Podoplanin in cancer cells is experimentally able to attenuate prolymphangiogenic and lymphogenous metastatic potentials of lung squamoid cancer cells

First, we examined the expression level of podoplanin in a variety of lung SCC cell lines. RT-PCR revealed that only EBC-1 cells showed no podoplanin expression (Figure 1A). This finding was confirmed by real-time PCR (data not shown). In addition, 21 days after cutaneous inoculation of EBC-1 cells in the dorsal area of BALB/c nu/nu mice, approximate 65% mice showed axillary lymph node metastasis, suggesting that EBC-1 cells possessed high lymphogenous metastatic potential and were available for validating the effect of podoplanin in lymphogenous metastasis. Then, EBC-1 cell-derived stable transformants with or without cDNA of human podoplanin were established. Eventually, 15 single clones were established, and the two of these (EBC1-Ps: EBC1-P4 and EBC1-P15) that had the highest expression of podoplanin mRNA were adopted (data not shown). On the contrary, 3 single clones with empty vectors were simultaneously established and two clones (EBC-Vs: EBC1-V1 and EBC1-V2) were randomly adopted as controls. Protein expression of exogenous podoplanin was confirmed by Western blot analysis (Figure 1B). These clones were used in a series of experiments in this study. Using these stable transformants, we examined the effects of podoplanin on EBC-1 migration and proliferation activities. As a result, podoplanin had no influence on the proliferation and migration of EBC-1 cells in vitro as evidenced by Figures 1C and 1D. Regarding the negative data obtained from the migration assay, we performed additional experiments to indicate a positive control, which revealed that stable overexpression of exogenous podoplanin could promote the migration activity in SAS cells, an oral SCC cell line (additional file 1). This finding strongly suggested that podoplanin had no influence on the migration activity of LSCCs in vitro. Moreover, no apparent changes in cell morphology, such as filopodia-like formation, could be observed in EBC1-Ps (Figure 1E). The morphological appearances were supported by the results of Western blot analysis, which demonstrated that the phosphorylated levels of ERM molecules (ezrin/radixin/moesin), which are implicated in cellular actin cytoskeleton rearrangement, were not affected by podoplanin overexpression in EBC-1 cells (Figure 1F). These findings except for the proliferation data are not consistent with those in previous reports [27], in which podoplanin could lead to enhanced migration activity and to inducement of filopodia-like formation based on remodeling of actin cytoskeleton.

Next, we examined the effect of podoplanin on the tumor growth and lymphogenous metastatic potential using a tumor implantation model as described in Methods. As a result, the growth activity of EBC1-P-derived tumors showed no significant difference compared to that of EBC1-V-derived tumors (Figure 2A). By contrast, EBC1-P cell-implanted mice exhibited a significantly and markedly reduced incidence of axillary lymph node metastasis compared to EBC1-V1 cell-implanted mice (Figure 2B). A representative lymph node specimen is shown in Figure 2C. These results suggested that podoplanin could help to reduce the potential of lymphogenous metastasis of LSCCs.

We hypothesized that the reduced potential of lymphogenous metastasis in EBC1-Ps was due to podoplanin-mediated inhibition of tumor-associated lymphangiogenesis. To validate our hypothesis, we performed an immunohistochemical study for blood and lymphatic vessels, using sections of implanted tumor tissues. First, we confirmed exogenous podoplanin expression in tumor tissue. As shown in Figure 3A, human podoplanin expression was immunohistochemically detected at the tumor cell surface in EBC1-P4-derived tumor tissue. Next, we evaluated tumor-associated lymphangiogenesis as described in Methods. Representative results of immunohistochemical staining for LYVE-1 are shown in Figure 3B. As evidenced by Figure 3B, many various-sized tumor cell nests circumscribed with the LYVE-1-positive brown signal were observed in EBC1-V1-derived tumor tissue (Figure 3B, EBC1-V1/LYVE-1, arrows). Since we had never seen such a LYVE-1-positive staining pattern, we used the alternative lymphatic endothelial marker podoplanin to perform an additional immunohistochemical examination to make sure the LYVE-1-positive brown signal was truly derived from lymphatic endothelial cells. As a result, we confirmed that mouse podoplanin had a staining pattern similar to that of LYVE-1, indicating that the tumor cell nests were circumscribed with lymphatic endothelial cells (additional file 2). In addition, to determine whether or not the tumor cell nests circumscribed with LYVE-1-positive lymphatic endothelial cells are in dilated lymphatic vessels, we performed Masson staining to visualize tumor-associated fibrous stroma. As shown in Figure 3C, no stromal component was revealed in tumor cell nests circumscribed with LIVE-1-positive endothelial cells with Masson staining, whereas the surrounding tumor tissue apparently showed Masson-staining-positive fibrous stroma, indicating that tumor cell nests were present in the dilated lymphatic vessels. This immunohistochemical evidence suggested that EBC-1 cells strongly possessed the ability to induce lymphangiogenesis and permeate lymphatic vessels, frequently with the result of lymphogenous metastasis. In accord with the lymphangiogenic character of EBC1 cell-derived tumor, we evaluated lymphangiogenic states, using several indexes including the vessel area and perimeter. The area and perimeter of LYVE-1-positive lymphatic vessels in viable tumor tissue were significantly lower in EBC1-P-derived tumors than in EBC1-V-derived tumors (Figure 3D). Surprisingly, the number of lymphatic vessels showed no significant difference (Figure 3D). We further evaluated blood angiogenic states. Blood vessels were immunohistochemically identified as CD31-positive vessels (Figure 4A), and were evaluated with the same indexes as in the case of lymphangiogenesis. The results showed that blood vessels were not significantly different for any indexes (Figure 4B). These histological findings suggest that podoplanin in LSCCs contributed to the inhibition of tumor-associated lymphangiogenesis.

To clarify the mechanisms by which tumor-associated lymphangiogenesis was suppressed in EBC1-P-derived tumor tissues, we examined the expression levels of proven pro-lymphangiogenic growth factors [7‐9, 13, 15, 16]. In our previous study, we reported that VEGF-A and VEGF-C mRNAs are apparently detectable by RT-PCR in wild-type EBC-1 cells. By contrast, the mRNA expressions of PDGF-B, VEGF-D, Angiopoietin (Ang)-2 and HGF were weak or undetectable [38]. Therefore, we quantitatively examined the expression levels of VEGF-A and VEGF-C in each clone in vitro. Real-time RT-PCR revealed that the expression levels of VEGF-C mRNA but not of VEGF-A were significantly reduced in EBC1-Ps compared to those in EBC1-Vs under culture conditions (Figure 5A). In addition, PDGF-B mRNA expression was weak but quantitatively able to be detected by real-time RT-PCR and showed no significant change among the clones (Figure 5A). Consistent with the gene expression pattern, the ELISA assay revealed that VEGF-C but not VEGF-A content in culture media was significantly reduced in EBC1-Ps, and that PDGF-BB content was undetectable (additional file 3). The expression levels of VEGF-D, Ang-2 and HGF were extremely weak or undetectable in each clone as well as in wild-type EBC-1 cells (Figure 5B). Next, we examined the EBC1 cell-derived mRNA levels of human VEGF-A, VEGF-C and PDGF-B in implanted tumor tissues as described in Methods. Real-time RT-PCR (Figure 5C) and RT-PCR (Figure 5D) using human specific primer sets revealed similar results compared to those from the in vitro experiments. We further examined the gene expression level of EBC1-derived human VEGF-C compared to that of mouse VEGF-C in implanted tumor tissue. As shown in Figure 5E, the human mRNA level was dramatically higher than the mouse mRNA level, suggesting that the level of EBC1-derived VEGF-C is predominant compared to that of host VEGF-C in the implanted tumor tissues. These findings may be insufficient to determine the significance of VEGF-C for inducing tumor lymphangiogenesis in our animal model; however, VEGF-C expression in EBC-1 cells is thought to be a key factor to induce tumor-associated lymphangiogenesis, and podoplanin-mediated downregulation of the VEGF-C gene might be a possible mechanism underlying the different levels of lymphangiogenic activity between EBC1-Vs and EBC1-Ps.

To identify the critical intracellular signal transductions by which podoplanin inhibited VEGF-C expression in EBC-1 cells, we examined the effects of intracellular signal inhibitors related to podoplanin-mediated downstream signals on VEGF-C expression. Recent studies have demonstrated that podoplanin is involved in intracellular signals of the Rho family composed of RhoA, Rac-1, and cdc42 [27, 29, 39]. Therefore, we focused on the relationships among podoplanin, Rho family-mediated downstream signaling molecules, ROCK/Rho kinase and JNK, and VEGF-C. Real-time RT-PCR revealed that the VEGF-C gene expression level in EBC1-Ps treated with sp600125, a JNK inhibitor, was significantly improved, nearly to the level found in EBC1-Vs (Figure 6A), whereas no change in the VEGF-C expression was observed in EBC1-Ps treated with the ROCK inhibitor Y-27632 (data not shown). Western blot analysis revealed that EBC1-Ps had higher JNK phosphorylation levels than EBC1-Vs, and sp600125-treated EBC1-Ps showed similar levels of phosphorylated JNK as in EBC1-Vs (Figure 6B). To further validate whether the increased activation of JNK-mediated downregulation of VEGF-C is dependent on podoplanin, the levels of phosphorylated JNK and VEGF-C mRNA were examined using siRNA methods. As a result, significant upregulation of VEGF-C mRNA (Figure 6C) and a decreased level of phosphorylated JNK (Figure 6D) were induced in EBC1-P4 cells treated with siRNA-podoplanin. Taken together, these findings suggested that the podoplanin-mediated increase in the level of activated JNK was a downregulation signal for the VEGF-C gene in EBC-1 cells. To enhance the credibility and universality of the podoplanin-JNK-VEGF-C axis in LSCCs, we further performed in vitro examinations, using H157, a lung SCC cell line with podoplanin expression (Figure 1A). Consistent with the results shown in Figure 6, VEGF-C gene expression and protein secretion were significantly upregulated in H157 cells treated with the JNK inhibitor sp600125 (Figure 7A). In addition, significant upregulation of VEGF-C mRNA (Figure 7B) and a decreased level of phosphorylated JNK (Figure 7C) were also induced in H157 cells treated with siRNA-podoplanin. On the contrary, we previously reported the intracellular signaling pathways, p42/44 MAPK and p38 MAPK, by which the VEGF-C gene is stimulated in the oral SCC cell line [38]. Although a 20 μM concentration of sp600125 reportedly had no influence on p42/44 MAPK and p38 MAPK activities in cultured cells [40], we confirmed the effect of sp600125 on the phosphorylation levels of p42/44 MAPK and p38 MAPK in EBC-1 and H157 cells. Consistent with the previous report [40], the phosphorylation levels of these molecules were hardly affected by sp600125 treatment at a 20 μM concentration in these cells (data not shown). Taken together, these findings suggest that JNK is a key pathway for inducing podoplanin-mediated downregulation of the VEGF-C gene in LSCCs.

The lung squamoid cancer cell lines EBC-1 and H157 were maintained with RPMI 1640 (Sigma-Aldrich Japan, Tokyo, Japan) supplemented with 100 units/mL penicillin/streptomycin and 10% fetal bovine serum (FBS). Human fibroblast (MRC5) and mouse fibroblast (NIH3T3) were purchased from the American Type Culture Collection and maintained with Dulbecco's Modified Eagle's Medium supplemented with 10% FBS. Human microvascular endothelial cells (HMVECs) were purchased from Kurabo Co. Ltd., Tokyo, Japan, and maintained with EGM2 medium (Kurabo). c-Jun N-terminal kinase (JNK) inhibitor sp600125 and ROCK/Rho kinase inhibitor Y-27632 were purchased from Promega K.K., Tokyo, Japan. Anti-human podoplanin mouse monoclonal antibody was purchased from AngioBio Co., Del Mar, CA. Anti-total JNK, anti-phospho-JNK, anti-phospho-ERM and anti-total ERM were purchased from Cell Signaling Technology, Beverly, MA. Anti-β-actin rabbit polyclonal antibody was purchased from Sigma-Aldrich Japan, Tokyo, Japan. Anti-murine lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1) rabbit polyclonal antibody was produced in our laboratory as described previously [38], and anti-mouse CD31 rabbit polyclonal antibody was purchased from Abcam Inc., Cambridge, MA.

Male BALB/c nu/nu mice (5 weeks old) were from Kyudo Co., Ltd. (Tosu, Saga, Japan). All animal experiments were done under approved protocols and in accordance with recommendations for the proper care and use of laboratory animals by the Committee for Animal, Recombinant DNA, and Infectious Pathogen Experiments at Kyushu University and according to the Law (No. 105) and Notification (No. 6) of the Japanese Government.

Total cellular RNA was extracted from culture cells or implanted tumor tissues with the ISOGEN system (Wako Pure Chemical Industries, Osaka, Japan) according to the manufacturer's instructions, and treated with RNase-free DNase I (Behringer Roche Applied Science Japan, Tokyo, Japan). Subsequently, aliquots (25 ng) of total RNA were reverse-transcribed and used for PCR templates. Real-time RT-PCR was performed for quantification of gene expression levels. Amplification of target genes was monitored in real-time, and gene expression levels were quantified using the Sequence Detection System, model 7000 (Applied Biosystems Ltd., Tokyo, Japan), according to the manufacturer's instructions for TaqMan methods. The oligonucleotide sequences of PCR primer sets and TaqMan probes were listed in additional file 4. The target gene expression level was normalized with GAPDH level and expressed as relative fold increases compared to mean level in control group.

The PCR primers incorporating Hind-III and Xho-I sites for amplification of human podoplanin are as follows: 5'-GATGTGGAAGGTGTCAGCTC-3'(forward) and 5'-GATCCTCGATGCGAATGCCT-3' (reverse). A PCR amplicon using cDNA from HMVECs was inserted into a plasmid vector, pCEP4, for mammalian cell expression according to general subcloning methods. The primer sequences for construction of plasmid templates are as follows: human and mouse vegf-c, 5'-GAAATTACAGTGCCTCTCTC-3' (forward) and 5'-CTAGTTCTTTGTGGGTCCAC-3' (reverse). Each PCR amplicon using cDNA from MRC5 or NIH3T3 was inserted into a plasmid with a pCRII TA cloning kit (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions. The sequence of the insert was examined using the CEQ 2000 Sequence Detection System (Beckman Coulter, Inc., Fullerton, CA), and complete matching was confirmed through comparison to those reported in GenBank (accession nos. AF 390106, NM 005429, and NM 009506 for human podoplanin, human vegf-c, and mouse vegf-c, respectively) was confirmed.

Under sufficient anesthesia by an i.p. injection of sodium pentobarbital, 1 × 10⁶ of EBC1-V1, EBC1-V2, EBC1-P4, and EBC1-P15 cells were subcutaneously injected into the dorsal region. After inoculation, tumor length and width were measured every 3 days for 3 weeks, and tumor volume was estimated by the formula V = π/6 × a ² × b, where a was the short axis and b the long [43]. 21 days after implantation, mice were sacrificed and the primary tumors and axillary lymph nodes (two lymph nodes per mouse) were harvested. Each harvested primary tumor tissue was sliced in two, and the slices were used as samples for histological and molecular biological experiments. Harvested axillary lymph nodes were subjected to HE staining. Through microscopic findings, metastatic status was divided into two cases--positive and negative--irrespective of the area of metastatic foci. The positive lymph nodes were counted, and the incidence of metastasis was expressed as the ratio of positive lymph nodes to the total number of lymph nodes in each group.

Cells were lysed with 200 ml of cell lysis buffer (Promega) containing a cocktail of protease inhibitors (Nacalai Tesque), and the supernatant of the lysed cells was recovered. An aliquot of 20 mg of proteins was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing condition, and were then transferred to a PVDF membrane. An hour after being blocked with PBS containing 5% non-fat milk, the membrane was incubated overnight, with each primary antibody diluted with PBS containing 5% BSA and 0.1% Tween 20, at 4°C. The dilution rate was determined according to the manufacturer's instructions. Following several washings with PBS containing 0.1% Tween 20, the membrane was treated for 2 hours with an appropriate HRP-labeled secondary antibody, HistoFine (DAKO, Glostrup, Denmark) at room temperature. The target proteins were visualized by a luminal chemiluminescent reagent, LumiGLO (Cell Signaling Technology). After that, the membrane was subjected to re-probing assay. Briefly, following stripping the binding antibodies using Stripping Solution (Nacalai Tesque, Kyoto Japan) according to the manufacture's instruction, the membrane was washed with PBS and subjected to the WB described above. The band intensity was measured with FMBIO (Hitachi Software Engineering Co. Ltd., Tokyo, Japan).

The WST-8 assay kit (Nacalai Tesque) was used to determine the proliferation rates of EBC1-Vs cells and EBC1-Ps cells, which were seeded at 1 × 10³ cells/well on 96-well plates (n = 12 each). Cell viability was evaluated every day for 3 days according to the distributor's protocol. The obtained data were expressed as relative fold increases of O.D. values compared to those 1 day after dissemination.

Boyden chamber cell invasion was assayed using a cell culture-chamber-insert system (BD, FALCON) with an 8 μm polyethylene terephthalate (PET) membrane. First, 1 × 10⁵ cells were seeded on the upper chamber in RPMI medium with 1% FBS. The RPMI medium with 1% or 10% FBS was added to the lower chamber. After 18 hours, cells that did not cross the membrane were scraped off the upper side of the membrane with a cotton swab. Cells that had transmigrated to the lower side were fixed with 70% ethanol and subjected to Giemsa staining. The membrane was excised from its support and mounted on a glass slide. Migrating cells in four independent microscopic visual fields (×100) were counted, and expressed as mean number per one field.

Angiogenesis and lymphangiogenesis in implanted tumor tissue were immunohistochemically evaluated. Detail immunohistochemcial procedures were described previously [38]. The lymphatic vessels and blood vessels were identified as LYVE-1-positive and CD31-positive vessels, respectively. Image J Software was used to count the total numbers and measure the areas and perimeters of vessels whose apparent luminal areas were framed by LYVE-1-positive or CD31-positive endothelial cells in viable tumors. The blood and lymphatic vessel number was expressed as that per unit of viable tumor area, and the vessel area was expressed as that per a vessel. All sections were independently evaluated by three persons (K.S., M.O., H.S.).

Human podoplanin siRNA was purchased from Invitrogen. The RNA sequence is as below; 5'-UAUAGCGGUCUUCGCUGGUUCCUGG-3'. Control siRNA High GC Duplex (Invitrogen) was used as a negative control. siRNA transfection was performed with lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions.

VEGF-C content in the culture media was measured using Quantikine Immunoassay systems for human VEGF-C (R&D Systems) according to the manufacturer's instructions. Three independent experiments were performed and the obtained data were statistically analyzed. VEGF-C content in the culture media was expressed as the amount of protein secreted from 100,000 cells during 24 hours.