A Search for Novel Cancer/Testis Antigens in Lung Cancer Identifies VCX/Y Genes, Expanding the Repertoire of Potential Immunotherapeutic Targets

Published OnlineFirst June 26, 2014; DOI: 10.1158/0008-5472.CAN-13-3725

Cancer Integrated Systems and Technologies Research

A Search for Novel Cancer/Testis Antigens in Lung Cancer Identiﬁes VCX/Y Genes, Expanding the Repertoire of Potential Immunotherapeutic Targets

Ayumu Taguchi1, Allen D. Taylor2, Jaime Rodriguez1,Muge€ Celikta¸s¸ 3, Hui Liu1, Xiaotu Ma4, Qing Zhang2, Chee-Hong Wong2, Alice Chin2, Luc Girard5,6, Carmen Behrens7, Wan L. Lam8, Stephen Lam8, John D. Minna5,6,9, Ignacio I. Wistuba1, Adi F. Gazdar5,10, and Samir M. Hanash3

Abstract Cancer/testis (CT) antigens are potential immunotherapeutic targets in cancer. However, the expression of particular antigens is limited to a subset of tumors of a given type. Thus, there is a need to identify antigens with complementary expression patterns for effective therapeutic intervention. In this study, we searched for genes that were distinctly expressed at a higher level in lung tumor tissue and the testes compared with other nontumor tissues and identiﬁed members of the VCX/Y gene family as novel CT antigens. VCX3A, a member of the VCX/Y gene family, was expressed at the protein level in approximately 20% of lung adenocarcinomas and 35% of squamous cell carcinomas, but not expressed in normal lung tissues. Among CT antigens with concordant mRNA and protein expression levels, four CT antigens, XAGE1, VCX, IL13RA2, and SYCE1, were expressed, alone or in combination, in about 80% of lung adenocarcinoma tumors. The CT antigen VCX/Y gene family broadens the spectrum of CT antigens expressed in lung adenocarcinomas for clinical applications. Cancer Res; 74(17); 1–12. 2014 AACR.

Introduction particular CT antigens is limited to only a subset of patients Many aberrantly expressed antigens are currently being with a particular tumor type. This is the case for members of targeted as therapeutic cancer vaccines (1, 2). Cancer/testis the MAGE family and NY-ESO-1, which are expressed in 10% – (CT) antigens are speciﬁcally expressed in immune-privi- to 50% and 10% to 30% of non smallcelllungcancers leged tissues, notably the testis and placenta, and exhibit (NSCLC), respectively (4, 5), and are currently being targeted aberrant expression in various types of cancer (3). Thus, they in clinical trials of vaccine immunotherapy (6). Moreover, are highly immunogenic, without self-tolerance, and are the effectiveness of immunotherapy targeting of MAGEA3 considered promising targets for cancer immunotherapy or NY-ESO-1 is so far limited (7, 8). ﬁ with fewer adverse effects. However, the expression of In this study, we identi ed VCX/Y genes as novel CT antigens. VCX/Y genes, mostly located on the X chromosome, may have arisen from a common ancestor to known CT antigens 1Department of Translational Molecular Pathology, The University of Texas SPANX and CSAG2 (9). VCX may be involved in mRNA stability MD Anderson Cancer Center, Houston, Texas. 2Fred Hutchinson Cancer (10). The expression levels of VCX/Y mRNAs were reduced in Research Center, Seattle, Washington. 3Department of Clinical Cancer Prevention, The University of Texas MD Anderson Cancer Center, Houston, normal, immortalized lung cell lines compared with lung Texas. 4Department of Molecular and Cell Biology, Center for Systems cancer cell lines. Reactivity against the protein product of Biology, The University of Texas Southwestern Medical Center at Dallas, the VCX3A gene was observed in approximately 20% of lung Dallas, Texas. 5Hamon Center for Therapeutic Oncology Research, The University of Texas Southwestern Medical Center at Dallas, Dallas, Texas. adenocarcinoma and 35% of squamous cell carcinoma (SCC) 6Department of Pharmacology, The University of Texas Southwestern tumors using tissue microarrays, whereas no expression was Medical Center at Dallas, Dallas, Texas. 7Department of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer detected in normal lung tissues. We further examined the Center, Houston, Texas. 8Department of Integrative Oncology, British expression patterns of multiple CT antigens in lung adeno- Columbia Cancer Research Centre, Vancouver, British Columbia, Canada. carcinomas. We found that one or more of four CT antigens 9Department of Internal Medicine, The University of Texas Southwestern Medical Center at Dallas, Dallas, Texas. 10Department of Pathology, The (XAGE1, IL13RA2, SYCE1, and VCX) were expressed in about University of Texas Southwestern Medical Center at Dallas, Dallas, Texas. 80% of lung adenocarcinoma tumors. Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Materials and Methods Corresponding Author: Ayumu Taguchi, Department of Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center, Cell culture and transfection 1515 Holcombe Boulevard, Houston, TX 77030. Phone: 713-563-8069; All lung cancer cell lines used in this study were cultured in Fax: 713-563-5746; E-mail: [email protected] RPMI1640 containing 10% FBS and 1% penicillin/streptomycin doi: 10.1158/0008-5472.CAN-13-3725 cocktail (Gibco). Human bronchial epithelial cells (HBEC30- 2014 American Association for Cancer Research. KT) were cultured with keratinocyte serum-free medium (Life

www.aacrjournals.org OF1

Taguchi et al.

Technologies, Inc.) that contained 50 mg/mL bovine pituitary when mRNA expression levels were < 0.5-fold lower than extract (Life Technologies) and 5 ng/mL EGF (Life Techno- those in adjacent normal tissues. logies). Human small airway epithelial (HSAEC1-KT) cells were cultured with Small AirwayEpithelialCellGrowth Gene expression analysis Medium (SAGM, Lonza) according to the manufacturer's Total RNA was extracted from cell lines and tissues using protocol. For the siRNA transfection experiments, two siR- the RNeasy Plus Mini Kit (Qiagen). RNA quality and concen- NAs that targeted VCX (#1: Hs_VCX_8 FlexiTube siRNA, tration were checked using the Experion Automated Electro- cat. #SI04173568, and #2: Hs_VCX_10 FlexiTube siRNA, cat. phoresis System (Bio-Rad) according to the manufacturer's #SI04187295) were obtained from Qiagen. PC-9 cells were protocol. Expression data were obtained using Illumina transfected at a final concentration of 50 nmol/L siRNA Human WG-6 v3.0 Expression BeadChips (Illumina) in the using Lipofectamine RNAiMAX (Life Technologies), accord- Genomics Core at The University of Texas Southwestern ing to the manufacturer's instructions. Seventy-two hours Medical Center at Dallas (Dallas, TX). Bead-summarized data after transfection, protein was harvested using RIPA buffer were obtained using Illumina BeadStudio software and pre- for Western blotting, and an MTS assay (CellTiter 96 Aque- processed using the R software package MBCB (Model-based ous One Solution Cell Proliferation Assay, Promega) was Background Correction for Beadarray) for background correc- performed according to the manufacturer's protocol. Three tion and probe summarization. Preprocessed data were quan- independent experiments for the MTS assay were performed tile normalized and log transformed. in triplicate. For treatment with 5-aza-20-deoxycytidine (5- Aza-dC; Sigma), cells were incubated with 2 mmol/L 5-Aza- Mass spectrometry analysis dC. Media were changed every 24 hours. After 5 days, total A mass spectrometry analysis of cell lines was performed RNA was extracted from cell lines using the RNeasy Plus as previously described (14). In brief, all cell lines were grown Mini Kit (Qiagen). in RPMI1640 (Pierce) that contained 10% of dialyzed FBS (Invitrogen), 1% penicillin and streptomycin cocktail, and Quantitative real-time reverse transcription-PCR 13C-lycine instead of regular lysine for seven passages, Complementary DNA samples were prepared using the according to the standard SILAC (stable isotope labeling High-Capacity cDNA Reverse Transcription Kit (Life Techno- by amino acids in cell culture) protocol (15). The same batch logies). A TaqMan PCR assay was performed with a 7500 Fast of cells was used to analyze the whole cell extract (WCE), Real-Time PCR System, TaqMan PCR master mix, commer- conditioned media, and cell surface proteins. WCEs, condi- VCX3A cially available primers, FAM-labeled probes for the tioned media, and cell surface proteins were fractionated by 18S gene (Hs04191277_gH), and VIC-labeled probes for reverse-phase chromatography using 2 mg, 1 mg, and 0.5 mg (Hs99999901_S1), according to the manufacturer's instruc- of total protein, respectively. Cell extracts were reduced tions (Life Technologies). Each sample was run in triplicate. and alkylated with iodoacetamidebeforechromatography. C VCX3A t values for the gene were calculated and normalized Protein digestion and identification by liquid chromatogra- C 18S DC DDC to t values for ( t). The t values were then calculated phy/tandem mass spectrometry was performed as described DC by being normalized to the t value for the control. previously (16, 17). The tandem mass spectra were searched The Cancer Genome Atlas dataset against a composite database of IPI human (v3.57) and IPI fi fi The Cancer Genome Atlas (TCGA) cancer methylomes, bovine (v3.43) proteins. A xed modi cation of 6.020129 mass units was added to lysine residues for database search- measured using Illumina Human Methylation 450 BeadChips, 13 were downloaded from the TCGA website, where the methyl- ing to account for incorporation of the C-lycine isotope. fi ation levels of 485577 CpG sites are represented as a b value of 0 To estimate the signi cance of the peptide and protein to 1 (b ¼ 1 for fully methylated sites and b ¼ 0 for fully matches, we used the tools PeptideProphet and ProteinPro- unmethylated sites). RNA-seq data of TCGA cancer samples phet. We selected all peptides that had been initially iden- fi were also downloaded. The normalized count from RSEM ti ed with a PeptideProphet probability minimum of 0.05 (11) for each gene was extracted from files with the extension and not shared between humans and bovines and submitted "rsem.genes.normalized_results." Between-sample quantile them to ProteinProphet. Protein Prophet infers a minimal normalization was performed using the R software package set of proteins that can explain the peptide evidence, assign- preprocessCore. The samples were classified as tumors (pri- ing a probability to each protein on the basis of the com- fi mary solid tumor, code: 01) or nontumors (solid tissue normal, bined peptide probabilities. The derived protein identi ca- fi code: 11) according to the sample naming conventions spec- tions were ltered at a maximum 5% error. The total number ified in the code tables of the TCGA and corresponding of spectral counts for each protein group in ProteinProphet barcodes. was used in the semiquantitative analysis. Each dataset was normalized to the total number of spectral counts of each Tissue samples compartment. Eighty-three pairs of treatment-na€ve lung adenocarcinoma tissues and adjacent nontumor lung tissues had been Tissue microarrays obtained with informed consent (12, 13). EGFR and KRAS The tissue microarrays used in this study comprised 463 mutations were screened as previously described (13). surgically resected NSCLC tumor specimens (305 adenocarci- Tumor tissues were considered TITF1/NKX2-1 negative nomas and 158 SCCs) collected under an institutional review

OF2 Cancer Res; 74(17) September 1, 2014 Cancer Research

Cancer/Testis Antigen VCX/Y for Lung Cancer Immunotherapy

board–approved protocol and archived as formalin-fixed, par- mine Ig-based immunoreactivity in plasmas from 96 female affin-embedded specimens in The University of Texas Special- subjects that included 12 and 17 subjects subsequently diag- ized Program of Research Excellence thoracic tissue bank at nosed with lung and colon adenocarcinoma, respectively. The University of Texas MD Anderson Cancer Center (Hous- One control was matched to colon adenocarcinoma cases and ton, TX). Tissue microarrays were prepared with a manual 2 to 5 controls were matched to individual lung adenocarci- tissue arrayer (Advanced Tissue Arrayer ATA100, Chemicon noma cases. The array study was performed as part of discov- International) using 1 mm diameter cores in triplicate for ery studies of biomarker candidates for the early detection tumors, as described previously (18). Histologic sections that of epithelial tumor types (19, 20). Plasma samples were ran- were 4 mm in thickness were then prepared for the subsequent domly allocated to the arrays within batches to eliminate immunohistochemical analysis. potential bias, and were each hybridized with an individual protein microarray at a dilution of 1:500 in probe buffer Western blot analysis and immunohistochemical (Synthetic Block in 150 mmol/L PBS, 0.1% Tween-20,) for analysis 1.5 hours at 4C. IgG reactivity was assessed quantitatively Monoclonal anti-mouse antibody raised against human using an indirect immunofluorescence protocol, and local VCX3A (Abnova) was used for the immunohistochemical and background-subtracted median spot intensities for down- Western blot analyses. b-Tubulin (Cell Signaling Technology) stream statistical analysis were generated using GenePix was used as a loading control for the Western blot analysis. Pro 6.1. Background-subtracted median spot intensities from Tissue microarray slides and lung tissue sections were immu- individual slides were quantile normalized, and individual nostained in a Dako Autostainer Plus automated machine. The clones were normalized to the mean and SD of the normal slides were deparaffinized and hydrated, and antigen retrieval samples before analysis. was performed using a decloaker with a Dako pH 6.0 target retrieval solution (Dako North America, Inc.). Intrinsic perox- idase activity was blocked using 3% methanol and hydrogen Results peroxide for 10 minutes, and serum-free protein block (Dako) Identification of potential CT antigens in lung cancer was used for 5 minutes to block nonspecific antibody binding. To discover potential CT antigens in lung cancer, we iden- Slides were then incubated with antibodies against human tified genes that were overexpressed in the testis compared VCX3A (1:1,000 dilution) for 30 minutes at ambient temper- with normal tissues using BioGPS (http://biogps.org/; ature. After being washed three times in Tris-buffered saline, GSE1133; ref. 21; Fig. 1A for the workflow). We focused on CT slides were incubated for 30 minutes with Dako Envision þ antigens on the X chromosome (CT-X antigens) because their Dual Link at ambient temperature. They were then washed gene expression is more testis-restricted than that of non-X CT three more times, incubated with Dako chromogen substrate antigens (22). We compared the mRNA expression profiles for 5 minutes, and counterstained with hematoxylin. Formalin- of 1,013 probes for 632 genes on the X chromosome between fixed, paraffin-embedded whole-section specimens, except for 5 testis and 47 other normal tissues, excluding the placenta, the omission of the primary antibodies, were used as negative ovary and neuronal tissue, as CT antigens may be expressed controls. For the immunostaining scoring, two board-certified in those tissues (22). Forty-five probes that corresponded to surgical pathologists participated in the analysis of this TMA 37 genes had more than 2-fold higher mRNA expression levels (H. Liu and J. Rodriguez). The scoring system evaluates the in testis than in nontestis tissues (P < 0.05, Mann–Whitney intensity of staining (0: negative, 1þ: weakly positive, 2þ: U test; Supplementary Table S1). A subset of 23 of these genes moderately positive, and 3þ: strongly positive) and percentage had not been previously recognized as CT antigens (CTData- of staining distribution in the tumor cells (from 0 to 100% of the base; http://www.cta.lncc.br/; ref. 23). cells) for each TMA core. Three TMA cores from the same Four genes, BEX1, NXF3, TCEAL3,andVCX2,werecon- patient were evaluated and the percentage of staining distri- sidered plausible candidates for encoding CT antigens bution was averaged to provide the final scoring. In this study, because their family members were predominantly localized tumors with 20% or more of the cells having 2þ or 3þ staining to the X chromosome, which is a typical characteristic of intensity were considered as positive. Before the TMA scoring, CT-X antigens (Supplementary Table S1; ref. 3). To deter- the two pathologists evaluated together at a double head mine whether these genes are expressed in lung cancer, we microscope the immunostaining pattern for VCX3A in control profiled the mRNA expression of four gene families (BEX, slides and in the TMA slides to agree on the criteria for intensity NXF, TCEAL,andVCX/Y) in 113 NSCLC (78 adenocarcinoma, and distribution. Then one of the pathologists (H. Liu) per- 3 adenosquamous cell carcinoma, 10 SCC, 12 large cell formed the scoring of the TMA slides. Finally, the two pathol- carcinoma, and 10 other NSCLCs), 29 SCLC, 2 carcinoid, 4 ogists (H. Liu and J. Rodriguez) met again to review the TMA mesothelioma, and 59 normal lung immortalized cell lines scoring together in the double head microscope and agreed on (30 normal bronchial epithelial cells and 29 normal small any question or discrepancy about the evaluation. airway epithelial cells). VCX/Y family genes were coexpressed at a higher level in a subset of NSCLC and SCLC cell lines Serological immunoreactivity against arrayed VCX3A thaninnormalbronchialepithelialcelllinesornormalsmall protein airway epithelial cell lines (Fig. 1B). VCX, VCX3A,andVCX3B A recombinant protein microarray (Version 5 Protoarrays, were predominantly expressed compared with VCX2 and Invitrogen) containing VCX3A protein was utilized to deter- VCY. BEX1, BEX2,andBEX4,andBEX5 were highly expressed

www.aacrjournals.org Cancer Res; 74(17) September 1, 2014 OF3

Taguchi et al.

A VCX VCX2 VCX3A VCX3B VCY VCX VCX2 VCX3A VCX3B VCY VCX VCX2 VCX3A VCX3B VCY B Adenocarcinoma Adenosquamous Carcinoid A549 5.18 4.56 5.83 4.73 4.28 H125 6.93 4.60 6.83 6.95 5.82 H720 4.82 4.64 5.68 5.23 4.75 Calu-3 4.95 4.84 5.42 4.78 4.70 H596 5.98 4.87 6.34 5.75 5.66 H727 5.54 4.78 6.70 6.13 5.35 Calu-6 8.49 5.10 8.51 8.30 4.96 H647 7.48 5.66 7.49 7.49 5.07 DFCI024 5.95 5.53 6.38 6.46 5.00 44,775 probes DFCI032 5.21 4.82 5.30 4.82 4.90 Methothelioma EKVX 6.19 5.07 6.56 6.54 5.44 SCC H2052 5.18 4.79 6.23 6.13 5.73 H1355 5.17 4.96 5.93 5.91 4.94 H157 5.05 4.92 6.07 4.89 5.18 H226 4.51 4.68 5.96 5.45 4.91 H1373 4.80 4.80 5.60 5.18 5.12 H1869 4.99 4.28 5.60 4.57 4.67 H28 4.75 4.92 5.24 5.89 5.21 H1395 7.07 5.77 7.26 7.36 4.84 H2170 6.65 5.60 6.96 7.23 5.53 H290 6.36 6.20 6.81 6.55 5.62 H1437 5.00 4.87 6.07 5.18 4.90 H520 7.13 4.37 7.09 7.13 5.20 On the X chromosome? H1563 5.85 5.15 6.24 5.59 4.97 HCC1313 8.03 5.85 8.08 8.28 5.38 H1568 6.64 5.20 6.84 6.88 4.58 HCC15 4.57 5.14 5.89 4.45 5.14 Bronchial epithelial cell H1573 8.55 6.95 8.71 8.52 6.04 HCC1897 5.39 4.53 6.28 5.70 4.66 HBEC2-KT 4.56 4.85 5.17 4.90 5.09 Yes H1648 6.25 5.10 6.68 6.41 5.39 HCC2450 6.18 4.72 6.75 6.44 5.56 HBEC3-KT 4.49 5.29 5.79 4.85 5.19 H1650 6.02 4.69 6.93 5.85 4.99 HCC2814 9.47 7.64 9.50 9.68 6.05 HBEC4-KT 4.84 4.70 5.65 6.13 4.92 H1651 6.42 4.73 6.64 7.08 4.87 HCC95 4.63 4.79 6.01 5.25 5.09 HBEC5-KT 4.95 4.83 5.88 4.32 4.95 H1666 5.49 5.34 6.06 5.78 4.95 HBEC6-KT 5.04 4.93 5.51 5.41 5.04 1,013 probes (632 genes) H1693 5.02 4.98 6.00 5.29 4.81 HBEC7-K 5.10 5.13 5.88 5.52 5.78 H1703 4.76 4.66 6.02 5.23 4.95 Large cell carcinoma HBEC10-K 5.15 4.96 5.45 4.73 5.00 H1734 4.78 4.96 5.76 5.43 5.23 H1155 13.21 11.41 13.38 13.32 9.02 HBEC11-KT 4.94 4.98 5.60 4.98 5.05 H1755 9.60 8.28 9.44 9.89 5.59 H1299 4.70 4.62 5.67 5.19 5.46 HBEC12-KT 5.31 4.37 5.26 5.44 4.91 H1781 5.63 5.21 6.34 6.01 5.08 H1570 5.94 4.70 6.20 5.94 4.94 HBEC13-KT 5.06 5.16 5.51 5.31 5.18 >2 higher mRNA levels in testis H1792 6.21 5.08 6.63 6.19 4.79 H2106 5.89 4.57 6.21 6.21 5.39 HBEC14-KT 5.29 4.64 4.83 5.63 5.13 H1793 5.33 5.33 6.19 5.80 5.15 H460 5.08 5.21 6.31 5.48 5.89 HBEC15-KT 5.30 4.69 5.86 5.38 4.81 compared with nontestis tissue? H1819 5.33 4.52 5.64 5.41 4.85 H661 11.16 10.35 11.36 11.40 10.12 HBEC16-KT 5.12 5.10 5.82 4.67 5.16 Yes H1838 5.09 4.82 5.78 5.37 5.55 HCC1359 5.21 4.96 6.05 5.06 5.12 HBEC17-KT 4.85 4.56 5.57 5.00 5.47 H1944 13.13 11.94 13.14 13.50 10.07 HCC1438 6.07 5.46 6.66 6.39 4.81 HBEC18-KT 4.50 5.22 5.64 4.44 5.03 H1975 5.24 5.04 5.63 5.24 5.12 HCC2374 5.67 5.37 6.06 6.05 5.44 HBEC24-KT 5.10 4.70 5.85 5.17 4.69 H1993 5.46 4.73 5.03 5.51 5.08 HCC3051 7.03 5.68 7.08 7.12 5.19 HBEC26-KT 5.02 4.62 5.61 4.73 4.41 H2009 7.32 5.44 7.96 7.55 5.60 HCC4017 5.17 4.74 5.88 4.95 4.79 HBEC27-KT 5.29 5.06 5.44 4.83 5.00 45 probes (37 genes) H2023 7.11 5.70 7.50 7.81 5.98 Hop-92 5.19 4.81 6.59 4.62 4.71 HBEC28-KT 5.44 4.76 5.67 4.92 4.76 H2030 4.68 4.61 5.75 5.86 5.06 HBEC29-KT 5.13 4.91 5.62 5.13 5.13 H2073 5.20 4.39 5.60 5.43 5.03 HBEC30-K 4.98 4.70 5.17 4.76 4.98 H2077 4.41 5.01 6.17 5.17 5.49 NSCLC HBEC31-K 4.60 4.34 5.34 5.04 4.68 H2085 4.76 4.63 5.56 5.79 5.10 Calu-1 5.90 4.67 6.65 6.25 4.88 HBEC32-KT 5.38 4.97 5.51 4.70 5.07 H2086 5.07 4.44 5.86 5.23 4.98 H1385 4.44 5.16 5.83 4.92 5.24 HBEC33-KT 4.39 4.89 5.06 5.32 5.19 Not known as cancer/testis antigen? H2087 4.62 4.97 5.62 5.11 5.04 H1770 6.30 5.10 6.16 6.42 5.47 HBEC34-KT 4.89 4.57 5.41 5.20 4.63 H2122 5.08 5.24 4.73 5.37 5.72 H2172 8.53 6.85 8.65 8.57 5.08 HBEC35-KT 5.15 4.74 5.26 5.26 4.89 Yes H2126 10.53 8.96 10.66 10.87 7.53 H2882 8.21 6.53 8.24 8.49 5.52 HBEC36-KT 5.06 5.03 5.52 5.34 4.68 H2228 4.82 4.43 5.71 5.18 5.38 H2887 6.48 4.36 6.49 6.53 5.03 HBEC37-KT 5.18 5.30 5.53 5.45 5.53 H2250 4.38 4.43 5.83 5.15 5.08 H292 4.75 4.80 5.58 4.29 5.15 HBEC38-KT 4.67 4.67 5.27 4.67 5.14 H2258 4.67 4.24 5.85 5.56 5.54 H3122 4.92 4.34 5.34 4.92 4.78 HBEC39-KT 5.18 4.96 5.33 5.33 6.00 H2291 4.79 4.73 5.43 5.09 5.07 HCC1171 5.80 4.78 6.20 6.38 5.78 23 genes H23 6.32 4.43 6.87 6.65 5.34 HCC2429 6.42 5.38 6.73 6.58 4.76 H2342 4.81 4.57 5.75 4.69 4.86 Small epithelial cell H2347 5.10 4.96 6.05 4.65 5.31 HSAEC1-K 5.05 4.21 5.62 5.51 4.67 H2405 4.68 5.01 5.54 5.24 4.99 SCLC HSAEC1-U 4.80 4.42 5.53 4.91 4.94 H322 5.44 5.15 6.23 5.75 5.18 H1184 5.96 4.62 5.93 5.99 5.05 HSAEC2-KT 5.40 4.61 5.78 5.06 4.61 Multiple family genes located H324 6.71 5.78 6.71 6.74 5.58 H1238 6.08 4.99 6.53 6.14 4.95 HSAEC2-UI 5.52 4.78 5.62 5.77 5.01 H3255 5.70 4.99 6.18 5.75 4.95 H128 5.32 4.60 5.86 5.25 5.01 HSAEC13-KT 5.12 4.76 5.29 5.56 4.60 on the X chromosome? H358 8.90 7.38 8.95 8.83 5.96 H146 6.31 4.59 6.94 6.85 4.71 HSAEC13-UI 4.94 5.34 5.40 5.01 4.88 Yes H441 4.95 4.72 4.84 5.35 5.73 H1514 5.31 4.70 6.02 5.83 5.66 HSAEC15-KT 4.64 4.76 5.62 5.27 5.59 H522 4.60 4.40 4.73 5.55 5.29 H1607 4.55 4.84 5.94 4.64 5.20 HSAEC15-UI 4.95 4.36 6.22 6.26 4.91 H650 8.52 6.49 8.27 8.64 6.00 H1672 10.26 5.27 10.18 10.52 6.50 HSAEC18-KT 4.76 4.58 6.07 5.01 4.44 H820 5.63 4.43 5.68 5.01 4.64 H187 5.01 4.92 5.75 5.65 5.31 HSAEC19-KT 4.51 4.59 5.45 5.13 4.78 4 genes H838 5.74 4.53 6.22 6.02 5.01 H209 5.51 8.79 6.27 5.20 7.05 HSAEC22-KT 5.24 4.46 5.47 4.76 5.12 H920 5.62 4.89 6.36 5.74 5.19 H2107 6.64 4.69 6.63 6.48 4.73 HSAEC22-UI 4.41 4.76 6.07 5.21 4.70 H969 8.95 8.09 8.88 9.04 7.78 H211 5.72 4.97 6.41 6.44 5.16 HSAEC24-KT 4.80 4.45 5.71 4.84 4.72 (BEX1, NXF3, TCEAL3, and VCX2) HCC1195 6.16 4.78 6.65 6.27 4.85 H2141 4.92 4.33 6.00 5.73 4.87 HSAEC24-UI 5.01 4.77 5.19 5.19 4.84 HCC1833 4.87 5.17 5.55 5.79 4.54 H2171 5.62 4.90 6.03 5.28 4.85 HSAEC30-KT 5.10 5.07 6.05 5.41 6.00 HCC193 7.92 5.35 8.25 8.17 5.27 H2195 5.24 4.41 5.94 5.29 4.99 HSAEC30-UI 5.11 4.74 5.47 5.21 4.64 HCC2108 5.27 5.54 6.02 5.32 5.19 H2227 9.49 6.96 9.63 9.69 6.58 HSAEC31-KT 5.45 5.15 5.53 4.90 4.92 HCC2279 4.94 4.99 5.85 4.85 5.12 H289 4.90 4.56 5.67 5.29 5.09 HSAEC31-UI 5.27 4.58 5.79 5.07 5.02 HCC2935 5.37 4.56 5.82 5.56 5.12 H345 5.33 4.58 5.95 4.90 4.40 HSAEC35-KT 4.29 4.88 5.59 5.09 4.96 HCC364 4.78 4.63 5.85 4.52 5.49 H378 5.87 4.94 6.30 6.12 4.94 HSAEC35-UI 4.74 5.07 5.23 4.94 4.90 HCC366 6.25 4.61 6.53 6.33 4.96 H510 12.90 11.43 13.03 13.00 9.84 HSAEC36-KT 4.72 4.41 5.41 5.15 5.09 HCC4006 5.35 4.60 5.67 4.76 5.13 H524 5.43 4.63 6.05 6.42 4.81 HSAEC36-UI 5.37 5.90 5.24 5.39 5.71 HCC4011 5.10 5.42 5.39 4.66 4.79 H526 6.52 4.63 6.71 6.30 5.53 HSAEC37-KT 4.78 4.46 5.74 5.21 5.26 Investigation of mRNA profiles of 4 HCC4018 6.57 6.11 6.95 6.95 5.53 H69 5.38 4.79 5.68 5.61 4.86 HSAEC37-UI 4.34 4.36 6.09 5.26 4.76 HCC4019 4.34 5.00 5.22 4.96 4.93 H748 7.26 14.02 10.96 13.16 11.60 HSAEC38-UI 4.74 5.71 5.83 4.94 5.44 gene families (BEX, NXF, TCEAL, and HCC44 4.92 5.00 6.09 5.10 5.00 H82 7.07 5.10 7.41 7.41 5.08 HSAEC45-KT 4.74 4.68 5.88 5.21 5.14 HCC461 5.23 5.10 5.13 4.90 5.13 H841 5.20 4.81 5.77 5.75 4.84 HSAEC45-UI 5.08 5.32 5.61 5.88 4.96 VCX) in 148 lung cancer and 59 normal HCC515 4.56 5.18 5.28 5.31 4.76 H865 6.20 4.89 6.57 6.24 5.01 HSAEC49-KT 4.88 4.76 5.25 4.93 5.20 HCC78 5.29 4.83 6.03 4.66 4.89 H889 8.82 8.48 8.83 8.98 6.71 HSAEC49-UI 5.42 4.84 5.80 5.42 4.75 HCC827 4.88 4.62 5.29 5.14 5.29 HCC33 5.68 4.72 5.75 5.32 5.01 lung immortalized cell lines HOP-62 4.51 4.65 5.66 4.74 4.88 HCC970 6.19 5.06 6.75 6.35 5.16 4 14 PC-9 14.03 12.89 14.18 14.24 9.96 5.01 6.00 7.03 8.03 9.02 9.96 ### ### ### ### SK-LU-1 6.08 5.20 5.94 5.30 5.24 (log intensity) 2

Figure 1. Identiﬁcation of VCX/Y genes as novel CT antigens in lung cancer. A, workﬂow for selecting novel CT antigen candidates in BioGPS. B, heatmap for gene expression levels of VCX/Y family genes in 148 lung cancer and 59 normal lung immortalized cell lines: adenocarcinoma (n ¼ 78), adenosquamous cell carcinoma (adenosquamous; n ¼ 3), SCC (n ¼ 10), large cell carcinoma (n ¼ 12), other NSCLC (n ¼ 10), SCLC (n ¼ 29), carcinoid (n ¼ 2), mesothelioma (n ¼ 4), bronchial epithelial (n ¼ 30), and small airway epithelial (n ¼ 29).

in some NSCLC cell lines and most SCLC cell lines but were (Supplementary Fig. S2A). Methylation data were available also expressed in normal cell lines, except for BEX5 (Sup- for 4, 1, 3, and 2 CpG sites in VCX, VCX2, VCX3A,andVCX3B plementary Fig. S1). These data suggested that BEX5 is a respectively, for lung adenocarcinomas (406 tumor and 32 potential CT antigen in lung cancer, particularly SCLC. Two nontumor tissues) and lung squamous cell carcinomas (252 members of the NXF gene family, NXF2 and NXF2B,are tumor and 43 nontumor tissues). The methylation levels of known CT antigens (CT39), and mRNA expression levels of all CpG sites were significantly lower in cancer tissues the other NXF family genes were not significantly different compared with nontumor tissues (Fig. 2A and Supplemen- between cancer and nontumor cell lines. Of the TCEAL tary Fig. S2B). We further evaluated the correlation between family genes, TCEAL2 was strongly expressed in most SCLC mRNA and methylation levels of VCX/Y family genes in cell lines, compared with nontumor cell lines indicating that cancer tissues. mRNA levels of VCX/Y genes were also it is a potential cancer testis antigen in SCLC, together with obtained from the TCGA RNA-seq data (326 adenocarcino- BEX5. mas and 217 SCCs for VCX; 377 adenocarcinomas and 248 The expression of CT-X genes is associated with promoter SCCs for VCX2; and 378 adenocarcinomas and 248 SCCs for demethylation and more broadly with genomic hypomethy- VCX3A and VCX3B). mRNA levels were significantly and lation (3, 6). Thus, we determined whether mRNA expression inversely correlated with methylation for all VCX/Y genes levels of VCX/Y genes were associated with hypomethylation. in lung adenocarcinomas (VCX, Spearman correlation coef- The methylation status of the 5 CpG sites in VCX,3inVCX2, ficient ¼0.22, P < 0.0001; VCX2, Spearman correlation 4inVCX3A,and5inVCX3B was analyzed in TCGA data coefficient ¼0.16, P ¼ 0.0025; VCX3A, Spearman

OF4 Cancer Res; 74(17) September 1, 2014 Cancer Research

Cancer/Testis Antigen VCX/Y for Lung Cancer Immunotherapy

A B 12 AdenoCa SCC AdenoCa VCX C 10 SCC

TN TN 8 15 1.0 6 0.8 4 10 0.6 2 Normalized count 0.4 0 P < 0.0001 P < 0.0001 0.0 0.2 0.4 0.6 0.8 1.0 0.2 Methylation (cg05524154) 5 12 VCX (cg05524154) relative mRNA 0.0 AdenoCa VCX2 2 1.0 10 SCC

8 Log 0 0.8 –+ – + – + – + – + – + – + – + 5-Aza-dC

6 VCX3A/18S of expression 0.6 4 H647 0.4 2 H1792 H1299 H1975 H3255

Normalized count HCC827 P < 0.0001 P < 0.0001 DFCI024 HCC4019 0.2 0 VCX2 (cg07380282) 0.0 0.2 0.4 0.6 0.8 1.0 Normal 0.0 Methylation (cg07380282) DE 1.0 12 AdenoCa VCX3A 0.8 10 SCC (kDa) siControlsiVCX #1siVCX #2 8 0.6 PC-9 HSAEC1-KTHBEC30-KT 37 VCX3A VCX3A 6 0.4 b-Tubulin P < 0.0001 P < 0.0001 4 b-Tubulin 50 0.2 2

VCX3A (cg13510648) Normalized count 0.0 0 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1.0 Methylation (cg13510648) F 0.8 12 0.8 AdenoCa VCX3B 0.6 10 SCC 0.6

0.4 8 0.4 P < 0.0001 P < 0.0001 6 0.2 0.2 VCX3B (cg07898500) 4

0.0 Absorbance (490 nm) 0.0 2 Normalized count Mock 0 0.0 0.2 0.4 0.6 0.8 1.0 siControl siVCX #1 siVCX #2 Methylation (cg07898500)

Figure 2. Regulatory mechanism and functional relevance of VCX/Y in NSCLC. A, methylation levels of representative CpG sites for VCX/Y genes. P values were calculated using the Mann–Whitney U test. N, nontumor tissue; T, tumor tissue. B, correlation of methylation levels and mRNA expression levels in tumor tissue. C, induction of VCX3A mRNA with 5-Aza-dC treatment in eight NSCLC cell lines. D, Western blot analysis of VCX3A protein in PC-9 and normal immortalized lung cell lines. b-Tubulin served as a loading control. E, knockdown of VCX with siRNA in PC-9 cells was verified by Western blotting. b-Tubulin served as a loading control. F, MTS assay of PC-9 cells treated with negative control siRNA or siRNAs against VCX. Columns, the average of triplicate samples from a representative experiment; bars, SD. correlation coefficient ¼0.21, P < 0.0001; and VCX3B, was inhibited in PC-9 cells by treatment with VCX siRNA Spearman correlation coefficient ¼0.18, P ¼ 0.0004), (Fig. 2E and F), indicating an oncogenic role for VCX/Y family whereas the correlation was not significant in SCC, except genes in NSCLC. for VCX3A (VCX, Spearman correlation coefficient ¼0.03, P ¼ 0.6983; VCX2,Spearmancorrelationcoefficient ¼ 0.02, P ¼ Tissue microarray-based analysis of VCX3A protein 0.8127; VCX3A, Spearman correlation coefficient ¼0.16, P expression in NSCLC ¼ 0.0099; and VCX3B, Spearman correlation coefficient ¼ Next, we determined VCX3A protein expression in NSCLC 0.07, P ¼ 0.2598; Fig. 2B), suggesting that VCX/Y gene tumors using a tissue microarray. VCX3A protein was strongly expression is epigenetically regulated in lung adenocarcino- expressed in normal testis (data not shown) but was negative ma and that additional regulatory mechanisms for VCX/Y or very weakly positive in normal lung tissue (Fig. 3A). VCX3A gene expression may occur in SCC that are not associated expression was predominantly found in the nucleus (Fig. 3B). A with the CpG sites tested. In addition, VCX3A gene expres- tumor was considered positive if 20% or more of tumor cells sion levels were determined in 8 NSCLC cell lines after were moderately or strongly stained. While nontumor adjacent treatment with the DNA demethylating agent 5-Aza-dC. tissues from 20 adenocarcinomas and 8 SCCs in the tissue VCX3A gene expression was clearly induced with 5-Aza-dC microarray were not stained with VCX3A (data not shown), 118 treatment in all 8 NSCLC cell lines, further supporting (25.5%) of 463 tumors were positively stained, unrelated to sex, the epigenetic regulation of VCX/Y gene expression (Fig. age (> or 65 years), smoking status, stage, or mutational 2C). VCX3A protein expression was confirmed by Western status for EGFR and KRAS. Reactivity was significantly higher in blotting in PC-9 cells, in which VCX/Y genes were highly SCC than in adenocarcinoma (37.3% and 19.3%, respectively; P overexpressed (Figs. 1B and 2D). Of note, cell proliferation < 0.0001, Fisher exact test; Table 1).

www.aacrjournals.org Cancer Res; 74(17) September 1, 2014 OF5

Taguchi et al.

Normal lung (alveolar) Normal lung (bronchus) A

B Negative Weakly positive Figure 3. VCX3A protein expression in NSCLC tissue microarray. A, immunohistochemical staining of VCX3A in normal lung (alveolar and bronchus) tissues. Scale bars, 200 mm. B, assessment of VCX3A expression levels in the nucleus of NSCLC tumor tissues. Scale bars, 200 mm.

Moderately positive Strongly positive

CT antigen expression in NSCLC cell lines at least one cell line, and 83, 69, and 83 CT antigens were Given that no single CT antigen is widely expressed in identified in WCE, media, and the cell surface, respectively NSCLC, we assessed whether a combination of CT antigens (Fig. 4A and Supplementary Table S3), consisting of 47 exhibited complementary expression and would thus in (40.9%) CT-X antigens and 68 (59.1%) non-X CT antigens. principle, provide a more effective approach to immuno- One hundred nine CT antigens were profiled at both the therapy than would single antigens. We first determined CT mRNA and protein levels. The mRNA expression levels of 10 antigen expression at both the mRNA and protein levels in CT antigens (CT45A5, GAGE7, IL13RA2, MAGEA10, 38 NSCLC cell lines, given the potential for discordance MAGEA4, MAGEB2, NY-ESO-1, PAGE5, SYCE1, and XAGE1) between RNA and protein expression (24). Of 255 known were significantly and positively correlated with protein ex- CT antigens in the CTDatabase (23), 182 genes were profiled pression levels in at least one compartment (P < 0.05, in38NSCLCcelllinesusingIlluminaWG-6v3.0gene Spearman correlation; Fig. 4B). In addition, mRNA levels expression microarrays (Supplementary Table S2), including were significantly higher in cell lines with protein expression 87 (47.5%) CT-X antigens and 96 (52.5%) non-X CT antigens. of these CT antigens compared with cell lines without Next, we determined the protein expression levels of CT proteinexpressionoftheseCTantigens(P < 0.05, Mann– antigens in the same NSCLC cell line set with quantitative Whitney U test). Protein expression levels were more widely in-depth proteomic profiling of three compartments: WCE, distributed than were mRNA levels, suggesting translational conditioned media, and cell surface proteins (14, 17). One regulation of CT antigens. All CT antigens, except for SYCE1, hundred fifteen (45.1%) of 255 CT antigens were identified in with significant correlation between mRNA and protein

OF6 Cancer Res; 74(17) September 1, 2014 Cancer Research

Cancer/Testis Antigen VCX/Y for Lung Cancer Immunotherapy

Table 1. VCX3A expression in NSCLC tissue microarray

VCX3A immunostaining

Positive, Negative, Characteristic Total n (%) n (%) Pc Sexa Male 210 56 (26.7) 154 (73.3) 0.4335 Female 216 50 (23.1) 166 (76.9) Agea >65 years 246 61 (24.8) 185 (75.2) 1.0000 65 years 180 45 (25.0) 135 (75.0) Smoking statusa Current 176 41 (23.3) 135 (76.7) 0.5784 Former 203 55 (27.1) 148 (72.9) Never 47 10 (21.3) 37 (78.7) Histologic type Adenocarcinoma 305 59 (19.3) 246 (80.7) <0.0001 SCC 158 59 (37.3) 99 (62.7) Stagea I 219 48 (21.9) 171 (78.1) 0.2125 II 101 33 (32.7) 68 (67.3) III 87 20 (23.0) 67 (77.0) IV 19 5 (26.3) 14 (73.7) Mutationb KRAS 51 13 (25.5) 38 (74.5) 0.4901 EGFR 22 5 (22.7) 17 (77.3) KRAS and EGFR wild 103 18 (17.5) 85 (82.5)

aInformation was not available for 37 samples. bMutations were not investigated in 287 samples. cP values were calculated by Fisher exact test or c2 test.

levels were CT-X antigens. Peptides encoded in the VCX mediate and low clusters, VCX, IL13RA2,orSYCE1 was highly gene family (VCX, VCX2, VCX3A, and VCX3B) are largely expressed in most tumors. Thus, we selected XAGE1, VCX, indistinguishable and were identified by mass spectrome- IL13RA2,andSYCE1 to form a CT antigen panel and com- try in the WCE of PC-9 cells (Fig. 4C). VCX/Y proteins are pared the mRNA expression levels of these four genes in highly homologous and have a paucity of arginine and tumor and adjacent nontumor tissues. The number of lysine residues in their sequences, resulting in few detectable tumors with CT antigen overexpression (more than 2-fold tryptic peptides for identification by mass spectrometry. A higher mRNA expression levels in cases vs. controls) was search for VCX peptides among other cell lines only yielded 55 (66.3%) for XAGE1, 26 (31.3%) for VCX, 10 (12.0%) for identificationofVCX3AinSW620colonadenocarcinoma IL13RA2, and 9 (10.8%) for SYCE1 (Fig. 5A, bottom cell line. heatmap; Table 2). One or more of the CT antigens in the panel were overexpressed in 66 (79.5%) lung adenocarcino- mRNA expression profiles of CT antigens in lung cancer mas. XAGE1 overexpression was significantly associated tissues with TITF1/NKX2-1–positive lung adenocarcinomas (P ¼ We determined the mRNA expression profiles of 10 CT 0.0054, Fisher exact test; Table 2 and Supplementary Table antigens with concordant mRNA and protein expression S4). While overexpression of VCX and SYCE1 was not asso- levels, together with VCX as a representative gene for the ciated with the adenocarcinoma patients' clinical character- VCX/Y family genes and MAGEA3, which is a current target istics, IL13RA2 overexpression was significantly less com- for vaccine immunotherapy. We evaluated 83 lung adeno- mon in current smokers (P ¼ 0.0436, c2 test). carcinomas and adjacent nontumor tissues (Supplementary The CT antigen panel was more broadly expressed than Table S4). Unsupervised hierarchical clustering of gene were NY-ESO-1 and MAGEA3 (Fig. 5A, bottom heatmap), expression in tumors revealed distinct CT antigen expres- which are currently being studied in clinical trials as vaccine sion patterns. Interestingly, lung adenocarcinomas yielded immunotherapy. Therefore, we determined the immunoge- threeclustersonthebasisofXAGE1 mRNA levels: high, nicity of the four CT antigens. We used two epitope pre- intermediate, and low (Fig. 5A, top heatmap). In the inter- diction algorithms, BIMAS (http://www-bimas.cit.nih.gov/;

www.aacrjournals.org Cancer Res; 74(17) September 1, 2014 OF7

Taguchi et al.

A C Surface Media (83) (69) SSSQPSPSDPK

VCX 19 8 7 23221 602

VCX2 42 23221 931 14 12 VCX3A 23221 681 15 VCX3B 1 2322 612

WCE (83)

Identified Not identified B Protein mRNA mRNA (normalized MS2 counts) (log intensity) (log intensity) Spearman 2 2 Gene Compartment correlation P value Mean ± SD Cell lines Mean ± SD Cell lines Mean ± SD P value Media 0.47 0.0031 2.7 ± 0.9 3 9.1 ± 0.1 35 4.6 ± 1.0 0.0002 CT45A5 Surface 0.35 0.0337 5.6 ± 4.6 2 8.5 ± 0.8 36 4.8 ± 1.4 0.0242 WCE 0.39 0.0163 4.1 ± 1.8 2 9.2 ± 0.1 36 4.7 ± 1.2 0.0028 Media 0.47 0.0032 24.9 ± 12.9 3 14.5 ± 0.1 35 5.5 ± 1.9 0.0002 GAGE7 Surface 0.37 0.0218 8.1 ± 7.1 2 14.4 ± 0.1 36 5.8 ± 2.4 0.0071 WCE 0.57 0.0002 20.3 ± 15.9 5 12.6 ± 2.4 33 5.3 ± 1.7 <0.0001 Surface 0.47 0.0032 18.1 ± 4.4 3 12.5 ± 0.3 35 5.3 ± 2.0 IL13RA2 0.0001 WCE 0.36 0.0286 4.3 ± 2.2 2 12.2 ± 0.4 36 5.5 ± 2.3 0.0156 Media 0.55 0.0004 9.3 ± 6.8 8 10.0 ± 1.4 30 4.7 ± 1.0 0.0009 MAGEA10 Surface 0.38 0.0172 4.8 ± 2.5 10 8.9 ± 3.7 28 4.7 ± 1.4 0.0093 WCE 0.60 <0.0001 7.4 ± 3.5 6 11.8 ± 0.4 32 4.7 ± 1.5 <0.0001 Media 0.74 <0.0001 44.0 ± 32.8 9 11.9 ± 0.7 29 4.5 ± 1.0 <0.0001 MAGEA4 Surface 0.69 <0.0001 22.4 ± 22.6 17 8.7 ± 3.6 21 4.3 ± 0.3 0.0001 WCE 0.77 <0.0001 36.4 ± 27.6 10 11.6 ± 0.4 28 4.4 ± 0.3 <0.0001 Surface 0.39 0.0163 3.0 ± 2.1 2 11.7 ± 0.4 36 4.9 ± 1.3 0.0028 MAGEB2 WCE 0.38 0.0197 1.4 ± 0.9 2 11.2 ± 1.0 36 4.7 ± 1.0 0.0057 Surface 0.49 0.0018 3.4 ± 1.9 5 9.1 ± 2.4 33 4.7 ± 1.2 0.0016 NY-ESO-1 WCE 0.37 0.0241 3.1 ± 2.2 4 8.7 ± 2.6 34 4.9 ± 1.5 0.0296 Media 0.47 0.003 16.7 ± 17.7 3 12.2 ± 4.6 35 4.6 ± 0.7 0.0002 PAGE5 WCE 0.39 0.0163 24.5 ± 6.7 2 12.2 ± 4.8 36 4.8 ± 1.4 0.0014 SYCE1 WCE 0.39 0.0165 2.8 ± 0.3 2 9.8 ± 0.2 36 5.1 ± 0.7 0.0028 Surface 0.50 0.0015 1.9 ± 1.4 6 13.8 ± 0.3 32 8.7 ± 4.1 0.0009 XAGE1 WCE 0.60 <0.0001 3.2 ± 4.0 9 13.8 ± 0.1 29 8.1 ± 4.0 < 0.0001

Figure 4. Transcriptomic and proteomic profiling of CT antigens in 38 NSCLC cell lines. A, Venn diagrams of CT antigens identified by mass spectrometric analysis in WCE, cell surface (surface), and conditioned media (media) of 38 NSCLC cell lines. B, CT antigens with significant correlation of expression levels between mRNA and protein. P values for comparison of mRNA expression levels in cell lines with or without CT antigen protein expression were calculated by Mann–Whitney U test. C, schema of human VCX/Y family proteins. Black bars indicate peptides identified in the PC-9 cell line. Presented numbers are based on NP_038480.2 (VCX), NP_057462.2 (VCX2), NP_057463.2 (VCX3A), and NP_001001888.2 (VCX3B).

ref. 25) and SYFPEITHI (http://www.syfpeithi.de/; ref. 26). 152; QESEVEEPL at 116, 136, and 156; EPLSQESQV at 122, Five peptides that consisted of nine amino acid residues for 142, 162, and 182; and QESQVEEPL at 126, 146, and 166. HLA XAGE1, 22 for VCX, 31 for IL13RA2, and 20 for SYCE1 were binding of EPLSQESEV and EPLSQESQV was predicted to be predicted to bind HLA molecules, with high scores, in both HLA-B51:01 restricted, and HLA binding of QESEVEEPL algorithms ( 100 in BIMAS and 20 in SYFPEITHI; Fig. 5B; and QESQVEEPL was predicted to be HLA-B40:01 restricted Supplementary Table S5). Interestingly, VCX contains iden- (Supplementary Table S5). Thus, these repeating amino acid tical immunogenic 9-mer peptides in different locations: sequences may amplify the immune response against VCX in EPLSQESEV at starting amino acid positions 112, 132, and patients with HLA-B51:01 or HLA-B40:01. As part of prior

OF8 Cancer Res; 74(17) September 1, 2014 Cancer Research

Cancer/Testis Antigen VCX/Y for Lung Cancer Immunotherapy

Table 2. Association of CT antigen overexpression status and clinical characteristics of 83 lung adenocarcinoma patients

All XAGE1 VCX IL13RA2 SYCE1 Clinical characteristics Total Yes No Pc Yes No Pc Yes No Pc Yes No Pc Yes No Pc Sex Male 24 20 4 0.7665 18 6 0.3183 9 15 0.4466 0 24 0.0562 2 22 1 Female 59 46 13 37 22 17 42 10 49 7 52 Age, y >65 54 46 8 0.0943 38 16 1 19 35 0.333 6 48 0.7339 8 46 0.1514 65 29 20 9 17 12 7 22 4 25 1 28 Smoking status Current 39 30 9 0.7805 30 9 0.1542 12 27 0.9459 1 38 0.0436 3 36 0.6823 Former 14 12 2 8 6 4 10 3 11 2 12 Never 30 24 6 17 13 10 20 6 24 4 26 Mutationa KRAS 33 26 7 0.6638 22 11 0.9907 10 23 0.915 3 30 0.6745 5 28 0.3392 EGFR 20 17 3 13 7 7 13 3 17 2 18 KRAS, EGFR wild 27 20 7 18 9 8 19 2 25 1 26 TITF1/NKX2-1 expression Positive 68 57 11 0.0706 50 18 0.0054 22 46 0.7665 8 60 1 8 60 1 Negative 15 9 6 5 10 4 11 2 13 1 14 Stageb I 36 30 6 0.7154 25 11 0.3337 13 23 0.7603 4 32 1 7 29 0.0816 II–IV 18 14 4 13 5 5 13 2 16 0 18

aMutations were not studied in three patients. bStaging information was not available for 29 patients. cP values were calculated by Fisher exact test or c2 test. studies that were aimed at identifying tumor antigens asso- cies.net/; ref. 27; Supplementary Table S6). For African-Amer- ciated with autoantibodies in epithelial tumors types, we icans, the sum of the allele frequency of HLA-A molecules was conducted discovery studies using recombinant and natural 36.1% and that of HLA-B was 20.1%; and for Caucasians, it was protein microarrays (19, 20). Given our discovery of novel CT 65.8% for HLA-A and 37.6% for HLA-B. antigens, we examined their association with an autoantibody response using protein microarrays. VCX3A, which we Discussion have found to be overexpressed in both lung and colon We identified the VCX/Y gene family as encoding novel CT cancer cell lines, was represented in recombinant protein antigens initially by mining gene expression data. These genes microarrays. We assessed its Ig-based reactivity in prediag- were initially identified as testis specific and were associated nostic sera from 29 plasmas from subjects diagnosed with with X-linked nonspecific mental retardation (28, 29). They lung or colon adenocarcinoma and 29 pools of matched exhibited in our study several characteristics of CT-X antigens: controls. Serum Ig autoantibody reactivity against VCX3A testis-specific expression, a multigene family on the X chro- was significantly higher among cases compared with con- mosome, overexpression in a subset of NSCLC cell lines and trols (P ¼ 0.020, Mann–Whitney U test) and a receiver tumor tissues, as confirmed by a tissue microarray analysis, no operating characteristic (ROC) analysis yielded an area overexpression in nontumor lung cell lines or tissues, and under the curve (AUC) of 0.678, suggesting the occurrence coexpression with family genes (3, 6). Our findings also suggest of an immune response against VCX3A in a subset of epigenetic regulation of VCX/Y gene expression, which is subjects with lung and colon adenocarcinoma and potential another typical characteristic of CT-X antigens. There is also usefulness of VCX3A autoantibodies as part of a diagnostic independent evidence that VCX gene expression is epigenet- biomarker panel (Fig. 5C). ically regulated (30). Moreover, the results of a recent genomic Because antigen presentation occurs in an HLA-restricted study suggested that VCX/Y and the known CT antigens manner, we determined the allelic distribution of HLA mole- SPANX and CSAG2 are derived from a common ancestor and cules with potential binding to our CT antigen panel, which that VCX/Y and SPANX are paralogs with a similar protein yielded 17 HLA class I molecules. The allelic frequencies of 13 structure that consists of C-terminal acidic repeats of variable HLA class I molecules were available for two U.S. populations lengths (9). To our knowledge, the relevance of VCX/Y to in the Allele*Frequencies database (http://www.allelefrequen- cancer has not been previously investigated. VCX/Y proteins

www.aacrjournals.org Cancer Res; 74(17) September 1, 2014 OF9

Taguchi et al.

Figure 5. CT antigen panel in 83 lung adenocarcinoma patients. A, an unsupervised hierarchical clustering of 10 selected CT antigens, MAGEA3, and VCX. Black boxes in the bottom heatmap indicate lung adenocarcinoma patients with CT antigen overexpression (more than 2-fold higher mRNA expression levels in cases vs. controls). B, potential epitopes of XAGE1, VCX, IL13RA2, and SYCE1. Black bars indicate 9-mer peptides with high scores in both BIMAS and SYFPEITHI. C, ROC curves for autoantibodies to VCX3A in prediagnostic plasmas of lung and colon adenocarcinoma patients (n ¼ 29) and matched controls (n ¼ 29) from the WHI observational study. P values were calculated using the Mann–Whitney U test.

are localized in the nucleus and may be involved in regulating identified by mass spectrometry in our study were common in ribosome assembly during spermatogenesis (31), inhibiting various XAGE1 isoforms. However, consistent with the findings mRNA decapping and regulating mRNA stability (10). In our of a previous study (33), XAGE1b-unique peptides were most study, VCX siRNA inhibited cell proliferation in PC-9 cells, commonly identified among XAGE1 isoform-unique peptides suggesting that VCX/Y genes have oncogenic roles. Therefore, in 38 NSCLC cell lines (data not shown). Five 9-mer XAGE1 further functional studies are warranted. peptides were identified as highly immunogenic on the basis of An integrative mRNA analysis and proteomic profiling of two distinct epitope prediction algorithms, BIMAS and SYF- 38 NSCLC cell lines revealed that mRNA and protein expres- PEITHI, and several additional immunogenic peptides have sion levels were significantly correlated with each other for been identified (34). 10 CT antigens. We analyzed 83 lung adenocarcinomas IL13RA2 is known to be overexpressed in solid tumors, samples to determine the gene expression profiles of these particularly glioblastoma multiforme. Although IL13RA2 has 10 CT antigens, VCX,andMAGEA3, which is currently being been associated with inflammation in lung epithelial cells, targeted in clinical trials. Our analysis resulted in a panel of its role in lung cancer has not been investigated (35). Okada four CT antigens (XAGE1, VCX, IL13RA2,andSYCE1); one or and colleagues recently reported the results of a phase I/II more of these was overexpressed in 80% of lung adenocar- trial using a-type 1 polarized dendritic cells loaded with cinomas, highlighting the potential of this combination for synthetic peptides for glioma-associated antigen epitopes, immunotherapy in lung cancer. including EphA2, IL13 receptor-a2, YKL-40, gp100, and XAGE1 expression and an associated autoantibody response polyinosinic-polycytidylic acid for HLA-A2(þ) patients with have been observed in NSCLCs (32), and several XAGE1 recurrent malignant glioma. Of 22 patients in the study, two transcript variants have been identified. Most XAGE1 peptides experienced objective clinical tumor regression. One patient

OF10 Cancer Res; 74(17) September 1, 2014 Cancer Research

Cancer/Testis Antigen VCX/Y for Lung Cancer Immunotherapy

with recurrent glioblastoma experienced a complete Authors' Contributions response (36). Conception and design: A. Taguchi, A.D. Taylor, A.F. Gazdar, S.M. Hanash Development of methodology: A. Taguchi, A.D. Taylor, M. Celiktas SYCE1 is a constituent of the synaptonemal complex Acquisition of data (provided animals, acquired and managed during meiosis (37); however, its role in cancer has not been patients, provided facilities, etc.): A. Taguchi, A.D. Taylor, J. Rodriguez, M.Celiktas,C.-H.Wong,A.Chin,L.Girard,W.L.Lam,S.Lam,I.I.Wistuba, well studied. Approximately 20% of lung adenocarcinomas A.F. Gazdar did not exhibit expression of the CT antigens in our panel Analysis and interpretation of data (e.g., statistical analysis, biostatistics, (Fig. 5A). XAGE1 (38) and IL13RA2 (39) are known to be computational analysis): A. Taguchi, A.D. Taylor, J. Rodriguez, H. Liu, X. Ma, Q. Zhang, C.-H. Wong, C. Behrens, A.F. Gazdar, S.M. Hanash induced by 5-Aza-dC. Our data indicate that similar epige- Writing, review, and/or revision of the manuscript: A. Taguchi, A.D. Taylor, netic regulation occurs in VCX/Y genes. Thus, for patients J.D. Minna, S. Lam, I.I. Wistuba, A.F. Gazdar, S.M. Hanash whose tumors do not express CT antigens, a combination of Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Taguchi, A. Chin, W.L. Lam, CT antigen-targeted therapy and DNA demethylating agents I.I. Wistuba, S.M. Hanash may be promising (6). Study supervision: A. Taguchi, S.M. Hanash In conclusion, we found that VCX/Y family genes are novel CT antigens in lung cancer on the basis of an analysis of gene fi Grant Support expression data; we further con rmed VCX3A expression at This work was supported by the Canary Foundation, the Lungevity Foun- the protein level in a tissue microarray analysis. Importantly, dation, the National Cancer Institute Early Detection Research Network, The we identified a panel of CT antigens, consisting of XAGE1, VCX, Department of Defense W81XWH-09-LCRP-CTRA, Lung Cancer SPORE (P50CA70907; J. Minna, A. Gazdar, L. Girard, I. Wistuba, C. Behrens, and IL13RA2, and SYCE1, which were expressed in most lung J. Rodriguez), and MD Anderson's Moon Shots Program. adenocarcinomas. This finding suggests that targeting this The costs of publication of this article were defrayed in part by the four-antigen panel will be a promising immunotherapy 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 approach for lung adenocarcinoma. fact.

Disclosure of Potential Conﬂicts of Interest Received December 28, 2013; revised April 29, 2014; accepted May 19, 2014; No potential conﬂicts of interest were disclosed. published OnlineFirst June 26, 2014.

References 1. Schlom J. Therapeutic cancer vaccines: current status and moving carcinoma and integration with mRNA expression. Genome Res forward. J Natl Cancer Inst 2012;104:599–613. 2012;22:1197–211. 2. Dougan M, Dranoff G. Immune therapy for cancer. Annu Rev Immunol 13. Thu KL, Vucic EA, Chari R, Zhang W, Lockwood WW, English JC, et al. 2009;27:83–117. Lung adenocarcinoma of never smokers and smokers harbor differ- 3. Simpson AJ, Caballero OL, Jungbluth A, Chen YT, Old LJ. Cancer/ ential regions of genetic alteration and exhibit different levels of testis antigens, gametogenesis and cancer. Nat Rev Cancer 2005; genomic instability. PLoS ONE 2012;7:e33003. 5:615–25. 14. Taguchi A, Politi K, Pitteri SJ, Lockwood WW, Faca VM, Kelly-Spratt K, 4. Suri A. Cancer testis antigens–their importance in immunotherapy et al. Lung cancer signatures in plasma based on proteome profiling of and in the early detection of cancer. Expert Opin Biol Ther 2006; mouse tumor models. Cancer Cell 2011;20:289–99. 6:379–89. 15. Ong SE, Mann M. A practical recipe for stable isotope labeling by 5. Chiriva-Internati M, Pandey A, Saba R, Kim M, Saadeh C, Lukman T, amino acids in cell culture (SILAC). Nat Protoc 2006;1:2650–60. et al. Cancer testis antigens: a novel target in lung cancer. Int Rev 16. Faca V, Pitteri SJ, Newcomb L, Glukhova V, Phanstiel D, Krasno- Immunol 2012;31:321–43. selsky A, et al. Contribution of protein fractionation to depth of 6. Fratta E, Coral S, Covre A, Parisi G, Colizzi F, Danielli R, et al. The analysis of the serum and plasma proteomes. J Proteome Res biology of cancer testis antigens: putative function, regulation and 2007;6:3558–65. therapeutic potential. Mol Oncol 2011;5:164–82. 17. Faca VM, Ventura AP, Fitzgibbon MP, Pereira-Faca SR, Pitteri SJ, 7. Vansteenkiste J, Zielinski M, Linder A, Dahabreh J, Gonzalez EE, Green AE, et al. Proteomic analysis of ovarian cancer cells reveals Malinowski W, et al. Adjuvant MAGE-A3 immunotherapy in resected dynamic processes of protein secretion and shedding of extra-cellular non-small-cell lung cancer: phase II randomized study results. J Clin domains. PLoS ONE 2008;3:e2425. Oncol 2013;31:2396–403. 18. Sun M, Behrens C, Feng L, Ozburn N, Tang X, Yin G, et al. HER 8. Kakimi K, Isobe M, Uenaka A, Wada H, Sato E, Doki Y, et al. family receptor abnormalities in lung cancer brain metastases A phase I study of vaccination with NY-ESO-1f peptide mixed and corresponding primary tumors. Clin Cancer Res 2009;15: with Picibanil OK-432 and Montanide ISA-51 in patients with .4829–37 cancers expressing the NY-ESO-1 antigen. Int J Cancer 2011;129: 19. Ladd JJ, Chao T, Johnson MM, Qiu J, Chin A, Israel R, et al. Autoan- 2836–46. tibody signatures involving glycolysis and splicesome proteins pre- 9. Hansen MA, Nielsen JE, Retelska D, Larsen N, Leffers H. A shared cede a diagnosis of breast cancer among postmenopausal women. promoter region suggests a common ancestor for the human VCX/Y, Cancer Res 2013;73:1502–13. SPANX, and CSAG gene families and the murine CYPT family. Mol 20. Ladd JJ, Busald T, Johnson MM, Zhang Q, Pitteri SJ, Wang H, et al. Reprod Dev 2008;75:219–29. Increased plasma levels of the APC-interacting protein MAPRE1, 10. Jiao X, Wang Z, Kiledjian M. Identification of an mRNA-decapping LRG1, and IGFBP2 preceding a diagnosis of colorectal cancer in regulator implicated in X-linked mental retardation. Mol Cell 2006; women. Cancer Prev Res 2012;5:655–64. 24:713–22. 21. Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D, et al. A gene 11. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA- atlas of the mouse and human protein-encoding transcriptomes. Proc Seq data with or without a reference genome. BMC Bioinformatics Natl Acad Sci U S A 2004;101:6062–7. 2011;12:323. 22. Hofmann O, Caballero OL, Stevenson BJ, Chen YT, Cohen T, Chua R, 12. Selamat SA, Chung BS, Girard L, Zhang W, Zhang Y, Campan M, et al. Genome-wide analysis of cancer/testis gene expression. Proc et al. Genome-scale analysis of DNA methylation in lung adeno- Natl Acad Sci U S A 2008;105:20422–7.

www.aacrjournals.org Cancer Res; 74(17) September 1, 2014 OF11

Taguchi et al.

23. Almeida LG, Sakabe NJ, deOliveira AR, Silva MC, Mundstein AS, involvement in regulation of ribosome assembly during spermatogen- Cohen T, et al. CTdatabase: a knowledge-base of high-throughput esis. Cell Res 2003;13:171–7. and curated data on cancer-testis antigens. Nucleic Acids Res 32. Nakagawa K, Noguchi Y, Uenaka A, Sato S, Okumura H, Tanaka M, 2009;37:D816–9. et al. XAGE-1 expression in non-small cell lung cancer and antibody 24. Schwanhausser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, response in patients. Clin Cancer Res 2005;11:5496–503. et al. Global quantification of mammalian gene expression control. 33. Sato S, Noguchi Y, Ohara N, Uenaka A, Shimono M, Nakagawa K, et al. Nature 2011;473:337–42. Identification of XAGE-1 isoforms: predominant expression of XAGE- 25. Parker KC, Bednarek MA, Coligan JE. Scheme for ranking potential 1b in testis and tumors. Cancer Immun 2007;7:5. HLA-A2 binding peptides based on independent binding of individual 34. Ohue Y, Eikawa S, Okazaki N, Mizote Y, Isobe M, Uenaka A, et al. peptide side-chains. J Immunol 1994;152:163–75. Spontaneous antibody, and CD4 and CD8 T-cell responses against 26. Rammensee H, Bachmann J, Emmerich NP, Bachor OA, Stevanovic S. XAGE-1b (GAGED2a) in non-small cell lung cancer patients. Int J SYFPEITHI: database for MHC ligands and peptide motifs. Immuno- Cancer 2012;131:E649–58. genetics 1999;50:213–9. 35. Nakashima H, Husain SR, Puri RK. IL-13 receptor-directed cancer 27. Gonzalez-Galarza FF, Christmas S, Middleton D, Jones AR. Allele vaccines and immunotherapy. Immunotherapy 2012;4:443–51. frequency net: a database and online repository for immune gene 36. Okada H, Kalinski P, Ueda R, Hoji A, Kohanbash G, Donegan TE, et al. frequencies in worldwide populations. Nucleic Acids Res 2011;39: Induction of CD8þ T-cell responses against novel glioma-associated D913–9. antigen peptides and clinical activity by vaccinations with {alpha}-type 28. Fukami M, Kirsch S, Schiller S, Richter A, Benes V, Franco B, et al. A 1 polarized dendritic cells and polyinosinic-polycytidylic acid stabi- member of a gene family on Xp22.3, VCX-A, is deleted in patients with lized by lysine and carboxymethylcellulose in patients with recurrent X-linked nonspecific mental retardation. Am J Hum Genet 2000;67: malignant glioma. J Clin Oncol 2011;29:330–6. 563–73. 37. Costa Y, Speed R, Ollinger R, Alsheimer M, Semple CA, Gautier P, et al. 29. Lahn BT, Page DC. A human sex-chromosomal gene family expressed Two novel proteins recruited by synaptonemal complex protein 1 in male germ cells and encoding variably charged proteins. Hum Mol (SYCP1) are at the centre of meiosis. J Cell Sci 2005;118:2755–62. Genet 2000;9:311–9. 38. Lim JH, Kim SP, Gabrielson E, Park YB, Park JW, Kwon TK. Activation of 30. Fukutomi S, Seki N, Koda K, Miyazaki M. Identification of methyl- human cancer/testis antigen gene, XAGE-1, in tumor cells is correlated ation-silenced genes in colorectal cancer cell lines: genomic with CpG island hypomethylation. Int J Cancer 2005;116:200–6. screening using oligonucleotide arrays. Scand J Gastroenterol 39. Takenouchi M, Hirai S, Sakuragi N, Yagita H, Hamada H, Kato K. 2007;42:1486–94. Epigenetic modulation enhances the therapeutic effect of anti-IL-13R 31. Zou SW, Zhang JC, Zhang XD, Miao SY, Zong SD, Sheng Q, et al. (alpha)2 antibody in human mesothelioma xenografts. Clin Cancer Res Expression and localization of VCX/Y proteins and their possible 2011;17:2819–29.

OF12 Cancer Res; 74(17) September 1, 2014 Cancer Research

A Search for Novel Cancer/Testis Antigens in Lung Cancer Identifies VCX/Y Genes, Expanding the Repertoire of Potential Immunotherapeutic Targets

Ayumu Taguchi, Allen D. Taylor, Jaime Rodriguez, et al.

Cancer Res Published OnlineFirst June 26, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-13-3725

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2014/06/27/0008-5472.CAN-13-3725.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2014/08/16/0008-5472.CAN-13-3725. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.