Interaction between Cancer Cells and Stromal Fibroblasts Is Required for Activation of the uPAR-uPA-MMP-2 Cascade in Pancreatic Cancer Metastasis

Cancer metastasis is a complex process, which results from the interaction of cancer cells with host cells and extracellular matrices (1). In this process, cancer cells and stromal cells exchange enzymes and cytokines and then constantly modify local extracellular matrix. This modified extracellular matrix interacts with cell-surface receptors and thereby promotes cell migration and invasion (2, 3).

Invasive tumor cells have a marked ability to degrade extracellular matrix via activation of matrix metalloproteinase (MMP)-2. MMP-2 is secreted as an inactive zymogen and requires distinct activation processes to be converted into an active MMP-2 (4). Two principal mechanisms are involved in MMP-2 activation. One proposed mechanism is through the MT1-MMP pathway, in which tissue inhibitor of metalloproteinase 2 (TIMP-2), a bifunctional molecule (5, 6), is capable of interacting with pro-MMP-2 via its NH₂-terminal domain and docking MT1-MMP via its COOH-terminal domain, thereby formatting a ternary complex. This complex is a prerequisite for effective activation of pro-MMP-2 by an adjacent TIMP-free active MT1-MMP (7, 8). Another mechanism of MMP-2 activation is through plasminogen activator/plasmin system, in which pro-urokinase plasminogen activator (pro-uPA) binds to its receptor, uPA receptor (uPAR), through a specific NH₂-terminal sequence of its noncatalytic chain (9–11). This binding results in uPA activation, accelerates the conversion of plasminogen to plasmin on the cell surface, and localizes these enzymes to focal contact sites (12–14). Although plasmin has been shown to principally activate MMP-1, MMP-3, and, to a certain extent, MMP-9 (15), increasing evidence proves that uPA and plasmin can activate pro-MMP-2 and thereby promoting tumor invasion and metastasis (16–19).

There is growing evidence indicating that interaction between tumor cells and surrounding stromal fibroblast (SF) plays a critical role in growth, invasion, metastasis, and angiogenesis of tumor (20–22). It has been shown that the invasive potential of pancreatic cancer cells can be greatly enhanced by coculturing with SF (23). Although an intimate relationship clearly exists between the growing tumor and surrounding stromal environment, the molecular mechanisms of tumor-stromal interaction in promoting cancer progression have not been well characterized. Several molecules have been identified participating in tumor-stromal interactions, including hepatocyte growth factor (21, 24), transforming growth factor β (24), and several MMPs (25, 26). In pancreatic cancer, uPA activation and uPAR overexpression are involved in the systemic dissemination of pancreatic cancer (27, 28). Recently, our preliminary study found that fibroblastic uPAR expression was significantly increased in metastatic pancreatic cancers, implying that the SF might participate in the metastatic process of pancreatic cancer via the uPA pathway. To evaluate the role of SF in promoting pancreatic cancer metastasis, we analyzed the expression patterns of uPA, MMP-2, and uPAR in human metastatic pancreatic cancer. Via an in vitro coculture system, by coculturing two pancreatic cancer cell lines, highly metastatic BxPc3 cells and low metastatic PaCa2 cells, with peritumor fibroblasts, we observed the effects of the cocultures on the expression of uPAR and evaluated the roles of cocultures in activation of the uPA-plasmin-MMP-2 cascade.

Tissue samples. A total of 20 pancreatic adenocarcinoma patients from Hepatobiliary Surgery Institute, Southwest Hospital, Third Military Medical University, were randomized and case-pair controlled selection (including age, tumor size, location, and classification) in this study. The tumor specimens included 6 metastatic pancreatic cancer (liver metastasis) specimens and 10 nonmetastatic pancreatic cancer specimens obtained from surgery, as well as 4 liver metastasis specimens obtained from formalin-fixed, paraffin-embedded tissues (Table 1). Each pancreatic tumor specimen was reviewed by pathologists. The protocol was approved by the Institutional Review Board and the patients gave written consent.

Antibodies and reagents. Antibodies and reagents included anti–MT1-MMP, TIMP-2, and MMP-2 (Oncogene Science); anti–plasminogen activator inhibitor (PAI-1 380), uPA-specific antibody 3471, and uPAR-specific antibody (399R American Diagnostica); gelatin, aprotinin, plasminogen, plasmin, glycine, and H-d-Val-Leu-Lys-pNA (S 2051 Sigma Chemical Co.); amiloride, active uPA, and MMP-2 (Calbiochem); mouse monoclonal antibodies anti-α₆β₁ (GoH3), anti-β₃ (LM609), and anti-β₅ (P1F5; PharMingen); and anti-β₁ (P4C10), anti-β₄ (3E1), anti-α₂ (P1E6), anti-α₃ (P1B5), and anti-α₅ (P1D6; Life Technologies, Inc.).

Immunohistochemistry. The details of the procedure have previously been described (19). Briefly, the deparaffinized sections were trypsinized (0.05% trypsin with 0.05% Triton X-100 in TBS) for 20 min and blocked with 10% goat serum in Superblock, where each section was incubated separately with monoclonal antibodies uPA at 20 μg/mL, uPAR at 10 μg/mL, MMP-9 at 8 μg/mL, and MMP-2 at 10 μg/mL at 4°C for 18 to 24 h. After washing four to five times (15 min each) with Triton-TBS, the slides were processed in the Ventana-automated stainer according to the manufacturer's instructions. The immunoperoxidase-3,3-diaminobenzidine–stained slides were subsequently counterstained with hematoxylin and mounted with a coverslip. Normal pancreases were used as controls.

Cell cultures. The metastatic human pancreatic carcinoma cell line BxPc3 and nonmetastatic human pancreatic carcinoma cell line PaCa2, described by Sawai H et al. (29), were purchased from the American Type Culture Collection. SF was isolated from the pancreatic carcinoma tissues (from surgery in our Institute) and epithelial cell contamination excluded by light microscopy. These cells were maintained in DMEM with 10% FCS.

Flow cytometry. Cultured cells were harvested by trypsinization and washed with PBS containing 1% normal goat serum. Cells were incubated with primary antibodies at 4°C for 1 h, followed by secondary antibodies conjugated to FITC for 30 min. The stained cells were resuspended in 100 μL of PBS and analyzed by Becton Dickinson FACSort.

Acid treatment of cells. The cells were treated with glycine buffer to deplete the membrane-bound uPA. Briefly, the cells were harvested with 0.25% trypsin and 2 mmol/L EDTA, treated with glycine buffer (pH 4.0) at 4°C for 3 min, and neutralized by incubation in Tris buffer at pH 7.0 for 10 min for further use.

Cell lysis and Western blotting. The cells were lysed in NP40 lysis buffer (1.5% NP40, 150 mmol/L NaCl, 0.2% SDS, 1 mmol/L EDTA, 20 mmol/L Tris-HCl, 1 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL leupeptin, 10 μg/mL aprotinin, 1 mmol/L Na₃VO₄, 50 mmol/L NaF). Protein concentrations were determined with a BCA protein assay kit (Pierce). For Western blotting, conditioned medium and cell lysates containing equal amounts of protein were separated by SDS-PAGE and transferred to nylon membranes. Membranes were probed with primary antibodies followed by peroxidase-labeled secondary antibodies and visualized by enhanced chemiluminescence detection system (Amersham) according to the manufacturer's instructions.

Gelatin zymography. MMP-2 and MMP-9 were analyzed by using 10% SDS-gelatin substrate gel. The conditioned medium was collected under serum-free conditions and subjected to a gelatin-SDS-PAGE electrophoresis. The gels were treated with 2.5% Triton X-100 at 37°C for 30 min to remove SDS and then incubated at 37°C for 16 h in substrate buffer (50 mmol/L Tris-HCl and 5 mmol/L CaCl₂ at pH 8.0). The gels were stained with 0.15% Coomassie blue R250 (Bio-Rad) in 50% methanol, 10% glacial acetic acid at room temperature for 20 min, and then destained in the same solution without Coomassie blue. The activities of enzymes were identified as clear gelatin-degrading bands against the blue background.

Colorimetric plasminogen activation assay. The cells were grown in DMEM with 10% plasminogen-depleted FCS in 100-mm dishes for 24 h. The activities of uPA in conditioned media were determined by converting plasminogen to plasmin by using a coupled colorimetric plasminogen activation assay. Briefly, the conditioned media were recovered and concentrated by Centricon concentrators (Amicon). Conditioned media of 10 μL were incubated with 2.8 μg of plasminogen in 65 μL of uPA buffer [100 mmol/L Tris-HCl (pH 8.8), 0.5% Triton X-100] in 96-well plates at room temperature for 4 h; then, the plasmin activity was determined by adding 25-μL substrates of plasmin containing 50 μg of H-d-Val-Leu-Lys-pNA (S-2251; Sigma), with absorbance of 405 nm using a Titertek multiscan plate reader. The conditioned media without plasminogen were used as control. Results were presented as plough units determined by a plasmin standard.

Reverse transcription-PCR analysis. The RNA was extracted from pancreatic cancer cells and SF by using RNA mini-Kit (Qiagen) and then reverse transcribed with Moloney murine leukemia virus reverse transcriptase in the presence of random primers. The reverse-transcribed cDNA was amplified further by 30 cycles of PCR in the presence of 10 pmol of sense and antisense primers. The primers used were as follows: for uPAR, 5′-ACAGGAGCTGCCCTCGCGAC-3′ and 5′-GAGGGGGATTTCAGGTTTAGG-3′; for GAPDH, 5′-ACGGATTTGGTCGTATTGGG-3′ and 5′-TGATTTTGGAGGGATCTCGC-3′. Each set of primers corresponded to sequences located on different exons to allow the detection of genomic DNA contamination. Each PCR cycle included a denaturation step at 94°C for 1 min, an annealing step at 61°C for 1 min, and an extension step at 72°C for 1 min. The PCR products were analyzed by electrophoresis on 2% agarose gel containing ethidium bromide and visualized under UV light. Adobe Photoshop software 7.0 was used to quantitate the densitometry of visualized bands and calculate the relative value units of densitometry.

Statistical analysis. Statistical analysis was done by using SPSS software for Windows version 10.0. Correlation analysis was conducted by Spearman's rho test. P < 0.05 was considered as statistically significant.

uPA expression and MMP-2 activation were correlated with liver metastasis of human pancreatic cancer. To determine the relationship between liver metastasis of pancreatic cancer and the expressions of MMPs and uPA, we applied immunohistochemistry to analyze the expressions of uPA, MT1-MMP, uPAR, and MMPs in 10 metastatic pancreatic cancers and 10 nonmetastatic pancreatic cancers. As shown in Table 1, in comparison with normal pancreatic tissues, there was increased expression of MMP-2 in all human pancreatic cancers, moderate expression of MMP-9 in four pancreatic cancers, and increased expression of MT1-MMP in two. The expression of MMPs was not correlated with pancreatic cancer metastasis (Table 3). In contrast, expression of uPA was found in 17 of 20 (85%) examined tumors, with failed expression in only three tumors (Table 1). The stronger staining for uPA was detected at cancer nests of the metastatic pancreatic cancers compared with nonmetastatic pancreatic cancers (Fig. 1A, left versus right). Of 10 cases of metastatic pancreatic cancer, 7 expressed high levels of uPA (Table 1). The expression of uPA was correlated with liver metastases of pancreatic cancers (P < 0.001, r = 0.714). These results suggest that up-regulation of uPA expression may be involved in pancreatic cancer metastasis.

We next detected the active forms of MMP-2 and MMP-9 in six metastatic and six nonmetastatic pancreatic cancer tissues by using gelatin zymograph. As shown in Fig. 2, the active forms of MMP-2 but not MMP-9 were found in liver metastases as well as in situ metastatic pancreatic cancers (Fig. 2A and B). In contrast, all nonmetastatic pancreatic cancers expressed equal levels of MMP-2 (Fig. 2C). These results suggest that MMP-2 is activated in the metastatic pancreatic cancers.

Fibroblastic uPAR expression was increased in metastatic pancreatic cancer. The expression patterns of fibroblastic uPA, MMP-2, and uPAR in pancreatic cancer tissues were analyzed by immunohistochemistry; the expressions of MMP-2 and uPA were also seen in fibroblasts (Table 2; Fig. 1A and D) but the expressions of MMP-2 and uPA in fibroblasts were unrelated to liver metastasis of pancreatic cancer (P = 0.177 and 0.065, respectively). In contrast, the fibroblastic uPAR expression was significantly increased in the metastatic pancreatic cancers compared with nonmetastatic pancreatic cancers (Fig. 1C, left versus right). Statistical analysis showed that the fibroblastic uPAR expression was significantly associated with the liver metastases of pancreatic cancers (P < 0.001, r = 0.857; Table 3), suggesting that fibroblasts may play a role in activating uPA cascades and promoting pancreatic cancer metastasis.

Activation of MMP-2 required direct interaction between tumor cells and fibroblasts in an in vitro culture system. A coculture was used to determine whether activation of MMP-2 required a direct cell-cell interaction. Specifically, two pancreatic cancer cell lines, the highly metastatic cell line BxPc3 and the low metastatic cell line PaCa2, were cocultured with SF and normal fibroblast (NF) for evaluating the activation of MMP-2 in the conditioned medium by using gelatin substrate zymography. As shown in Fig. 3, pancreatic cancer cells and fibroblasts produced equivalent levels of pro-MMP-2 and pro-MMP-9. However, when pancreatic cancer cells were cocultured with SF, an active form of MMP-2 was detected in BxPc3/SF cocultures but not in PaCa/SF (Fig. 3A, top, lane 4 versus lane 3) or BxPc3/NF (bottom, lane 3 versus lane 2) cocultures, although MMP-2 expression was eventually unchanged. These results indicate that MMP-2 is activated by the coculture of pancreatic cancer BxPc3 cells with SF.

In the coculture, pancreatic cancer cells or SF can produce some soluble cell factors to regulate gene expression of the apposing cells. To determine whether this regulatory mechanism resulted in MMP-2 activation in BxPC3/SF cocultures, BxPc3 cells were indirectly cocultured with SF by using a Transwell filter. We found that MMP-2 was not activated in the indirect coculture of BxPc3/SF (Fig. 3A, bottom, lane 5). These results suggest that MMP-2 activation requires the interaction of cancer cells with SF.

The uPA-plasminogen pathway was responsible for MMP-2 activation in the coculture system. To identify which pathway was responsible for MMP-2 activation in BxPc3/SF coculture, we examined expressions of MT1-MMP, TIMP-2, and uPA in cell lysates by Western blotting and evaluated uPA activities in the conditioned media using a coupled colorimetric plasminogen activation assay. Two pancreatic cancer cells expressed equal levels of MT1-MMP, TIMP-2, and uPA. The expressions of MT1-MMP and TIMP-2 were not changed in BxPc3/SF coculture and PaCa2/SF coculture (Fig. 3B, rows 1 and 2, lane 4 versus lane 5). However, uPA activities determined by plasmin as well as the protein levels of uPA were significantly increased in BxPc3/SF coculture but not in PaCa2/SF coculture (Fig. 3B, bottom, lane 4 versus lane 5 and Fig. 3C). uPA was not activated in the indirect BxPc3/SF coculture or in BxPc3/NF coculture. These results indicate that the uPA-plasminogen pathway, but not MT1-MMP, is activated in BxPc3/SF coculture, which requires a direct interaction of SF with cancer cells.

We next used two approaches to confirm whether MMP-2 activation is uPA-plasminogen dependent. First, we added amiloride (an uPA inhibitor), aprotinin (a specific inhibitor of plasmin), and anti–MT1-MMP antibodies to BxPc3/SF coculture and found that addition of 25 μmol/L amiloride to the BxPc3/SF coculture prevented uPA activation, decreased the conversion of plasminogen to plasmin (Fig. 3C), and suppressed active MMP-2 (Fig. 3D, top, lane 3), but anti–MT1-MMP did not suppress activation of MMP-2 (Fig. 3D, top, lane 2). Furthermore, addition of 40 μg/mL aprotinin to BxPc3/SF coculture also suppressed MMP-2 activation (Fig. 3D, top, lane 4), suggesting that the uPA-plasmin cascade is essential for MMP-2 activation in this coculture. Second, because uPA and MMP-2 were not activated in PaCa2/SF coculture, we exogenously added active uPA to PaCa2/SF coculture to observe the conversion of plasminogen to plasmin and the activation of MMP-2. Our preliminary study showed that 50 nmol/L uPA was the optimal condition for stimulating MMP-2 activation. In this study, we found that exogenously adding active uPA into PaCa2/SF coculture promoted the conversion of plasminogen to plasmin (Fig. 3C) as well as the activation of MMP-2 (Fig. 3D, bottom, lane 3). These results suggest that MMP-2 activation is dependent on the uPA-plasminogen pathway.

Up-regulation of uPAR in the SF of BxPc3/SF coculture was responsible for activation of uPA. Interaction of pro-uPA with uPA receptor on the cell surfaces leads to uPA activation (30). To determine whether uPAR is involved in uPA activation, we observed the protein levels of uPAR in SF, pancreatic cancer cells, and the coculture by Western blotting. As shown in Fig. 4A, SF, PaCa2, and BxPc3 expressed equivalent levels of uPAR (Fig. 4A, lanes 1-3, respectively). However, expression of uPAR was increased in the coculture of BxPc3 cells with SF but not in the coculture of PaCa2 with SF (Fig. 4A, lane 4 versus lane 6), suggesting that the coculture of BxPc3/SF up-regulates uPAR expression.

In the coculture, pancreatic cancer cells or SF can produce some soluble cell factors to regulate the gene expression of the apposing cells. To determine whether this regulatory mechanism resulted in the up-regulation of uPAR in BxPC3/SF coculture, BxPc3 cells were indirectly cocultured with SF by a Transwell filter or directly cocultured with SF to evaluate uPAR expression by reverse transcription-PCR. We found that, in the indirect BxPc3/SF coculture, the up-regulated mRNA expression of uPAR was found in SF but not in BxPc3 cells (Fig. 4B, lane 5 versus lane 6). Similarly, the up-regulated mRNA expression of uPAR was also found in the direct BxPc3/SF coculture (Fig. 4B, lane 4). In contrast, mRNA expression of uPAR was not increased in the direct and indirect BxPc3/NF cocultures (Fig. 4B, lanes 2 and 3). These results have confirmed our previous results of Western blotting (Fig. 4A) and also suggest that the metastatic pancreatic cancer BxPc3 cells may produce some soluble cell factors to up-regulate uPAR expression in SF but not in NF.

In the coculture, the activation of uPA and MMP-2 requires a direct cell-cell contact between SF and cancer cells, whereas uPAR expression is on the contrary. To further explain this phenomenon, we observed the expression of PAI-1 in these cocultures by Western blotting. As shown in Fig. 4C, PAI-1 showed no change in direct or indirect cocultures, suggesting that the mechanism for tumor cells activating uPA may be more complicated.

Integrin α₆β₁ and uPAR coordinately activated uPA and resulted in MMP-2 activation. It has been shown that integrins are involved in MMP-2 activation (31). We then analyzed integrin profiles in two pancreatic cancer cell lines by flow cytometric analysis. As shown in Fig. 5A, the expressions of α₆ and β₁ were significantly increased in BxPc3 cells rather than in PaCa2 cells. The expression levels of α₃, β₄, α_v, α₅, β₃, and β₅ integrins were equivalent in both cell lines. These results indicate that integrin α₆β₁ is up-regulated in the pancreatic cancer BxPc3 cells.

To determine whether integrin α₆β₁ played a role in activation of MMP-2 in this complicated coculture, BxPc3 cells were incubated with the specific α₆β₁-blocking antibody GoH3 and then cocultured with SF in the presence of GoH3. MMP-2 activation and uPA activities were determined in BxPc3/SF coculture via Western blotting and plasminogen activation assay. Our previous study showed that 10 μg/mL GoH3 was the optimal condition for functional inhibition of α₆β₁. In this study, we found that preincubation of GoH3 with BxPc3 cells decreased plasmin production by 50% (Fig. 5B) and subsequently inhibited MMP-2 activation in BxPc3/SF coculture (Fig. 5C), suggesting that α₆β₁ functions in the activation of uPA and MMP-2 in BxPc3/SF coculture.

In this study, we investigated the role of SF in activating the uPA-plasmin-MMP-2 cascade and regulating the invasive behaviors of pancreatic cancer cells. We found that expressions of uPA and fibroblastic uPAR were increased in metastatic pancreatic cancer. The metastatic pancreatic cancer expressed the active forms of MMP-2, but the nonmetastatic pancreatic cancer did not In the in vitro coculture, the coculture of SF with metastatic pancreatic cancer BxPc3 cells resulted in activation of MMP-2 and up-regulated expression of uPAR. In this coculture, the uPA-plasminogen cascade, but not the MT1-MMP-TIMP-2 pathway, was involved in MMP-2 activation that required direct interaction between SF and cancer cells. We also found that, in the coculture, the interaction between integrins of cancer cells and the uPARs of SF may be involved in the activation of the uPAR-uPA-MMP-2 cascade. Our results suggest that SF regulates the activation of the uPA-plasminogen-MMP-2 cascade, thereby promoting pancreatic cancer metastasis.

Overexpressions of uPA and MMPs have been a consistent finding in adenocarcinomas of the colon, lung, breast, prostate, and ovary and melanoma (9, 21–27), but relatively little attention has been paid to the pancreatic cancer. We found that uPA and MMP-2 were expressed in most pancreatic tumors. However, uPA /uPAR expression and MMP-2 activation were significantly increased in metastatic pancreatic cancers compared with nonmetastatic pancreatic cancers. In the metastatic pancreatic cancers, the fibroblastic uPAR expression was significantly increased (Fig. 1C, left; Table 3) and associated with the liver metastases of pancreatic cancers (P = 0.000, r = 0.857; Table 3), suggesting that fibroblasts may play a role in promoting pancreatic cancer metastasis.

In the in vitro coculture, the activation of MMP-2 in the highly metastatic pancreatic cancer BxPc3 cells requires the participation of peritumor fibroblasts and activation of uPA-plasmin cascade. In fact, MMP-2 was not activated in either metastatic pancreatic cancer BxPc3 cells or nonmetastatic pancreatic cancer PaCa2 cells. When these pancreatic cancer cells were cocultured with SF, the activation of MMP-2 was only detected in BxPc3/SF coculture but not in PaCa/SF coculture, although the coculture did not enhance the levels of MMP-2 expression. In BxPc3/SF coculture, MMP-2 activation was linked to the activation of the uPA-plasmin cascade because chemical inhibitors of either uPA or plasmin prevented MMP-2 activation. In addition, addition of active uPA to PaCa2/SF coculture with no active uPA detected resulted in MMP-2 activation. These findings are in agreement with these studies showing that uPA-plasmin cascade is involved in pro-MMP-2 processing (16–18, 32, 33) and uPA helps release active MMP-2 from the matrix-bound pro-MMP-2 (34). Besides the uPA-plasmin cascade, the MT1-MMP/TIMP-2 pathway is thought to be another important mechanism to activate pro-MMP-2 (35, 36). However, in our cocultures, neither MT1-MMP nor TIMP-2 was increased in BxPc3/SF coculture. Addition of TIMP-2 or anti–MT1-MMP to the coculture failed to activate MMP-2. Therefore, in the BxPc3/SF coculture, MT1-MMP and TIMP-2 may format ternary complexes without leaving TIMP-free active MT1-MMP to activate MMP-2, suggesting that MMP-2 activation via uPA-plasmin cascade may be a mechanism involved in the role of SF in regulating pancreatic cancer metastasis.

Our observations also suggest that a direct interaction between pancreatic cancer cell and peritumor fibroblast is required for activating the uPA/plasmin cascade. In the cocultures, uPA was supplied by pancreatic cancer cells as well as SF, but uPAR and pro-MMP-2 were expressed principally in SF. Yet, uPAR expression was up-regulated in SF when cocultured with pancreatic cancer BxPc3 cells by using direct and indirect cocultures. These results are consistent with the findings by Seghezzi et al. (37). However, uPA activation was only found in the direct coculture but not in the indirect coculture of BxPc3 cells with SF. In this indirect coculture, the pro-uPA was also secreted in the condition media and the PAI-1 expression was not changed. Therefore, in the cocultures, there exists a more complicated mechanism for tumor cells soliciting their surrounding stromal cells to activate uPA.

It has been known that MMP-2 is secreted and stored in extracellular depots as precursor zymogens and activated by specific proteases at cell surface. The clustering of integrin β₁ by fibronectin and binding of MMP-2 to integrin α_vβ₃ facilitate MMP-2 activation (31). Previous reports have indicated that uPAR can physically interact with multiple integrins, including β₁, β₂ (38, 39), and α₅β₁ (40), and to regulate cell migration to fibronectin (41). In the cocultures, we found that BxPc3 cells expressed an increased level of integrin α₆β₁. In the cocultures, uPA activation was significantly increased in α₆β₁-positive BxPc3/SF coculture but not in α₆β₁-negative PaCa2/SF coculture. In addition, addition of the α₆β₁ function-blocking antibodies suppressed uPA activation in BxPc3/SF coculture. Therefore, in the cocultures, the interaction between integrins of cancer cells and the uPARs of the SF may be a mechanism to promote activation of the uPAR-uPA-MMP-2 cascade.

In conclusion, during the process of migration of pancreatic cancers, the SF not only supports cancer cell growth of the primary tumor but also participates in activation of the uPA-plasminogen-MMP-2 cascade, thereby promoting pancreatic cancer invasion and metastasis. Thus, we can envision an in vivo scenario where the α₆β₁-positive pancreatic cancer cells interact with stromal cells, which results in up-regulation of uPAR, and integrin α₆β₁ interacts with uPAR, which results in activation of the uPA-plasminogen-MMP-2 cascade, facilitating the activated uPA-plasminogen-MMP-2 cascade focused on the focal adhesions of integrin α₆β₁ to matrix. The activated uPA and MMP-2 may help the dissemination of these cells from the primary tumor.

We thank David Kramer for helpful discussions and critical reading of the manuscript and Dr. Guo-dong Liu for editing the manuscript.