Glycodelin: a New Biomarker with Immunomodulatory Functions in Non–Small Cell Lung Cancer Marc A

Published OnlineFirst April 21, 2015; DOI: 10.1158/1078-0432.CCR-14-2464

Biology of Human Tumors Clinical Cancer Research Glycodelin: A New Biomarker with Immunomodulatory Functions in Non–Small Cell Lung Cancer Marc A. Schneider1,2, Martin Granzow3, Arne Warth2,4, Philipp A. Schnabel5, Michael Thomas2,6, Felix J.F. Herth2,7, Hendrik Dienemann2,8, Thomas Muley1,2, and Michael Meister1,2

Abstract

Purpose: In recent years, immune therapeutic strategies Results: Glycodelin mRNA expression was significantly elevat- against non–smallcelllungcancer(NSCLC)basedontis- ed in tumors (n ¼ 336) compared with matched normal tissue sue-derived biomarkers, for example PD1/PD-L1 (CD274), (P < 0.0001). Overall survival (OS) was significantly reduced in have evolved as novel and promising treatment options. NSCLC with high glycodelin mRNA levels in women but not in However, the crosstalk between tumor and immune cells is men. Glycodelin was detected in the sera of patients, and the levels poorly understood. Glycodelin (gene name PAEP), initially correlated with recurrence and metastatic disease. Knockdown of described in the context of pregnancy and trophoblastic glycodelin with siRNAs in NSCLC cell lines resulted in significant implantation, is a secreted immunosuppressive glycopro- upregulation of immune system modulatory factors such as tein with an as-of-yet largely unknown function in lung PDL1, CXCL5, CXCL16, MICA/B, and CD83 as well as prolifer- cancer. ation stimulators EDN1 and HBEGF. Furthermore, decreased Experimental Design: In this study, we characterized migration of tumor cells was observed. the expression and role of glycodelin in NSCLC through Conclusions: Altogether, the comprehensive characterization mRNA and protein expression analyses, functional knock- of glycodelin in NSCLC provides strong support for its use as a down experiments, and correlations with clinicopathologic biomarker with immune modulatory function. Clin Cancer Res; parameters. 21(15); 3529–40. 2015 AACR.

Introduction To improve the poor survival of lung cancer patients, chemo- therapeutic treatments are progressively supplemented with pre- With approximately 1.6 million cases worldwide, lung cancer is cision therapy based on genetic analyses. Sequencing of lung the leading cause of cancer-related death (1). Historically, lung carcinoma demonstrated a high somatic mutation rate of TP53, cancer is divided into two groups: non–small cell lung cancer KRAS, and EGFR (3). First-line and second-line tyrosine kinase (NSCLC), which accounts for more than 80% of all cases, and inhibitors that target the EGFR or anaplastic lymphoma kinase small-cell lung cancer (SCLC; ref. 2). (ALK) improve progression-free survival (4–6). Surface markers and secreted chemokines of altered cells attract lymphocytes, natural killer cells (NK) or cytotoxic T cells. The tumor microenvironment is strongly defined by extracellular 1Translational Research Unit, Thoraxklinik at University Hospital Hei- matrix (ECM) reorganization and chemokine secretion of tumor delberg, Heidelberg, Germany. 2Translational Lung Research Center cells that leads to a barrier for lymphocytes (7). Infiltration of Heidelberg (TLRC-H), Member of the German Center for Lung NSCLCs with lymphocytes is associated with better disease-free Research (DZL), Heidelberg, Germany. 3Institute of Human Genetics, University of Heidelberg, Heidelberg, Germany. 4Institute of Patholo- survival (8). Thus, immune therapy has the potential to stimulate gy, University of Heidelberg, Heidelberg, Germany. 5Institut fur€ Allge- the patient's immune system response and attack cancer cells meine und Spezielle Pathologie, University of the Saarland, Homburg/ more effectively. Manipulation of NK cells and natural killer T Saar, Germany. 6Department of Thoracic Oncology, Thoraxklinik at University Hospital Heidelberg, Heidelberg, Germany. 7Department of cells (NKT) is considered to be a useful tool for cancer immuno- Pneumology and Critical Care Medicine, Thoraxklinik at University therapy (9, 10). 8 Hospital Heidelberg, Heidelberg, Germany. Department of Surgery, Glycodelin (gene name PAEP), initially described as placental Thoraxklinik at University Hospital Heidelberg, Heidelberg, Germany. protein 14 (PP14) or progesterone-associated endometrial pro- Note: Supplementary data for this article are available at Clinical Cancer tein (PAEP), is strongly associated with trophoblastic invasiveness Research Online (http://clincancerres.aacrjournals.org/). (11). Downregulation of glycodelin leads to an increased activa- Corresponding Author: Michael Meister, Thoraxklinik at University Hospital Hei- tion of the maternal immune system and can result in abortion delberg, Amalienstrasse 5, Heidelberg 69126, Germany. Phone: 49-6221-396-1650; during the first trimester (12). Glycodelin is differentially glyco- Fax: 49-6221-396-1652; E-mail: [email protected] sylated depending on gender and function (13). Glycodelin A, the doi: 10.1158/1078-0432.CCR-14-2464 immunosuppressive form of glycodelin, functions as a prolifer- 2015 American Association for Cancer Research. ation-suppressor and apoptosis inducer of T cells, monocytes, B

www.aacrjournals.org 3529

Schneider et al.

trol siRNA-transfected and PAEP siRNA pool-transfected samples Translational Relevance per cell line. Detailed descriptions are provided in the Supple- Glycodelin is well described as an immune system mod- mentary Materials and Methods. ulator during pregnancy. In this work, we demonstrate the – expression and secretion of glycodelin in non small cell lung Quantitative Real-Time PCR cancer (NSCLC), its potential role as an immune system Real-time quantitative PCR (qPCR) was performed in accor- modulator, and the feasibility of using glycodelin as a bio- dance with MIQE-guidelines (19) using a LightCycler 480 Real- PAEP marker for NSCLC. Glycodelin mRNA (gene name ) was Time PCR Instrument in a 384-well plate format (Roche). Gene- overexpressed in approximately 80% of all tumors compared specific primers and probes (Universal ProbeLibrary; Roche) were PAEP with normal tissue. Knockdown of resulted in a dereg- used in combination with ABsolute QPCR Mix (Thermo Scien- ulation of immune system modulators, for example, PD-L1, a tific). Ct values were calculated with LightCycler 480 software current immune therapeutic target. We demonstrate the detec- version 1.5 using the second derivative maximum method tion of glycodelin in tissues and sera of NSCLC patients. The (Roche). Detailed descriptions are provided in the Supplementary fact that the glycodelin serum levels correlated with the Materials and Methods. patients' response to treatment suggests its use as a biomarker for follow-up of NSCLC. A diagnostic tool for glycodelin quantification in pregnancy is commercially available and Immunohistochemistry might be conveniently for use in lung cancer diagnoses. To compare the expression of whole glycodelin and glycodelin A, the polyclonal N-20 antibody (sc-12289, Santa Cruz Biotech- nology) and the monoclonal A87-B/D2 (Glycotope) antibody, raised against glycodelin A purified from mid-trimester amniotic fluid, were used. Detailed descriptions are provided in the Sup- cells and NK cells (14). Atypical glycodelin expression is observed plementary Materials and Methods. in hormone-related breast cancer, endometrial cancer, and ovarian cancer (15). Recent studies have demonstrated glycodelin Detection of glycodelin in human sera expression in melanoma cells and lung cancer (16, 17). Gene- Human sera were collected before any disease-specific treat- PAEP expression data from our group revealed that is expressed in ment and stored at 80C within 2 hours after venipuncture. The a lung cancer cell line as well (18). NSCLC validation cohort included pretherapeutic sera of 159 To elucidate the role of glycodelin in NSCLC, we comprehen- randomly selected NSCLC patients (clinical stage I–IV). The sively analyzed its expression in resection specimens as well as its chronic obstructive pulmonary disease (COPD) control cohort function in cell culture-based experiments. Furthermore, expres- included 90 randomly selected sera of COPD I–IV patients. The sion data were correlated with clinical data, including patient glycodelin levels of the sera were detected using an enzyme-linked outcomes. We demonstrate that glycodelin is highly expressed in immunosorbent assay kit (ELISA BS-20-30; Bioserv Diagnostics) fi NSCLC. It regulates tumor cell expression pro les to undergo with 50 mL of each serum in two technical replicates. The readout immune system defense and can be used as a biomarker to track and standard curve were performed with ELISA Reader (Tecan tumor progression, recurrence, or metastatic spread. Group Ltd.). The results of ELISA were visualized with GraphPad Prism 5. Detailed descriptions are provided in the Supplementary Materials and Methods Materials and Methods. Tissue sample collection, characterization, and preparation Tissue samples were provided by Lung Biobank Heidelberg, a Statistical analyses member of the accredited Tissue Bank of the National Center for Data of qPCR analyses were statistically analyzed under Tumor Diseases (NCT) Heidelberg, the BioMaterialBank Heidel- REMARK criteria (20) with SPSS 22.0 for Windows (IBM). The berg and the Biobank platform of the German Center for Lung evaluation of discriminatory values for PAEP expression in tumor Research (DZL). All diagnoses were made according to the 2009 that best differentiated between groups of patients with good and WHO classification for lung cancer by at least two experienced poor survival prognosis was performed with the critlevel proce- pathologists. Detailed descriptions are provided in the Supple- dure using ADAM statistical software package (German Cancer mentary Materials and Methods. Research Center, DKFZ, Heidelberg, Germany; ref. 21). Binary variables were built using these cutoffs. The endpoint of the study Total RNA isolation and cDNA synthesis was overall survival (OS). Therefore, survival time was calculated Total RNA was isolated from tissue using an RNeasy Mini Kit from the date of surgery until the last date of contact or death. (Qiagen) according to the manufacturer's instructions. For RNA Univariate analysis of survival data was performed according to isolation from cell lines, the AllPrep DNA/RNA Mini Kit (Qiagen) Kaplan and Meier (22). The log-rank test was used to test the was used. Total RNA was transcribed to sscDNA with a Transcrip- significance between the groups. A P value of less than 0.05 was tor First Strand cDNA Synthesis Kit (Roche). Detailed descriptions considered significant. Multivariate survival analysis was per- are provided in the Supplementary Materials and Methods. formed using the Cox proportional hazards model. Receiver Operating Characteristic (ROC) analyses were performed to Gene-expression analyses examine the diagnostic accuracy of glycodelin. ELISA data were For gene-expression analysis, we used a human genome expres- statistically analyzed with GraphPad Prism 5. The non-parametric sion microarray (Affymetrix U133 Plus 2.0, Affymetrix) covering Mann–Whitney U test (23) as well as the Kruskal–Wallis test (24) approximately 47,000 transcripts and variants. The microarray were used to investigate significant differences between the experiment was performed with three biologic replicates of con- patient groups.

3530 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin in NSCLC

Cell culture expressed (Ct value < 38) in 90% of all tumors, but only in 54% of The cell line 2106T was generated from human squamous cell the corresponding normal tissues (Table 2). One hundred forty- carcinomas (SCC) as described previously (18) and cultivated in four patients (40%) exhibited PAEP exclusively in tumor but not DMEM/Ham's F-12 (Life technologies) with 10% FCS (PAA, GE in normal tissue. Twenty-six (7%) patients did not express PAEP Healthcare Life Sciences) for not more than 20 passages. The either in tumor or in normal tissue and were excluded from further H1975 cell line was purchased from the ATCC (Wesel, Germany) analyses. and cultivated in DMEM/Ham's F-12 with 10% FCS. MeWo The comparison of tumor and normal tissue revealed that PAEP melanoma cell line was purchased from cell lines service (CLS, expression levels were significantly upregulated in more than 80% Eppelheim, Germany) and cultivated with EMEM supplemented of all SCCs (P < 0.0001) and more than 75% of all adenocarci- with 2 mmol/L L-glutamine, 1% nonessential amino acids, 1 nomas (ADC; P < 0.0001) compared with matched normal tissue mmol/L sodium pyruvate, and 10% FCS (premix from CLS). For (Fig. 1A and D). No significant gender-specific expression of PAEP the indicated experiments, cell lines were cultured for a maximum was observed between male and female patients in our study. The of 24 hours with complete serum-free growth medium. Detailed median expression of PAEP mRNA in ADC was significantly descriptions are provided in the Supplementary Materials and higher than in SCC (P < 0.0001, Fig. 1B). Both cancer types Methods. exhibited a high-expression range of PAEP in tumor tissue, rang- ing from 20-fold downregulation to a more than 8,000-fold siRNA experiments upregulation of expression compared with paired normal tissue. The following siRNAs were used (Qiagen): Hs_PAEP_1, Regarding the different stages of patients' disease, median PAEP Hs_PAEP_2, Hs_PAEP_3, Hs_PAEP_5 (Gene Solution siRNA, expression was upregulated in stage II (P < 0.05 for ADC) and #1027416). For all assays, all four siRNAs were pooled and used stage III for both ADC and SCC (Fig. 1C). Although PAEP was with a resulting RNAi concentration of 10 nmol/L. Allstars neg- expressed in a broad range of more than 16 qPCR cycles in tumor ative control siRNA (Qiagen, #1027280) was used as a non- tissue, normal tissue exhibited lower expression with less varia- silencing control siRNA. Detailed descriptions are provided in tion (Fig. 1D), suggesting a tumor-specific function of glycodelin. the Supplementary Materials and Methods. Expression of PAEP in NSCLC is a prognostic factor in females Cytokine array Statistical analyses were performed with patients expressing PAEP n ¼ 2106T, H1975, and MeWo cells were transfected with control in NSCLC ( 336). Patients with SCC and increased PAEP or PAEP pooled siRNA for 72 hours. The medium was replaced by mRNA levels of exhibited a reduced OS rate (Fig. 1E). fi P ¼ complete serum-free growth medium for 24 hours. Supernatants However, the difference in survival was not signi cant ( PAEP were collected and used for cytokine arrays (C-series, customer 0.139). In contrast, the intratumoral expression level of fl designed, RayBiotech). Most regulated cytokines and chemokines had no in uence on the prognosis of ADC patients (Fig. 1F). The were selected from the Affymetrix gene-expression analyses. The survival of NSCLC patients was strongly associated with gender. fi assay was performed according to the manufacturer's instructions. Females exhibited a signi cantly reduced OS rate when expressing PAEP P ¼ Signals were evaluated using Image Studio Lite (LI-COR). Experi- a high level ( 0.014, Fig. 1G). On the other side, there was ments were performed using at least two biologic replicates with no correlation between expression level and survival in male two technical replicates each. patients (Fig. 1H). Further analyses indicated that lymph node status is a prognostic factor in women (P ¼ 0.016, Supplementary PAEP Migration assay Fig. S1). In a multivariate analysis, including expression, For the migration assay, 2106T (2.5 104) and H1975 (3 104) histology, gender and lymph node status, only gender (female vs. fi cells were seeded onto a 24-well plate. The next day, cells were male) and lymph node status (pN2 vs. pN0) were signi cant PAEP transfected with control siRNA or the PAEP pooled siRNAs for 72 prognostic factors (Table 2). High expression (high vs. low, hours. Afterwards, cells were serum-starved overnight (16 hours), see Supplementary Materials and Methods) was associated with a treated with 10 mg/mL mitomycin C (Applichem) for 2 hours and moderately increased HR. However, the statistical analysis P ¼ detached with Accutase-solution (Promocell). Approximately 5 revealed only a trend ( 0.19). 104 cells were applied with 300 mL serum-free DMEM onto Glycodelin as well as immunosuppressive glycodelin A is ThinCerts Cell Culture Inserts (Greiner bio-one) with a pore size expressed in NSCLC tissue of 8 mm. At the bottom side, 500 mL of DMEM completed with 10% To verify the glycodelin protein expression in NSCLC, IHC was FBS was prepared. After 24 hours, cells that migrated through the performed on formalin-fixed and paraffin-embedded (FFPE) tis- membrane were detached from the underside of ThinCerts with sue slides derived from patients (n ¼ 13) that were previously Accutase-solution. Cells were washed with PBS, and LDH activity shown to have a high glycodelin mRNA expression. Four repre- was determined with the Cytotoxicity Detection Kit (LDH; Roche) sentative samples (two ADC and two SCC) are presented in Fig. 2. as a measure of relative cell number. The relative migration of The expression pattern of glycodelin varied between the NSCLC control siRNA-transfected cells was set to 100%. Experiments were subtypes. In general, glycodelin staining exhibited greater hetero- performed three times with at least three biologic replicates each. geneity in ADC (Fig. 2A, patient 1 and 2). In contrast, glycodelin was expressed more homogenously, but at a lower level in almost Results all tumor cells of SCC (Fig. 2A, patients 3 and 4). Antibody PAEP is strongly overexpressed in adenocarcinomas and specificity was tested and validated in controls using a blocking squamous cell carcinomas peptide (Supplementary Fig. S2A). PAEP mRNA expression was investigated in 362 NSCLC sam- A monoclonal antibody was used to compare the expression of ples. Patients' characteristics are presented in Table 1. PAEP was an immunomodulating form of glycodelin, glycodelin A, with

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3531

Schneider et al.

Table 1. Patients' characteristics Gene-expression analyses Validation cohort NSCLC Validation cohort COPD Parameter n (%) Parameter n (%) Parameter n (%) Median age 65 (38–88) Median age 64 (41–87) Median age 64 (44–84) Gender 362 Gender 159 Gender 90 Male 250 (69) Male 105 (66) Male 55 (61) Female 112 (31) Female 54 (34) Female 35 (39) Histology Histology Stage Adeno 211 (58) Adeno 102 (64) GOLD I 4 (4) Squamous 151 (42) Squamous 35 (22) GOLD II 10 (11) Adenosquamous 4 (3) GOLD III 21 (23) Therapy Pleomorph 1 (1) GOLD IIII 53 (59) OP 212 (59) Large Cell 1 (1) n.d. 2 (2) OP/RT 13 (4) NSCLC 16 (10) OP/CT 100 (28) OP/RT/CT 37 (10) Therapy No therapy 6 (4) P stage RFA 1 (1) IA 37 (10) RT 4 (3) IB 90 (25) CT 52 (33) IIA 70 (19) RT/CT 34 (21) IIB 51 (14) OP 36 (23) IIIA 105 (29) OP/RT 5 (3) IIIB 9 (2) OP/CT 14 (9) OP/RT/CT 6 (4) ECOG 0 320 (88) C stage 1 32 (9) IA 16 (10) 2 4 (1) IB 11 (7) n.d. 8 (2) IIA 7 (4) IIB 10 (6) IIIA 21 (13) IIIB 19 (12) IVA 26 (18) IVB 49 (31) P stage IA 10 (16) IB 11 (18) IIA 5 (8) IIB 13 (21) IIIA 15 (25) IIIB 3 (5) IVA 2 (3) IVB 2 (3) ECOG 0 102 (64) 1 47 (30) 2 7 (4) n.d. 2 (1) NOTE: Bold numbers indicate the total number of cases in each column. Abbreviations: OP, surgery; CT, chemotherapy; RT, radiotherapy; RFA, radiofrequency ablation; ECOG, Eastern Cooperative Oncology Group; GOLD, Global Initiative for Chronic Obstructive Lung Disease; n.d., no data.

Table 2. PAEP expression in lung tissue that of total glycodelin in NSCLC. Glycodelin was expressed by qPCR statistics most tumor cells (Fig. 2B, top). However, we observed a various n Parameter % staining pattern with the anti-glycodelin A antibody (Fig. 2B, PAEP expression 362 100 bottom). Some signals of both antibodies were completely over- Tumor 325 90 Normal 196 54 lapping. In some cases, glycodelin A expression differed from that Exclusively in tumor 144 40 of total glycodelin or could not be detected. Exclusively in normal 15 4 Normal lung alveolar cells exhibited low expression of glyco- Expressed in both 181 50 delin (Fig. 2A, third column). All stained tumor tissues were Not expressed 26 7 infiltrated by CD8-positive T cells (Fig. 2A, last column). Glyco- Multivariate analysis of OS delin expression could also be observed in the tumor cells of P Variable HR (95% CI) infiltrated lymph nodes (Supplementary Fig. S2C). PAEP tumor (DCt < 9.8 vs. DCt > 9.8) 1.31 (0.88–1.95) 0.19 Histology (adeno vs. squamous) 1.11 (0.73–1.69) 0.62 Gender (female vs. male) 0.59 (0.38–0.93) 0.02 Secreted glycodelin can be detected in the sera of NSCLC pN (1 vs. 0) 1.40 (0.89–2.20) 0.15 patients with primary tumor, recurrence, or metastatic spread pN (2 vs. 0) 2.55 (1.69–3.86) <0.001 Glycodelin is known to be a secretory protein. Therefore, we NOTE: Bold values indicate significant events. investigated the possibility of detecting the protein in the sera of

3532 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin in NSCLC

Figure 1. PAEP is highly overexpressed in lung cancer and leads to a decreased OS. A, PAEP expression in 336 lung tumor and corresponding normal tissues was quantiﬁed by qPCR and tumor versus normal tissue expression. B to D, normalized tumor Ct values were visualized with GraphPad Prism. E to H, the percentage of OS is depicted for tumors with normalized Ct values <9.8 or >9.8; , median; T, tumor; N, normal; , P < 0.05; , P < 0.0001.

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3533

Schneider et al.

A Glycodelin CD8 Tumor tissue enlarged Tumor adjacent normal tissue Tumor tissue

Figure 2. Glycodelin as well as glycodelin A is strongly expressed in lung tumor tissue. A and B, representative immunohistologic stainings of lung tumor and normal tissues of glycodelin and glycodelin A are shown. A, CD8 was additionally stained to detect inﬂammatory cells. B, a glycodelin A–speciﬁc antibody was used to detect the Squamous cell carcinoma Adenocarcinoma immunomodulatory, glycosylated form of glycodelin.

B Tumor 1 Tumor 2 Tumor 3 Bronchus Monoclonal A87-B/D2 Polyclonal N-20 Monoclonal A87-B/D2 Polyclonal

NSCLC and non-cancer patients. In the detection cohort, patients PAEP (Fig. 3A), with a median concentration of 8.9 ng/mL and a shown in Fig. 1A were divided into subgroups of high (DCt < 3, maximum of 228.4 ng/mL. The glycodelin serum concentrations n ¼ 30), average (6.8 13, n ¼ 15). Serum samples, which signiﬁcantly increased compared with non-cancer patients. The were collected before surgery, were tested for glycodelin. Patients detection of glycodelin in sera of NSCLC patients was veriﬁed in with benign hamartomas (n ¼ 15) were included as a non- a large cohort of samples randomly picked from the lung cancer control. The highest amounts of glycodelin were detected biobank of the Thoraxklinik. In addition to patients who were in the sera of patients who exhibited high gene expression of treated by surgery alone or in combination with adjuvant

3534 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin in NSCLC

Figure 3. Glycodelin is secreted from tumors. A to E, glycodelin concentration was quantified by ELISA. A, patients were grouped related to the qPCR expression level of PAEP and serum level of glycodelin was determined before surgery. B, elevated glycodelin serum levels in NSCLC detection cohort were confirmed in a validation cohort and compared with patients with COPD. C, 10 randomly selected adenocarcinoma and corresponding normal lung tissue pieces were homogenized and glycodelin concentrations were measured. D, serum level was measured before and after excision of tumor. E, glycodelin concentration in serum was quantified before surgery and during follow-up ward rounds. Patient 3 showed a recurrence of primary tumor. An adrenal gland metastasis of Patient 4 was resected after ward round four. F, primary tumor and metastasis of patient 4 was stained for glycodelin; T, tumor; N, normal; R, recurrence; M, metastasis; WR, ward round; þ, tumor-dependent death. chemo-and/or radiotherapy ("early disease cohort"), this cohort significantly higher compared to that of the COPD control also included inoperable tumor stages IIIB–IV treated by chemo- group. Females with affected lymph nodes had significant higher and/or radiotherapy. We then compared the glycodelin serum glycodelin serum concentrations than with unaffected lymph levels with those of a randomly selected COPD cohort (Fig. 3B nodes (Supplementary Fig. S3A). This is in contrast with males and Table 1). The median glycodelin serum concentration was where no difference in serum concentrations was detected in

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3535

Schneider et al.

Figure 4. Knockdown of glycodelin leads to a reorganization of genes involved in tumor immunosurveillance, environment modulation, and cell migration in NSCLC cell lines. A to H, 2106T and H1975 were transfected with a pool of four different PAEP-speciﬁc siRNAs for 72 hours. A and B, medium was changed and 24 hours later, supernatants as well as cells were collected and glycodelin was detected by western blot. b-Actin and tubulin were used as loading controls. (Continued on the following page.)

3536 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin in NSCLC

regard to the lymph node status. ROC analyses of the NSCLC MeWo cells exhibited the highest mRNA expression of PAEP (data cohort demonstrated an AUC of 0.774 (Supplementary Fig. not shown) and glycodelin secretion of several investigated cell S3B). Serum glycodelin was not detectable in some patients lines (Supplementary Fig. S5F). In addition to 2106T and H1975, despite a high PAEP mRNA expression (Fig. 3A, first row). This we also observed secretion of glycodelin in the supernatant of the situation was especially true for patients with smaller tumors large cell carcinoma cell line H460 and the breast cancer-derived (data not shown). Therefore, we homogenized tumors of four cell line MDA-MB-436. No secretion was observed for the mono- patients (two with high PAEP mRNA tumor expression and high cytic cell lines THP1 and Jurkat or the NK cell line KHYG-1. serum glycodelin concentration and two with high PAEP mRNA The role of glycodelin as a regulator of the immune system is tumor expression and low serum glycodelin concentration; well characterized during pregnancy (25) and hormone-related Supplementary Fig. S3C). All homogenized tumors exhibit high cancer (15). Many studies of glycodelin A have reported immu- intratumoral glycodelin concentrations despite low serum levels nosuppressive functions. To gain further insight into the poten- found in patients 3 and 4. Glycodelin concentrations in the tial molecular functions of glycodelin in tumor cells, we per- homogenates of randomly selected tumors from 10 ADC formed microarray gene-expression profiles of the three cell patients were significantly higher (P ¼ 0.039) compared with lines 2106T, H1975, and MeWo after silencing of PAEP. The matched normal tissue (Fig. 3C). To further substantiate the highest regulated genes (Supplementary Table ST1) were val- tumor as the source of glycodelin in patient serum, glycodelin idated by qPCR. Results are illustrated in Fig. 4C and Supple- was measured in the sera after complete tumor resection by mentary Fig. S4B. Knockdown of PAEP led to strong mRNA surgery. As presented in Fig. 3D, the relative glycodelin con- upregulation of immune system-regulating ligands (MHC class centrations in the sera decreased significantly (P ¼ 0.0005) after I polypeptide-related sequence proteins (MICA/B), C–X–C the resection of the tumor. In a further approach, the glycodelin motif chemokine 5 and 16 (CXCL5 and CXCL16), cluster of serum concentration was measured during the clinical follow-up differentiation 83 (CD83), cluster of differentiation 274 and of patients (Fig. 3E and Supplementary Fig. S3D). As demon- programmed cell death 1 ligand 2 (CD274 and PDCD1LG2, strated in Fig. 3C, all four patients exhibited a strong decrease in also known as PDL1/2) and to deregulation of cytoskeleton and glycodelin secretion in the second ward round after tumor cell matrix genes [claudin-1 (CLDN1), keratin 17 (KRT17)and resection. Glycodelin levels in relapse-free patients 1 and 2 keratin-associated protein 2–3(KRTAP2-3), urokinase-type remained stable at approximately 1.5 ng/mL after surgery. In plasminogen activator (PLAU)], as well as to an increase of contrast, we observed increasing glycodelin concentrations in signal pathway–regulating proteins [e.g., endothelin-1 (EDN1), patients with recurrence or metastatic spread (patient 3 and 4). prostaglandin G/H synthase 1 and 2 (PTGS1/2, also known as IHC of the primary tumor confirmed glycodelin expression in COX1/2) and proheparin-binding EGF-like growth factor patient 4 (Fig. 3F). Patient 2106, the donor of the cell line used (HBEGF)]. A cytokine and chemokine array was used to validate in this study, exhibited an increasing glycodelin serum concen- the gene-expression data. The secretion of chemokines CCL5, tration during follow-up until tumor-independent death (Sup- CXCL16, and CXCL5 was increased, whereas secretion of cluster plementary Fig. S2D). of differentiation 14 (CD14), osteopontin (SPP1), and IL11 was The presence of glycodelin in patients' sera suggests that mea- downregulated (Fig. 4D and Supplementary Table ST1). Matrix surement of glycodelin levels might be a useful tool for disease metalloproteinase-9 (MMP9) expression was strongly decreased monitoring and recurrence detection in NSCLC. in 2106T. Upregulation of PDL1 (in 2106T and H1975) and PTGS2 mRNA (in H1975) after PAEP silencing was confirmed at Knockdown of glycodelin in cancer cell lines leads to a the protein level by Western blot analyses (Fig. 4E). An example of reorganization of genes involved in tumor cytokine array data is presented in Supplementary Fig. S5G. We immunosurveillance, environment modulation, and cell observed a change in cell structure after knockdown of PAEP (Fig. migration 4F and Supplementary Fig. S4C) in 2106T, H1975, and MeWo Glycodelin expression and secretion by lung cancer cells has not cells. Silencing of PAEP resulted in a significant reduction of been observed so far. To investigate a functional role of glycodelin migration for 2106T (P < 0.0001) and H1975 (P < 0.0001, Fig. in NSCLC, PAEP was silenced by siRNA-transfection in two 4G). MeWo cells did not exhibit any migration in this assay. These NSCLC cell lines (Fig. 4A and B) and in the MeWo cell line data indicate that glycodelin expression is involved in the regu- (Supplementary Fig. S4A). Downregulation by RNAi resulted in lation of tumor immunology. strongly decreased intracellular and secreted protein levels of glycodelin. Interestingly, culture medium with 2% FCS alone caused an antibody signal (Fig. 4A and B, first lane). This signal Discussion resulted from FCS, because culture medium without supplements In the present study, we demonstrated the overexpression of as well as complete serum-free growth medium did not exhibit glycodelin mRNA (PAEP) in NSCLC in a large patient cohort. For this band (Supplementary Fig. S5D). For siRNA transfection, a the first time, we reported the expression and secretion of glyco- pool of four different PAEP siRNAs was used. Every single siRNA delin in several lung cancer cell lines. Furthermore, we demon- strongly reduced glycodelin expression (Supplementary Fig. S5E). strated an immunomodulatory function of glycodelin in NSCLC.

(Continued.) C, expression of indicated genes was determined by qPCR. Experimentswere performed with at least three biologic replicates on three different days with three technical repeats each. D, medium was replaced by DMEM þ supplements and supernatants were collected 24 hours later. Relative protein expression of indicated proteins was determined using a cytokine nitrocellulose array. Experiments were performed with at least two biologic replicates on different days with two technical replicates each. E, expression of indicated proteins was detected by Western blot analysis. b-Actin was used as loading control. F, pictures of 2106T and H1975 were taken after knockdown of PAEP. G, cells were serum-starved overnight. The next day, a migration assay was performed three times on different days with at least two biologic replicates each for 24 hours; , P < 0.05; , P < 0.005.

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3537

Schneider et al.

Knockout of glycodelin in NSCLC cell lines led to the differential Tumors might use various strategies to bypass the immune regulation of immune system-relevant genes that exhibit potential system. Tumors remodel their microenvironment. In our study, as targets for NSCLC therapeutics (26), for example, PD-L1 we demonstrated that the knockdown of glycodelin results in the (CD274), which was upregulated after silencing of PAEP. Our increased expression of MMP9 in H1975 and the decreased findings provide evidence that glycodelin detection in human expression of PLAU in 2106T and H1975. Both proteins are serum might be useful to monitor the response of NSCLC patients involved in the regulation of the ECM. In this regard, the dense to tumor treatment. matrix areas around lung tumors have been shown to strongly To date, the expression of glycodelin has been well described reduce T-cell migration (7). Increasing glycodelin concentrations mainly in hormone-related tumors such as breast and ovarian have been linked to deregulated PLAU and MMP9 activity and cancer (15). Zadran and colleagues (27) recently reported that inhibition of trophoblast invasiveness (11). This finding is sup- glycodelin mRNA (PAEP) was one of the most differentially ported by observations in melanoma cells, where cell migration regulated genes in a small cohort of lung ADC patients compared and invasion were strongly inhibited after glycodelin silencing with healthy patients. In our study, we demonstrated that glyco- (16). In our study, we observed the expression of glycodelin in delin was highly expressed in a large cohort of NSCLC samples. metastatic lymph nodes and distant metastases. Similarly, we PAEP mRNA levels in tumors were elevated in approximately 80% observed that glycodelin knockdown changed cell morphology of all patients compared with normal tissue and were upregulated and led to the decreased migration of NSCLC cell lines. We in advanced tumor stages. High PAEP expression was associated observed that glycodelin was secreted by breast cancer, melanoma with shorter survival in SCC. In particular, women seemed to and several lung cancer cell lines. Therefore, glycodelin expression exhibit worse survival, which was observed independently of might be a general mechanism by which other cancers regulate tumor pathology but was possibly influenced by lymph node tumor immunology. status. Because glycodelin A is regulated by progesterone in Tumors escape immunosurveillance through the regulation of females during the menstruation cycle and pregnancy (28), there chemokine secretion and the presentation of cell surface mole- might be a hormone-regulated mechanism that triggers glycode- cules. Immunoediting is a well-known important factor within lin expression in NSCLC. We aim to clarify the mechanism of this context (32). In our work, we observed that several immune glycodelin regulation in a future study and we will investigate the system modulators were upregulated in cell lines after siRNA role of EGFR and KRAS, which are the most common driver knockdown of PAEP. MICA and MICB, both MHC-I–related mutations in lung cancer (3). chains (MIC), are ligands for NKGD2, a receptor on the surface þ In our immunohistochemical stainings of NSCLC tumors, we of NK cells and CD8 T cells (33). For lung cancer, MICA and observed the inhomogeneous expression of glycodelin in tumor MICB are reportedly recognized by tumor-infiltrating T cells (34). nests. Glycodelin was highly expressed in tumors that were The expression of immune system-defending factors CD83 as well þ infiltrated by CD8 T cells. Two studies reported better survival as chemokines CXCL5 and CXCL16 is strongly increased after for NSCLC patients with a high rate of infiltrating macrophages knockdown of PAEP. CXCL5 and CXCL16 have been shown to be þ and T cells (8, 29). We were not able to detect PAEP mRNA involved in angiogenesis (35, 36) and attracting CD4 immune expression in approximately half of all normal tissues. In the case cells in NSCLC (37), similar to PTGS2/COX2 (38). CD83 expres- þ of a very low expression level, this could be attributed to the sion influences CD4 T-cell development by stimulating dendritic glycodelin expression of bronchial epithelial cells. We detected cells (39). Altogether, glycodelin knockout leads to the reorgani- higher glycodelin protein concentrations in the serum samples of zation of immune system-stimulating and immune system- NSCLC compared with patients with COPD. In women, glyco- silencing factors and plays an important role in the immune delin is normally expressed during the menstrual cycle. Because system regulation of NSCLC. the median age of lung cancer patients in our large validation The fetomaternal interface that is present during pregnancy has cohort was 64, the secretion of glycodelin in the context of evolved immunomodulatory properties to allow the fetus to menstruation cycle can be largely excluded. A valid AUC can be invade the uterus without an attack from the maternal immune interpreted as the probability that the classifier will assign a system (12). Parallels between cancer and pregnancy have been higher score to a randomly chosen positive example than to a described in detail (40). As the knockdown of PAEP in our study randomly chosen negative example. The ROC analyses of NSCLC regulated immune system-relevant genes in particular, we postu- cohort indicated that glycodelin was only conditionally suitable late that glycodelin expression and secretion are essential for lung for early detection of NSCLC. In this context, we observed that a cancer cells to escape from the immune system. high glycodelin expression in tumor tissues did not necessarily In a prospective study, we plan to analyze glycodelin serum mean that glycodelin was also measurable in patients' sera. levels in the palliative setting to investigate whether glycodelin Several factors such as tumor size, position, vascularization, might be used as a surrogate for the success of chemotherapy spread, etc., might influence the secretion of glycodelin into the and/or radiotherapy. blood. Therefore, the sensitivity of the glycodelin detection in In summary, we have demonstrated the expression of glyco- serum was too low for early detection of NSCLC. However, our delin at the mRNA and protein levels in NSCLC. From our results data demonstrate that the measurement of glycodelin in the and the known immune modulating functions of glycodelin, it is serum might be useful for the clinical follow-up of surgically feasible to propose an immune system defense mechanism of treated patients. We observed not only a correlation between the lung tumors through the expression and secretion of glycodelin. glycodelin serum level and the tumor response to treatment, but Glycodelin produced by tumor cells might reduce the effect of also at recurrence at least in most patients who pretherapeuti- current therapies using immune-checkpoint inhibitors (e.g., PD- cally expressed glycodelin. Other studies have suggested the use L1–targeted therapies), as it is able to remodel the host immune of glycodelin as a serum marker for colon cancer (30) and system and the tumor immune system defense. Targeting glyco- ovarian cancer (31). delin in NSCLC therapy might be a new mechanism to overcome

3538 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin in NSCLC

tumor immunosurveillance. Our data strongly suggest that the Administrative, technical, or material support (i.e., reporting or organizing detection of glycodelin in the tumor and serum seems to be data, constructing databases): M. Thomas, T. Muley helpful in designing tumor-specific immunotherapies and in Study supervision: F.J.F. Herth, H. Dienemann, M. Meister Other (histopathology and immmunohistochemistry: methods, results, and monitoring the response to treatment during clinical follow-up interpretation): P.A. Schnabel of NSCLC patients. Acknowledgments Disclosure of Potential Conflicts of Interest The authors thank Chang Xu, Martin Fallenbuechel, Jessica Eschenbach, No potential conflicts of interest were disclosed. Christa Stolp, and Andrea Bopp for expert technical assistance and Dr. Michael Elsaesser for the serum samples of pregnant women. The tissue samples were Authors' Contributions provided by the tissue bank of the National Center for Tumor Diseases (NCT, Conception and design: M.A. Schneider, F.J.F. Herth, H. Dienemann, Heidelberg, Germany) in accordance with the regulations of the tissue bank and M. Meister the approval of the ethics committee of the University of Heidelberg. Development of methodology: M.A. Schneider, P.A. Schnabel, F.J.F. Herth, Grant Support M. Meister fi Acquisition of data (provided animals, acquired and managed patients, This study was nancially supported in part by The Federal Ministry of provided facilities, etc.): M.A. Schneider, A. Warth, P.A. Schnabel, F.J.F. Herth, Education and Research, Germany and the German Center for Lung Research. T. Muley, M. Meister The costs of publication of this article were defrayed in part by the Analysis and interpretation of data (e.g., statistical analysis, biostatistics, payment of page charges. This article must therefore be hereby marked advertisement computational analysis): M.A. Schneider, M. Granzow, P.A. Schnabel, in accordance with 18 U.S.C. Section 1734 solely to indicate M. Thomas, T. Muley, M. Meister this fact. Writing, review, and/or revision of the manuscript: M.A. Schneider, M. Granzow, A. Warth, P.A. Schnabel, M. Thomas, F.J.F. Herth, T. Muley, Received September 23, 2014; revised January 19, 2015; accepted April 20, M. Meister 2015; published OnlineFirst April 21, 2015.

References 1. Ferlay J SI, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, et al. abortion and partially upregulated in mole pregnancy. J Histochem Cyto- GLOBOCAN 2012 v1.0, Cancer incidence and mortality worldwide: IARC chem 2008;56:477–85. cancerbase no. 11 [Internet]. Lyon, France: International Agency for 13. Morris HR, Dell A, Easton RL, Panico M, Koistinen H, Koistinen R, et al. Research on Cancer; 2013. 2013. Gender-specific glycosylation of human glycodelin affects its contraceptive 2. Giaccone G, Herbst RS, Manegold C, Scagliotti G, Rosell R, Miller V, et al. activity. J Biol Chem 1996;271:32159–67. Gefitinib in combination with gemcitabine and cisplatin in advanced non– 14. AlokA,KarandeAA.Theroleofglycodelinasanimmune-modulating small cell lung cancer: a phase III trial–INTACT 1. J Clin Oncol 2004;22: agent at the feto-maternal interface. J Reprod Immunol 2009;83: 777–84. 124–7. 3. Govindan R, Ding L, Griffith M, Subramanian J, Dees ND, Kanchi KL, et al. 15. Seppala M, Koistinen H, Koistinen R, Hautala L, Chiu PC, Yeung WS. Genomic landscape of non–small cell lung cancer in smokers and never- Glycodelin in reproductive endocrinology and hormone-related cancer. smokers. Cell 2012;150:1121–34. Eur J Endocrinol 2009;160:121–33. 4. Rosell R, Carcereny E, Gervais R, Vergnenegre A, Massuti B, Felip E, et al. 16. Ren S, Liu S, Howell PM Jr, Zhang G, Pannell L, Samant R, et al. Functional Erlotinib versus standard chemotherapy as first-line treatment for Euro- characterization of the progestagen-associated endometrial protein gene in pean patients with advanced EGFR mutation-positive non-small-cell lung human melanoma. J Cell Mol Med 2010;14:1432–42. cancer (EURTAC): a multicentre, open-label, randomised phase 3 trial. 17. Kunert-Keil C, Steinmuller F, Jeschke U, Gredes T, Gedrange T. Immuno- Lancet Oncol 2012;13:239–46. localization of glycodelin in human adenocarcinoma of the lung, squa- 5. Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al. Gefitinib mous cell carcinoma of the lung and lung metastases of colonic adeno- or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med carcinoma. Acta Histochem 2011;113:798–802. 2009;361:947–57. 18. Gottschling S, Jauch A, Kuner R, Herpel E, Mueller-Decker K, Schnabel PA, 6. Shaw AT, Kim DW, Nakagawa K, Seto T, Crino L, Ahn MJ, et al. Crizotinib et al. Establishment and comparative characterization of novel squamous versus chemotherapy in advanced ALK-positive lung cancer. N Engl J Med cell non–small cell lung cancer cell lines and their corresponding tumor 2013;368:2385–94. tissue. Lung Cancer 2012;75:45–57. 7. Salmon H, Franciszkiewicz K, Damotte D, Dieu-Nosjean MC, Validire P, 19. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, et al. The Trautmann A, et al. Matrix architecture defines the preferential localization MIQE guidelines: minimum information for publication of quantitative and migration of T cells into the stroma of human lung tumors. J Clin Invest real-time PCR experiments. Clin Chem 2009;55:611–22. 2012;122:899–910. 20. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM, et al. 8. Al-Shibli KI, Donnem T, Al-Saad S, Persson M, Bremnes RM, Busund LT. REporting recommendations for tumour MARKer prognostic studies Prognostic effect of epithelial and stromal lymphocyte infiltration in non– (REMARK). Eur J Cancer 2005;41:1690–6. small cell lung cancer. Clin Cancer Res 2008;14:5220–7. 21. Abel U, Berger J, Wiebelt H. CRITLEVEL: an exploratory procedure for the 9. Terme M, Ullrich E, Delahaye NF, Chaput N, Zitvogel L. Natural killer cell- evaluation of quantitative prognostic factors. Methods Inf Med 1984;23: directed therapies: moving from unexpected results to successful strategies. 154–6. Nat Immunol 2008;9:486–94. 22. Dinse GE, Lagakos SW. Nonparametric estimation of lifetime and disease 10. Motohashi S, Ishikawa A, Ishikawa E, Otsuji M, Iizasa T, Hanaoka H, et al. A onset distributions from incomplete observations. Biometrics 1982;38: phase I study of in vitro expanded natural killer T cells in patients with 921–32. advanced and recurrent non–small cell lung cancer. Clin Cancer Res 2006; 23. Mann HB, Whitney DR. On a Test Whether one of Two Random Variables is 12:6079–86. Stochastically Larger than the other. Ann Math Stat 1947;18:1–64. 11.LamKK,ChiuPC,ChungMK,LeeCL,LeeKF,KoistinenR,etal. 24. Kruskal WH, Wallis WA. Use of ranks in one-criterion variance analysis. Glycodelin-A as a modulator of trophoblast invasion. Hum Reprod J Am Stat Assoc 1952;47:583–621. 2009;24:2093–103. 25. Lee CL, Lam KK, Koistinen H, Seppala M, Kurpisz M, Fernandez N, et al. 12. Toth B, Roth K, Kunert-Keil C, Scholz C, Schulze S, Mylonas I, et al. Glycodelin-A as a paracrine regulator in early pregnancy. J Reprod Immu- Glycodelin protein and mRNA is downregulated in human first trimester nol 2011;90:29–34.

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3539

Schneider et al.

26. Davies M. New modalities of cancer treatment for NSCLC: focus on 34. Groh V, Rhinehart R, Secrist H, Bauer S, Grabstein KH, Spies T. Broad immunotherapy. Cancer Manag Res 2014;6:63–75. tumor-associated expression and recognition by tumor-derived gamma 27. Zadran S, Remacle F, Levine RD. miRNA and mRNA cancer signatures delta T cells of MICA and MICB. Proc Natl Acad Sci U S A 1999; determined by analysis of expression levels in large cohorts of patients. 96:6879–84. Proc Natl Acad Sci U S A 2013;110:19160–5. 35.IsozakiT,ArbabAS,HaasCS,AminMA,ArendtMD,KochAE,etal. 28. Rutanen EM, Koistinen R, Seppala M, Julkunen M, Suikkari AM, Huhtala Evidence that CXCL16 is a potent mediator of angiogenesis and ML. Progesterone-associated proteins PP12 and PP14 in the human is involved in endothelial progenitor cell chemotaxis: studies in endometrium. J Steroid Biochem 1987;27:25–31. mice with K/BxN serum-induced arthritis. Arthritis Rheum 2013;65: 29. Kawai O, Ishii G, Kubota K, Murata Y, Naito Y, Mizuno T, et al. Predom- 1736–46. inant infiltration of macrophages and CD8(þ) T Cells in cancer nests is a 36. Pickens SR, Chamberlain ND, Volin MV, Gonzalez M, Pope RM, Mandelin significant predictor of survival in stage IV non–small cell lung cancer. AM II, et al. Anti-CXCL5 therapy ameliorates IL-17-induced arthritis by Cancer 2008;113:1387–95. decreasing joint vascularization. Angiogenesis 2011;14:443–55. 30. Govindarajan R, Parthasarathy S. Glycodelin: a possible new biological 37. Darash-Yahana M, Gillespie JW, Hewitt SM, Chen YY, Maeda S, Stein I, marker in colorectal cancer. J Clin Oncol 24:18 2006 (suppl; abstr 20081). et al. The chemokine CXCL16 and its receptor, CXCR6, as markers and 31. Jeschke U, Bischof A, Speer R, Briese V, Richter DU, Bergemann C, et al. promoters of inflammation-associated cancers. PLoS ONE 2009;4: Development of monoclonal and polyclonal antibodies and an ELISA for e6695. the determination of glycodelin in human serum, amniotic fluid and cystic 38. Sharma S, Yang SC, Zhu L, Reckamp K, Gardner B, Baratelli F, et al. Tumor fluid of benign and malignant ovarian tumors. Anticancer Res 2005;25: cyclooxygenase-2/prostaglandin E2-dependent promotion of FOXP3 þ þ 1581–9. expression and CD4 CD25 T regulatory cell activities in lung cancer. 32. Bremnes RM, Al-Shibli K, Donnem T, Sirera R, Al-Saad S, Andersen S, et al. Cancer Res 2005;65:5211–20. The role of tumor-infiltrating immune cells and chronic inflammation at 39. Fujimoto Y, Tu L, Miller AS, Bock C, Fujimoto M, Doyle C, et al. CD83 þ the tumor site on cancer development, progression, and prognosis: empha- expression influences CD4 T-cell development in the thymus. Cell sis on non–small cell lung cancer. J Thorac Oncol 2011;6:824–33. 2002;108:755–67. 33. Gonzalez S, Lopez-Soto A, Suarez-Alvarez B, Lopez-Vazquez A, Lopez- 40. Holtan SG, Creedon DJ, Haluska P, Markovic SN. Cancer and pregnancy: Larrea C. NKG2D ligands: key targets of the immune response. Trends parallels in growth, invasion, and immune modulation and implications Immunol 2008;29:397–403. for cancer therapeutic agents. Mayo Clin Proc 2009;84:985–1000.

3540 Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research

Glycodelin: A New Biomarker with Immunomodulatory Functions in Non −Small Cell Lung Cancer

Marc A. Schneider, Martin Granzow, Arne Warth, et al.

Clin Cancer Res 2015;21:3529-3540. Published OnlineFirst April 21, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-14-2464

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2015/04/22/1078-0432.CCR-14-2464.DC1

Cited articles This article cites 38 articles, 10 of which you can access for free at: http://clincancerres.aacrjournals.org/content/21/15/3529.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/21/15/3529.full#related-urls

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/21/15/3529. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.