Identification of Epstein-Barr Virus–Induced Gene 3 as a Novel Serum and Tissue Biomarker and a Therapeutic Target for Lung Cancer

Lung cancer is the leading cause of cancer deaths in the world (1). Although the overall 5-year survival rate of patients with non–small cell lung cancer (NSCLC) is only 10% to 15%, that of patients with stage IA NSCLC exceeds 80% (2). Several tumor markers such as carcinoembryonic antigen (CEA), serum cytokeratin 19 fragment (CYFRA21-1), and progastrin-releasing peptide (Pro-GRP) are clinically available for lung cancer diagnosis. However, these markers are not entirely useful for screening early-stage cancers because of their relatively low sensitivity and specificity (3, 4). Therefore, the development of novel diagnostic tools is necessary.

Because cell surface proteins are considered more accessible to immune mechanisms and drug delivery systems, identification of cancer-specific cell surface and secretory proteins may be an effective approach to develop novel diagnostic biomarkers such as serum biomarkers and antibody-based therapies (5). We have screened genes encoding molecules that are upregulated in lung cancer tissues but not expressed in normal human tissues. For these experiments, we carried out cDNA microarray analysis of 27,648 genes or expressed sequence tags (EST) in combination with purification of tumor cells by laser microdissection. In addition, we compared the microarray data with the expression profiles of 31 normal tissues (27 adult and 4 fetal organs; refs. 6–10). To verify the biological and clinicopathologic significance of the respective gene products, we carried out tumor tissue microarray analysis of clinical lung cancer materials as well as loss of function assays, using siRNA (11–29). We also used ELISA to determine whether the genes exhibiting a tumor-specific transmembrane/secretory protein are potential diagnostic serum biomarkers (30–35). We also screened for epitope peptides recognized by human histocompatibility leukocyte-A0201/A2402–restricted cytotoxic T lymphocyte from the list of selected cancer–testis antigens (36–41). This systematic approach revealed that the Epstein-Barr virus–induced gene 3 (EBI3) was frequently transactivated in primary lung cancer.

EBI3 encodes a 34-kDa secretory glycoprotein with a signal peptide and 2 fibronectin type III domains (42). The fibronectin type III repeat region is an approximately 100 amino acid domain that contains binding sites for DNA, heparin, fibrin, and cell surface proteins (43). EBI3 had possible roles in maintenance of pregnancy or immunotolerance of the human maternal body toward the fetus (44). A few studies indicated the expression of EBI3 in Hodgkin's lymphoma and adult T-cell lymphoma/leukemia (45, 46). However, the significance of EBI3 activation in human carcinogenesis and its potential as a therapeutic target as well as a serological/prognostic biomarker were not clarified.

We here describe a crucial role of EBI3 activation in human lung cancer development and progression as well as its potential as a novel biomarker and/or molecular therapeutic target for lung cancer.

The human lung cancer cell lines used in this study included 8 adenocarcinoma (ADC; A427, A549, LC319, PC-3, PC-9, PC-14, NCI-H1373, and NCI-H1781), 1 bronchioalveolar carcinoma (BAC; NCI-H358), 6 squamous cell carcinoma (SCC; NCI-H226, NCI-H520, NCI-H2170, NCI-H1703, EBC-1, and RERF-LC-AI), 1 large cell carcinoma (LCC; LX1), and 6 small cell lung cancers (SCLC; DMS114, DMS273, SBC-3, SBC-5, NCI-H196, and H446). A human bronchial epithelial cell line (BEAS-2B) was used as a control (Supplementary Table S1). All cells were grown in monolayer in appropriate medium supplemented with 10% fetal calf serum (FCS) and maintained at 37°C in humidified air with 5% CO₂. Primary lung cancer samples had been obtained earlier with informed consent as described elsewhere (6, 10). All tumors were staged on the basis of the postsurgical pathologic tumor node metastasis stage classification of the International Union Against Cancer (47). In addition, a total of 414 NSCLC tissues and adjacent normal lung tissues have been obtained with clinicopathologic data from patients who underwent surgery at Saitama Cancer Center (Saitama, Japan). Summary of patient background was described in Supplementary Table S2. This study and the use of all clinical materials mentioned were approved by institutional ethical committees.

Serum samples were obtained in 2000–2008 with informed consent from 636 individuals comprising 463 training set and 173 validation set for serum EBI3 detection by ELISA. The training set includes serum samples from 274 patients with lung cancers, 63 patients with chronic obstructive pulmonary disease (COPD), and 126 healthy volunteers. These serum samples from NSCLC patients were obtained at Kanagawa Cancer Center Hospital (Yokohama, Japan; n = 121) and Hiroshima University (Hiroshima, Japan; n = 78) and those from SCLC patients were enrolled as a part of the Japanese Project for Personalized Medicine (BioBank Japan; n = 75). Serum samples from COPD were obtained from BioBank Japan and those from healthy volunteers were obtained from Hiroshima University. The independent validation set for confirming the reproducibility of serum EBI3 as a cancer detection biomarker included serum samples from 101 NSCLC and 31 SCLC patients from BioBank Japan and those from 41 healthy volunteers from Hiroshima University. Summary of patient background is described in Supplementary Table S3. Serum was obtained at the time of diagnosis and stored at −150°C. The use of all clinical materials mentioned was approved by institutional ethical committees.

A total of 3-μg aliquot of mRNA from each sample was reversely transcribed to single-stranded cDNAs by using random primer (Roche Diagnostics) and SuperScript II (Invitrogen). Semiquantitative reverse transcriptase PCR (RT-PCR) experiments were carried out with the following sets of synthesized primers specific to EBI3 or β-actin (ACTB) as an internal control: EBI3, 5′-TGTTCTCCATGGCTCCCTAC-3′ and 5′-AGCTCCCTGACGCTTGTAAC-3′; ACTB, 5′-GAGGTGATAGCATTGCTTTCG-3′ and 5′-CAAGTCAGTGTACAGGTAAGC-3′. PCRs were optimized for the number of cycles to ensure product intensity to be within the linear phase of amplification.

Human multiple tissue blots covering 16 tissues (BD Biosciences, Clontech) were hybridized with an [α-³²P]-dCTP–labeled, 404-bp PCR product of EBI3, prepared as a probe by primers 5′-TGTTCTCCATGGCTCCCTAC-3′ and 5′-CTACTTGCCCAGGCTCATTG-3′. Prehybridization, hybridization, and washing were done following the manufacturer's recommendations. The blots were autoradiographed with intensifying screens at −80°C for 7 days.

Immunocytochemical analyses were carried out as previously described by using a commercially available goat polyclonal anti-human EBI3 antibody (Santa Cruz Biotechnology; catalogue no. sc-26797; ref. 14). On immunocytochemical analyses, we confirmed that the antibody was specific for EBI3 protein, using NSCLC cell lines that endogenously expressed EBI3 as well as cell lines derived from NSCLC or bronchial epithelia that did not express it.

Tumor tissue microarrays were constructed with 414 formalin-fixed primary lung cancers as described elsewhere (48). The tissue area for sampling was selected by visual alignment with the corresponding hematoxylin and eosin–stained section on a slide. Three, 4, or 5 tissue cores (diameter 0.6 mm; height 3–4 mm) taken from a donor tumor block were placed into a recipient paraffin block, using a tissue microarrayer (Beecher Instruments). A core of normal tissue was punched from each case, and 5-μm sections of the resulting microarray block were used for immunohistochemical analysis.

To investigate the EBI3 protein expression in clinical samples that had been embedded in paraffin blocks, we stained sections by using EnVision⁺ kit/horseradish peroxidase (DakoCytomation) for immunostaining according to the manufacturer's instructions (12). In brief, antigens were retrieved by heating the sections in Target Retrieval Solution, Citrate pH 6 (DakoCytomation). A goat polyclonal anti-human EBI3 antibody (Santa Cruz Biotechnology; catalogue no. sc-26797) was added to each slide after blocking of endogenous peroxidase and proteins. The sections were incubated with horseradish peroxidase–labeled anti-goat immunoglobulin G as the secondary antibody. Substrate/chromogen was added, and the specimens were counterstained with hematoxylin. On immunohistochemical analyses, we confirmed that the antibody was specific for EBI3 protein by antigen blocking assays, using EBI3 antigen peptides (catalogue no. sc-26797P; Santa Cruz Biotechnology) that were used for immunization of goats to produce polyclonal anti-human EBI3 antibodies (details were shown in the legend of Supplementary Fig. S1).

Three independent investigators semiquantitatively assessed EBI3 positivity without prior knowledge of clinicopathologic data. The intensity of EBI3 staining was evaluated by using the following criteria: strong positive (scored as 2+), brown staining in more than 50% of tumor cells completely obscuring cytoplasm; weak positive (1+), any lesser degree of brown staining appreciable in tumor cell cytoplasm; and absent (scored as 0), no appreciable staining in tumor cells. Cases were accepted as strongly positive if 2 or more investigators independently defined them as such.

We constructed a sandwich-type ELISA system as described previously (32). In brief, a 96-well microplate (catalogue no. 439454; Nalge Nunc International) was coated with a goat polyclonal anti-human EBI3 antibody (Santa Cruz Biotechnology; catalogue no. sc-26797) overnight at 4°C. Wells were blocked with PBS (pH 7.4) containing 5% bovine serum albumin and 0.05% Tween 20 for 2 hours and then 3-fold–diluted sera were added and incubated for 2 hours. A biotinylated polyclonal antibody specific for EBI3 using Biotin Labeling Kit-NH2 (Dojindo Laboratories) was added as a detection antibody and incubated for 2 hours, followed by reaction with avidin-conjugated peroxidase (P347; DakoCytomation) using a Substrate Reagent (R&D Systems). The reaction was stopped by adding 50 μL of 2N sulfuric acid. Color intensity was determined by a photometer at a wavelength of 450 nm, with a reference wavelength of 570 nm. Standard curve of color intensity was made by using immunogen EBI3 peptide for anti-EBI3 antibody production (catalogue no. sc-26797; Santa Cruz Biotechnology). The color intensity gained by applying 20 nmol/L of the EBI3 peptide was tentatively defined as 1 U/mL. Levels of 3 conventional tumor markers (CEA, CYFRA21-1, and Pro-GRP) in serum were measured by ELISA with a commercially available enzyme test kit (CEA: Hope Laboratories; CYFRA21-1: DRG Instruments GmbH; and Pro-GRP: TFB Inc.) according to the supplier's recommendations.

siRNA duplexes (Dharmacon, Inc.; 600 pmol/L) were transfected into lung cancer cell lines A549 and LC319, using 30 μL of Lipofectamine 2000 (Invitrogen) following the manufacturer's protocol. The transfected cells were cultured for 7 days. Cell numbers and viability were evaluated by Giemsa staining and triplicate MTT assays (Cell Counting Kit-8 solution; Dojindo Laboratories). To confirm suppression of EBI3 expression, semiquantitative RT-PCR was carried out with synthesized primers specific to EBI3 described above. The target sequences of the synthetic oligonucleotides for RNA interference (RNAi) were as follows: control 1 (si-CNT/ON-TARGET plus; Dharmacon Inc.; pool of 5′-UGGUUUACAUGUCGACUAA-3′; 5′-UGGUUUACAUGUUUUCUGA-3′; 5′-UGGUUUACAUGUUUUCCUA-3′; and 5′-UGGUUUACAUGUUGUGUGA-3′); control 2 (si-LUC/luciferase: Photinus pyralis luciferase gene, 5′-CGUACGCGGAAUACUUCGA-3′); and siRNAs against EBI3-1 and EBI3-2 (si-EBI3-#1, 5′-CAAUGAGCCUGGGCAAGUA-3′; si-EBI3-#2, 5′-UCACGGAUGUCCAGCUGUU-3′).

To establish COS-7 cells stably expressing EBI3, plasmids expressing either EBI3 (pcDNA3.1-EBI3-myc/His) or mock plasmids (pcDNA3.1-myc/His) were transfected into COS-7 cells that did not express endogenous EBI3, using FuGENE 6 Transfection Reagent (Roche Diagnostics) according to the manufacturer's protocol. Transfected cells were cultured in Dulbecco's Modified Eagle's Media containing 10% FBS and geneticin (0.4 mg/mL) for 14 days; then 50 individual colonies were trypsinized and screened for stable transfectants by a limiting dilution assay. Expression of EBI3 protein was determined in each clone by Western blotting and immunostaining. COS-7 transfectants that could stably express EBI3 were seeded onto 6-well plates (1 × 10⁴ cells/well) and maintained in medium containing 10% FCS and 0.4 mg/mL geneticin. After 120 hours cell viability was evaluated by MTT assay. Colonies were also stained at the same time. All experiments were done in triplicate.

Statistical analyses were carried out by the StatView statistical program (SAS). Survival curves were calculated from the date of surgery to the time of death due to NSCLC or to the last follow-up observation. Kaplan–Meier curves were calculated for each relevant variable and for EBI3 expression; differences in survival times among patient subgroups were analyzed by using the log-rank test. The sample size for comparing 2 survival curves was confirmed by using a statistical power level of 90% and a 2-sided level of 5% on the basis of observed probability. The risk factors associated with the prognosis of patients with NSCLC were evaluated by using Cox's proportional hazard regression model with a step-down procedure. Only variables that were statistically significant in univariate analysis were evaluated by multivariate analysis. The criterion for removing a variable was the likelihood ratio statistic, which was based on the maximum partial likelihood estimate (default P value of 0.05 for removal from the model).

Differences in the serum EBI3 levels among tumor groups, healthy volunteers, and patients with COPD were analyzed by Mann–Whitney U tests. Serum EBI3 levels before and after surgery were analyzed by the paired t test. The correlations between serum biomarkers were calculated by Spearman's rank correlation. Serum levels of EBI3, CEA, CYFRA21-1, and Pro-GRP were analyzed by drawing receiver operating characteristic (ROC) curves for their capability of distinguishing between patients with lung cancers and healthy volunteers. Statistical comparison between the area under the ROC curve (AUC) of EBI3 and that of other conventional serum markers was done by MedCalc software (MedCalc Software), which calculates the difference in AUC and standard errors (SE), followed by statistical calculation of the P value based on the method proposed by Hanley and McNeil (49).

To screen novel molecule targets that can be used for detection of cancer at an early stage and for development of novel treatment strategies, we first carried out genome-wide gene expression profile analysis of 120 lung carcinomas, using a cDNA microarray (6–10). Among 27,648 screened genes or ESTs, we identified elevated expression (5-fold or higher) of EBI3 transcripts in the great majority of lung cancer tissues examined. We subsequently confirmed its overexpression by semiquantitative RT-PCR experiments in 11 of 15 lung cancer tissues and in 14 of 22 lung cancer cell lines (Fig. 1A). We subsequently carried out immunocytochemical analysis to examine the subcellular localization of endogenous EBI3 protein in 4 lung cancer cell lines as well as BEAS-2B cells. We found that EBI3 was detected in the cytoplasm of EBI3-positive lung cancer cell lines with granular appearance whereas it was not detected in cells in which the EBI3 transcript was not detected (Fig. 1B). Because EBI3 is a secretory protein, we applied ELISA to evaluate the EBI3 protein levels in the culture media of the same set of cell lines. The amount of detectable EBI3 in culture media was concordant to the expression levels of EBI3 detected by semiquantitative RT-PCR and immunocytochemistry, suggesting that anti-EBI3 antibody is capable of specifically recognizing the secretory EBI3 protein (Fig. 1C).

Northern blot analysis using an EBI3 cDNA fragment as a probe identified a 1.3-kb transcript that was highly expressed only in placenta (Fig. 2A). We examined the expression of the EBI3 protein in 5 normal tissues (liver, heart, kidney, lung, and placenta) and lung cancer tissues, using anti-EBI3 antibodies. EBI3-positive staining was mainly observed in the cytoplasm of placenta and lung cancer cells; however, it was not detectable in 4 normal tissues (Fig. 2B). We confirmed that the positive signal by anti-EBI3 antibody obtained in lung cancer tissues was markedly diminished by preincubation of the antibody with an EBI3 antigen peptide used for immunization of a goat, which independently indicated its specificity to the EBI3 protein (Supplementary Fig. S1). We also evaluated EBI3 staining in NSCLC and adjacent normal lung tissues, confirming the EBI3 protein to be positively stained in the majority of NSCLC tissues but not in their corresponding normal lungs (Fig. 2C).

To investigate the biological and clinicopathologic significance of EBI3 in pulmonary carcinogenesis, we carried out immunohistochemical staining on tissue microarray containing primary lung tumor tissue sections obtained from 414 patients with NSCLC, who had undergone surgery. We classified a pattern of EBI3 expression on the tissue array ranging from absent (scored as 0) to weak/strong positive (scored as 1+ to 2+; Fig. 2D). The number of NSCLC tissues scored as 0, 1, and 2 was 55 (13.3%), 156 (37.7%), and 203 (49.0%), respectively. As shown in Table 1, male gender (P = 0.0001 by Fisher's exact test), no smoking history (P = 0.0051), larger tumor size (pT_2–4; P = 0.0012), and non-ADC histology (P = 0.0003) were significantly associated with strong EBI3 positivity. As shown in Figure 2E, the survival periods of patients with NSCLC showing absent/weak-positive EBI3 staining was significantly longer than those that of patients with strong-positive EBI3 staining (P = 0.0014; log-rank test). To determine the prognostic significance of the clinical characteristics and EBI3 expression in patients with NSCLC, we carried out Cox's proportional hazard regression analysis on the parameters listed in Table 2. Multivariate analysis determined that EBI3 positivity (P = 0.0439) as well as age, pathologic tumor stage, and pathologic node stage were independently useful as prognostic factors for surgically treated patients with NSCLC.

Because the in vitro findings by ELISA evaluating the EBI3 protein levels in the culture media of lung cancer cells had suggested application of EBI3 as a serum biomarker, we investigated whether the EBI3 protein is detectable in the sera of patients with lung cancer. ELISA experiments detected the EBI3 protein in serologic samples obtained from 274 patients with lung cancer, 63 patients with COPD, and 126 healthy volunteers. The mean serum levels of EBI3 in patients with lung cancer, patients with COPD, and healthy volunteers were 6.1 ± 5.2 (mean ± 1 SD), 1.9 ± 2.6, and 1.5 ± 1.6 U/mL, respectively. The serum levels of EBI3 were significantly higher in patients with lung cancer than in those with COPD or in healthy donors (P < 0.0001 and P < 0.0001, respectively, Mann–Whitney U test), whereas the difference between healthy individuals and patients with COPD was not significant (P = 0.246). When classified according to histologic type, the serum levels of EBI3 were 6.0 ± 5.0 U/mL in patients with lung ADC, 6.1 ± 4.4 U/mL in patients with lung SCC, and 6.3 ± 6.2 U/mL in patients with SCLC (Fig. 3A); the differences between the 3 histologic types were not significant (ADC vs. SCC, P = 0.4857; ADC vs. SCLC, P = 0.8309; and SCC vs. SCLC, P = 0.4707). Using ROC curves drawn with the data of the 274 patients with lung cancer and the 126 healthy volunteers, the cutoff level in this assay was set to provide optimal diagnostic accuracy and likelihood ratios (minimal false-negative and false-positive results) for EBI3, that is, 5.5 U/mL with a sensitivity of 46.7% (128 of 274) and specificity of 97.6% (123 of 126). According to tumor histology, the proportion of serum EBI3–positive cases was 47.1% for ADC (95% CI, 39.2%–54.9%; 73 of 155), 54.5% for SCC (95% CI, 39.8%–69.2%; 24 of 44), and 41.3% for SCLC (95% CI, 30.2%–52.5%; 31 of 75). The proportion of serum EBI3–positive cases was 3.2% (2 of 63) for COPD. Validation analysis using another independent set of serum samples obtained from patients with lung cancer and healthy volunteers also confirmed significant elevation of EBI3 in the sera of patients with lung cancer (Fig. 3B).

We also compared the serum EBI3 values with the expression levels of EBI3 in primary tumors in the same set of 16 NSCLC cases whose serum had been collected before surgery (8 patients with EBI3-positive tumors and 8 with EBI3-negative tumors). The levels of serum EBI3 showed good correlation with the expression levels of EBI3 in primary tumors (Fig. 3C). We then conducted ELISA by using 20 pairs (before surgery and 2 months after surgery) of serum samples from patients with lung cancer to monitor the levels of serum EBI3 in these patients. The concentration of serum EBI3 decreased significantly after surgical resection of the primary tumors (Fig. 3D). The results independently support the high specificity and potentiality of serum EBI3 as a biomarker for detecting cancer at an early stage and for monitoring the disease.

To evaluate the clinical usefulness of serum EBI3 levels as a tumor detection biomarker, we also measured the serum levels of 3 conventional tumor markers (CEA for ADC, CYFRA21-1 for SCC, and Pro-GRP for SCLC patients) in the same set of serum samples from patients with cancer and control individuals by ELISA. AUC of EBI3 was significantly larger than that of CEA in lung ADC (difference between areas, 0.131; SE = 0.0401; 95% CI, 0.0521–0.0210; P = 0.0011); it was also larger than that of CYFRA21-1 in lung SCC (difference between areas, 0.151; SE = 0.0572; 95% CI, 0.0388–0.0263; P = 0.0083), although it was not as good as that of Pro-GRP in SCLC (difference between areas, 0.164; SE = 0.0359; 95% CI, 0.0935–0.0234; P < 0.0001, Fig. 3E). The correlation coefficient between EBI3 and the 3 conventional tumor markers (CEA, CYFRA21-1, and Pro-GRP) was not significant, indicating that measuring both EBI3 and the conventional tumor markers in serum can improve the overall sensitivity of lung cancer detection. Indeed, a combination assay using EBI3 and CEA in serum could improve the overall sensitivity of ADC detection and diagnosis until 65.8% (102 of 155). The sensitivity of CEA alone was 34.2% (53 of 155) and that of EBI3 was 47.1% (73 of 155). The false-positive rate of the combined assay for the 2 tumor markers among healthy volunteers (control group) was 4.8% (6 of 126). The false-positive rate for either CEA or EBI3 individually was 2.4% (3 of 126). The combination assay using EBI3 and CYFRA21-1 in serum could improve the overall sensitivity of SCC detection until 65.9%. For diagnosing SCC, the sensitivity of CYFRA21-1 alone was 38.6% (17 of 44) and that of EBI3 was 54.5% (24 of 44). The false-positive rate by combination of the 2 tumor markers among healthy volunteers (control group) was 4.8% (6 of 126). The combination assay using EBI3 and Pro-GRP could improve the overall sensitivity of SCLC detection until 72.0% (54 of 75). For diagnosing SCLC, the sensitivity of Pro-GRP alone was 60.0% (45 of 75) and that of EBI3 was 41.3% (31 of 75). The false-positive rate by combination of the 2 tumor markers among healthy volunteers was 4.0% (5 of 126).

To assess whether EBI3 plays a role in growth or survival of lung cancer cells, we evaluated the knockdown effect of endogenous EBI3 expression by siRNA together with 2 different control siRNA (siRNA for CNT and LUC). Treatment of 2 different NSCLC cells, A549 or LC319, with the effective siRNA reduced the expression of EBI3 (Fig. 4A) and resulted in significant inhibition of cell viability and colony numbers as measured by MTT and colony formation assays (Fig. 4B and C). The results suggest that upregulation of EBI3 contribute to the growth or survival of cancer cells.

To disclose the potential role of EBI3 overexpression in tumorigenesis, we transfected either EBI3-expressing plasmids or mock plasmids into COS-7 cells, which normally express very low endogenous EBI3, and established cells that stably expressed EBI3. Two established COS-7 cell lines expressing exogenous EBI3 (COS-7-EBI3-#A and COS-7-EBI3-#B; Fig. 5A) exhibited significant rapid cell growth compared with control mock cells (COS-7-MOCK-M1 and COS-7-MOCK-M2; Fig. 5B). Furthermore, there was a tendency of COS7-EBI3 cells to form larger colonies than the control cells (Fig. 5C). These data strongly suggest a potential oncogenic effect of EBI3.

We carried out gene expression profile analysis by using cDNA microarrays for screening genes encoding transmembrane/secretory proteins that are upregulated in the majority of lung cancers (5). With subsequent systemic screening systems using tumor tissue microarray, RNAi, high-throughput ELISA, and bioinformatics, we showed that EBI3 is a useful serum and cancer tissue biomarker. EBI3 is also a potential target for the development of novel drugs for treating lung cancer.

EBI3 encodes a secretory glycoprotein; it dimerizes with p28 or interleukin (IL)-12 p35-related subunits to form IL-27 or IL-35 cytokines, respectively (42, 43). IL-35 is a new member of the IL family and has an immunosuppressive function of regulatory T cells in mouse inflammatory tissues; however, the precise function is not well understood in the human immune system (50).

In this study, we showed that EBI3 was expressed in placenta and was also highly expressed in the majority of surgically resected samples from patients with NSCLC. Strong EBI3 expression in NSCLC tissues is associated with poor prognosis. To the best of our knowledge, this is the first study to report that EBI3 expression has a strong prognostic value with regard to human NSCLC. We also found that inhibition of endogenous EBI3 expression by siRNA resulted in marked reduction in lung cancer cell viability. Concordantly, mammalian cells expressing exogenous EBI3 exhibited a significant increase in cellular proliferation. Because EBI3 is a secretory protein, presence of an autocrine/paracrine loop could be an explanation for the underlying mechanism of the growth-promoting effect of EBI3. Although the detailed function of EBI3 in lung carcinogenesis, including identification of its cell surface receptor, should be elucidated in future studies, our results implied that EBI3 may be associated with cancer growth and a highly malignant phenotype of lung tumors. Because EBI3 is considered to be a cancer antigen, it might be a good target for the development of cancer immunotherapy such as cancer vaccines as well as therapeutic strategies selectively targeting EBI3 activity, which could be essential for cancer cell growth.

We constructed an ELISA system to measure serum levels of EBI3 and showed that serum EBI3 was significantly higher in patients with lung cancer than in patients with COPD and in healthy volunteers. Furthermore, we verified our results in an independent set of serum samples. The positivity of serum EBI3 seems to be associated with the presence of lung tumors because the concentration of serum EBI3 was significantly decreased after surgical resection of primary tumors, and the serum EBI3 levels showed good correlation with its expression levels in primary tumor tissue in these patients. Importantly, AUC for serum EBI3 levels was significantly larger than that for CEA and CYFRA21-1, suggesting that EBI3 is a better diagnostic marker for lung cancer. When we classified all lung cancer serum samples used in this study (training and validation sets) into clinical stages of lung cancer, the sensitivity of serum EBI3 for cancer detection was 44.4% for stage I and II lung ADC, 50.0% for stage I and II lung SCC, and 35.2% for limited-stage SCLC (Supplementary Fig. S2; Supplementary Table S4). These results indicate that the sensitivity of serum EBI3 as a single-tumor marker for early stage lung cancer is likely to be better than that of CEA and CYFRA21-1, although further validation in larger scale samples with various clinical stages is necessary. Interestingly, the correlation coefficient between serum EBI3 and conventional markers was not significant, and combination assays using both EBI3 and CEA/CYFRA21-1/Pro-GRP improved diagnostic accuracy with a minimal false-positive rate. Our data presented here show a potential usefulness of EBI3 as a serologic biomarker for lung cancers. It should also be noted that we observed EBI3 activation in other types of cancers, such as pancreatic cancer (data not shown), thus suggesting it's diagnostic and therapeutic application for other tumor types.

In conclusion, we have identified EBI3 as a potential serum and tissue biomarker for diagnosis of lung cancer. This molecule is a possible candidate for the development of new anticancer drugs.

No potential conflicts of interest were disclosed.

The authors thank BioBank Japan for providing the serum samples.

This work was supported in part by Grant-in-Aid for Scientific Research (B) and Grant-in-Aid for Scientific Research on Innovative Areas from The Japan Society for the Promotion of Science. Y. Daigo is a member of Shiga Cancer Treatment Project supported by Shiga Prefecture (Japan).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.