Brachyury, a Driver of the Epithelial–Mesenchymal Transition, Is Overexpressed in Human Lung Tumors: An Opportunity for Novel Interventions against Lung Cancer

Lung cancer is the leading cause of cancer-related death worldwide (1). Depending on the type and stage of disease, current therapeutic options for lung cancer patients involve surgery, chemotherapy, radiotherapy, and more recently developed targeted therapies against, for example, the EGF receptor (EGFR; refs. 2, 3). Despite advances in therapeutic interventions, the overall 5-year survival rate for all lung cancer stages combined remains at only 16%; the rate of tumor recurrence is high even in the group of patients with early-stage lung cancer (stage I), in which the recurrence rate is approximately 50% within 5 years from the time of diagnosis (4). Complicating the outcome of chemotherapy or targeted therapies for the management of lung cancer, drug resistance mechanisms are often exhibited by the tumor cells, either as an intrinsic property or one that is acquired along with treatment (5). Ideally, a novel therapeutic intervention against lung cancer should address tumor recurrence while simultaneously minimizing tumor resistance for improved response to therapy.

The epithelial–mesenchymal transition (EMT) has recently emerged as a process of relevance during carcinoma progression and metastasis (6–8). During tumor EMT, epithelial tumor cells lose the expression of proteins involved in cell-to-cell adhesion, such as E-cadherin, and gain expression of proteins typically associated with mesenchymal cells, including Fibronectin, N-cadherin, and Vimentin (9, 10). The phenotypic switch also results in enhanced tumor cell motility and invasiveness and, as a consequence, tumor cells undergoing EMT are thought to be able to detach from the primary tumor and to initiate the cascade of events leading to the establishment of metastases (11). Several recent studies have also shown yet another interesting aspect of tumor EMT, which involves the acquisition of cancer stem-like features by the tumor cells (12), including tumor resistance to a range of therapeutic interventions (13–17).

We recently characterized the T-box transcription factor Brachyury as a driver of EMT in human carcinoma cells (18–21). Brachyury was shown to induce the expression of molecules associated with the mesenchymal phenotype, tumor cell motility, and invasiveness in vitro, as well as metastatic propensity in xenograft models of lung cancer (19). We also showed that multiple human tumor tissues and carcinoma cell lines have elevated levels of Brachyury mRNA, in contrast to most human normal tissues in which Brachyury mRNA is rarely detected (18, 19). The expression of Brachyury mRNA was also shown in primary lung tumor tissues, predominantly in tumors of higher stages (stages II–IV) than among those of stage I or histologically normal lung. In this study, we sought to characterize Brachyury as a potential target for lung cancer therapy by analyzing its protein expression levels in primary lung tumors and various human normal tissues. By using a Brachyury-specific, murine monoclonal antibody (mAb), we show, for the first time, Brachyury protein expression in human lung tumors, including adenocarcinomas and squamous cell carcinomas. In addition, genetic and epigenetic processes that may contribute to the expression of Brachyury in human tumor tissues were evaluated. It is also reported here, for the first time, that overexpression of Brachyury in human lung carcinoma lines positively correlates with resistance to EGFR kinase inhibition. Moreover, we show that Brachyury-positive lung cancer cells can be effectively lysed by Brachyury-specific CTLs, further supporting the development of Brachyury-based cancer vaccine approaches for the treatment of human lung cancer.

Thirty-nine patients with histologically diagnosed primary lung cancer were enrolled in the Interinstitutional Multidisciplinary BioBank (BioBIM) of the Department of Laboratory Medicine and Advanced Biotechnologies, IRCCS San Raffaele Pisana, Rome, Italy, in collaboration with the Surgical and Pathology Department of San Giovanni Addolorata Hospital and Medical Oncology Unit of the “Tor Vergata” Clinical Center, Rome, Italy. Lung tumor tissue samples were collected at the time of surgery (Tables 1, 2). Twenty-four histologically normal lung tissues adjacent to tumors were also obtained from lung cancer patients. No patient received neoadjuvant chemotherapy or radiation therapy previous to surgery and tissue collection. In addition, 34 samples corresponding to 11 types of normal tissues obtained from noncancer subjects have been evaluated in this study. Informed consent was obtained from each participating subject; the study was carried out under the appropriate institutional ethics approvals and in accordance with the principles embodied in the Declaration of Helsinki.

Sections of paraffin-embedded, formalin-fixed tissues were tested for Brachyury (Brachyury homolog, T) antigen expression using the avidin–biotin complex method as previously described (22). Briefly, tissue sections were deparaffinized in xylene, rehydrated in a series of graded ethanol, and treated with 0.3% H₂O₂ in methanol to block endogenous peroxidase activity. Microwave-citrate buffer antigen retrieval method was carried out to unmask the antigen. The sections were blocked in 10% horse serum (Invitrogen) for 1 hour at room temperature and then incubated overnight at 4°C with a mouse anti-Brachyury mAb (ab57480; Abcam) at a 1:100 dilution. In addition, a positive control antibody (mouse anti-Cytokeratin mAb; BD) and an isotype-matched mouse mAb (MOPC 21; Sigma-Aldrich) were used to verify accurate staining method. Antibodies specific for E-cadherin and Vimentin were purchased from BD Biosciences. Immunostaining was carried out using the Vectastaining ABC kit (Vector Laboratories) following the manufacturer's instructions; color was developed with 3,3′-diaminobenzidine (DAB) peroxidase substrate (Vector Laboratories). Sections were counterstained with hematoxylin, dehydrated in ethanol, cleared in xylene, and mounted under a coverslip using Permount (Fisher Scientific).

Two pathologists independently evaluated the tumor and normal tissue samples in a blinded, randomized way. For each slide, 3 to 5 random fields were evaluated; for each field, the percentage of DAB-positive tumor cells was calculated as: [(number of DAB-positive tumor cells/total number of tumor cells) × 100], and the relative staining intensity was scored as weak (+) for pale brown intensity, moderate (++) for intermediate brown intensity, and strong (+++) for intense, dark brown immunoprecipitate. For normal tissues, the percentage of reactivity was individually evaluated for each cell type and calculated as: [(number of DAB-positive cells/total number of cells of the same type) × 100]. Staining was recorded as negative if less than 5% of the tumor or parenchymal cells from normal tissues stained positive for Brachyury expression.

The human lung cancer cell lines used in this study were obtained from the American Type Culture Collection and propagated in RPMI-1640 medium with 2 mmol/L glutamine, 1× solution of penicillin/streptomycin (Mediatech, Inc.) and 10% FBS (Invitrogen). H460 cells were stably transfected with a control short hairpin RNA (shRNA)- or a Brachyury-specific shRNA-encoding vector; A549 cells were stably transfected with a control pCMV or a vector encoding for the full-length human Brachyury protein (pBrachyury), as previously described (19).

For detection of Brachyury in human tumor tissues, total protein extracts from frozen tissues were prepared in lysis buffer: 10 mmol/L HEPES pH = 8.0, 350 mmol/L NaCl, 0.1 mmol/L EDTA pH = 8.0, Complete Protease Inhibitor Cocktail (Roche Applied Science). For separation of cytoplasmic and nuclear protein fractions, tissues were lysed in a hypotonic buffer (10 mmol/L HEPES pH = 8.0, 50 mmol/L NaCl, 1 mmol/L EDTA pH = 8.0, 0.2% Triton X-100, 500 mmol/L Sucrose, Complete Protease Inhibitor Cocktail) followed by 30 seconds of centrifugation at 16,000 g at 4°C. Soluble, cytosolic proteins were collected; nuclear proteins were extracted in lysis buffer from the pellet fractions. Protein extracts (total, cytoplasmic, and nuclear) were purified by precipitation with chloroform/methanol, followed by resuspension in PBS/Tween 0.05% buffer. For immunoblotting, 30 μg of tissue protein extracts were separated by SDS-PAGE, blotted onto nitrocellulose membranes, and subsequently incubated with primary anti-Brachyury (Abcam) and anti-Actin (C-2, Santa Cruz Biotechnology Inc.) mAbs. A secondary horseradish peroxidase–conjugated anti-mouse Ab was subsequently used; chemiluminescent signal was detected with an ECL kit (EuroClone).

Commercially available normal and tumor lung tissue cDNA panels were analyzed (TissueScan Lung Cancer qPCR Arrays I and II; Origene Technologies, Inc.) by using the Gene Expression Master Mix and the following TaqMan Gene Expression Assays (Applied Biosystems): Brachyury (Hs00610080), Twist 1 (Hs00361186), Snail (Hs00195591), PAP (Hs00173475_m1), CEACAM5 (Hs00944023_m1), and GAPDH (4326317E). PCR was carried out according to the manufacturer's recommendations on the 7300 Real-Time PCR System (Applied Biosystems). Mean C_t values for target genes were normalized to mean C_t values for the endogenous control GAPDH [−ΔC_t = C_t (GAPDH) − C_t (target gene)]. The ratio of mRNA expression of target gene versus GAPDH was defined as 2^−ΔC_t.

In vitro invasion assays were conducted as previously described (19). Cells on the bottom side of the filters were counted in 5 random ×100 microscope objective fields.

Real-time PCR reactions were carried out with 5 ng of genomic DNA using a TaqMan Copy Number Assay (Hs01038891_cn) with the reference assay for RNase P (Applied Biosystems). Copy numbers were calculated using the CopyCaller software (Applied Biosystems).

The Brachyury promoter (–2,000 to +500) was analyzed for the presence of CpG islands and suitable restriction sites. Methylation status of an identified CpG island was evaluated by using the Promoter Methylation PCR kit (Affymetrix). Briefly, genomic DNA (3 μg) was digested with the StuI restriction enzyme, followed by purification of methylated DNA fragments by binding to the methyl CpG-binding protein 2 (MeCP2, also designated as MBP). Eluted DNA was subjected to PCR amplification using the following primers: forward 5′-TCAGGAGGGTCCGGCGTCAG-3′ and reverse 5′-GCACACCTCGCCCATTCGCT-3′. PCR products (expected size: 581 bp) were resolved on 1.5% agarose gels; the presence of a band on the gel indicates methylation of the Brachyury promoter in the genomic DNA sample.

Indicated tumor cells were seeded in 96-well plates, allowed to attach overnight, and treated with various doses (1–50 μmol/L) of AG1478 (Sigma-Aldrich) for 2 days. Viable cells in culture were assayed by the MTT assay as previously described (19). Assays were carried out in triplicate; data shown are representative averages, with SE. Survival for treated wells was calculated as a percentage of the values representing wells of untreated cells.

Brachyury-specific CTLs were generated from the blood of a normal donor (designated as CTL line I) or 2 different prostate cancer patients postvaccination with a PSA-TRICOM-based vaccine (designated as CTL lines II and III), as previously described (18). No Brachyury-specific T cells could be detected in these patients before vaccination. Tumor cells (lung H226, H441, and H1703 and pancreatic ASPC1) were labeled with 50 μCi of ¹¹¹Indium-labeled oxyquinoline and subsequently incubated with Brachyury-specific CTLs at the indicated effector-to-target ratios. For cold target inhibition assays, unlabeled K562 cells transfected with A2.1 were added to the labeled targets (10×) either unpulsed or pulsed with 20 μg/mL of the Brachyury peptide. Following a 16-hour incubation, ¹¹¹In released was measured by gamma counting and the percent specific lysis was calculated as previously described (18).

The specificity of a murine mAb directed against the human Brachyury protein was first shown by its ability to specifically recognize the 48.2-kDa recombinant His6-Brachyury protein purified from baculovirus-directed overexpression in insect cells (Fig. 1A). To determine the ability of the mAb to recognize the Brachyury protein in whole-cell protein lysates obtained from human lung carcinoma cells, a tumor cell pair was used consisting of H460 cells that were stably transfected with a control shRNA- or a Brachyury-specific shRNA-encoding vector. Evaluation of the Brachyury protein by immunoblotting with the anti-Brachyury mAb showed a single protein band of expected molecular weight in the H460 control. shRNA cells (Fig. 1B) with reduced intensity (∼70% reduction) in H460 cells stably inhibited for Brachyury expression. These results confirmed the specificity of the mAb and its ability to selectively detect Brachyury protein in human tumor cells.

Brachyury protein expression was detected in 16 of 39 (41%) of primary lung tumor tissues analyzed, including 10 of 21 adenocarcinomas (48%) and 3 of 12 (25%) squamous carcinomas (Table 1). For each of the 16 lung tumor tissue samples that were positive for Brachyury expression in this study, Table 2 indicates the tumor grade, tumor–node–metastasis stage, and the positivity observed in tumor cells as well as in the histologically normal lung tissue adjacent to the tumor, in terms of percentage of Brachyury-positive cells as well as the intensity of the staining (see Materials and Methods for a description of the scoring method used). Moreover, staining of tumor cells in the nuclear versus cytoplasmic compartment is indicated. As shown, the percentage of lung tumor cells that stained positive with the anti-Brachyury mAb ranged from 10% to 90%; in all cases, staining was localized in the nuclear compartment although cytoplasmic staining was also observed in 13 of 16 positive tissues. Although the intensity of the nuclear staining ranged from weak (+) to intense (+++), the cytosolic staining was scored as weak (+) in all but one case. Similar staining pattern has been reported for the Twist protein in the cytosol and nuclei of gastric carcinoma cells (23). Representative images of lung tumor sections positive for Brachyury protein expression, corresponding to an adenocarcinoma (patient 2, Fig. 1C–i), a squamous carcinoma (patient 12, Fig. 1C–ii), and a bronchioloalveolar carcinoma, mucinous type (patient 16, Fig. 1C–iii) showed the intense staining of Brachyury protein in the nucleus of tumor cells, and the weak, more diffuse pattern of cytoplasmic staining in the same cells. Intense Brachyury staining was also observed in the cytosol, but not the nuclei, of tumor cells invading into blood vessels (patient 10, Fig. 1C–iv). No data are available at this time with regard to the mechanism that promotes cytosolic compartmentalization of Brachyury or the biologic significance, if any, of this observation. A previous report (24) has shown that the activity of the EMT driver Snail is downregulated by phosphorylation-mediated cytosolic localization, a phenomenon that will be further investigated in the context of Brachyury.

In all 16 cases evaluated, no stromal cells stained positive for Brachyury expression in the lung tissue distal to the Brachyury-positive tumor mass. However, microscopic evaluation of the 16 cases of lung tumors that were positive for Brachyury expression showed the presence of Brachyury-positive cells in the stroma adjacent to the tumor in 7 of the 16 cases (Table 2 and Fig. 1C–v). Detection of Brachyury-positive cells in the tissue adjacent to the tumor seemed to be independent of the level of Brachyury positivity observed at the tumor site. For example, Brachyury-positive cells have been observed in the stroma adjacent to tumor from patients 2 and 3, in which 80% of the tumor cells scored strongly positive for Brachyury, as well as in patient 9 in which only 10% of the cells in the tumor were Brachyury positive (Table 2). Figure 1C–v shows a representative image of an area of transition between tumor and stroma for patient 9, showing a focal area of strong positivity for Brachyury expression in the tumor mass and in the stroma adjacent to the tumor. The anti-Brachyury mAb was also tested for its ability to detect Brachyury protein in lysates derived from human lung tumor tissues. Although no reactivity was observed with cytosolic protein fractions (data not shown), a single band at the expected molecular weight was observed in nuclear fractions derived from the primary tumor of 2 lung cancer patients evaluated (patients 7 and 17, Fig. 1D). In addition, histologically normal lung tissue adjacent to the tumor obtained from patient 17 showed intense Brachyury positivity in Western blot (Fig. 1D). The potential explanations as to why tissue directly adjacent to lung cancer cells is positive in some cases whereas distal tissue and lung tissue from noncancer patients are always negative will be discussed below. There were also 4 cases (patients 5, 7, 9, and 10) in which strong but focal staining of a minority of chondrocytes in distal bronchial cartilage was detected with the anti-Brachyury mAb (data not shown).

To evaluate whether the expression of Brachyury correlates with expression of EMT-related proteins in vivo, 4 cases of lung cancer with various levels of Brachyury were evaluated for the expression of epithelial E-cadherin and mesenchymal Vimentin by immunohistochemistry (IHC) in serial tissue sections. Two representative cases are shown in Fig. 2A. In one example (adenocarcinoma patient 2, top panels), the epithelial tumor compartment (indicated with white arrowheads) showed intense expression of Brachyury and simultaneous expression of epithelial E-cadherin and mesenchymal Vimentin. Another example of an adenocarcinoma (patient 6, bottom panels) showed expression of Brachyury in the epithelial tumor compartment (white arrowheads), with weak expression of E-cadherin and no expression of Vimentin in the tumor cells. In both cases, single disseminated cells within the adjacent stroma (indicated with black arrowheads) stained positive for Brachyury but negative for E-cadherin. These results thus indicated that the pattern of expression of Brachyury and the EMT markers, E-cadherin and Vimentin, is not necessarily consistent among various lung tumor specimens and points out at a potentially complex connection between these markers in lung cancer.

To investigate whether a genomic rearrangement could account for the enhanced expression of Brachyury in lung tumors, a qPCR-based method was used to evaluate Brachyury gene copy number variations in Brachyury-positive versus Brachyury-negative lung tumor tissues. Five of 5 Brachyury-negative and 4 of 5 Brachyury-positive tumors revealed a normal diploid status for the Brachyury gene; anomalous Brachyury gene copy number (3 copies) was only detected in one Brachyury-positive case (Supplementary Table S1). These results would rule out gene amplification as a mechanism responsible for Brachyury upregulation in lung tumors.

The methylation status of a specific CpG island on the Brachyury promoter was also investigated. As shown in Fig. 2B and Supplementary Table S1, the Brachyury promoter was methylated in 2 of 5 Brachyury-positive tumors and 1 of 5 Brachyury-negative tumors, which showed a PCR amplification product at the expected size (Fig. 2B). These results suggested that methylation of this particular region of the Brachyury promoter is unlikely to be involved in the regulation of Brachyury expression in lung cancer.

The expression of Brachyury protein was also evaluated in a set of human normal tissues obtained from adult noncancer subjects. Expression of Brachyury was negative among most normal tissues analyzed, including lung, heart, brain, liver, kidney, spleen, skeletal muscle, adrenal gland and skin, with the major exception of testis that was positive in 3 of 3 cases and normal thyroid that seemed positive in 4 of 6 cases evaluated (Supplementary Table S2). In testis, cells in the germinal epithelium of the seminiferous tubules stained weakly positive for the Brachyury protein (Fig. 3A–i). The staining was prevalently nuclear although weak cytoplasmic staining was also observed in the same cells. The expression of Brachyury protein in normal testis was in agreement with our previous studies showing Brachyury mRNA in cDNA samples prepared from human normal testis (18). Here we also report, for the first time, the expression of Brachyury protein in some biopsies of human normal thyroid tissues. As shown in Fig. 3A–ii, follicular cells stained strongly positive for Brachyury expression in the nucleus and, to a lesser extent, the cytoplasmic compartment. In contrast, as depicted in representative images in Fig. 3A, all other human normal tissues evaluated were negative for Brachyury expression in all cell types, including normal lung (Fig. 3A–iii, iv), normal spleen (Fig. 3A–v), and normal skeletal muscle (Fig. 3A–vi).

Examples of other previously characterized transcription factors able to drive EMT in various tumor models include the zinc finger protein Snail (25, 26) and the helix-loop-helix (HLH) transcription factor Twist (27, 28). Using a commercial panel of cDNAs obtained from 40 lung tumor tissues and 8 histologically normal lung tissues obtained from lung cancer patients, we comparatively evaluated the expression of mRNA encoding for Brachyury, Snail, and Twist. As shown in Fig. 3B, whereas the expression of Brachyury mRNA was selectively enhanced in lung tumor tissues (21 of 40 positive, 52.5%) versus normal lung obtained from lung cancer patients (1 of 8 positive, 12.5%), the expression of Snail or Twist mRNA was equally elevated in lung normal or tumor tissues and, in the case of Snail, the average expression among normal lung tissues was approximately 10-fold higher than that of lung tumors. Similar results were obtained at the protein level; all 3 molecules were detected in protein lysates from normal testis, used as a positive control, whereas the expression of Snail and Twist, but not Brachyury, was also detected in lysates from human normal lung (Fig. 3C).

We have also compared the expression of Brachyury mRNA in normal testis and thyroid tissues to that of 2 tumor-associated antigens that have been extensively evaluated in clinical studies without any evidence of autoimmunity. As shown in Fig. 3D, mRNA expression levels of prostatic acid phosphatase (PAP) and carcinoembryonic antigen (CEA) are higher to those of Brachyury in normal thyroid.

The expression of Brachyury mRNA was comparatively evaluated by real-time PCR in multiple human primary lung tumors and various lung carcinoma cell lines. As shown in Fig. 4A and Supplementary Table S3, 42 of 80 lung tumors (52.5%) and 7 of 10 lung cancer cell lines (70%) evaluated were positive for Brachyury mRNA. Interestingly, when comparing the levels of expression in the lung tumor samples and the various lung carcinoma lines used for experimental preclinical studies, a comparable range of Brachyury mRNA expression can be observed that spans a 4-log range, suggesting that available human tumor cell line models reflect the heterogeneity seen in patients and may be useful in developing strategies attempting to target the EMT process.

EGFR kinase inhibitors are commonly used for the management of lung cancer. We next investigated whether expression of Brachyury in lung carcinoma lines could have a significant impact on their ability to withstand EGFR inhibition mediated by Tyrphostin AG1478, a cell-permeable, reversible, ATP-competitive inhibitor of the EGFR tyrosine kinase. Two human lung carcinoma line pairs with high versus low levels of Brachyury expression were generated by stable transfection of (i) A549 cells with a control pCMV or a pBrachyury vector or (ii) H460 cells with a control shRNA versus a Brachyury shRNA. As shown in Fig. 4B, A549 pBrachyury cells had a survival advantage compared with control A549 pCMV cells in response to treatment with various doses of the EGFR inhibitor, AG1478. Similarly, H460 cells inhibited for the expression of Brachyury were found to be more susceptible to treatment with various doses of inhibitor, as compared with control cells. To further characterize the role of Brachyury on the resistance of lung carcinoma cells to EGFR inhibition, Brachyury-low A549 cells were exposed in culture to a nonlethal concentration (1 μmol/L) of AG1478. Long-term exposure of A549 to AG1478 markedly induced (>3-log) the expression of Brachyury mRNA above the level found in dimethyl sulfoxide (DMSO)-treated cells (Fig. 4C). In addition, A549 tumor cells that survived exposure to AG1478 had significantly enhanced invasive capacity in vitro, compared with DMSO-treated or untreated cells (Fig. 4D).

We have previously shown that human CD8⁺ Brachyury-specific T cells could be expanded in vitro from peripheral blood mononuclear cells of cancer patients and normal donors by using a 9-mer peptide of Brachyury that specifically binds to the HLA-A2 molecule (18). Here, various Brachyury-specific T-cell lines have been used to lyse, in vitro, H226, H441, and H1703 lung carcinoma cells. As shown in Fig. 5A, normal donor–derived Brachyury-specific T cells were able to lyse H441 tumor cells that express Brachyury and are HLA-A2 positive, but not H226 cells which express elevated levels of Brachyury but lack the appropriate MHC class I expression. Using a different Brachyury-specific T-cell line derived from a prostate cancer patient, a cold target inhibition assay with H441 tumor cells using peptide-pulsed K562-A2.1 cells as competitors showed the epitope specificity of the lysis (Fig. 5B). Using tetramer-isolated, CD8⁺ Brachyury-specific T cells derived from a different prostate cancer patient, a high level of specific lysis was observed with H441 and H1703 lung carcinoma cells (both Brachyury positive/HLA-A2 positive), as compared with the control HLA-A2–negative ASPC1 cells (Fig. 5C).

In this study, we have shown for the first time the expression of Brachyury protein, a transcriptional regulator of human carcinoma EMT, in 41% of primary lung carcinomas, including adenocarcinomas and squamous cell carcinomas, which together account for approximately 80% of all non–small cell lung cancer diagnoses. Brachyury positivity varied from 10% to 90% of the tumor cells and, in a fraction of cases, staining was also seen in scattered cells localized in the stroma adjacent (but not distant) to the tumor. These studies were carried out by IHC and confirmed by Western blot analysis using a mAb against Brachyury. We have also contrasted here for the first time the highly tumor-associated pattern of Brachyury expression, compared with Snail and Twist, 2 previously described drivers of EMT, which were expressed at equally high levels in normal and lung cancer tissues. It is also shown here, for the first time, a negative association between Brachyury expression in lung cancer cell lines and their susceptibility to EGFR inhibition. In addition, data are presented on the analysis of copy number and promoter methylation status of the Brachyury gene, in multiple lung tumor specimens.

The presence of Brachyury-positive tumor cells in human lung cancer tissues could be indicative of tumor cells undergoing EMT in vivo, which are expected to have enhanced migratory and invasive potential and greater drug resistance. It is important to emphasize, however, that the process of EMT is one of plasticity in which epithelial tumor cells can be rendered mesenchymal-like by a number of tumor environmental factors, such as interleukin (IL)-8 (21) and TGF-β (9, 29), leading to greater migration and invasion. Thus, one cannot predict when analyzing a primary tumor by IHC or PCR, which reflects the state of individual tumor cells at only one point in time, whether a subpopulation of cells had previously undergone the EMT process and had already metastasized. Our results indicated no simple association between the expression of Brachyury and the EMT markers E-cadherin and Vimentin in lung tumors. It should be pointed out that EMT is a dynamic process and that IHC depicts only one point in time of this process. The clinical relevance of Brachyury expression in lung cancer is unknown at this time and further studies need to define whether its expression has any impact on lung cancer progression or survival. In support of the role of Brachyury in human carcinoma progression, for example, the expression of Brachyury has been recently identified as a poor prognostic factor in early-stage (T1-2N0M0, Dukes A) colon cancer (30).

The identification of tumor cells undergoing EMT in vivo has been reported, for example, in colon cancer in which tumor cells with features indicative of EMT have been observed as dissociated tumor cells at the invasive front (31, 32). In this study, we have shown that Brachyury positive, single disseminated cells can be detected in the lung stroma adjacent to the tumor in a fraction of cases analyzed, although no cellular components reacted with the anti-Brachyury mAb in the normal stroma distal to the tumor. Although further studies are needed to characterize the origin of these Brachyury-positive cells in the transition tumor stroma, one possibility is that they correspond to invasive cells that detached from the tumor mass and invaded the surrounding tissue. Another possibility is that soluble mediators secreted by the tumor such as IL-8 (21) are able to induce normal cellular components of the adjacent stroma to upregulate the expression of Brachyury.

From a clinical point of view, identification of the mechanisms that control the expression of Brachyury in lung cancer could lead to improved therapeutic options for patients with Brachyury-positive tumors. Unlike with chordomas, in which duplication of the Brachyury gene is commonly observed (33), most lung tumors analyzed showed no deviations from the normal diploid status. Moreover, epigenetic control mediated by methylation of the Brachyury promoter was also ruled out, as no correlation was observed between Brachyury levels and methylation of its promoter. These results indicate that the expression of Brachyury in lung cancer might result from a diverse mutational or epigenetic mechanism, or via signaling initiated in the extracellular compartment by components of the tumor microenvironment.

It is important to point out that a previous study using a polyclonal Ab against Brachyury has reported that a variety of human carcinomas were negative for Brachyury protein expression, including lung carcinomas (34). In our study, we have used a mAb that was extensively evaluated by IHC and Western blot, showing its ability to (i) specifically react with purified Brachyury protein, (ii) react with a band at the expected molecular weight in lysates from lung carcinoma cell lines, (iii) react with a band at the expected molecular weight in lysates from lung tumor tissues, and (iv) be able to stain lung tumors by IHC. Differences in epitope specificity as well as in the avidity of the used antibodies could potentially explain the difference in results between these studies. In addition, the presence of Brachyury protein in lung tumor cell lines was shown here by using Brachyury-specific T-cell lines that were able to lyse, in an MHC-restricted manner, human lung carcinoma lines in vitro.

The present article also reinforces the tumor-associated character of Brachyury. Expression of Brachyury protein was undetectable in most human normal tissues analyzed here, with the exception of testis that was previously reported to be positive at the mRNA (18) and protein levels (34) and, for the first time, in normal thyroid tissue (4 of 6, 67%). As the testis is known to be immune privileged, it is anticipated that the expression of Brachyury protein in this organ will not pose a problem in the context of a Brachyury-based immunotherapeutic approach. With regard to the expression of Brachyury in normal thyroid tissue, we have shown here that the tumor-associated antigens PAP and CEA, which have been investigated extensively in clinical studies with no evidence of thyroid dysfunction, are expressed at higher levels in thyroid tissue compared with the expression of Brachyury mRNA.

In addition to its role as a driver of EMT, we have recently characterized another important function of Brachyury in human carcinoma cells. Brachyury expression positively correlates with the acquisition of tumor stemness features, including the ability to self-renew and to withstand treatment with various chemotherapy agents as well as radiation (Huang and colleagues; manuscript in preparation). In this study, we have shown for the first time the effect of Brachyury expression on resistance to EGFR inhibition. EGFR is overexpressed in approximately 60% of lung cancers and activation of this receptor has been shown to be associated with poor clinical outcome (35). Pharmacologic inhibition of EGFR is being explored as a therapeutic option for lung cancer patients, with 2 EGFR kinase small molecule inhibitors, erlotinib and gefitinib, currently approved for use in the clinic (35). Unfortunately, despite an initial response, most treated patients develop resistance to EGFR kinase inhibition (36). As shown here for the first time, overexpression of Brachyury in lung carcinoma cells also associates with resistance to cell death mediated by EGFR kinase inhibition. Both lung cancer lines used in our study, A549 and H460, express wild-type EGFR and harbor a K-Ras mutation; however, their sensitivity to EGFR inhibition was different. Whereas H460 cells with high Brachyury levels were resistant to EGFR inhibition with AG1478, Brachyury low A549 cells were more responsive to the same treatment. Another interesting observation was the acquisition of an EMT-like phenotype by A549 cells exposed to AG1478 for an extended period of time. In view of these data, it is possible that Brachyury-mediated EMT could be a major factor contributing to the development of resistance to EGFR kinase inhibitors.

One strategy to eliminate mesenchymal-like, Brachyury-positive tumor cells is via immune targeting with Brachyury-based cancer vaccines. The Brachyury protein is shown here to fulfill a fundamental requisite as a target for vaccine strategies against the tumor, owing to its highly tumor-associated pattern of expression. Moreover, we have shown that Brachyury-specific T cells can be expanded from the blood of cancer patients (18) and as we have shown here, those T cells can lyse, in vitro, Brachyury-positive lung carcinoma cells in an MHC-restricted manner. A phase I clinical study with a Brachyury vaccine has now been initiated in patients with carcinomas. The results presented here support the inclusion of lung cancer patients in this phase I clinical study as well as in future phase II clinical studies using Brachyury-based vaccines.

No potential conflicts of interest were disclosed.

The authors thank Margie Duberstein for technical assistance and Debra Weingarten for editorial assistance.

The work was supported by Intramural Research Program of the Center for Cancer Research, National Cancer Institute, NIH and partially supported by Italian Ministry of Instruction, University and Research (MIUR): MERIT, grant code B11J10000180008.

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.