The human lung cancer cell lines, RERF-LC-KJ, RERF-LC-MS, HARA and HARA-B, were purchased from the Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan). All cells were cultured in RPMI-1640 medium with 10% fetal calf serum (FCS) at 37°C in a humidified atmosphere of 5% CO₂.

RERF-LC-MS and RERF-LC-KJ cell lysates were labelled with the Cy3 and Cy5 protein labeling dyes (GE Healthcare Bio-Sciences AB, Uppsala, Sweden), respectively. The labelled samples were mixed and applied to a 24-cm immobilized pH gradient gel strip (IPG-strip pH 4.0–7.0) for first dimension separation. For the second dimension separation, the IPG-strips were placed on the top of the SDS-PAGE gels. After electrophoresis, gels were scanned with a laser fluoroimager (Typhoon Trio, GE Healthcare Bio-Sciences AB, Uppsala, Sweden). Quantitative analysis of protein spots was carried out with Decyder-DIA software (GE Healthcare Bio-Sciences AB). The samples for the spot-picking gel were prepared without labelling. The spot-picking gel was scanned after staining with deep purple total protein stain reagent (GE Healthcare Bio-Sciences AB). The antigen spots of interest were picked using an Ettan Spot Picker (GE Healthcare Bio-Sciences AB). Proteins were extracted by solubilizing the picked gel pieces using 88 mM sodium periodide for nitrocellulose panning experiment.

Picked gel pieces were in-gel digested with trypsin overnight. The digested peptides were dried and resuspended in 10 μl of 0.1% trifluoroacetic acid, following which they were purified using ZipTip μC₁₈ pipette tips (EMD Millipore, Billerica, MA, USA). The digested peptides were analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF/MS; AutoflexII, Bruker Daltonics Inc., Billerica, MA, USA). Peptide mass fingerprints were used for searching public protein primary sequence databases to identify proteins. The Mascot search engine (http://www.matrixscience.com) was initially used to query the entire theoretical tryptic peptide.

To generate monoclonal antibodies, panning of the scFv phage display library was performed using nitrocellulose membrane blots as previously described (12). Briefly, a portion of each protein extracted from the 2D-DIGE spots was immobilized onto a nitrocellulose membrane using the Bio-Dot Microfiltration apparatus (Bio-Rad Laboratories, Hercules, CA, USA), and these were then incubated with the blocking solution (10% skimmed milk, 25% glycerol) for 2 h. The non-immune scFv phage display library (19) was applied to each well of the Bio-Dot Microfiltration apparatus (10¹² CFU/well). After 2–3-h incubation, each well was washed ten times with Tris-buffered saline containing 0.05% Tween-20 (TBST). Bound phage was then eluted with 100 mM triethylamine. The eluted phage was used to infect log phase E. coli TG1 cells and cells were grown for 1 h at 37°C. Output phage titer was measured by counting the number of infected cells on Petrifilm (3M Corporate, St. Paul, MN, USA). The panning cycle was repeated four times.

For each identified target protein, 30 individual phage-infected TG1 clones were picked and grown separately to propagate phages, which were then purified by precipitation with polyethylene glycol. The purified phages (10¹² CFU/well) were incubated with the respective target proteins, which were extracted from the 2D-DIGE protein spots and immobilized using the Bio-Dot Microfiltration apparatus (Bio-Rad Laboratories) as described above. Phages bound to each target protein were visualized using an HRP-conjugated anti-M13 monoclonal antibody (GE Healthcare Bio-Sciences AB). For specificity assay, human recombinant kinase insert domain receptor (KDR), tumor necrosis factor receptor 1 (TNFR1) (R&D Systems Inc., Minneapolis, MN, USA), caspase-8 and importin-α were used as antigens.

Human lung cancer and normal TMAs (Super BioChips Laboratories, Seoul, Korea) were deparaffinized in xylene and rehydrated in ethanol. After heat-induced epitope retrieval using the Target Retrieval Solution pH 9.0 (Dako, Glostup, Denmark), endogenous peroxidase was blocked with 0.3% H₂O₂ for 5 min. The slides were then incubated with an scFv-displaying phage (1012 CFU/ml) for 30 min. After washing three times with TBST, the slides were incubated for 30 min with Envision⁺ Dual Link (Dako). Finally, the slides were washed three times with TBST and treated with 3,3′-diaminobenzidine, and then counterstained with Mayer’s hematoxylin. For statistical analysis, study samples were divided into high and low expression groups based on the following two criteria. In terms of distribution, the percentage of positive cells in a population of all tumor cells was scored as 0 (0%), 1 (1–50%), and 2 (51–100%). In terms of quantity, the signal intensity was scored as 0 (no signal), 1 (weak), 2 (moderate) or 3 (marked). Cases with a total score of ≥3 were classified into the high expression group.

For overexpression and knockdown of gene expression, cells were transfected with an expression plasmid containing the cDNA of the gene or with a gene-specific siRNA, respectively. Transfection of cells with OSBPL5 and CALU expression plasmids (Life Technologies, Carlsbad, CA, USA) was carried out using Lipofectamine LTX (Life Technologies) according to the manufacturer’s instructions. Briefly, RERF-LC-MS cells were plated in 100-mm dish. The next day, 15 μg of plasmid was mixed with 15 μl plus reagent in 3 ml Opti-MEM. After 15-min incubation, 37.5 μl lipofectamine LTX was added, incubated for 30 min, and then the DNA-lipofectamine complex was added to the cells. The cells were used in invasion assay after 24-h incubation. Transfection of cells with gene specific siRNA (Qiagen, Montgomery Country, MD, USA) was carried out using HiPerFect Transfection Reagent (Qiagen) according to the manufacturer’s instructions. Briefly, RERF-LC-KJ, HARA-B and HARA cells were plated in a 100-mm dish. The next day, 50 nM siRNA was mixed with 40 μl HiPerFect Transfection reagent in 3 ml Opti-MEM. After 10-min incubation, the complex was added to the cells. Transfected cells were used in invasion assay after 24-h incubation. The cells treated with only transfection reagents are regarded as mock cells. Expression levels of OSBPL5 and CALU genes in cells, transfected either with the respective cDNA-expression plasmid or with the indicated siRNA, were determined using the RT-PCR method.

Invasion assay was performed using a 96-well BME cell invasion assay kit (Trevigen Inc., Gaithersburg, MD, USA). The upper chambers of the 96-well cell culture inserts were washed with serum-free medium, coated with 50 μl of basal membrane extract (BME) and then dried overnight at 37°C. One million cells in serum-free media were added to the upper chambers and 150 μl of medium containing 10% FCS was added to the lower chambers. The invasion chambers were kept for 72 h at 37°C in the cell culture incubator. Non-invasive cells on the upper insert membranes were removed by gentle rubbing. Invasive cells on the lower insert membranes were stained with calcein-AM solution, and were assayed by measuring the fluorescence intensity using ARVO MX (Perkin-Elmer, Waltham, MA, USA).

In order to identify metastasis-related proteins in lung cancer, we performed 2D-DIGE analysis of lung cancer cells with high LN metastatic potential (RERF-LC-KJ) (20,21) and lung cancer cells with non-metastatic potential (RERF-LC-MS) (22,23). Fig. 1 shows the fluorescent image of a representative 2D-gel containing proteins expressed in these cells. Quantitative analysis identified 15 protein spots whose intensities altered >2-fold in RERF-LC-KJ cells compared to in RERF-LC-MS cells. Proteins from these spots were then identified by MALDI-TOF/MS (Table I). Thus, a portion of each extracted protein was immobilized by dot-blotting onto a nitrocellulose membrane and this membrane was used for 4-cycle panning of a non-immune scFv-displaying phage library. The output/input ratio (titer of the recovered phage library after the panning/titer of the library before the panning) was increased as the panning round was repeated (Table II). This elevated output/input ratio indicated the enrichment of antigen-binding scFv clones. A total of 30 clones (for each target protein) were randomly picked from the fourth panning output and their bindings to respective antigens were verified by phage dot blot ELISA. Results shown in Fig. 2 demonstrated that each one of the 15 target proteins were able to bind to multiple number of scFv antibody phages. From these positive clones, we selected the ones displaying highest affinities for evaluating their respective specificity. Fig. 3 showed the specificity evaluation results, which demonstrated that all selected scFv clones specifically recognized their respective target proteins, but not KDR, TNFR1, caspase-8 and importin-α, which were used as negative control antigens. Thus, by using the antibody proteomics technology described in this study, we isolated monoclonal antibodies to 15 metastasis-related target proteins and validated their specificity.

In order to identify and select the metastasis-related proteins in lung cancer from a large pool of candidate proteins, expression profiles of the identified proteins were determined by TMA analysis using the phage antibodies. The TMA used in this study contained tissues from 46 lung cancer cases with information on LN metastasis. Examination of the expression profile of each antigen revealed that glucosidase II, unnamed protein product, glycyl-tRNA synthetase, chaperonin containing TCP1 subunit 8 and eukaryotic initiation factor 4AII were not expressed in the clinical samples of lung tumor tissues, suggesting that these proteins were only produced in the cancer cell lines. However, ten other proteins were expressed in the clinical samples of lung tumor tissues (Table III). Results summarized in Table III also show that the expression ratio of OSBPL5 and CALU, among all the expressed candidate proteins, were significantly higher in the LN metastasis-positive cases than in the metastasis-negative cases (p=0.0156 and 0.0055, respectively). Moreover, 15 cases out of a total of 46 lung tumor cases were OSBPL5 and CALU double-positive, and 12 of them (80% of OSBPL5 and CALU double-positive cases) were LN metastasis-positive. Therefore, the correlation analysis between protein expression and clinicopathological characteristic revealed significant association between OSBPL5 and CALU expression and LN metastasis in lung tumors.

To delineate the functions of OSBPL5 and CALU in metastatic lung cancer, we analyzed the effects of gene overexpression and gene knockdown on lung cancer cell invasiveness, a main characteristic of metastasis. First, we transfected RERF-LC-MS cells with either OSBPL5 expression plasmid pCMV-OSBPL5 or with CALU expression plasmid pCMV-CALU, and confirmed that OSBPL5 or CALU, respectively, was indeed overexpressed in these cells. Test results for the invasiveness of cells, as shown in Fig. 4A, clearly indicate that the RERF-LC-MS cells overexpressing either OSBPL5 or CALU were significantly more invasive than the cells transfected with the control plasmid. Next, we transfected RERF-LC-KJ cells with OSBPL5 siRNA or CALU siRNA and then examined the invasiveness of cells in which these genes were knocked down. As shown in Fig. 4B, the invasiveness of cells transfected with either OSBPL5 or CALU siRNA was significantly lower compared to that of the mock group, while there was no difference in cell proliferation (data not shown). Knockdown of expression of either OSBPL5 or CALU by transfection of RERF-LC-KJ cells with the respective siRNA did not diminish the invasiveness of cells completely, suggesting that the expression of OSBPL5 and CALU are probably partly responsible for the increased invasiveness of cells. An inhibitory effect of the cell invasiveness by OSBPL5 gene-knockdown was also observed in the invasive lung cancer cells (HARA-B), which were derived from a bone lesion formed after the intracardiac inoculation of HARA cells (Fig. 5). These results suggested that OSBPL5 and CALU might play a critical role in facilitating invasiveness of lung cancer cells.

To further determine the usefulness of OSBPL5 and CALU as diagnostic or therapeutic targets, we analyzed the expression levels of these proteins in normal lung tissues. TMA analysis using lung cancer and normal lung tissues showed that OSBPL5 and CALU were specifically expressed in the lung tumor tissues (Table IV). Our observation that both OSBPL5 and CALU were specifically expressed in the lung tumor tissues suggested that these two proteins might have some functional roles in lung cancer cells. Thus, they could either help in elucidating the underlying mechanism of lung cancer or serve as targets for developing therapies against lung cancer.