Background
Lung cancer is a highly prevalent disease and is one of the leading causes of death worldwide. This neoplasia is usually detected in advanced stages and it has a 5-year survival rate of 20% [
1]. Lung adenocarcinoma (AC) and lung squamous cell carcinoma (SCC) are the most common histological subtypes of lung cancer and they are generally smoking-related [
2]. Tobacco contributes to the onset of lung carcinoma by inducing the expression of several cytokines including the molecule TGF-β, which is secreted by stromal fibroblasts [
3]. TGF-β is a ubiquitous and pleiotropic cytokine that plays a dual role in cancer development. While it acts as a tumor suppressor in the early stages of the disease, at later stages of tumor development it contributes to malignant transformation through the activation of cell proliferation, metastasis and tumor angiogenesis [
4,
5]. Indeed, the production of TGF-β by tumor and stromal cells in response to radiotherapy and chemotherapy contributes to treatment resistance [
6], and TGF-β inhibition in these cases improves treatment responses, particularly in models of solid carcinomas such as breast cancer [
7].
The presence of lymph node metastasis is strongly associated with low survival rates in cancer patients [
8,
9], even in those diagnosed at early stages of the disease [
10]. Tumor metastasis largely depends on the interaction between cancer cells and the tumor stroma. While host cells have tumor-suppressing capacities, malignancy induces several changes in the stroma (
e.g., tumor hypoxia) that eventually promote cell proliferation, invasion and metastasis [
11]. Significantly, cytokines such as TGF-β play a key role in the transformation of the stroma during tumor development. Moreover, we have shown previously that TGF-β-induced factors are associated with worse overall prognosis in non-small-cell lung cancer (NSCLC) patients [
12].
The lymphatic vessels constitute the main route by which solid carcinomas access the lymph nodes. Several studies have demonstrated that lymphangiogenesis is positively correlated with lymph node spread and adverse NSCLC prognoses [
9]. Furthermore, both tumor and immune cells have been captured by electron microscopy in transit through channels formed in lymphatic endothelial cell (LEC) monolayers [
13], although the molecular mechanisms by which tumor and immune cells enter lymphatic capillaries remain unknown. Lymphatic metastasis of NSCLCs may be facilitated by the specific morphological characteristics of the lymphatic endothelium. These vessels present an interrupted basal membrane [
14] and their inter-endothelial junctional complexes are distributed in a dispersed button-like disposition [
15]. Therefore, as it has been described for leucocytes, cell transit across these specific capillaries appears to be indolent [
16]. Nevertheless, inflammation induces changes in the phenotype of the initial lymphatic vasculature [
17] that elicit integrin-dependent mechanisms for an efficient recruitment of inflammatory cells [
18,
19].
As cancer is considered an inflammatory disease [
20], it is important to determine whether integrins and their receptors also participate in tumor cell intravasation into the lymphatic vasculature. In fact, several studies have proposed an association between increased integrin expression in tumors and enhanced metastasis to the lymph nodes [
21,
22], and we previously demonstrated that hypoxia and nicotine promote the chemotaxis and adhesion of lung carcinoma cells to lymphatic endothelial cells [
23,
24]. In the present study, we examined the relationship between TGF-β exposure and tumor cell metastasis to the lymph nodes, and we sought to determine whether this relationship is mediated by integrin-dependent mechanisms.
Materials and methods
Cell culture and treatments
The human NSCLC cell lines H157, A549 and H1299, as well as cryopreserved primary Lung-Derived Human Lymphatic Microvascular Endothelial Cells (HMVEC-LLy, Lonza (Walkersville, MD, USA), were grown as described previously [
12]. The cell lines were authenticated by PCR amplification of genomic DNA using specific primers for the specific CDKN2A mutation (c.205 G > T, in exon 2) and a KRAS mutation (c.34 G > C, in exon 2), and they were identified by the subsequent sequencing of the PCR products.
NSCLC cells were cultured in serum-free RPMI with 2 ng/ml human recombinant TGF-β (R&D Systems, Minneapolis, USA) for 24 h or 5 days. The medium was replaced and fresh cytokine was added every 48 h. For TGF-β blocking experiments, tumor cells were incubated with 10 mM of the TGF-βRI chemical inhibitor, SB431542 hydrate (Sigma-Aldrich, Steinheim, Germany), or 200 μg/ml of the TGF-β inhibitory peptide P144 (Polypetide Group, Strasbourg, France), 30 min before TGF-β treatment. Integrin αvβ3 blockade in H157 cells was achieved by adding 10 μg/ml of αvβ3-blocking antibody (MAB1976Z, Millipore, Billerica, MA, USA) 30 min before performing the assay. FAK was inhibited by incubation overnight with 1 μM PF-573228 (Sigma-Aldrich, Steinheim, Germany).
Cell adhesion assays
Analysis of H157 cell adhesion to the lymphatic endothelium was performed as described previously [
24]. Briefly, 3 × 10
4 H157 cells were labeled for 20 min at 37°C with 10 μM calcein-AM (Sigma-Fluka, Steinheim, Germany), seeded on LEC monolayers and allowed to attach for 30 min at 37°C. Non-adherent cells were washed out and cell fluorescence was measured on a BMG Polar star Galaxy plate reader (Lab Technologies, Barcelona, Spain), using an excitation wavelength of 485 nm and a 520 nm emission filter.
Cell transmigration assays
A total of 4 × 10
4 LECs were seeded on 8 μm pore-size filters in modified Boyden chambers (BD Biosciences, San José, CA, USA) as described previously [
19]. Next, 7 × 10
4 H157 cells in 150 μl of serum-free RPMI medium were added and allowed to migrate for 24 h at 37°C towards the complete media added to the lower side of the filters. Transmigration efficiency was calculated as described previously [
19].
The L1CAM and CD31 integrin receptors were blocked by pre-incubation of tumor cells or endothelial cells with blocking antibodies (20 μg/ml) for 1 h before carrying out the transmigration assays. The antibodies against human L1CAM (L1-9.3, directed against the L1CAM homotypic binding region, and L1-35.9, directed against the L1CAM RGD binding region) have been described previously [
25]. The CD31 antibody was purchased from Sigma Aldrich (Steinheim, Germany).
RNA isolation and PCR array
Total RNA was extracted with Trizol (Gibco, Carlsbad, CA, USA) according to the manufacturer’s instructions. For the PCR array, cDNA synthesis was carried out using 1 μg of total RNA and the RT2 First Strand Kit (SABiosciences, Qiagen Dusseldorf, Germany). Gene expression was profiled using the ECM and Adhesion Molecules RT2 Profiler™ PCR Array (SABiosciences, Qiagen Dusseldorf, Germany), according to the manufacturer’s instructions.
Tumor cell transfection
H157 cells (1 × 106 cells/ml) were transfected with 20 μg of a scrambled RNA or a HuSHTM shRNA Plasmid Panels-29mer targeting integrin β3 (Origene, Rockville, MD, USA) in Opti-MEM medium (Invitrogen, Barcelona, Spain) using a Biorad Gene Pulsar I electroporator (Biorad, Berkeley, USA). Stable β3 integrin-silenced clones or cells expressing a non-specific scrambled RNA sequence were selected by culturing cells in the presence of 1.5 μg/ml puromycin-dihydrochloride antibiotic (Sigma-Aldrich, Steinheim, Germany). To generate GFP-expressing cells, H157 cells (2 × 105) were transfected with 1 μg of the pEGFP-C1 plasmid (Clontech, Mountain-View, CA, USA) using FuGENE 6 Transfection Reagent (Roche, Barcelona, Spain), following the manufacturer’s instructions. Transfection efficiency was confirmed by flow cytometry and fluorescent microscopy, respectively.
Western blot
Total cell protein extracts were prepared using RIPA buffer as described previously [
12]. Membranes were blocked for 1 h with 10% non-fat milk or 5% BSA in TBS containing 0.1% Tween-20, and then incubated overnight at 4°C with the primary antibody at the dilutions recommended by the manufacturer.
The primary antibodies against FAK and phospho-FAK (Tyr397) were purchased from Cell Signaling (Danvers, MA, USA), and the anti-β-actin from Sigma-Aldrich (Steinheim, Germany). HRP-conjugated anti-rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was used as the secondary antibody. Blots were developed using Lumi-Light Plus Reagent (Roche, Barcelona, Spain), and the autoradiograms were scanned using a GS-800 calibrated densitometer and analyzed using Quantity One software (Biorad, Berkeley, CA,USA).
Orthotopic mouse model of NSCLC
All protocols involving animal experiments were approved by the Experimentation Ethics Committee of the University of Navarra. Female athymic nude mice (5-6 weeks old) were purchased from Harlan Laboratories and GFP-H157 cells (1 × 10
6) in PBS containing 10 μg of Matrigel (BD Biosciences, San José, CA, USA) were injected in a total volume of 20 μl into the left lung of these nude mice as described previously [
26]. Each mouse was then injected intra-peritoneally with either vehicle (PBS) or 200 μg of the TGF-β inhibitor peptide P144 daily. Mice were sacrificed 28 days after treatment or upon exhibiting symptoms of cachexia. Primary tumors and brachial and axillary lymph nodes from both sides were extracted, fixed in Bouin solution and paraffin-embedded for histopathological analysis.
Immunohistochemistry and confocal microscopy imaging
Endogenous peroxidase activity was quenched in formalin-fixed paraffin-embedded tissue sections (3 μm thick) and they were then exposed to microwaves. Non-specific binding was blocked by incubation for 30 min in 5% goat serum in TBS, before the sections were incubated overnight at 4°C with antibodies against GFP (1:1000, Abcam, Cambridge, UK) or β3 integrin (1:1000, Chemicon/Millipore, Billerica, MA, USA). The sections were then incubated for 30 min at room temperature with Envision polymer (Dako) to increase the signal intensity. Peroxidase activity was visualized with diaminobenzidine, and the sections were counterstained with hematoxylin and mounted in DPX mounting medium (BDH Chemical, Poole, UK). GFP staining was scored qualitatively and expressed as the proportion of positive cells (0–100%), as described previously [
12].
Cells were seeded onto 35 mm glass-bottom culture dishes (MatTek Corporation, Ashland, MA, USA) for confocal microscopy (Ultraview ERS, PerkinElmer, Waltman, MA, USA) and the images from stacks (0.5 μm deep) were captured every 2 min over 2 h using a 63× water objective, and they were analyzed using Ultraview ERS (PerkinElmer, Waltman, MA, USA) and FIJI (Image J, Bethesda, MA, USA) software.
Primary tumor growth analysis
Tumor growth was quantified using FIJI software (Image J) on microphotograph images obtained on a Zeiss Axio Imager M1 microscope (Carl Zeiss, AG, Oberkochen, Germany) from fixed samples. The methods and parameters used for micro-CT image acquisition and image reconstruction have been described elsewhere [
27].
Statistical analysis
Normally distributed data were analyzed using a Student’s t-test or ANOVA followed by post-hoc analyses. Data with a non-parametric distribution were analyzed using the Kruskal-Wallis and Mann–Whitney U-tests. Mouse survival was analyzed using the log-rank test. Differences were considered significant at p < 0.05. All analyses were performed using SPSS 15.0 or Graph Pad Prism 5 software.
Discussion
The induction of angiogenesis, invasion and metastasis by TGF-β in advanced stages of cancer has been well demonstrated [
29]. Accordingly, the inhibition of TGF-β-mediated signaling has aroused great interest in the scientific community as a potential therapeutic approach to cancer treatment. Small molecule inhibitors such as the TGF-βRI inhibitors LY2157299, SB-50124 and SM16, monoclonal antibodies such as lerdelimumab, metelimumab, fresolimumab and IMC-TR1, and anti-sense mRNA molecules such as trabedersen and lucanix have yielded promising results in preclinical research and clinical trials. However, none of these compounds have yet been approved for clinical use due to the severe side effects observed in some patients, including cardiac toxicity, gastro-intestinal symptoms, fatigue, skin rash and epistaxis [
33].
While much has been written on the role of TGF-β in metastasis, there is little information on the mechanisms that govern the movement of tumor cells from tissues into the lymphatic flow and towards the lymph nodes. We demonstrate that TGF-β pretreatment increases the chemotaxis, adhesion and transmigration of H157 cells, a cell line derived from squamous cell lung carcinoma, across monolayers of primary lymphatic endothelial cells of the lung. This dynamic change is accompanied by an increase in the expression of metastasis-related genes and a switch from amoeboid to mesenchymal-like cellular movement. Mesenchymal cell movement has been associated with the formation of focal adhesion contacts, a process in which integrins play a prominent role [
34].
TGF-β triggers a complex network of signaling cascades that appear to involve cross-talk between integrins and TGF-β [
35]. We observed an increase in the expression of several integrins at both the mRNA and protein levels that was particularly notable in the case of β3 integrin. This observation is consistent with previous reports describing TGF-β-induced increments in β3 integrin mRNA and protein expression, and αvβ3 surface expression in human lung fibroblasts via a β3 integrin, c-Src- and p38 MAPK-dependent pathway [
36].
The expression of αvβ3 integrin in tumor cells has been associated with poor prognosis and increased metastasis in several carcinoma types, including osteosarcoma, pancreas and breast cancers [
37‐
39]. In the present study, we observed decreased tumor cell adhesion and transmigration across monolayers of lymphatic endothelial cells when β3 integrin was blocked or silenced in tumor cells. Blockade of the β3 integrin ligands L1CAM and CD31 reduced tumor cell transmigration, supporting the role of active adhesion mechanisms in tumor cell transit across lymphatic endothelial cells in our experimental conditions. Indeed, previous works described binding of αvβ3 integrin as expressed by melanoma cells to blood vascular endothelium via endothelium-expressed L1CAM [
40,
41]. Furthermore, hypoxia has been show to induce L1CAM-mediated breast cancer cell adhesion to tumor microvasculature [
42].
The role of β3 integrin in metastasis is not restricted to cell adhesion and it is also involved in the regulation of TGF-β bioavailability. In fact, the TGF-β-mediated induction of β3 integrin has been described as part of a positive feed-back loop in which β3 integrin facilitates TGF-β activation [
43] by binding to the RGD domains in the complexes formed between TGF-β and the Latent Associated Peptide (LAP). This activation contributes to TGF-β-stimulated cancer metastasis in mammary epithelial cells [
43]. The active cross-talk between TGF-β and integrins is triggered in tumors in response to hypoxia, oxidative stress or therapy, and it promotes tumor survival. For example, radiotherapy increases αvβ3 integrin expression as a survival mechanism in NSCLC H157 and H460 cell lines and consequently tumor growth is reduced by a combination of radiotherapy and treatment with the β3 integrin antagonist Cilengitide [
44]. We observed increased survival and decreased tumor size in mice injected with β3 integrin-deficient cells as compared with those injected with β3 integrin-competent cells. Moreover, the effects of the TGF-β inhibitory peptide P144, which significantly enhances survival and attenuates tumor growth, were more dramatic in mice injected with β3-integrin-deficient cells. Treatment with P144 has been shown to inhibit tumor growth [
45], angiogenesis [
46] and metastasis [
47], and to potentiate the efficacy of anti-tumor immunotherapy [
48] in several animal tumor models.
When we analyzed lymph node affectation, we found that the inhibition of stromal TGF-β with P144 greatly diminished the appearance of tumor cells in the lymph nodes of animals injected with untreated H157 cells. These results are consistent with previous findings highlighting the role of stromal-produced TGF-β in the establishment of metastasis from primary tumors [
49]. Remarkably, silencing of β3 integrin in the same tumors also reduced tumor cell transit to the lymph nodes to half the levels observed in mice injected with β3 integrin-competent cells.
Surprisingly, in vitro pretreatment of cells with TGF-β did not increase further metastasis to the lymph nodes of H157 NSCLC cells in comparison with the already high basal metastatic counts (80%) due perhaps to an excessively long end point for these experiments. In addition, TGF-β-pretreated tumor cells were resistant to separate targeting of β3 integrin silencing or stromal TGF-β inhibition with P144. This resistance may be explained by the acquisition incremented competences to bind and activate TGF-β exemplified by the increased expression of other integrins, such as αvβ5 and α4β1, and extracellular matrix degrading proteases such as MMPs.
Therefore, although the interplay between integrin β3 and TGF-β and between tumor and stromal cells in these animals remains to be fully elucidated, it is suggesting the fact that the phenotype of TGF-β1
-/- mice is fully reproduced in mice with mutations in the RGD binding motif in the amino acidic sequence of LAP [
50]. Given the role of integrin β3 in TGF-β-mediated proteolytic activation (33) and the binding of P144 to TGF-β, we propose that these two molecules are in competition for TGF-β binding. Thus, when integrin β3 expression is low (control cells), P144 can bind more efficiently to TGF-β and exert its inhibitory activity. However, after TGF-β exposure incremented numbers of integrin β3 molecules expressed on the cell membrane bind to and activate TGF-β, thereby competing P144 binding to its target. In keeping with this hypothesis, mice injected with tumor cells that were pretreated with TGF-β but in which integrin β3 expression was silenced responded to P144 treatment with significantly impaired metastasis to the lymph nodes. These findings suggest that TGF-β pretreated cells are primed for subsequent activation by stromal TGF-β to increase their metastatic potential.
This is not the first time combined treatments that include TGF-β inhibitors have been proposed. Indeed, several studies have demonstrated that the administration of TGF-β inhibitors in combination with immune-stimulating vaccines or cytotoxic agents improve the efficacy of current TGF-β-based therapies [
51]. However, in the case of integrin inhibiting peptides, caution is advised as for example, the inhibition of β1 integrin in models of mammary carcinoma activates the expression the β3 integrin and TGF-β mediated metastasis [
52]. Accordingly, the correct integrin/TGF-β interaction must be identified before embarking upon complex therapeutic approaches.
Acknowledgements
This work was supported by the Spanish Ministry of Health Grant PI10/02131 (Institute Carlos III) and by a Grant from the Foundation for Applied Medical Research (FIMA), Spain. We thank Dr Mark Sefton from BiomedRed who provided medical writing services.
Supplementary experimental procedures
Western blots - Primary antibodies against TGF-β1 (1:300) and TGF-βRII (1:600) were purchased from R&D Systems (Minneapolis, USA), while anti-p-Smad2 (1:2000) was obtained from Chemicon and anti-Smad 2/3 (1:2000) from Cell Signaling.
Migration, chemotaxis and flow cytometry were performed as described previously (23).
Competing interests
The authors do not have any financial competing interest to disclosure.
Authors’ contributions
ES: performed de in vitro experiments, treated animals, extracted data and analyzed it. SG: performed the animal experiments, histologies and video recording of the cells. JD, helped to set up the animal model and supported all the experimentation performed with P144 inhibitory peptide. XM, helped with statistics and confocal imaging. RP, contributed to all the in vitro analysis of cell migration and adhesion to different substrates. PA designed and supervised all the experiments in which anti L1-CAM antibodies were used. AR: design the project and experimental approach, supervised all the results and wrote the manuscript. All authors read and approved the final manuscript.