Serglycin Is a Theranostic Target in Nasopharyngeal Carcinoma That Promotes Metastasis

Published OnlineFirst February 2, 2011; DOI: 10.1158/0008-5472.CAN-10-3557

Cancer Tumor and Stem Cell Biology Research

Serglycin Is a Theranostic Target in Nasopharyngeal Carcinoma that Promotes Metastasis

Xin-Jian Li1, Choon Kiat Ong5, Yun Cao1,2, Yan-Qun Xiang3, Jian-Yong Shao1,2, Aikseng Ooi9, Li-Xia Peng1, Wen-Hua Lu1, Zhongfa Zhang11, David Petillo9, Li Qin4, Ying-Na Bao1, Fang-Jing Zheng1, Claramae Shulyn Chia8, N. Gopalakrishna Iyer6, Tie-Bang Kang1, Yi-Xin Zeng1, Khee Chee Soo7, Jeffrey M. Trent10, Bin Tean Teh5,9, and Chao-Nan Qian1,3,5,10

Abstract Nasopharyngeal carcinoma (NPC) is known for its high-metastatic potential. Here we report the identification of the proteoglycan serglycin as a functionally significant regulator of metastasis in this setting. Comparative genomic expression profiling of NPC cell line clones with high- and low-metastatic potential revealed the serglycin gene (SRGN) as one of the most upregulated genes in highly metastatic cells. RNAi-mediated inhibition of serglycin expression blocked serglycin secretion and the invasive motility of highly metastatic cells, reducing metastatic capacity in vivo. Conversely, serglycin overexpression in poorly metastatic cells increased their motile behavior and metastatic capacity in vivo. Growth rate was not influenced by serglycin in either highly or poorly metastatic cells. Secreted but not bacterial recombinant serglycin promoted motile behavior, suggesting a critical role for glycosylation in serglycin activity. Serglycin inhibition was associated with reduced expression of vimentin but not other epithelial–mesenchymal transition proteins. In clinical specimens, serglycin expression was elevated significantly in liver metastases from NPC relative to primary NPC tumors. We evaluated the prognostic value of serglycin by immunohistochemical staining of tissue microarrays from 263 NPC patients followed by multivariate analyses. High serglycin expression in primary NPC was found to be an unfavorable independent indicator of distant metastasis-free and disease-free survival. Our findings establish that glyco- sylated serglycin regulates NPC metastasis via autocrine and paracrine routes, and that it serves as an independent prognostic indicator of metastasis-free survival and disease-free survival in NPC patients. Cancer Res; 71(8); 1–11. 2011 AACR.

Introduction lymph nodes (LN) or even distant organs at the time of diagnosis (6). However, the molecular mechanisms underlying Nasopharyngeal carcinoma (NPC) is a common malignancy NPC metastasis are poorly understood. in southern China and Southeast Asia (1, 2). NPC has the Serglycin is a proteoglycan consisting of a core protein to highest metastasis rate among head and neck cancers (3–5), which negatively charged glycoaminoglycan (GAG) chains of with the majority of the patients having metastases to regional either chondroitin sulfate or heparin are attached (7, 8). The core protein containing 158 amino acid residues can be divided into 3 domains: a signal peptide domain, an N-terminal domain Authors' Affiliations: 1State Key Laboratory of Oncology in South China, with unknown function, and a C-terminal domain (9). The Departments of 2Pathology and 3Nasopharyngeal Carcinoma, Sun functions of serglycin in various cells depend on the type and 4 Yat-sen University Cancer Center, Guangzhou, Guangdong; Division of size of the GAG chains decorating the core protein (8, 10–19). Pharmacoproteomics, Institute of Pharmacy and Pharmacology, Univer- sity of South China, Hengyang, Hunan, China; 5NCCS-VARI Translational Serglycin mRNA or protein has been detected in normal Research Program, 6Department of Surgical Oncology, and 7Division of hematopoietic, endothelial, and embryonic stem cells (11, 14, 8 Medical Sciences, National Cancer Center Singapore; Department of 20–23). Serglycin is thought to be important for homeostasis General Surgery, Singapore General Hospital, Singapore; 9Laboratories of Cancer Genetics and 10Genome Biology, Van Andel Research Institute of positively charged components (e.g., proteases) in storage (VARI), Grand Rapids, Michigan; and 11Center for Systems and Computa- granules due to its negatively charged GAG chains (24–28). tional Biology, The Wistar Institute, Philadelphia, Pennsylvania In cytotoxic lymphocytes or natural killer T cells, a macro- Note: Supplementary data for this article are available at Cancer Research molecular complex of granzyme B and perforin complexed Online (http://cancerres.aacrjournals.org/). with serglycin induces the apoptosis of target cells (29–34). Corresponding Author: Chao-Nan Qian, Department of Nasopharyngeal Carcinoma, Sun Yat-sen University Cancer Center, 651 Dongfeng East Serglycin has been associated with tumorigenesis in acute Road, Guangzhou, Guangdong 510060, P.R. China. Phone: 86-20- myeloid leukemia (AML) and myeloma, and it is found to be a 87343457; Fax: 86-20-87343624. E-mail: [email protected] selective marker for distinguishing AML from Philadelphia doi: 10.1158/0008-5472.CAN-10-3557 chromosome-negative chronic myeloproliferative disorders. It 2011 American Association for Cancer Research. is also highly expressed by multiple myeloma cell lines (35, 36).

www.aacrjournals.org OF1

Li et al.

Although the involvement of serglycin in tumor metastasis has Histologic evaluation and immunohistochemical been speculated (37), the exact role of serglycin in NPC staining remains unknown. Mouse lymph nodes were routinely fixed and sectioned at 5 The epithelial–mesenchymal transition (EMT), a funda- mm throughout the lymph nodes. To evaluate the microme- mental process in embryonic development, is involved in tastases in mouse lymph nodes, one section in every 20 the metastasis and progression of tumors (38, 39). Activation sequential sections was selected for hematoxylin and eosin of the EMT program, accompanied by an increase in the staining. mesenchymal marker vimentin and loss of the epithelial For immunohistochemical (IHC) staining of serglycin and marker E-cadherin, endows carcinoma cells with enhanced vimentin, paraffin-embedded tissues were sectioned at 5 mm migratory and invasive properties that facilitate dissemination and IHC staining was performed as described previously (42). to permissive niches (39). Briefly, the sections were incubated with a rabbit anti-human serglycin polyclonal antibody (Cat No.: HPA000759, Sigma- Materials and Methods Aldrich; working dilution 1:50) or mouse anti-human E-cadherin monoclonal antibody (Abcam) overnight at 4C, or Cell culture and cellular growth rate mouse anti-human Vimentin monoclonal antibody (Neomar- Human NPC cell line CNE-2 and its clones (S18, S22, and kers; working dilution 1:100) for 30 minutes at room tem- S26, cultured in less than 50 passages), and SUNE-1 and its perature. An EnVision kit (DAKO) was used to detect the clone 5-8F were maintained in Dulbecco's modified Eagle's primary antibodies followed by 3,3-diaminobenzidine sub- medium supplemented with 10% FBS at 37C. Cellular growth strate visualization and counterstaining with hematoxylin. curves were plotted by using the cellular viability values The intensity of IHC staining in the tumor cells was scored assessed by the MTS method (Cell Titer 96 Aqueous One independently by 2 pathologists by using the semiquantitative Solution Cell Proliferation Assay solution; Promega). IRS (immunoreactive score) scale according to Remmele and Stegner (43), which takes into account both the intensity of the In vitro migration and invasion assays color reaction (no staining ¼ 0; weak staining ¼ 1; moderate Wound healing assays and transwell assays were used to staining ¼ 2; strong staining ¼ 3) and the percentage of evaluate the migration and invasion abilities of the cells (40). stained cells (0% ¼ 0; 1%–10% ¼ 1; 11%–50% ¼ 2; 51%–80% ¼ See the Supplementary Methods for more details. 3; 81%–100% ¼ 4). The average value from the 2 referees was used as the final score. Detection of serglycin in conditioned medium A total of 2 106 cells were plated in 100-mm culture plates Quantitative PCR and incubated for 48 hours in a regular medium, then replaced The methodology of quantitative PCR is described in the with a serum-free medium and incubated for an additional 24 Supplementary Materials and Methods. The sequences of PCR hours. Ten milliliters of conditioned medium was collected primers used for amplification of serglycin and glyceralde- and concentrated to a volume of 200 mL by using Amicon Ultra hyde-3-phosphate dehydrogenase (GAPDH) were as follows: centrifuge filters (10 kDa molecular weight cutoff pore size; GAPDH forward, 50-AAGGTCATCCCTGAGCTGAA-30; GAPDH Millipore). Twenty microliters of the concentrated condi- reverse, 50-TGACAAAGTGGTCGTTGAGG-30; serglycin for- tioned medium was subjected to SDS-PAGE and blotted ward, 50-TATCCT ACGCGGAGAGCCAGGTAC-30; serglycin with anti-human serglycin (Cat no. H00005552-M03, Abnova) reverse, 50-TTCCGTTAGGAAGCCACTCCC AGATC-30. The antibodies. experiments were performed in triplicate.

Human tissues and tissue microarray Lentiviral transduction studies All the human tissue samples were obtained from the Cell lines stably expressing either serglycin short hairpin Department of Pathology, Sun Yat-sen University Cancer Cen- RNA (shRNA) or a scrambled nontarget shRNA were establi- ter (SYSUCC), with prior patient consents and the approval of shed by a BLOCK-iT Lentiviral Pol II miR RNAi system the Institutional Clinical Ethics Review Board at SYSUCC. The (Invitrogen) according to the manufacturer's instruc- tissue microarrays (TMA) contained qualified primary NPC tions. The targets of human serglycin shRNAs are 50- samples from 263 pathologically diagnosed patients at SYSUCC CTGGTTCTGGAATCCTCAGTT-30 and 50-CGCTGCAATC- between December 30, 1997, and September 6, 2002. Of these CAGACAGTAAT-30. See the Supplementary Methods for patients, 199 were men and 64 women, with a median age of 46 more details. years (ranging from 17 to 77 years). All patients received radiotherapy, with doses of 70 to 74 Gy to the primary tumor, Immunoblotting 60 to 64 Gy to the involved areas of the neck, and 50 Gy to the The primary antibodies, including mouse anti-human ser- uninvolved areas of the neck. For the patients with late-stage glycin monoclonal antibody (Abnova), mouse anti-human E- disease (stages III and IV), 2 to 3 cycles of induction or cadherin monoclonal antibody (Abcam), rabbit anti-human concurrent platinum-based chemotherapy was given. The vimentin polyclonal antibody (Cell Signalling Technology), patients were followed up regularly. TMAs were constructed and mouse anti-human b-actin monoclonal antibody (Sigma as described previously (41) and the methodology is described Aldrich) were used at a dilution of 1:1,000. See the Supple- in the Supplementary Methods. mentary Methods for more details.

OF2 Cancer Res; 71(8) April 15, 2011 Cancer Research

Serglycin Regulates Metastasis of Nasopharyngeal Carcinoma

Deglycosylation assay SYSUCC. Xenograft tumors for genomic expression profiling Concentrated conditioned medium containing secreted were generated by subcutaneous injection of the cancer cells serglycin protein was treated with an enzymatic Glycoprotein in the mice. The spontaneous LN metastasis model has been Deglycosylation Kit (EMD Chemicals) according to the man- published previously (40). Briefly, a total of 1 105 cells were ufacturer's instructions. Briefly, 20 mL of concentrated condi- injected into the left hind footpad of each mouse to generate a tioned medium, mixed with denaturation buffer (2% SDS, 1 primary tumor. After 7 to 8 weeks, the popliteal LN of the left mol/L b-mercaptoethanol) and reaction buffer (250 mmol/L hind foot was collected on the terminal day for routine tissue sodium phosphate buffer, pH 7.0), was incubated at 100C for processing. 5 minutes. After adding 1 mL each of N-glycosidase F, a2- 3,6,8,9-neuraminidase, endo-a-N-acetylgalactosaminidase, Genomic expression profiling b1,4-galactosidase, and b-N-acetylglucosaminidase, plus 2.5 The methodology of gene expression profiling is described mL of TRITON X-100 (15% solution), the conditioned medium in the Supplementary Methods. To identify genes that were was incubated at 37C for 48 hours. Immunoblots were carried differentially expressed between high- and low-metastasis out to determine the efficiency of serglycin deglycosylation. cells/xenografts, the log2 transformed expression value of each gene in S18 cells/xenograft was subtracted by the mean Animals and spontaneous lymph node metastasis assay of log2 transformed expression value of the corresponding Female athymic mice between 5 and 6 weeks of age were gene in the 3 low-metastasis cells/xenografts (i.e., CNE2, S22, obtained from Shanghai Institutes for Biological Sciences S26). The results from this calculation were then sorted; the (Shanghai, China). All the animal studies were conducted in 25 most upregulated and 25 most downregulated genes accordance with the principles and procedures outlined in the are presented in Figure 1A and B as heat maps. The gene guidelines of Institutional Animal Care and Use Committee at expression data for this study have been uploaded to the

Figure 1. Serglycin expression A B C and glycosylation in high- metastasis clones. Gene 1,000,000 expression profiling was CNE-2 S22 S26 S18 CNE-2 S22 S26 S18 100,000 performed on cultured cells of the SPINK6 SPINK6 SERGLYICN SERGLYCIN rglycin 4 lines (A), and on the VTN UGT1A10 evel 10,000 corresponding xenograft tumors AKR1C1 UGT1A7 SPINK5L3 UGT1A5 1,000 grown subcutaneously in the nude MFAP5 UGT1A9

PSG4 UGT1A4 mRNA le 100 mice (B). Red and blue, DAB2 UGT1A1 respectively, indicate upregulation CCL26 UGT1A3 Relative ser 10 and downregulation of a gene RNF182 UGT1A8 CGA UGT1A6 1 relative to the average of the 3 KRT19 AKR1C1 SPINK4 SPINK5L3 low-metastasis lines (CNE-2, S22, MYPN ID2 and S26). C, the mRNA levels of FABP4 FAM102B KANK4 WNT5A serglycin (normalized to GAPDH) WNT5A DAB2 in the 4 lines were confirmed by SLC12A3 NNMT EPDR1 ABCG2 D quantitative real-time PCR, PPP2R2B CGA showing that serglycin expression DIO2 RNF182 Serglycin FYN VTN 300 kDa in CM was significantly higher in S18 HEG1 C21orf63 Serglycin C21orf63 AKR1C3 130 kDa cells than in the other lines. MAP2K6 SH3BP5 in WCL Moreover, high level of serglycin FXYD3 IRF6 β-Actin KCNS3 S100A14 mRNA was also found in another COL17A1 ABCA12 high-metastasis clone 5-8F in TXNIP DSC3 LY6D SLITRK5 contrast to its parental low- SCEL BNC1 DSG3 SERPINB2 metastasis NPC cell line SUNE-1. CSTA KLK10 Column, mean; error bar, SD TSPAN1 KCNS3 ESRP1 SCEL (from triplicates). D, the serglycin RHBDL1 KRT17 protein levels in these lines as TMC4 KCTD12 PPP1R14C LAMC2 determined by Western blotting. SERPINB4 ST6GALNAC1 E E, the concentrated conditioned SEPP1 LOC642587 ANKRD22 DSG3 medium of S18 cells was NDUFA4L2 SHISA2 deglycosylated at 37Cfor ABP1 KRT5 CLCA2 SYK 48 hours by an enzymatic ST6GALNAC1 S100A2 deglycosylation kit. After partial S100A14 KRT14 IRF6 KRT6A 250 kDa deglycosylation, a protein band of KRT6A S100A9 CALB1 S100A8 150 kDa 80 kDa was detected by LOC642587 CALB1 immunoblotting of serglycin, in 100 kDa contrast to the control sample with Differential expression value: 75 kDa no enzyme. CM, conditioned ≤ -5, -2.5, 0, 2.5, ≥5 medium; WCL, whole cell lysate.

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 OF3

Li et al.

A B Primary NPC Liver metastases g scores erglycin staining erglycin Se Primary Liver metastases NPCs from NPC (n = 48) (n = 16) C 450 400 CNE-2 350 * 300 * CNE-2 M3 250 200 150 100

mber of cells/field 50 0 Num

D 1,200 1,000 800 600

tive serglycin serglycin tive 400 mRNA level mRNA level m

Relat 200 0 CNE-2 CNE-2 M3

Figure 2. Serglycin expression is upregulated in liver metastases and in migrated CNE-2 cells. A, IHC staining was used to evaluate the protein expression levels of serglycin in primary NPC tissues and in liver metastases from NPC. The ranges of serglycin expression (brown) were shown, with prominent staining in metastatic lesions. Scale bars, 50 mm. B, statistical analysis (Mann–Whitney test) revealed a significant increase of serglycin expression in liver metastases from NPC relative to expression in primary NPC (*, P < 0.001). C, a 2-chamber assay (transwell assay) with and without Matrigel was used to evaluate migration and invasion abilities, respectively. Parental CNE-2 cells and CNE-2 M3 cells were studied. Sequential screening successfully selected the CNE-2 M3 cells that had increased migration and invasion abilities. Column, mean; bars, SD (from triplicates). *, P < 0.001 relative to parental CNE-2 cells (Student's t test). D, serglycin mRNA levels in CNE-2 and CNE-2 M3 cells were determined by quantitative real-time PCR normalized to GAPDH. Serglycin expression was dramatically upregulated in the CNE-2 M3 cells, consistent with the previous findings.

Gene Expression Omnibus, with the accession number the serglycin and vimentin staining scores, or between the GSE24154. serglycin and E-cadherin staining scores. The median of the IHC score value was used as the cutoff value to divide Statistical methods the patients into high- and low-serglycin expression groups. The Single comparisons were performed by Student's t test, censoring time distribution was estimated by the Kaplan–Meier Mann–Whitney test, or c2 test (2-tailed; P < 0.05 was considered method and P values were calculated by log rank analysis. significant). The Spearman correlation test (2-tailed) was used Cox's regression model was used for multivariate survival to calculate the correlation coefficient (r)andP value between analysis. Predictors were judged to be significant at P < 0.05.

OF4 Cancer Res; 71(8) April 15, 2011 Cancer Research

Serglycin Regulates Metastasis of Nasopharyngeal Carcinoma

AB C 1.8 n i i 70,000 161.6 60,000 1.4 50,000 1.2 CM 40,000 nm) 1.0 0.8 Scrambled 30,000 490

mRNA level WCL TS activity

(A 0.6

elative serglyc 20,000 SG KD1 M M 004.4 R 10,000 β-Actin 0.2 0 SG KD2 0.0 123456 Days

D E 900 Scrambled 800

mbled SG KD1 700 SG KD2

Scra 0.5 mm 600 500 400 * * # # KD 1 300

SG 200 100 Number of S18 cells/field 0 KD 2 SG

0 h 24 h 36 h

Figure 3. Suppression of serglycin inhibits the migration and invasion abilities of S18 cells. S18 cells were transduced with lentiviruses expressing scrambled shRNA or shRNAs targeting serglycin at different loci (KD1 and KD2). A, serglycin mRNA levels (normalized to GAPDH) determined by quantitative real- time PCR in the serglycin knockdown cells were significantly reduced. B, serglycin protein levels in both conditioned medium (CM) and whole cell lysate (WCL) were determined by immunoblotting. Suppression of serglycin in SG KD1 and SG KD2 cells completely eliminated the secretion of serglycin protein into the medium. C, suppression of serglycin did not alter the cellular growth rate. D, suppression of serglycin dramatically reduced the migration ability of the cells as determined by a wound-healing assay. Yellow dashed lines denote the margins of the wound. E, a 2-chamber assay was used for further evaluation of the invasion/migration ability of the S18 cells. Suppression of serglycin significantly reduced the cells’ migration and invasion abilities. Column, mean; bars, SE (from triplicates). Student's t test was used for the statistical analyses. *, P < 0.001 relative to that of scrambled shRNA; #, P ¼ 0.0027 for SG KD1 and 0.0061 for SG KD2 when compared with that of scrambled shRNA.

Results the second most upregulated gene in S18 xenograft (Fig. 1B). Serglycin expression is elevated in NPC cells with higher To confirm the findings described above, quantitative metastasis potential real-time PCR was performed to quantify the mRNA levels We have previously isolated and established cellular of serglycin among the 4 lines, with consistent results clones having different metastatic abilities from the par- (Fig. 1C). Moreover, a high level of serglycin expression ental NPC cell line, CNE-2 (40). Among these clones, clone was also detected in the high-metastasis clone 5-8F isolated 18 (S18) had the highest metastatic ability, whereas clone 22 from another low-metastasis parental NPC cell line SUNE-1 (S22) and clone 26 (S26), and the parental line CNE-2, had (ref. 44; Fig. 1C). Moreover, S18 and 5-8F cells could secrete low-metastatic abilities. To explore the underlying molecu- serglycin into culture medium, whereas the other low- lar mechanism(s), we performed genomic expression profil- metastasis cell lines could not (Fig. 1D). The specificity of ing of these 4 cell lines from in vitro cultured cells. In the the anti-human serglycin antibody was confirmed and high-metastasis clone S18, the serglycin gene (SRGN)was shown in Supplementary Figure 1. After partial deglycosyla- the second most highly upregulated (Fig. 1A). Considering tion of secreted serglycin, a protein with a lower molecular the potential differences between in vivo and in vitro con- weight (approximately 80 kDa) could be detected (Fig. 1E), ditions, xenograft tumors from these 4 cell lines were also confirming the glycosylation of the serglycin protein collected for gene expression profiling. Again, serglycin was secreted by high-metastasis cells.

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 OF5

Li et al.

A B C Figure 4. Secreted serglycin enhances migration and invasion 2.0 abilities of low-metastasis S26 y cells. A, low-metastasis S26 cells 1.5 were transduced with lentiviruses nm) 1.0 expressing serglycin cDNA or an

490 Vector empty vector as control. The

CM (A 0.5 serglycin expression levels were 250 kDa MTS activity M SG 0.0 determined by Western blotting. SG Protein levels of serglycin in 123456 130 kDa WCL conditioned medium (CM) and Days whole cell lysate (WCL) were β-Actin increased by stable ectopic expression of serglycin cDNA in S26 cells. B, S26 cells stably overexpressing serglycin cDNA or D E the control vector were subjected to MTS assays. There was no 450 statistical difference between 600 * * these 2 cellular populations. C, a 400 high level of secreted serglycin 500 350 was detected by immunoblotting 400 in the concentrated conditioned 300

lls per field medium of the S26 cells carrying s per insert 300 serglycin plasmid. Twenty-four 250 hours of stimulation of wild-type 200 200 S26 cells with the 2 conditioned media resulted in an increase of 100 150 migration ability (D) and an Migrated ce nvaded cells 0 In 100 increase of invasion ability (E) of the S26 cells. Column, mean; bars Vector SG Vector SG SE (from triplicates). *, P < 0.01 with Student's t test.

Serglycin is upregulated in liver metastases from NPC serglycin protein was confirmed via immunoblotting (Fig. 3B). and in highly migratory/invasive NPC cells Yet, knocking down serglycin in S18 cells did not alter the To evaluate the clinical relevance of our hypothesis that growth rate of the cells (Fig. 3C), implying that serglycin did serglycin regulates NPC metastasis, we collected archival not have a role in controlling cellular growth. The migration tissues of liver metastases from NPC (n ¼ 16) for comparison ability of S18 cells was significantly suppressed after the loss of with primary NPC tissues (n ¼ 48). IHC staining showed that secreted serglycin (Fig. 3D and E). The invasion ability of the protein levels of serglycin were significantly higher in the S18 cells was also significantly reduced by knocking down liver metastases (Fig. 2A and B), implying that serglycin could serglycin (Fig. 3E). These results proved that secreted serglycin be important in clinical scenarios. could promote NPC cell migration and invasion in an auto- To confirm that upregulation of serglycin in highly migra- crine mode. tory cells could be repeatedly found, a 2-chamber (Boyden chamber) assay was used to isolate highly migratory cells from Secreted serglycin promotes the motility of the parental CNE-2 line. After 3 sequential passages of screen- low-metastasis NPC cells ing, highly migratory/invasive cells (CNE-1 M3) were isolated To validate the motility-enhancing activity of serglycin, we (Fig. 2C), and serglycin was found to be highly expressed in generated an S26 cell line stably overexpressing serglycin in these cells (Fig. 2D). the conditioned medium (Fig. 4A), without altering the cellular growth (Fig. 4B). Lower serglycin secretion suppresses the migration and The conditioned medium of S26 cells overexpressing invasion of NPC cells without influencing growth rate serglycin or the control vector was subjected to 50 con- To examine the causal role of serglycin in motility of NPC centration (from 10 mL to 200 mL), and concentrated ser- cells, we engineered cell lines from S18 that stably expressed glycin protein was then confirmed by immunoblotting either shRNAs targeting serglycin expression (SG KD1 and SG (Fig. 4C). The concentrated conditioned medium was used KD2) or a scrambled nontarget shRNA. The suppression of to stimulate migration and invasion by wild-type S26 cells. serglycin expression at the mRNA level was confirmed by The number of migrated and invaded cells were significantly quantitative real-time PCR (Fig. 3A). A decrease in secreted increased after the stimulation (Fig. 4D and E). These results

OF6 Cancer Res; 71(8) April 15, 2011 Cancer Research

Serglycin Regulates Metastasis of Nasopharyngeal Carcinoma

migration of the cells (Supplementary Fig. 2), suggesting A B that the glycosylation of serglycin is critical for promoting cellular motility.

Expression of serglycin mediates the metastasis rate of NPC in vivo The effect of serglycin in vivo was determined by using an 0.5 mm 50 µm LN metastasis model (40). As expected, serglycin knockdown cells showed a significant reduction in metastasis rate, and C P = 0.036 serglycin-overexpressing cells showed an increased metastasis P = 0.032 60 rate (Fig. 5). These results confirmed that serglycin was 7 50 16 16 involved in the control of NPC metastasis in vivo. Nonmetastasis 40

of mice Metastasis 30 52 Serglycin expression is associated with the epithelial– 42 43 20 mesenchymal transition N Number 10

0 To explore the relationship of serglycin and EMT, IHC Scrambled SG KD1 SG KD2 staining of serglycin, vimentin, and E-cadherin was performed in 48 primary human NPC tissue samples. The results indi- D E cated that serglycin expression level positively correlated to the expression level of vimentin (Fig. 6A and B), and inversely correlated to the expression level of E-cadherin (Fig. 6A and C). To confirm this finding, the protein levels of several EMT markers in 4 cellular populations were evaluated. A high level 0.5 mm 50 µm of secreted serglycin accompanied elevated levels of mesenchymal markers vimentin, N-cadherin, and fibronectin, and a

F P = 0.022 reduced level of epithelial protein E-cadherin in the high-

60 metastasis cells (Fig. 6D). The suppression of serglycin 9 50 20 resulted in a lower vimentin level, but did not influence the Nonmetastasis 40 levels of other EMT proteins (Fig. 6E). All these data suggest Metastasis er of mice 30 52 that serglycin is correlated with EMT by mediating the 20 42 vimentin level. Numbe 10

0 Vector SG overexpressionSerglycin High-level serglycin expression is an independent, overexpression unfavorable prognostic indicator for NPC Finally, to evaluate the prognostic values of serglycin Figure 5. In vivo metastasis rates of serglycin knockdown cells and expression, we performed IHC staining for serglycin in a set serglycin-overexpressing cells. Subcutaneous injection of S18 cells stably of tissue microarrays containing 263 NPC samples (Supple- expressing serglycin shRNAs (SG KD1 and KD2) or scrambled shRNA into mentary Table 1; Fig. 7A and B). The correlations between the left hind footpad of nude mice led to metastasis to the left popliteal LN serglycin expression and clinicopathologic characteristics are in some animals. A, image of a popliteal LN 38 days after injection of S18 cells expressing a scrambled shRNA. B, enlarged image of the framed area presented in Supplementary Table 2. High serglycin expres- in (A), showing metastatic S18 cells invading the normal lymphoid tissue. sion was significantly correlated with the occurrence of dis- C, the proportion of popliteal LN metastasis after inoculation. Knocking ease progression or distant metastasis. Multivariate analyses down serglycin in S18 cells significantly reduced the metastasis rate from revealed that a high level of serglycin expression was an 88.1% (52 of 59) to 72.4% (42 of 58). D, S26 cells stably overexpressing serglycin or a control vector were inoculated into the left hind footpad of independent, unfavorable prognostic indicator for disease- the mouse. The image shows a metastatic popliteal LN 38 days after free survival and distant metastasis-free survival (Supplemen- injection. E, enlarged image of the framed area in (D), showing metastatic tary Table 3). The survival curves of disease-free survival and serglycin-expressing S26 cells in the subcapsular area of the LN invading distant metastasis-free survival are presented in Figure 7C the normal lymphoid tissue. F, the proportion of popliteal LN metastasis and D. Taken together, these analyses revealed that high after inoculation of S26 cells. Overexpression of serglycin in S26 cells significantly increased the metastasis rate from 67.7% (42 of 62) to 85.2% serglycin expression level in NPC significantly correlated with (52 of 61). adverse patient outcomes.

Discussion suggest that serglycin could also promote NPC motility via a In this study, we have identified a metastasis-enhancing paracrine mode, by which secreted serglycin from a rare glycoprotein, serglycin, by using high-throughput gene expres- clone could stimulate the motility of cells unable to secrete sion profiling analyses of CNE-2 clones having low- or high- serglycin. Interestingly, prokaryotic recombinant serglycin metastasis potential. The first advantage of our approach is to without glycosylation modification could not promote minimize the influences from confounding factors. When

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 OF7

Li et al.

A Case 1 Case 2 B n = 47 11 r = 0.305 P = 0.037 9 7 5 entin score glycin 3 Vime

Serg 1 (1)-1 -1 1 3 5 7 9 11 Serglycin score C 13 n = 46 11 r = - 0.388 9 P = 0.008 7

Vimentin 5 3

E-cadherin score 1 -1 -1 1 3 5 7 9 11 Serglycin score

S18 cells -cadherin E E-

Serglycin CM D Serglycin Vimentin

E-cadherin Serglycin Serglycin CM

CM Fibronectin E-cadherin E-cadherin WCL Vimentin N-cadherin Vimentin Snail N-cadherin N-cadherin W WCL Twist Fibronectin Fibronectin β-Actin β-Actin β-Actin

Figure 6. Serglycin expression modulates vimentin expression in NPC cells. A, continuous sections of human NPC tissue were subjected to IHC staining with antibodies against serglycin, vimentin, and E-cadherin. The high expression of serglycin in the tumor tissue in case 1 was accompanied by an elevated level of vimentin and absence of E-cadherin. Conversely, the low expression of serglycin in the tumor tissue of case 2 was accompanied by absence of vimentin and an elevated level of E-cadherin. Scale bars, 50 mm. B, using the same methods, 47 primary NPC tissues satisfactorily stained with antibodies against vimentin and serglycin were scored and plotted. A significant positive correlation between serglycin and vimentin was shown. C, For E-cadherin staining, 46 NPC tissues were qualified for analysis. A significant negative correlation between serglycin and E-cadherin was found. D, immunoblotting of serglycin and EMT markers showed a high level of secreted serglycin in the conditioned medium of high-metastasis S18 and 5-8F cells, which was accompanied by high levels of vimentin, N-cadherin, and fibronectin, and a low level of E-cadherin. Conversely, low expression of secreted serglycinin low-metastasis cells (CNE-2, S26, and SUNE-1) was accompanied by low levels of vimentin, N-cadherin, and fibronectin, and a high level of E-cadherin. E, knocking down serglycin expression by shRNAs (SG KD1 and SG KD2) dramatically suppressed the vimentin level in a whole cell lysate of S18 cells, but the levels of other EMT markers were not influenced.

different cell lines (or tumor tissues) derived from different signal from this aggressive clone will be overwhelmed by the patients are used for comparisons, many confounding factors majority of low-metastasis populations. However, tumor might appear, including gender, race, age, and other disease metastases usually arise from rare clones in the tumor (45). backgrounds. All these factors will be perfectly balanced when Using the identified culprit clone for comparison can therefore we compare cellular clones derived from a single patient. The easily reveal the key molecules regulating metastasis. second advantage is to amplify the impact of critical cellular Among the most upregulated genes in S18, serglycin is population on metastasis. As reported previously (40), among second on the list under both in vitro and in vivo conditions, 29 clones isolated from the CNE-2 line, only 2 clones possessed implying the heavy involvement of this glycoprotein in reg- high-metastasis ability, with S18 being the most aggressive. ulating NPC metastasis. To our knowledge, no published data Obviously, if we extract the mRNA from the parental line, the have directly linked serglycin with tumor metastasis. This

OF8 Cancer Res; 71(8) April 15, 2011 Cancer Research

Serglycin Regulates Metastasis of Nasopharyngeal Carcinoma

A B Serglycin Serglycin low expression high expression 1.0

0.8

Figure 7. Elevated serglycin level Low 0.6 correlates with shorter disease- free survival and distant magnification 0.4 metastasis-free survival in NPC patients. TMA analyses of a cohort 0.2 of 263 NPC patients diagnosed at Death event M0 were conducted. A, high and

High 0.0 low levels of serglycin protein Proportion of uncensored cases

expression in TMA are shown magnification 02040 60 80 100120140 under both low and high Follow-up time (months) magnifications of a light microscope. Scale bars, 100 mm. CD B, the median follow-up time for 1.0 1.0 Serglycin low expression (n = 138) this cohort of patients was 83 Serglycin low expression (n = 138) months. C, the disease-free 0.8 0.8 survival (DFS) rate was significantly higher in the low- Serglycin high expression (n = 125) 0.6 0.6 serglycin group. D, the distant Serglycin high expression (n = 125) metastasis-free survival (DMFS) rate was also significantly higher in 0.4 0.4 the low-serglycin group. DFS probability 0.2 P = 0.003 DMFS probability 0.2 P = 0.001

0.0 0.0

0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Time to progression (months) Time to metastasis (months)

study has discovered that serglycin is involved in regulating Disclosure of Potential Conflicts of Interest NPC metastasis by promoting the migratory and invasive abilities of NPC cells via autocrine and paracrine modes, with No potential conflicts of interest were disclosed. autocrine mechanism seeming to be more important. Sergly- cin expression can also mediate the level of vimentin, which Acknowledgments not only is a marker of EMT but also has an important role in regulating cellular migration (46). Moreover, serglycin is an We thank David Nadziejka, Grand Rapids, Michigan, for critical reading of independent, unfavorable prognostic indicator for distant the manuscript. metastasis-free survival and disease-free survival in NPC patients. Grant Support As shown in Figure 1A and B, other genes are also highly expressed in S18 cells, implying that multiple factors could be This work was supported by grants from the State Key Program of National Natural Science Foundation of China (Grant 81030043 to C-N. involved in regulating NPC metastasis. When serglycin expres- Qian, X-J. Li, Y. Cao, Y-Q. Xiang, L-X. Peng, W-H. Lu, Y-N. Bao, F-J. Zheng), sion was knocked down in high-metastasis S18 cells, a 15% the National High Technology Research and Development Program of reduction in the metastasis rate was achieved, suggesting that China (863 Program; Grant 20060102A4002 to X-J. Li, L-X. Peng, W-H. Lu, C-N. Qian, J-Y. Shao, Y-X. Zeng), the Major State Basic Research targeting multiple factors is a hope for complete suppression Program of China (973 Project; Grant 2006CB910104 to Y-X. Zeng), the of NPC metastasis. Youth Science Fund of the National Natural Science Foundation of China In summary, our study shows that serglycin plays a pivotal (Grant 81000946 to L. Qin), the Van Andel Foundation (A. Ooi, D. Petillo), the National Cancer Center Research Foundation of Singapore (C.K. Ong, A. role in regulating NPC metastasis by way of enhancing cellular Ooi, C.S. Chia, N.G. Iyer, B.T. Teh), and the Singapore Millennium Founda- migration, cellular invasiveness, vimentin expression level, tion(C.K.Ong,B.T.Teh). and in vivo spread of cancer cells. Moreover, a high level of The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in serglycin expression can potentially be used to predict shorter accordance with 18 U.S.C. Section 1734 solely to indicate this fact. disease-free survival and shorter metastasis-free survival of NPC patients. Targeting serglycin could be a novel option for Received September 30, 2010; revised December 29, 2010; accepted January prevention of NPC metastasis. 13, 2011; published OnlineFirst February 2, 2011.

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 OF9

Li et al.

References 1. Wee JT, Ha TC, Loong SL, Qian CN. Is nasopharyngeal cancer really a 23. Maillet P, Alliel PM, Mitjavila MT, Perin JP, Jolles P, Bonnet F. "Cantonese cancer"? Chin J Cancer 2010;29:517–26. Expression of the serglycin gene in human leukemic cell lines. Leu- 2. Yu MC, Yuan JM. Epidemiology of nasopharyngeal carcinoma. Semin kemia 1992;6:1143–7. Cancer Biol 2002;12:421–9. 24. Abrink M, Grujic M, Pejler G. Serglycin is essential for maturation of 3. Ahmad A, Stefani S. Distant metastases of nasopharyngeal carci- mast cell secretory granule. J Biol Chem 2004;279:40897–905. noma: a study of 256 male patients. J Surg Oncol 1986;33: 25. Braga T, Grujic M, Lukinius A, Hellman L, Abrink M, Pejler G. Serglycin 194–7. proteoglycan is required for secretory granule integrity in mucosal 4. Geara FB, Sanguineti G, Tucker SL, Garden AS, Ang KK, Morrison mast cells. Biochem J 2007;403:49–57. WH, et al. Carcinoma of the nasopharynx treated by radiotherapy 26. Henningsson F, Hergeth S, Cortelius R, Abrink M, Pejler G. A role for alone: determinants of distant metastasis and survival. Radiother serglycin proteoglycan in granular retention and processing of mast Oncol 1997;43:53–61. cell secretory granule components. FEBS J 2006;273:4901–12. 5. Lee AW, Law SC, Foo W, Poon YF, Cheung FK, Chan DK, et al. 27. Ringvall M, Ronnberg E, Wernersson S, Duelli A, Henningsson F, Retrospective analysis of patients with nasopharyngeal carcinoma Abrink M, et al. Serotonin and histamine storage in mast cell secretory treated during 1976–1985: survival after local recurrence. Int J Radiat granules is dependent on serglycin proteoglycan. J Allergy Clin Oncol Biol Phys 1993;26:773–82. Immunol 2008;121:1020–6. 6. Guigay J. Advances in nasopharyngeal carcinoma. Curr Opin Oncol 28. Lieberman J. The ABCs of granule-mediated cytotoxicity: new weap- 2008;20:264–9. ons in the arsenal. Nat Rev Immunol 2003;3:361–70. 7. Kolset SO, Zernichow L. Serglycin and secretion in human mono- 29. Veugelers K, Motyka B, Goping IS, Shostak I, Sawchuk T, Bleackley cytes. Glycoconj J 2008;25:305–11. RC. Granule-mediated killing by granzyme B and perforin requires a 8. Pejler G, Abrink M, Wernersson S. Serglycin proteoglycan: regulating mannose 6-phosphate receptor and is augmented by cell surface the storage and activities of hematopoietic proteases. Biofactors heparan sulfate. Mol Biol Cell 2006;17:623–33. 2009;35:61–8. 30. Raja SM, Metkar SS, Honing S, Wang B, Russin WA, Pipalia NH, et al. 9. Kolset SO, Tveit H. Serglycin–structure and biology. Cell Mol Life Sci A novel mechanism for protein delivery: granzyme B undergoes 2008;65:1073–85. electrostatic exchange from serglycin to target cells. J Biol Chem 10. Woulfe DS, Lilliendahl JK, August S, Rauova L, Kowalska MA, Abrink 2005;280:20752–61. M, et al. Serglycin proteoglycan deletion induces defects in platelet 31. Galvin JP, Spaeny-Dekking LH, Wang B, Seth P, Hack CE, Froelich aggregation and thrombus formation in mice. Blood 2008;111: CJ. Apoptosis induced by granzyme B-glycosaminoglycan com- 3458–67. plexes: implications for granule-mediated apoptosis in vivo. J Immu- 11. Toyama-Sorimachi N, Kitamura F, Habuchi H, Tobita Y, Kimata K, nol 1999;162:5345–50. Miyasaka M. Widespread expression of chondroitin sulfate-type ser- 32. Dressel R, Raja SM, Honing S, Seidler T, Froelich CJ, von Figura K, glycins with CD44 binding ability in hematopoietic cells. J Biol Chem et al. Granzyme-mediated cytotoxicity does not involve the mannose 1997;272:26714–9. 6-phosphate receptors on target cells. J Biol Chem 2004;279:20200– 12. Humphries DE, Wong GW, Friend DS, Gurish MF, Qiu WT, Huang C, 10. et al. Heparin is essential for the storage of specific granule proteases 33. Metkar SS, Wang B, Aguilar-Santelises M, Raja SM, Uhlin-Hansen L, in mast cells. Nature 1999;400:769–72. Podack E, et al. Cytotoxic cell granule-mediated apoptosis: perforin 13. Duelli A, Ronnberg E, Waern I, Ringvall M, Kolset SO, Pejler G. Mast delivers granzyme B-serglycin complexes into target cells without cell differentiation and activation is closely linked to expression of plasma membrane pore formation. Immunity 2002;16:417–28. genes coding for the serglycin proteoglycan core protein and a 34. Raja SM, Wang B, Dantuluri M, Desai UR, Demeler B, Spiegel K, et al. distinct set of chondroitin sulfate and heparin sulfotransferases. J Cytotoxic cell granule-mediated apoptosis. Characterization of the Immunol 2009;183:7073–83. macromolecular complex of granzyme B with serglycin. J Biol Chem 14. Schick BP, Gradowski JF, San Antonio JD. Synthesis, secretion, and 2002;277:49523–30. subcellular localization of serglycin proteoglycan in human endothelial 35. Niemann CU, Kjeldsen L, Ralfkiaer E, Jensen MK, Borregaard N. cells. Blood 2001;97:449–58. Serglycin proteoglycan in hematologic malignancies: a marker of 15. Toyama-Sorimachi N, Miyasaka M. A sulfated proteoglycan as a novel acute myeloid leukemia. Leukemia 2007;21:2406–10. ligand for CD44. J Dermatol 1994;21:795–801. 36. Theocharis AD, Seidel C, Borset M, Dobra K, Baykov V, Labropoulou 16. Toyama-Sorimachi N, Sorimachi H, Tobita Y, Kitamura F, Yagita H, V, et al. Serglycin constitutively secreted by myeloma plasma cells is a Suzuki K, et al. A novel ligand for CD44 is serglycin, a hematopoietic potent inhibitor of bone mineralization in vitro. J Biol Chem cell lineage-specific proteoglycan. Possible involvement in lym- 2006;281:35116–28. phoid cell adherence and activation. J Biol Chem 1995;270: 37. Schick BP. Serglycin proteoglycan deletion in mouse platelets phy- 7437–44. siological effects and their implications for platelet contributions to 17. Uhlin-Hansen L, Wik T, Kjellen L, Berg E, Forsdahl F, Kolset SO. thrombosis, inflammation, atherosclerosis, and metastasis. Prog Mol Proteoglycan metabolism in normal and inflammatory human macro- Biol Transl Sci; 93:235–87. phages. Blood 1993;82:2880–9. 38. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal 18. Winberg JO, Kolset SO, Berg E, Uhlin-Hansen L. Macrophages transitions in development and disease. Cell 2009;139:871–90. secrete matrix metalloproteinase 9 covalently linked to the core 39. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transi- protein of chondroitin sulphate proteoglycans. J Mol Biol 2000;304: tion. J Clin Invest 2009;119:1420–8. 669–80. 40. Qian CN, Berghuis B, Tsarfaty G, Bruch M, Kort EJ, Ditlev J, et al. 19. Zernichow L, Dalen KT, Prydz K, Winberg JO, Kolset SO. Secretion of Preparing the "soil": the primary tumor induces vasculature reorga- proteases in serglycin transfected Madin-Darby canine kidney cells. nization in the sentinel lymph node before the arrival of metastatic FEBS J 2006;273:536–47. cancer cells. Cancer Res 2006;66:10365–76. 20. Schick BP, Ho HC, Brodbeck KC, Wrigley CW, Klimas J. Serglycin 41. Yao X, Qian CN, Zhang ZF, Tan MH, Kort EJ, Yang XJ, et al. Two proteoglycan expression and synthesis in embryonic stem cells. distinct types of blood vessels in clear cell renal cell carcinoma have Biochim Biophys Acta 2003;1593:259–67. contrasting prognostic implications. Clin Cancer Res 2007;13:161–9. 21. Elliott JF, Miller CL, Pohajdak B, Talbot D, Helgason CD, Bleackley 42. Qian CN, Furge KA, Knol J, Huang D, Chen J, Dykema KJ, et al. RC, et al. Induction of a proteoglycan core protein mRNA in mouse T Activation of the PI3K/AKT pathway induces urothelial carcinoma of lymphocytes. Mol Immunol 1993;30:749–54. the renal pelvis: identification in human tumors and confirmation in 22. Kulseth MA, Mustorp SL, Uhlin-Hansen L, Oberg F, Kolset SO. animal models. Cancer Res 2009;69:8256–64. Serglycin expression during monocytic differentiation of U937–1 cells. 43. Remmele W, Stegner HE. Recommendation for uniform definition of Glycobiology 1998;8:747–53. an immunoreactive score (IRS) for immunohistochemical estrogen

OF10 Cancer Res; 71(8) April 15, 2011 Cancer Research

Serglycin Regulates Metastasis of Nasopharyngeal Carcinoma

receptor detection (ER-ICA) in breast cancer tissue. Pathologe 45. Chiang AC, Massague J. Molecular basis of metastasis. N Engl J Med 1987;8:138–40. 2008;359:2814–23. 44. Song LB, Yan J, Jian SW, Zhang L, Li MZ, Li D, et al. [Molecular 46. Ivaska J, Pallari HM, Nevo J, Eriksson JE. Novel functions of vimentin mechanisms of tumorgenesis and metastasis in nasopharyngeal in cell adhesion, migration, and signaling. Exp Cell Res 2007;313: carcinoma cell sublines]. Ai Zheng 2002;21:158–62. 2050–62.

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 OF11

Serglycin Is a Theranostic Target in Nasopharyngeal Carcinoma that Promotes Metastasis

Xin-Jian Li, Choon Kiat Ong, Yun Cao, et al.

Cancer Res Published OnlineFirst February 2, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-10-3557

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2011/02/02/0008-5472.CAN-10-3557.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/2011/04/01/0008-5472.CAN-10-3557. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.