Laminin 5 expression protects against anoikis at aerogenous spread and lepidic growth of human lung adenocarcinoma

Authors


Abstract

Adenocarcinoma of the lung is characterized by frequent aerogenous spread (AE) and advancement along the alveolar wall (BAC growth). To elucidate the mechanism of AE metastasis and BAC growth in human lung adenocarcinoma, we established an in vivo orthotopic animal model and an in vitro culture. Investigation of expression levels of integrins, laminins and Type IV collagens, which are the major regulating molecules for cell attachment and anoikis was carried out and a clear correlation between the expression level of laminin 5 (LN5) and the BAC growth was observed using an orthotopic animal model. Introduction of LN5 cDNA to A549 cells increased anoikis resistance in an expression dependent manner. Cells with LN5 overexpression resisted with anoikis after treatment with PI3K-Akt and ERK inhibitors. The amount of phosphorylated focal adhesion kinase (FAK) was also higher in LN5 overexpressing cells. Major tyrosine residues of the EGF receptor at 1068, 1086 and 1173, except at 1148, remained phosphorylated only in the LN5 overexpressing cells even without EGF stimulation, that indicates the ligand independent activation of EGF receptor. BAC growth ratio and AE was confirmed to be significantly correlated with LN5 expression in surgically resected human lung adenocarcinomas by immunohistochemistry. Our results indicate that the activation of the EGF receptor by overexpressing LN5-integrin-FAK signaling pathway may play a crucial role in BAC growth and AE metastasis in human lung adenocarcinoma. © 2005 Wiley-Liss, Inc.

Adenocarcinoma, a major histologic type of lung carcinomas, can be divided into several subtypes on the basis of its histological structures.1 The most characteristic pathobiological features of lung adenocarcinoma are a bronchiolo-alveolar carcinoma (BAC) growth pattern and aerogenous spread (AE), which is morphologically defined as cancer that metastasizes through the alveolar spaces.

Both characteristic features, BAC growth pattern and AE metastasis, are observed frequently in goblet cell type, Type II and Clara cell type adenocarcinomas. They must be achieved through the following pathway: (i) cancer cells grow and spread on the basement membrane (BM) of the lung (BAC growth pattern); (ii) the cancer cells detach from the BM and survive in an anchorage-independent manner in the alveolar space (AE metastasis); and (iii) the cancer cells re-attach and grow on the BM at another site in the lung (BAC growth pattern). During these metastatic steps, the anchorage-independent survival of the cells in the alveolar space is thought to be the most critical step for cancer survival. The mechanisms responsible for the BAC growth pattern and AE metastasis in lung adenocarcinoma have not been fully elucidated. Part of the difficulty in advancing this research is the lack of an established animal model for assessing the BAC growth pattern and the AE metastasis model.

During the course of BAC growth, cancer cells produce and bind to the BM, which is composed of Type IV collagen, proteoglycans and laminins. Among the components of BM, laminin 5, 10 and 11, are the major laminin isoform found in lung epithelial BM.2, 3, 4 Interactions between epithelial cells and laminins provide a stable anchorage of epithelial cells to the BM and the migration of epithelial cells during regeneration.1, 2, 5, 6, 7 Laminins bind to integrins, their specific cell surface receptors, and evoke cell surviving or antiapoptotic signals through the action of focal adhesion kinase (FAK), a non-receptor protein tyrosine kinase in focal adhesion.8, 9, 10

Survival, growth and differentiation in normal epithelial cells are maintained in an anchorage-dependent manner. The detachment of normal epithelial cells from the BM causes apoptosis, a process known as anoikis.11, 12 The survival and growth of cancer cells can be maintained in an anchorage-independent manner. Two major cell survival and growth signaling pathways, the PI3K-Akt pathway and the ERK pathway, are regulated by integrin-regulated tyrosine kinases.12, 13, 14 Although this anchorage-independent growth pattern is a characteristic of various types of cancer cells, little is known about the mechanism of anchorage-independent survival during the AE metastasis of lung adenocarcinomas.

To elucidate the mechanism of BAC growth and AE metastasis in lung adenocarcinoma, we established an animal model for evaluating BAC growth and AE metastasis orthotopically using human lung adenocarcinoma cell lines. We further investigated the intracellular signaling pathway involved in this characteristic pattern of human lung adenocarcinoma metastasis.

Material and methods

Cells and antibodies

Six human lung adenocarcinoma cell lines, A549, PC-14, LC-2/ad, RERF-LC-KJ, NCI-H322 and NCI-H358, were kindly provided by Dr. M. Noguchi (Department of Pathology, Tsukuba University, Japan). The original histologies of these adenocarcinoma cell lines were well-differentiated adenocarcinoma for A549, NCI-H322 and NCI-H358, moderately differentiated adenocarcinoma for LC-2/ad, poorly differentiated adenocarcinoma for PC-14 and unspecified for RERF-LC-KJ. The A549 cell line was maintained in DMEM/F12 supplemented with 10% FBS. All the other cell lines were maintained in RPMI 1640 supplemented with 10% FBS. The cells were incubated at 37°C in a humidified cabinet in an atmosphere of 5% CO2 in air.

Mouse anti-human integrin β1, β4, Fluorescein isothiocyanate (FITC)-conjugated α6, phycoerythrin (PE)-conjugated β1 and rat anti-human PE-conjugated β4 subunit monoclonal antibodies were purchased from Chemicon International, Inc. (Temecula, CA) and integrin α6 subunit monoclonal antibody was purchased from Cymbus Biotechnology Ltd. (Hants, UK). The function blocking rat anti-human integrin α6 subunit monoclonal antibody was purchased from BD Bioscience (Franklin Lakes, NJ). Rabbit polyclonal antibodies to phospho-Akt, Akt, ERK and a mouse monoclonal antibody to phospho-ERK were obtained from Cell Signaling Technology, Inc. (Beverly, MA). Rabbit polyclonal antibodies to EGFR pY1148, pY1068, and FAK pY397 were purchased from Biosource International, Inc. (Camarillo, CA). A mouse monoclonal antibody to EGFR pY1173 was from Kyowa Hakko Kogyo Co., Ltd. (Tokyo, Japan). A rabbit polyclonal antibody to FAK was purchased from Upstate Biotechnology (Lake Placid, NY). Rabbit polyclonal antibody to EGFR and a goat polyclonal antibody to actin were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). A mouse monoclonal antibody to LN5 was obtained from Chemicon International, Inc. The PI3K inhibitor LY294002 and the MEK inhibitor PD98059 were purchased from Promega Corporation (Madison, WI). AG1478 was purchased from Calbiochem (San Diego, CA).

Establishment of orthotopic implantation of human lung adenocarcinoma in SCID mouse model

Male SCID mice, approximately 6 weeks old, were obtained from CLEA JAPAN (Hamamatsu, Japan). The mice were free of known pathogens at the time of the study and were housed in sterilized filter-topped cages and fed autoclaved food and water.

The mice were anesthetized using diethyl ether. A 1-cm ventral midline incision was made in the neck to expose the trachea, which was then punctured using a 24-gauge intravenous catheter (TERUMO, Tokyo, Japan). The catheter was inserted into the trachea, and 1 × 106 tumor cells in 100 μl of serum-free medium were directly inoculated through the inserted needle into the bronchioloalveolar cavity. The skin incision was then closed using a polypropylene suture.

The mice were sacrificed 8 weeks after the orthotopic implantation of the human lung adenocarcinoma cells. The lungs were removed, sliced at 2–3 mm intervals, fixed in methanol and embedded in paraffin. The cut sections were then stained with hematoxylin and eosin (H&E) for the morphological study and with periodic acid Schiff (PAS) to detect the basement membrane.

To evaluate the BAC growth pattern, the lepidic growth ratio was defined as the number of tumors in which all of the cancer cells exhibited a BAC growth pattern when stained with PAS divided by the total number of tumors.

RNA extraction and quantitative real-time RT-PCR

Total RNA was isolated using TRIzol reagent (Life Technologies, Inc., Frederick, MD). All samples were treated with RNase-free DNase (QIAGEN, Hilden, Germany) during RNA isolation, according to the manufacturer's protocol. The purity and concentration of the RNA were determined by spectrometry at 260 nm.

Real-time RT-PCR was carried out by using the SYBR green system. Five micrograms of total RNA was reverse transcribed by using oligo(dT)20 primers and the Thermoscript RT-PCR System (Life Technologies, Inc., Frederic, MD) and purified using the QIAquick PCR Purification Kit (QIAGEN). The PCR were carried out using a Light-Cycler – RNA Master SYBR Green I kit (Roche Diagnostics, GmbH, Mannheim, Germany). PCR amplification was carried out using a capillary tube in a final reaction volume of 10 μl. Real-time detection of the amplified cDNA was carried out using a Light-Cycler Instrument (Roche Diagnostics). The following oligonucleotides were used for the PCR: forward laminin α3 (LAMA3) chain, TTCAAGTTCCCGGCAGTCTC; reverse laminin α3 chain, ATGATTGGCCTCCCTCTTGG; forward laminin α5 (LAMA5) chain, CTGAGCATCCTGGAGAACC; reverse α5 chain, TTCTTGTCACGCAGAGACAC; forward laminin β1 (LAMB1) chain, AGCACTGTAACGGCTCTGAC; reverse laminin β1 chain, ACACTGTGGCGTCTCAATGC; forward laminin β2 (LAMB2) chain, AGCCAACGGCACTTTGCTAC; reverse laminin β2 chain, TGCATGTGCAGCGGTGACAG; forward laminin β3 (LAMB3) chain, CGTTGTCCCTTCCGAGAGAC; reverse laminin β3 chain, TGTCAGGTCAGGCAACGAAG; forward laminin γ1 (LAMC1)chain, CTGCCATCAACCAGACCATC; reverse laminin γ1 chain, TTCTTGAGCAGCCTGTGAAG; forward laminin γ2 (LAMC2)chain, 5′-GCAGAGCACAAGAAGCACTG-3′; reverse laminin γ2 chain, 5p-TCCATCTGCTGTCACATTGG-3′; forward Type IV collagen, 5′-CTTCCTGGGATTGATGGAGTT-3′; and reverse Type IV collagen, 5′-GTCCAGGTATACCCACCAAT-3′. The sequence specificity of the primers was confirmed by performing homology searches of the NCBI database using BLASTN software. The RT-PCR products of each molecule were used as external PCR standards. Serial 102-fold dilutions of these RT-PCR products, corresponding to 1 × 108 to 1 × 102 copies/μl, were amplified in parallel with the experimental samples. Using the LightCycler software, the amplification curves of the experimental samples were then spotted against these standard curves to generate an estimated gene-specific mRNA copy number.

Retroviral expression vectors and gene transfer

Full-length pBg2C cDNA was cloned between the BAMHI and EcoRI sites. To remove the cDNA as a BamHI/BamHI fragment, the EcoRI site was blunted and a BamHI linker, CGGATCCG, was inserted, generating pBg2C-2. The insertion of the BamHI linker enabled the EcoRI restriction site to remain intact in pBg2C-2. The resulting BamHI/BamHI fragment was inserted into the BamHI site in a pLNCX2 vector, downstream of the CMV promoter. The retrovirus was prepared in ϕ2 and PA317 packaging cells with coprecipitates of calcium phosphate. A549 was transduced with the retroviral vectors in the presence of 10 μg/ml of polybrene. Transfected cell populations were selected using growth medium supplementation with 800 μg/ml of G418.

Anoikis assay

All of the cell lines were cultured to 70–80% confluence in regular tissue culture dishes before starting the anoikis assay. After the cells had reached 70–80% confluence, they were treated with 10 mM EDTA dissolved in PBS and washed 3 times with PBS. The cells were then plated in 0.9% soft agar-coated dishes in a serum-free DMEM/F12 or RPMI 1640 medium. For the viability assays, 1 × 106 cells were suspended in soft agar-coated dishes. After culturing for various time periods, the cells were collected from the soft agar-coated dishes by pipetting. Viability was determined using the trypan blue exclusion method.

The PI3K inhibitor LY294002, the MEK inhibitor PD98059 and the EGFR specific inhibitor AG1478 were dissolved in DMSO. In experiments where cells were treated with these various inhibitors, the same volumes of DMSO were added to the controls.

Flow cytometry

The expression of integrin in the cells was evaluated using fluorescence-activated cell sorting (FACS) analysis. Various cells were cultured in regular tissue culture dishes until they reached 80% confluence. After detachment using 10 mM EDTA, the cells were harvested from the dishes by pipetting to yield single cell suspensions. Mouse anti-human integrin β1, β4, α6 and α4 subunit monoclonal antibodies, and goat F(ab′)2 anti-mouse IgG coupled to fluorescein were applied and the resulting cell suspensions were subjected to flow FACS analysis using a FACSCalibur (Becton Dickinson, CA). PE conjugated integrin α6, FITC conjugated integrin β1 and β4 antibodies were used for 2-color analysis. A minimum of 10,000 cells was examined for each sample.

Western blot analysis

For the Western blot analysis, the cells were solubilized in a 1% NP-40 lysis buffer (20 mM Tris-HCl, pH = 7.5; 150 mM NaCl; 1 mM EDTA; 1 mM EGTA; 2.5 mM sodium pyrophosphate; 1 mM β glycerophosphate; 1 mM sodium vanadate; and a protease inhibitor cocktail [Roche Diagnostics, GmbH]) on ice. The cell lysates were clarified by centrifugation at 18,500g for 30 min at 4°C. The protein concentration was determined using the Bradford reagent (Bio-Rad Laboratories, Hercules, CA). The proteins were resolved by SDS-PAGE and transferred to a PVDF membrane. For the Western blot analysis, nonspecific sites on the membrane were blocked by incubation for 1 hr at room temperature in a solution containing 5% skim milk in Tris-buffered saline/Tween buffer. All of the membranes were then incubated overnight at 4°C with primary antibodies in the same blocking solution. The membranes were then washed and incubated for 1 hr at room temperature with horseradish peroxidase-conjugated secondary antibodies (Zymed Laboratories, Inc., South San Francisco, CA). The immune complexes were detected using a Western blotting enhanced chemiluminescent detection system (Amersham Bioscience) according to the manufacturer's instructions.

The intensity of individual immunoreactive protein bands was determined by scanning the developed X-ray films and measuring the optical density and area of the bands using the Atto Densitograph software library (Atto Corporation, Tokyo, Japan).

Surgical cases and immunohistochemical staining

Eighty-seven adenocarcinomas of the lung were obtained from a retrospective analysis of all adenocarcinomas resected at the National Cancer Center Hospital East (Chiba, Japan). The histologic subtypes were as follows: goblet cell carcinoma, 22 cases; Type II or Clara with AE, 23 cases; Type II or Clara without AE, 19 cases; and poorly differentiated adenocarcinoma, 23 cases.

Immunohistochemical staining for LN5 was carried out using the avidin-biotin-peroxidase complex method. A protease antigen retrieval/unmasking method was used to enhance the detection of LN5 expression in the paraffin-embedded tissues.

Scoring for LN5 expression in the surgical specimens was carried out by calculating the average percentage of basement membrane that stained brownish, regardless of the intensity, in the lepidic fields of the cancer growth. The grade of LN5 expression was evaluated as follows: negative (−), 0–50% of the basement membrane was immunoreactive; positive (+), >51% of the basement membrane was immunoreactive. The scoring of the lepidic growth rate of the tumors in the mice was carried out by calculating the ratio of the number of lepidic tumors, in which the cancer cells had advanced along the alveolar wall, to the total number of lung tumors in each mouse. The BAC growth ratio in the surgically resected human lung adenocarcinoma tissues was calculated as the ratio of the area exhibiting a BAC growth pattern to the total area occupied by the tumor. AE was defined as the presence of cancer cells existing in an alveolar space with no epithelial attachment.

Statistical analysis

The correlations between cell survival and the expression of each laminin and integrin were evaluated by determining the Pearson rank correlation coefficients. Differences in the expression of LN5 and the BAC growth ratio among the surgical cases were evaluated using the Student's unpaired t-test, and associations between expression of LN5 and AE were analyzed using the χ2 test and a 2-way table.

Results

BAC growth in an orthotopic xenotransplanted SCID mice model using human lung adenocarcinoma cell lines

All 6 human lung adenocarcinoma cell lines produced intrapulmonary tumors exhibiting various growth patterns. The presence of a BAC growth pattern in the animal models injected with each of the 6 human lung adenocarcinoma cell lines was evaluated by measuring the advancement of the cancer growth along the BM of the alveolar wall using PAS stain and LN5 immunohistochemistry (Fig. 1a,b). The ratios of tumor cells exhibiting a BAC growth pattern compared to the total number of tumor cells (lepidic growth ratio) were as follows: RERF-LC-KJ, 0.20; LC-2/ad, 0.23; PC-14, 0.32; A549, 0.67; NCI-H358, 0.69; and NCI-H322, 0.75.

Figure 1.

The incidence of lepidic growth according to the expression of mRNA of LAMC2 in various human lung adenocarcinoma cells. Representative light micrographs of lung tumor inoculated into bronchial alveolar space stained with PAS (a) and LN5 (b). The A549 cells show lepidic growth without destructing the alveolar wall. The correlation between lepidic growth ratio and mRNA of type IV collagen and LAMC2 among various human lung adenocarcinoma cell lines (c). Although Type IV collagen does not show correlation with lepidic growth ratio (p = 0.1584, r2 = 0.4285), LAMC2 shows correlation with lepidic growth ratio (p = 0.0218, r2 = 0.7692). When statistical analysis was carried out including Clone 3 and 5, correlation between lepidic growth ratio and LAMC2 remains significant (p = 0.0018, r2 = 0.8259). The data shown are median ± SEM resulting from at least 3 samples.

The typical morphology of human cancer cells undergoing lepidic growth is a thickening of the BM as shown in Figure 1a, indicating the production of BM components by the cancer cells. We then quantified the expression levels of LAMA3, LAMA5, LAMB1, LAMB2, LAMB3, LAMC1, LAMC2 and Type IV collagen, the major components of BM using real-time RT-PCR and found that the expression levels of these components varied among different cell lines. Although no correlations between the mRNA level of Type IV collagen, LAMA3, LAMA5, LAMB1, LAMB2, LAMB3 or LAMC1 and the lepidic growth ratio were observed, the lepidic growth ratio was correlated with the expression of LAMC2, which is a component of LN5 (r2 = 0.9196, p = 0.0025) (Fig. 1c).

Anoikis induction in various human lung adenocarcinoma cell lines in vitro

Because of the size of cancer cells and alveolar space in mice, evaluation of AE spreading in human lung cancer cells in the orthotopic animal model was difficult. We then investigated the survival rate after anoikis induction in all cell lines to evaluate the potential ability of cells to detach from the BM and undergo AE and compare the survival rate with lepidic growth pattern in animal model. The cell survival rate was well correlated with the lepidic growth ratio in the orthotopic animal model. We next compared the expression of cell attachment molecules (laminin 5, 10, 11 and integrins) and cell survival. The expression levels of the laminins were not correlated with cell survival except for the expression of LAMC2 (r2 = 0.8322, p = 0.0112) (Fig. 2a). Our data suggested that LN5 expression was correlated with lepidic growth and AE metastasis in animal models. We evaluated the presence of cell surface integrins α3, α6, β1 and β4, which are known to be receptors for LN5. A FACS analysis showed that only the surface expression levels of the α6 integrin subunit were correlated with the survival rate (r2 = 0.6949, p = 0.0392). Because integrin α6 couple with β1 or β4 as a receptor of LN5,2 2-color analysis with FACS were evaluated. This experiment showed that “the sum of MFI,” which means equation image, of integrin α6β4 was correlated with cell survival (r2 = 0.6989, p = 0.0381) whereas integrin α6β1 was not (r2 = 0.6207, p = 0.0627).

Figure 2.

The correlation between survival rate on anoikis assay and various factors including laminin and integrin. The correlation between survival rate and subunit of LAMC2 and integrin α6. Six human lung adenocarcinoma cell lines were cultured in suspension for 72 hr. Cells were collected and the viability of each cell line was determined by trypan blue exclusion method. The number of X-axis shows the mRNA copies of mRNA of LAMC2. The mRNA copies per 100 ng of total mRNA was quantified by Light Cycler. (a) The X axis of each integrin subunit shows the median fluorescence intensity (MFI) determined by flow cytometry analysis. (b) The p-value and r2 of each subunit was shown in the each graph. The data represent the median resulting from at least 3 experiments. (c,d) The X axis of each “sum of integrin” was calculated with equation image determined by 2-color flow cytometry analysis.

These data indicate that cell survival in human lung adenocarcinoma cells is correlated with LN5 expression and with the presence of the α6 integrin receptor on the cell surface.

Anoikis assay and lepidic growth rate of A549 and LAMC2-transfected clones

To examine the contribution of LAMC2 to anoikis resistance, we tested whether the overexpression of LAMC2 protects A549 cell lines from anoikis. The target construct, LAMC2, was introduced into A549 cells by retroviral gene transfer. Five clones with different expression levels of LAMC2 were obtained and used for the anoikis assay. The amount of LAMC2 mRNA per 100 ng of total RNA in each cell line was as follows: parent A549, 6.30 × 103 copies; Clone 1, 8.57 × 103 copies; Clone 2, 1.09 × 104 copies; Clone 3, 3.81 × 104 copies; Clone 4, 3.88 × 104 copies; and Clone 5, 6.47 × 104 copies. In addition, the LN5 expression levels in Clones 1–5 were confirmed using immunocytochemistry after cell suspension for 72 hr. Figure 3c is a representative figure of LN5 immunocytochemical staining of A549 parent and Clone 5 cells. The membranous staining of LN5 and integrin α6 was observed more in Clone 5 than A549 but no difference was observed with integrin β1 and β4 in A549 and Clone 5. When the rates of LAMC2 expression and anoikis were compared, a significant correlation between LAMC2 expression and the antiapoptotic rate was observed (r2 = 0.8377, p = 0.0105) (Fig. 3a).

Figure 3.

Anoikis assay and lepidic growth ratio of LAMC2 transferred clones of A549. Correlation between survival rate and mRNA of LAMC2 (a). Five clones were obtained by retroviral transfection of LAMC2. These cells were cultured in suspension for 72 hr and viability was determined by trypan blue exclusion method. The X-axis shows the mRNA copies of LAMC2 per 100 ng total RNA. This result shows the correlation between survival rate and expression of mRNA of LAMC2 among these 6 cell lines (p = 0.0105, r2 = 0.8377). The data shown are median ± 2 SD resulting from at least 3 experiments. The correlation between lepidic growth ratio and mRNA of LAMC2 in transfected clones was shown in (b). These data shown are median ± SEM resulting from at least 3 experiments. LN5, integrin α6, β4 and β1 subunit expression in A549 and Clone 5 was immunocytochemically examined (c). LN5 and integrin α6 expression was observed predominantly at basement membrane more in Clone 5 than that in A549 parent clone. The expression of integrin β4 and 1 did not differed in A549 and Clone 5.

To compare LAMC2 expression and lepidic growth in vivo, we examined the intrabronchial propagation of 3 different representative LAMC2 expressing clones (parent A549, 6.30 × 103 copies; Clone 3, 3.81 × 104 copies; and Clone 5, 6.47 × 104 copies/100 ng of total RNA) in mice. The lepidic growth ratios of these cells were as follows: A549, 0.57; Clone 3, 0.72; and Clone 5, 0.93 (Fig. 3B). These results indicate that the expression of LAMC2 might be necessary or play an important role in both the anti-anoikis activity and BAC growth seen in human lung adenocarcinomas.

Function of focal adhesion kinase, Akt, ERK and EGFR in LN5 overexpressing A549 Cell lines

To investigate the outside-in integrin signaling mechanism involving LN5 overexpression in lung adenocarcinoma cells, we compared the LN5 expression level and the activity of major integrin signaling molecules that have been linked to cell survival.5 The amount of total FAK, Akt, ERK, EGFR and phosphorylated FAK, Akt, ERK and EGFR at various phosphorylation sites is shown in Figure 4. The amount of total FAK was not significantly different among the 3 cell lines that were investigated. Although the phosphorylation of FAK decreased markedly after cell suspension in all 3 cell lines, the phosphorylation of FAK recovered in Clone 5 after 72 hr of suspension. When FAK activation was calculated as the ratio of the density of the phosphorylated FAK band to that of the total FAK bands in each sample, the phosphorylation rates of FAK after 72 hr of cell suspension were 0.13, 0.26 and 0.64 in A549, Clone 3, and Clone 5, respectively (Fig. 4).

Figure 4.

LAMC2 overexpressing A549 cells re-activate ERK and Akt survival pathways and partially activate EGFR during cell detachment. A549, Clone 3, and Clone 5 were cultured in serum free suspension and collected at various time period to prepare cell lysates. Western blotting with phospho-tyrosine FAK, phospho-Serine Akt, phospho-ERK, and phospho-tyrosine EGFR was carried out. Total protein levels were determined with anti-FAK, anti-Akt, anti-ERK, and anti-EGFR antibodies. Western blotting with actin was carried out to confirm equal protein loading. These cell lines with 24 hr serum free medium starvation attached onto plastic dishes were used as a control (described “ad” in this figure). To quantify the results for these cells, the bands were scanned by densitometry. The ratio of each phospho-FAK, phospho-Akt and phospho-ERK was calculated and described below the bands.

The activity of the PI3K-Akt pathway is also regulated by the phosphorylation of Akt. Akt was phosphorylated in each of the 3 cell lines when the cells were cultured on culture dishes. After the cells were detached from the dishes, the Akt in these cells was rapidly dephosphorylated, but rephosphorylation was observed after 72 hr of cell suspension. All 3 cell lines were dephosphorylated at 1 hr after cell suspension and gradually became rephosphorylated by 24 hr after suspension. Finally, hyperphosphorylation was observed in Clones 5 and 3 (3.24 and 1.39, respectively) when the cells were kept in suspension for 72 hr, compared to the phosphorylation rates of these cells grow in a serum-free medium. In comparison, a phosphorylation rate of up to 0.72 was observed in A549 cells (Fig. 4).

The phosphorylation levels of ERK in Clones 3 and 5 significantly increased at 72 hr after cell suspension, as shown in Figure 4. The regulation of ERK during cell detachment was different from that of Akt. ERK was transiently dephosphorylated but was thoroughly rephosphorylated after 1 and 24 hr of cell suspension. This pattern of phosphorylation was particularly apparent in Clones 3 and 5 (the phosphorylation rates after 72 hr of cell suspension were as follows: A549, 0.55; Clone 3, 2.54; Clone 5, 1.64) (Fig. 4). These experiments was repeated at least 3 times and the same tendency was accepted.

These data indicate that LN5 expression is correlated with the activation of FAK, Akt and ERK after 72 hours of cell suspension and suggest that LN5 triggers a cell-survival signaling mechanism under anchorage-independent conditions.

We then investigated the status of EGFR activation, because integrin-mediated adhesion induces the phosphorylation of EGFR tyrosine residues.15, 16 Phospho-specific antibodies enabled us to determine that 3 EGFR tyrosine residues (1068, 1086, 1173) remained phosphorylated after cell suspension for 72 hr, whereas phosphorylation of the 1148 residue was not detected at 24 hr after cell suspension without serum in these cells (Fig. 4). This phenomenon was particularly obvious in the Clone 5 cell line.

We also measured EGF gene expression in the 3 cell lines and found no difference in EGF gene expression among attached and unattached cell lines (data not shown).

To test whether the increased activation of PI3K-Akt, ERK or EGFR in these cells was responsible for their resistance to anoikis, we inactivated the PI3K, MEK and EGFR in these cells using a specific PI3K inhibitor (LY294002), a MEK inhibitor (PD98059) and an EGFR inhibitor (AG1478). The addition of LY294002 and PD98059 induced apoptosis in a dose-dependent fashion in parental A549 and Clone 3 cells, but the Clone 5 cells markedly resisted apoptosis, compared to the A549 and Clone 3 cells. The dual inhibition of the PI3K and ERK pathway using LY294002 and PD98059 induced apoptosis in Clone 5, compared to the results for the A549 and Clone 3 cells (Fig. 5). Similar results were observed using AG1478, an EGFR-specific inhibitor, which induced a large degree of apoptosis in Clone 5 cells than in A549 or Clone 3 cells (Fig. 5). Because integrin α6β4 is a receptor for LN5 and was correlated with cell survival in our study (Fig. 2), we treated with the function blocking antibody to the integrin α6. Treatment with the antibody showed no inhibitory effect in A549 cells. In contrast, blockage of cell survival was observed in Clone 3 and 5 treated with the antibody. The cell survival rate at concentration of 5 μg/ml in Clone 3 and 5 cells are almost same levels with the cell survival rate of A549 cells.

Figure 5.

Effect of PI3K, MEK, and EGFR tyrosine kinase inhibitors on A549, Clone 3 and Clone 5 cells. Cells were cultured in suspension in the absence or presence of inhibitors LY294002 (10 μM and 20 μM), PD98059 (20 μM and 50 μM), LY294002 and PD98059 (10 μM + 20 μM and 20 μM + 50 μM for each) and AG1478 (1 μM and 5 μM) individually for 72 hr. Inhibition of integrin α6 by function blocking antibody (2 μg/ml and 5 μg/ml) for 72 hr was also carried out. Cells were then collected and the viability was determined by the trypan blue exclusion method. The data shown are median ± 2 SD resulting from 3 experiments.

These data suggest that the PI3K and ERK pathways were upregulated, presumably by the increase in activated FAK and EGFR via the integrin signaling caused by LN5.

Immunohistochemical characteristics of lung adenocarcinoma

Figure 6 shows the LN5 immunohistochemical staining results in surgically resected human lung adenocarcinoma tissues. LN5 was predominantly observed at the BM of the cancer cells, similar to the situation seen for A549 cells in the animal model (Fig. 6). The results of the immunohistochemical study are summarized in Table I. The BAC growth ratio was significantly higher in LN5-positive tumors than in LN5-negative tumors (0.75 and 0.24, respectively) (p < 0.0001). When we evaluated the correlation between AE and LN5 expression, AE was observed more frequently in LN5-positive tumors than in LN5-negative tumors (70.5% and 27.9%, respectively); this difference was significant (p = 0.0002). These data, combined with the results obtained from the surgical specimens, indicate that LN5 expression in lung adenocarcinoma is significantly correlated with a BAC growth pattern and AE metastasis.

Figure 6.

LN5 immunohistochemical staining of human lung adenocarcinoma. Representative figure of LN5 immunohistochemical staining of surgically resected human lung adenocarcinoma. LN5 was observed predominantly at basement membrane of cancer cells.

Table I. LNS Immunoreactivity and BAC Growth Ratio and the Presence of AE
Laminin 5 expressionNegative1Positive2p-value
  • 1

    A total of 0–50% of the basement membranes were immunoreactive.

  • 2

    Less than 51% of the basement membrane were immunoreactive.

BAC growth ratio (mean SD)0.24 (0.21)0.75 (0.24)<0.0001
AE12/43 (27.9%)31/44 (70.5%)0.0002

Discussion

Our results demonstrated that the morphological BAC growth pattern observed in our animal model was similar to the characteristic BAC growth pattern of human adenocarcinoma. When we focused on the expression of LN5, a major component of basement membranes, a correlation between the presence of LN5 and the BAC growth ratio was observed in surgically resected human adenocarcinomas (Fig. 6). Similar results were observed between BAC growth (lepidic growth ratio) and LN5 mRNA expression in our animal model (Fig. 1c). Introduction of LN5 in A549 cells also protected human adenocarcinoma cell lines from anoikis in a dose of LN5 expression dependent manner. This phenomenon was compatible with the correlation between AE and LN5 expression seen in surgically resected human lung adenocarcinomas. An analysis of FAK, which activates PI3K-Akt and ERK downstream signaling, showed that the degree of FAK phosphorylation in LAMC2-transformed A549 cells was obviously higher than that in parental A549 cells. In addition, LN5 introduction could also induce a survival signal through the action of EGFR via EGF independent manner, that was compatible with previously reported mechanism as an integrin induced EGFR activation.15, 16 These data suggest that the PI3K and ERK signaling pathways, which contribute to anoikis prevention, may share a common upstream regulation mechanism involving the autophosphorylation of EGFR and FAK via LN5-activated integrin. Although we have also tried to introduce antisense of LAMC2 gene into A549, none of the clone without expression of LN5 was obtained because of marked cell apoptosis. This phenomenon indicates that LN5 expression may affect cell survival in certain kind of human lung adenocarcinoma cell lines.

One of the major signaling pathways downstream of integrin, an LN5 receptor, is the PI3K-Akt pathway.13 We therefore investigated the possibility that LN5-dependent survival signals in lung adenocarcinoma cells are mediated through the PI3K-Akt pathway. We found that Akt phosphorylation rapidly disappeared when cells were kept in suspension but that rephosphorylation occurred after 72 hr of cell suspension. Furthermore, the expression of phosphorylated Akt in LN5-overexpressing A549 was higher than that in A549. This data suggest that the PI3K-Akt pathway is one of the major pathways that mediate survival signals from LN5 in tumor cells. Another major signaling pathway downstream of integrin is the ERK pathway.17 Similar to Akt, hyperphosphorylation was observed at 72 hr after cell suspension, and the level of phosphorylated ERK in LN5-overexpressing A549 was higher than that in A549. The reason why ERK phosphorylation sustained whereas Akt was dephosphorylated within 1 hr of cell suspension is not clear. One possible reason is the existence of a K-ras mutation in codon 12 (Gly to Ser),18 which causes Ras activation and the initiation of a constitutive active signaling cascade, including ERK.

The inhibition of Akt phosphorylation by a PI3K-specific inhibitor induced cell death, but this effect was less pronounced in LAMC2-overexpressing A549 than in A549. Similar results were obtained when ERK phosphorylation was inhibited using an MEK inhibitor. These results indicates that the PI3K-Akt and ERK signaling pathways may transmit cell survival signals and that these pathways are intensified by the expression of LN5.

Several reports indicate that EGFR and integrins form a macromolecular complex that does not contain EGF and that integrins induce the phosphorylation of EGFR.15, 16 In our study, the phosphorylation of EGFR tyrosine residues 1068, 1086, 1173 was detected using phospho-specific antibody staining, whereas the phosphorylation of tyrosine residue 1148 was not detected after 72 hr of culture in a serum-free suspension. This phosphorylation pattern for EGFR tyrosine residues is the same as that in a former report; in addition to the lack of phosphorylation of tyrosine residue 1148, a major site that is phosphorylated in response to EGF, a very low level of EGF mRNA was observed, suggesting a lack of endogenous EGF. These data suggest that EGFR associates with integrin, which is stimulated by the secretion of LN5 by cells in suspension. A discrepancy in the time course of the integrin-EGFR complex formation was observed between a former report and the present results. In the former report, EGFR tyrosine phosphorylation was downregulated within 30 min, but our data shows that phosphorylation lasted for at least 72 hr. We believe that this discrepancy is due to a difference in the cell lines that were used in the experiments. Although the former report used human cell line ECV304, we used human lung adenocarcinoma cell line A549. In fact, AG1478 induced apoptosis in 3 cell lines in a dose-related manner. Interestingly, the results of the anoikis assay using AG1478 were different from those using the PI3K and MEK inhibitors. Unlike the PI3K and MEK inhibitors, AG1478 induced a higher rate of apoptosis in LAMC2-overexpressing cells than in A549. Similar results were observed when both PI3K and MEK inhibitors were used at same time. Together with the expression of phosphorylated EGFR results, these data suggest the possibility that the origin of the PI3K and ERK signal pathways is switched from FAK via integrin expression in A549 cells to EGFR, which is associated with integrin expression in LAMC2-overexpressing A549 cells, and that this phenomenon may be a peculiar LN5-induced protection mechanism against anoikis.

Gefitinib, an EGFR tyrosine kinase inhibitor, has been reported to exhibit antitumor activity in patients with non-small cell lung cancer.19, 20, 21 Although EGFR is more frequently expressed in squamous cell carcinoma, gefitinib seems to have a higher response rate in patients with adenocarcinoma. To date, no molecule markers for predicting gefitinib sensitivity have been identified.22, 23 The present results indicate that LN5-dependent EGFR activation play a crucial role in the survival of adenocarcinoma cells by conferring anoikis resistance and that this pathway may be of therapeutic value in treatments involving EGFR tyrosine kinase inhibitors. Because the amount of phosphorylated EGFR could not be evaluated in the surgically resected adenocarcinoma specimens because of the fixation technique, we were unable to investigate the correlation between LN5 expression and phosphorylated EGFR expression in surgically resected adenocarcinomas. Further investigation of the significance of the interaction between LN5 and EGFR activation in surgically resected adenocarcinoma is needed.

Acknowledgements

We thank Y. Okuhara, C. Okumura and S. Sasaki for excellent technical assistance. Our study was supported in part by a Grant-in-Aid for Cancer Research from the Japanese Ministry of Health and Welfare and by a Grant-in-Aid for the Second Term Comprehensive 10-Year Strategy for Cancer Control, also from the Ministry of Health and Welfare. K.K, S.M., M.G., SC.Z, T.S are recipients of Resident Research Fellowships from the Foundation for the Promotion of Cancer Research (Tokyo, Japan).

Ancillary