpain inhibitor (PD105606), and protein trafficking inhibitors (monensin and chloroquine). Treatment with the proteasomal inhibitor MG132, and more prominently with the calpain inhibitor PD105606, restored β₁ integrin levels in AS-TG2 cells (Supplementary Fig. S1B–D), suggesting that TG2 interferes with calpain-mediated β₁ integrin degradation in EOC cells. In contrast, inhibition of lysosomal proteases (leupeptin) did not significantly alter integrin levels in AS-TG2 cells (Supplementary Fig. S1E). These observations may explain the selective down-regulation of β₁ integrin levels in cells transfected with AS-TG2 and identify a novel function of TG2 in epithelial cancer cells.

I.p. metastasis characteristic of EOC requires modifications of tumor cells to facilitate interaction with the peritoneal stroma and mesothelium. In this report, we show overexpression of TG2 in ovarian tumors and its secretion in malignant ascites fluid. We show that TG2 mediates ovarian cancer cell adhesion to fibronectin and stimulates directional cell motility, these processes being mediated by TG2 via interaction and stabilization of β₁ integrin. Using an i.p. xenograft model, we show that TG2 knockdown decreases the pattern of diffuse tumor spread, implicating it as a mediator of i.p. metastasis.

Having identified TG2 as an overexpressed transcript in primary EOC cells (15), we show here that >75% of ovarian tumors overexpress the protein. TG2 is not expressed on the surface ovarian epithelium but is present in stage I and II ovarian tumors. TG2 up-regulation has been reported in glioblastoma, pancreatic, breast, and lung cancer, and a multitude of functions has been invoked for it (21, 27, 28). We focus here on the induced stabilization of cell adhesion of TG2, as this is critical to the establishment of i.p. metastases, where cancer cells are required to “stick” to the oncomatrix to establish peritoneal implants. Adhesion to fibronectin and chemotaxis was decreased in EOC cells by knockdown of TG2 and enhanced by stable overexpression of the protein. This phenotype was preserved in vivo, where the pattern of distribution of i.p. implants was altered by TG2 knockdown. Control animals injected with SKOV3-pcDNA3.1 cells developed i.p. implants widely disseminated on the mesentery and the peritoneal/hepatic gutters, resembling the human disease. Animals injected with SKOV3/AS-TG2 cells developed large tumors in the retroperitoneal space and few or no mesenteric metastatic foci, suggesting that TG2 plays an important role in mediating development of peritoneal metastases. The volume of dominant masses was not significantly different in animals injected with AS-TG2 cells compared with controls, whereas peritoneal seeding was considerably decreased (Fig. 4; Table 1; Supplementary Table S2). An association between TG2 and an invasive clinical phenotype was recently described for pancreatic cancer, activation of focal adhesion kinase being considered the main involved mechanism (17). Other functions of the enzyme, and especially its protective role against apoptosis, shown in other systems, might also play a role (27–30).

The altered metastatic phenotype observed by us is due to deficient interaction between tumor cells and the ECM in the absence of TG2. It has been previously shown that TG2 in complex with β₁ integrin enhances cell adhesion to fibronectin (31) and that TG2 interacts directly with the 42-kDa domain of fibronectin, consisting of modules I₆ II_1,2 I_7-9 (31, 32). We have observed interaction of TG2 and β₁ integrin in cells, but interestingly, TG2 and β₁ integrin do not interact in vitro (33). Our unpublished observations also show that addition of exogenous TG2 (recombinant TG2 or protein purified from guinea pig liver) fails to increase ovarian cancer cell adhesion to fibronectin, suggesting that the mere presence of TG2 in the extracellular milieu is not sufficient to facilitate cell adhesion. Interestingly, we found diminished expression of β₁ integrin subunit on the surface of cells where TG2 was stably down-regulated. Further, degradation of β₁ integrin was more rapid in AS-TG2 cells compared with controls and inhibition of calpain-mediated integrin degradation, but not of lysosomal proteases, restored its expression level. These observations suggest that, in the absence of TG2, β₁ integrin is cleared rapidly via calpain-induced proteolysis. These data highlight a novel function for transglutaminase, distinct from its previously suggested role as a “coreceptor” for fibronectin (33). As β₁ integrin can complex several α subunits, modulating cell-matrix interactions, TG2 down-regulation may affect integrin complex formation, leading to deficient cell adhesion to different components of the ECM. By directly modulating integrin expression, TG2 could affect the invasiveness of ovarian tumors and clinical outcome (26, 34, 35) as well as sensitivity to chemotherapy (36–38). Integrin signaling has been linked to cancer metastasis (39–41), and disruption of integrin function by competing antibodies or peptides has been proposed as a possible cancer treatment (42–44). The data presented here provide compelling evidence that TG2 may be a target, as its interaction with and regulation of integrins directly promote cell adhesion, which is an important step in peritoneal metastasis.

We also found that TG2 is secreted abundantly in malignant ascites fluid. This is likely due to accumulation from tumor and/or mesothelial cells, as primary ovarian cancer cells and cancer cell lines secrete TG2. Although TG2 lacks a leader sequence, it is secreted into the extracellular space through an as yet unknown mechanism (45). The expression of TG2 in tumor cells supports its detection in ascites fluid. We cannot exclude the contribution of mesothelial cells to the composition of malignant ascites; however, the absence of TG2 in inflammatory ascites of pleural fluid makes it likely that TG2 is secreted directly by the tumor cells. Other factors critical to i.p. metastasis are secreted abundantly in ascites, fibronectin, lysophosphatidic acid, hyaluran, and vascular endothelial growth factor, making the i.p. milieu favorable for cancer cell growth (11, 46). The presence of TG2 in the peritoneal fluid may play a role in remodeling the oncomatrix by cross-linking of ECM proteins.

The important conclusion of this study is that TG2 is highly expressed in human ovarian tumor cells and is secreted in malignant ascites. At the interface between cancer cells and the stroma, TG2 alters the pattern of tumor growth and dissemination in the peritoneal space by modulating β₁ integrin expression and function.

Grant support: American Cancer Society grants IRG-84-002-19 and MRSG 107613, Indiana University Cancer Center Ovarian Cancer Pilot Grant, Showalter Research Fund, and Marsha Rivkin (D. Matei).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. David Donner (University of California at San Francisco, San Francisco, CA) for helpful advice and critical review of the manuscript, Dr. Janusz Tucholski (University of Alabama) for plasmids, and Dr. Jyianou Rao (University of California at Los Angeles, Los Angeles, CA) for ascites specimens.