FAM198B Is Associated with Prolonged Survival and Inhibits

1 FAM198B is Associated with Prolonged Survival and Inhibits Metastasis in Lung 2 Adenocarcinoma via Blockage of ERK-mediated MMP-1 Expression 3 4 Chia-Ying Hsu1,2, Gee-Chen Chang3,4, Yi-Ju Chen5, Yi-Chiung Hsu6,7, Yi-Jing Hsiao1,2, 5 Kang-Yi Su1,2,8, Hsuan-Yu Chen2,7,9, Chien-Yu Lin2,7, Jin-Shing Chen10, Yu-Ju Chen5, 6 Qi-Sheng Hong1,2, Wen-Hui Ku11, Chih-Ying Wu12, Bing-Ching Ho1,2, Ching-Cheng 7 Chiang1, Pan-Chyr Yang2,13,14, Sung-Liang Yu1,2,8,15,16,17 8 9 1Department of Clinical and Laboratory Sciences and Medical Biotechnology, 10 National Taiwan University College of Medicine, Taipei, Taiwan, 2Center of Genomic 11 Medicine, National Taiwan University College of Medicine, Taipei, Taiwan, 3Faculty 12 of Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan, 13 4Division of Chest Medicine, Department of Internal Medicine, Taichung Veterans 14 General Hospital, Taichung, Taiwan, 5Institute of Chemistry, Academia Sinica, Taipei, 15 Taiwan, 6Department of Biomedical Sciences and Engineering, National Central 16 University, Taoyuan, Taiwan, 7Institute of Statistical Science, Academia Sinica, Taipei, 17 Taiwan, 8Department of Laboratory Medicine, National Taiwan University Hospital, 18 Taipei, Taiwan, 9Graduate Institute of Medicine, Kaohsiung Medical University, 19 Kaohsiung, Taiwan, 10Division of Thoracic Surgery and Department of Surgery, 20 National Taiwan University Hospital and National Taiwan University College of 21 Medicine, Taipei, Taiwan, 11Taipei Institute of Pathology, Taipei, Taiwan, 22 12Department of Pathology & Laboratory Medicine, Taichung Veterans General 23 Hospital, Taichung, Taiwan, 13Department of Internal Medicine, National Taiwan 24 University Hospital, Taipei, Taiwan, 14Institute of Biomedical Sciences, Academia 25 Sinica, Taipei, Taiwan, 15Department of Pathology and Graduate Institute of 26 Pathology, National Taiwan University College of Medicine, Taipei, Taiwan, 27 16Graduate Institute of Clinical Medicine, National Taiwan University College of 28 Medicine, Taipei, Taiwan, 17Institute of Medical Device and Imaging, College of 29 Medicine, National Taiwan University, Taipei, Taiwan 30 31 Running title: FAM198B is a novel tumor suppressor in lung cancer 32 33 Keywords: FAM198B, tumor suppressor, cell invasion, N-glycosylation, prognosis 34 35 Financial support 36 This study was supported by grants from the Ministry of Science and Technology, 37 Taiwan, (NSC 98-2314-B-002-120-MY3, NSC 102-2911-I-002-303, MOST 38 103-2911-I-002-303, MOST 104-2911-I-002-302, 104R8400). Mathematics in

1 Biology Group of the Institute of Statistical Science Academia Sinica and Taiwan 2 Biosignature Project of Lung Cancer supported data analysis work. 3 4 Corresponding author 5 Sung-Liang Yu, Department of Clinical Laboratory Sciences and Medical 6 Biotechnology, National Taiwan University College of Medicine; No. 1 Chang-Te 7 Street, Taipei 10048, Taiwan. 8 Telephone: +886-2-2312-3456 ext. 88697 9 Fax: +886-2-2395-8341 10 E-mail: [email protected] 11 12 Conflicts of interest 13 The authors declare no potential conflicts of interest. 14 15 TRANSLATIONAL RELEVANCE 16 Lung adenocarcinoma is the most common malignant tumor worldwide, and the 17 outcome of patients is still unsatisfactory with low survival rates. To reduce cancer 18 mortality caused by recurrence and metastasis, an in-depth understanding of the 19 mechanisms involved in cancer progression is urgently needed. Herein, we 20 demonstrate that FAM198B is a novel tumor suppressor that inhibits cancer 21 metastasis via attenuating pERK/MMP-1 signaling axis, and high FAM198B 22 expression is positively associated with overall survival in lung adenocarcinoma 23 patients in public database and in a 95-Taiwanese cohort. Interestingly, we find that 24 FAM198B is an N-glycoprotein, and the glycosylation can increase the protein 25 stability of FAM198B and is necessary for the metastasis-suppression activity. 26 Collectively, FAM198B represents a novel prognostic marker for predicting survival 27 of lung adenocarcinoma. 28 29 30 31 32 33 34 35 36 37 38

1 ABSTRACT 2 Purpose: The comprehensive understanding of mechanisms involved in the tumor 3 metastasis is urgently needed for discovering novel metastasis-related genes for 4 developing effective diagnoses and treatments for lung cancer. 5 Experimental design: FAM198B was identified from an isogenic lung cancer 6 metastasis cell model by microarray analysis. To investigate the clinical relevance of 7 FAM198B, the FAMB198B expression of 95 Taiwan lung adenocarcinoma patients 8 was analyzed by quantitative real-time PCR and correlated to patients’ survivals. The 9 impact of FAM198B on cell invasion, metastasis and tumor growth was examined by 10 in vitro cellular assays and in vivo mouse models. Additionally, the 11 N-glycosylation-defective FAM198B mutants generated by site-directed mutagenesis 12 was used to study protein stability and subcellular localization of FAM198B. Finally, 13 the microarray and pathway analysis were used to elucidate the underlying 14 mechanisms of FAM198B-mediated tumor suppression. 15 Results: We found that the high expression of FAM198B was associated with 16 favorable survival in Taiwan lung adenocarcinoma patients and in a lung cancer 17 public database. Enforced expression of FAM198B inhibited cell invasion, migration, 18 mobility, proliferation and anchorage-independent growth and FAM198B silencing 19 exhibited opposite activities in vitro. FAM198B also attenuated tumor growth and 20 metastasis in vivo. We further identified MMP-1 as a critical downstream target of 21 FAM198B. The FAM198B-mediated MMP-1 downregulation was via inhibition of 22 the phosphorylation of extracellular signal-regulated kinase (ERK). Interestingly 23 Deglycosylation nearly eliminated the metastasis suppression activity of FAM198B 24 due to a decrease of protein stability. 25 Conclusions: Our results implicate FAM198B as a potential tumor suppressor and to 26 be a prognostic marker in lung adenocarcinoma. 27 28 INTRODUCTION 29 Lung cancer is the leading cause of cancer-related deaths worldwide and lung 30 adenocarcinoma is the predominant histological subtype of lung cancer in woman, 31 never-smokers, and younger adults (1-3). Cancer development is a multi-phase 32 process resulting from genomic instability, transcriptional alterations, cancer stemness, 33 abnormal metabolic pathways, epigenetic alteration, tumor-promoting inflammation 34 and evasion of the immune system, which drive the progression of cancer (4-8). 35 Proteolytic degradation of the extracellular matrix (ECM) and the basement 36 membranes surrounding the primary tumor by the matrix metalloproteinase (MMPs) 37 is a critical step for tumor angiogenesis, invasion and metastasis (9). The MMPs have 38 served as potential prognostic markers and therapeutic targets in cancer (10).

1 Previously, we established an isogenic lung cancer metastasis cell model and 2 discovered several metastasis-related gene (11-18). However, the comprehensive 3 understanding of mechanisms involved in the tumor metastasis remains to further 4 explore. Thus, the discovery and understanding of novel metastasis-related genes are 5 crucial for developing more effective diagnoses and treatments for lung cancer. 6 The human family with sequence similarity 198, member B (FAM198B) is a 7 novel gene with unknown function and predicted to be a membrane-bound 8 glycoprotein localized on Golgi apparatus (19, 20). FAM198B might play a role in the 9 regulation of mouse and Xenopus embryonic development and is a downstream target 10 of the FGF receptor signaling pathway (21, 22). Nearly one-half of all known 11 eukaryotic proteins are N-glycosylated, which is a ubiquitous posttranslational 12 modification (23) and the alteration of glycosylation has been reported to be 13 associated with tumor proliferation, invasion, metastasis, angiogenesis, receptor 14 activation and intracellular or cell-matrix interactions (24-26). 15 In this study, we identified a novel potential tumor suppressor, FAM198B, from 16 an isogenic lung cancer metastasis cell model by expression microarrays. The clinical 17 relevance of FAM198B was analyzed in both public databases and Taiwanese lung 18 adenocarcinoma patients. The tumor suppression activity of FAM198B was 19 characterized by in vitro and in vivo metastasis and tumorigenesis assays. Microarray 20 and pathway analyses were used to investigate the molecular signaling of FAM198B 21 and the effect of glycosylation on FAM198B stability was also investigated. 22 23 MATERIALS AND METHODS 24 Human lung tumor specimens and RNA extraction 25 A total of 95 frozen tissues were collected from lung adenocarcinoma who underwent 26 complete surgical resection at the Taichung Veterans General Hospital (Taichung, 27 Taiwan) between May, 2000 and June, 2009. After surgical resection, half tumor 28 specimen was immediately frozen within 30 min and half was formalin-fixed, 29 paraffin-embedded (FFPE). The average tumor content and necrosis were 53.09 ± 30 21.10 and 10.35 ± 13.19, respectively (mean ± SD, n=92 for tumor content, n=90 for 31 necrosis). Total RNA was extracted from the frozen tumor tissue specimens by using 32 the TRIzol Plus RNA Purification Kit (Thermo Fisher Scientific, Waltham, MA). The 33 qRT-PCR of TATA-box binding protein (TBP) was used to assess the quality of RNAs. 34 Ct 40 was used as the cutoff value to define undetectable or detectable and specimens 35 with Ct less than 40 were enrolled in this study (27-29). This study was conducted in 36 accordance with the Declaration of Helsinki and Good Clinical Practice. Institutional 37 Review Boards approved all aspects of the study. All participants provided written 38 informed consent.

1 2 In vivo animal models 3 For the in vivo metastasis assay, 1 x 105 cells were suspended in 0.1 ml of PBS and 4 intravenously injected into the lateral tail vein of six-week-old SCID mice. All mice 5 were sacrificed 4 weeks post-injection, and the mouse lungs were removed and fixed 6 in 10% formalin. The lung surface tumor foci were counted under a dissecting 7 microscope. The mouse lungs were further embedded in paraffin, sectioned into 4μm 8 layers and stained with hematoxylin and eosin (H&E) for histological analysis. For 9 the subcutaneous tumor growth assay, 1 x 106 cells were suspended in 0.1 ml of PBS 10 and implanted subcutaneously into the dorsal region of six-week-old null mice. Tumor 11 growth was examined thrice a week, and tumor volume was estimated by the formula 12 LW2/2, where L is the length and W is the width of the tumor. Tumor weight was 13 measured at the end of the study. For the orthotopic tumor implantation assay, 5 x 104 14 cells were suspended in 20 μl of PBS containing 10 ng Matrigel (R&D Systems, 15 McKinley Place, NE) and inoculated into left lung of 6-week-old null mice. The mice 16 were sacrificed 45 days after implantation. The mouse lungs were removed and ﬁxed 17 in 10% formalin, and the size and number of lung tumor colonies were measured by 18 microscopic examination. All animal procedures were performed with the approval of 19 the Institutional Animal Care and Use Committee of the National Taiwan University. 20 21 Statistical analysis 22 Overall survival curves were calculated by Kaplan–Meier analysis and the log-rank 23 test was performed to test the difference between survival curves. Each cutoff point 24 for overall survival and disease-free survival for definition of the high/low-FAM198B 25 expression groups is listed in Supplementary Table S1, S3 and S4. Cox proportional 26 hazards regression analysis with stepwise selection method was used to evaluate 27 independent prognostic factors. Covariates of the regression model were FAM198B, 28 age, gender, stage, EGFR status, KRAS status, smoking history, and histologic 29 subtypes. For in vitro or in vivo studies, Student’s t-test was used to compare the 30 difference between two groups. All tests were two-tailed and P values < 0.05 were 31 considered significant. 32 33 Additional methods 34 Detailed methods on cell lines and culture conditions, plasmids and transfection, 35 viruses and transduction, real-time quantitative polymerase chain reaction, 36 immunoprecipitation and immunoblot, in vitro invasion and migration assays, 37 single-cell tracking migration assay, cell proliferation assay and 38 anchorage-independent growth assay, microarray and pathway analysis, glycosidase

1 assay, identification of intact N-glycopeptides by LC-MS/MS, protein stability assay, 2 and immunoﬂuorescence analysis are described in the Supplementary Methods. 3 4 RESULTS 5 Discovery of FAM198B in an isogenic lung cancer metastasis cell model 6 To investigate which novel candidate genes contribute to metastasis of lung 7 cancer, the genome-wide RNA expression of a well-established isogenic metastasis 8 cell line model, including low metastatic CL1-0 cells and high metastatic CL1-5 cells, 9 was profiled by microarray analyses (30, 31). Through comparison of the 10 differentially expressed genes between CL1-0 and CL1-5 cells, we found a gene, 11 FAM198B, which mRNA expression of CL1-5 cells was 130-fold lower than that of 12 in the parental CL1-0 cells. The differential expression of FAM198B was confirmed 13 by real-time quantitative RT-PCR (qRT-PCR) in CL1-0 compared with CL1-5 cells 14 (86.15 ± 14.08 and 1.00 ± 0.27, P<0.05) (Supplementary Figure S1). 15 16 Down-regulation of FAM198B is associated with poor overall survival in lung 17 adenocarcinoma patients 18 Next, we assessed whether the FAM198B expression is associated with survival 19 of lung adenocarcinoma patients. We first conducted a survival analysis using the 20 publicly available Memorial Sloan Kettering Cancer Center (MSKCC) microarray 21 dataset (32). There was a significant association between FAM198B expression and 22 overall survival in the dataset of 104 lung adenocarcinoma patients. Patients with low 23 FAM198B expression had a worse overall survival compared with those with high 24 FAM198B expression (Figure 1A and Supplementary Table S1). 25 To further validate this correlation in Asia population, 95 lung adenocarcinoma 26 patients were enrolled in Taiwan to detect the FAM198B expression in tumor 27 specimens by qRT-PCR. The clinical characteristics of these patients are summarized 28 in Supplementary Table S2. Patients with low FAM198B expression had worse 29 overall survival than those with high FAM198B expression (Figure 1B and 30 Supplementary Table S3), but no significance was found between FAM198B 31 expression and progression-free survival at any defined cutoff of FAM198B 32 expression (Supplementary Table S4). Result of the multivariate Cox proportional 33 hazard regression analysis with stepwise selection method showed that that high 34 FAM198B expression (hazard ratio [HR] =0.41, 95% confidence interval [CI]= 35 0.18-0.91; P=0.029), late stage (HR=4.16, 95% CI =1.79-9.70; P=0.001) and EGFR 36 mutation (HR=0.44, 95% CI=0.20-1.00; P=0.049) were independent prognostic 37 factors of the overall survival. (Table 1). It suggested that FAM198B may function as 38 a tumor suppressor in lung adenocarcinoma.

1 2 FAM198B inhibits cancer invasion and anchorage-independent growth in vitro 3 Based on the clinical finding, we then explored the potential role of FAM198B in 4 the invasiveness and mobility of lung cancer cells. First, the endogenous FAM198B 5 expressions of 11 lung cancer cell lines were determined by qRT-PCR (Supplementary 6 Figure S2) and the high and low FAM198B-expressing cells were identified for 7 further functional experiments. To determine whether FAM198B can modulate cell 8 invasion and migration, FAM198B was ectopically expressed in three endogenously 9 low FAM198B-expressing cell lines, CL1-5, H226, and A549. We found that 10 FAM198B significantly inhibits the invasion and migration abilities of these lung 11 cancer cell lines (Figure 2A). 12 Five, shFAM198B#2, #5, #7, #8, and #9, out of nine shFAM198B vectors 13 purchased from The RNAi Consortium effectively inhibited FAM198B expression 14 assayed by HEK293T cell transfection and immunoblotting assay (Supplementary 15 Figure S3A). The knockdown of FAM198B in two endogenously high 16 FAM198B-expressing cell lines, CL1-0 and EKVX, by transduction of 17 shFAM198B#5 lentivirus significantly enhanced the activities of invasion and 18 migration (Figure 2B). The FAM198B suppressive activity was also confirmed in 19 CL1-0 cells by the other two constructs, shFAM198B#2 and shFAM198B#7 20 lentiviruses (Supplementary Figure S3B, left panel). Next, we performed a single-cell 21 tracking assay to assess the effect of FAM198B on cell mobility directly. The 22 migration rate and directionality of CL1-5/FAM198B cells were greatly reduced from 23 4.49 ± 3.05 to 1.57 ± 0.16 and from 0.55 ± 0.14 to 0.29 ± 0.07, respectively, compared 24 with the CL1-5/mock control cells. In contrast, CL1-0/shFAM198B#5 cells showed 25 an increase in cell migratory rate and directionality (6.06 ± 0.78 and 0.38 ± 0.10, 26 respectively) compared with CL1-0/shLacZ cells (1.85 ± 0.29 and 0.15 ± 0.06, 27 respectively) (Figure 2C). 28 To explore whether FAM198B modulates other cancer phenotypic functions, we 29 performed 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and 30 anchorage-independent growth assays to assess the role of FAM198B on cell 31 proliferation and colonogenesis. We found that the cell proliferation rate and 32 colony-forming ability of stably FAM198B-expressing CL1-5 cells were significantly 33 reduced compared with the mock control cells (Figure 2D-E, left panel). Conversely, 34 knockdown of FAM198B increased the cell proliferation of CL1-0 cells compared 35 with the shLacZ control group by three different shFAM198B lentiviruses (Figure 2D, 36 right panel and Supplementary Figure S3B, middle panel). Similarly, inhibitory 37 activity of FAMB198B on colony formation was found (Figure 2E, right panel). 38 These data implied that FAM198B might act as a tumor suppressor to inhibit the

1 growth and invasion of lung cancer cells in vitro. 2 3 FAM198B inhibits lung tumor metastasis and tumorigenesis in vivo 4 To further confirm the metastasis suppression activity of FAM198B in vivo, a 5 mouse metastasis assay was performed, in which the stably FAM198B-expressing 6 CL1-5 cells or mock control cells were intravenously injected into the lateral tail vein 7 of SCID mice. Mice were sacriﬁced after 4 weeks, and lung metastases were counted 8 by a dissection microscope. The number of metastatic nodules in the mouse lungs was 9 significantly reduced in the FAM198B-expressing group compared to the mock 10 control group (170.13 ± 50.24 and 604.50 ± 58.77, n=8 for each group, Figure 3A). 11 Furthermore, to mimic the entire lung metastasis process, an orthotopic lung 12 implantation experiment was performed (16). Tumor cells were orthotopically 13 inoculated into the left lobe of mice lung, and we examined the metastatic nodules 14 formed on the mice right lung. FAM198B group formed not only smaller orthotopic 15 tumor growth (0.73 ± 0.31 and 2.84 ± 0.69, respectively) but also decreased the 16 number of lung metastatic tumors (2.16 ± 1.32 and 4.66 ± 1.03, respectively) 17 compared with the mock control group (Figure 3B and 3C). The effect of FAM198B 18 on mouse survival was evaluated using another orthotopic implantation experiment. 19 As Figure 3D shown, FAM198B-expressing group had a longer survival compared 20 with the mock control group (medium survival 39 v.s. 27 days; P=0.003). Next, we 21 used the subcutaneous xenograft mouse models to assess the impact of FAM198B in 22 tumor growth in vivo. The stably FAM198B-expressing CL1-5 cells or mock control 23 cells were subcutaneously injected into the flanks of nude mice, and the tumor 24 volumes were measured thrice a week. At 30 days after implantation, the mice were 25 sacriﬁced. The FAM198B-expressing group exhibited a significant reduction in both 26 tumor volume and weight compared with those of the mock control group (Figure 3E). 27 These data implicated that FAM198B inhibits tumorigenesis and metastasis of lung 28 cancer in vivo. 29 30 FAM198B is an N-linked glycosylated protein 31 Occasionally, we noticed two bands in the immunoblot of overexpressed 32 FAM198B-V5. The position of the minor lower band is close to the predicted 33 molecular weight, 64 kDa. The presence of multiple protein bands is often due to 34 post-translational modification such as glycosylation (33). Thus, we used the publicly 35 available neXtProt database (http://www.nextprot.org/) and NetNGlyc server 36 (http://www.cbs.dtu.dk/services/NetNGlyc/) to predict the possible post-translational 37 modifications of FAM198B, particularly for glycosylation sites. We found two 38 potential N-glycosylation sites at asparagine 98 and 289 and two potential

1 O-glycosylation sites at serine 137 and 149. All of these glycosylation sites are 2 located in the predicted extracellular domain of the Golgi apparatus membrane 3 (Figure 4A) (20). After Endo H digestion, the intermediate band converted to a lower 4 band, suggesting that the intermediate band is a high mannose-type glycoprotein. The 5 molecular weight of both upper and intermediate bands was simultaneously decreased 6 after PNGase F treatment (Figure 4B, left panel). These results indicated that 7 FAM198B might contain hybrid or complex N-linked glycosylation. O-glycosidase 8 treatment only resulted in a slight molecular shift of both the upper and intermediate 9 bands compared with the combined treatment of O-glycosidase and PNGase F (Figure 10 4B; right panel). Furthermore, cells were cultured in presence of an N-glycosylation 11 inhibitor, tunicamycin, for 24 h, and the FAM198B glycosylation was severely 12 impaired and similar to that observed in PNGase F treatment (Figure 4C). 13 To examine which predicted site is responsible for FAM198B glycosylation, first, 14 two potential N-glycosylation sites of FAM198B were mutated from asparagine to 15 glutamine, individually or simultaneously (N98Q, N289Q and 2NQ), and two 16 potential O-glycosylation sites were mutated from serine to alanine (S137A, S149A 17 and 2SA). The mutant proteins were transiently expressed in CL1-5 cells and 18 subjected to immunoblotting assays. We found that the glycosylation was strongly 19 impeded in FAM198B/N289Q and FAM198B/2NQ and slightly in FAM198B/N98Q. 20 Conversely, the molecular shift of FAM198B/S137A, FAM198B/S149A and 21 FAM198B/2SA was not obvious (Figure 4D). To clarify whether Asn98 and Asn289 22 are the major N-glycosylation sites, the transiently FAMB198-expressing CL1-5 cells 23 were treated with tunicamycin and analyzed by immunoblotting. Figure 4E indicated 24 that the N-glycosylation inhibitor treatment resulted in an obvious molecular weight 25 shift in wild type FAM198B and FAM198B/N98Q but not in FAM198B/N289Q and 26 FAM198B/2NQ. By contrast, the O-glycosidase treatment resulted in a similar and 27 minor shift pattern in wild type, N98Q, N289Q and 2NQ mutants of FAM198B 28 (Figure 4F). 29 Furthermore, to identify the glycan structure on Asn98 and Asn289, V5-tagged 30 FAM198B proteins were purified from CL1-5/FAM198B cells and analyzed by 31 nanoscale liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). 32 After identifying by Byonic software and manually confirming each MS/MS spectrum, 33 we found that high mannose types, Man(7-10)GlcNAc(2) presented on Asn289 34 (Figure 4G). Additionally, we identified hybrid and complex-type of tri-antennary 35 glycoforms with sialic acid and core- or terminal fucose on Asn98 while another 36 complex poly-LacNAc with core-fucosylation located on Asn289 (Supplementary 37 Figure S4B; Supplementary Table S5). Most interestingly, we also identified another 38 unpredicted N-linked glycosylation site, Asn322, which had highly fucosylation

1 (Supplementary Figure S4A; Supplementary Table S5). To examine whether the 2 Asn322 was occupied by a glycosyl residue, we mutated Asn322 from asparagine to 3 glutamine (N322Q) and generated a triple-mutant (3NQ) containing N98Q, N289Q 4 and N322Q. Then FAM198B/N322Q or FAM198B/3NQ proteins were transiently 5 expressed in CL1-5 cells and immunoblotted by anti-V5 antibody. Figure 4H 6 indicated that the glycosylation was slightly impeded in FAM198B/N322Q and 7 markedly destroyed in FAM198B/3NQ. These results supported that Asn98, Asn289 8 and Asn322 are three major N-glycosylation sites of human FAM198B. 9 10 N-glycosylation stabilizes FAM198B 11 To determine the biological function of N-glycosylation, we examined the 12 subcellular localization and expression of FAM198B by immunofluorescence and 13 immunoblot assays. N-glycosylation-defective mutations did not influence cellular 14 localization, Golgi apparatus, but the expression of N-glycosylation-defective 15 FAM198B mutants was much lower than wild type FAM198B in transiently 16 FAM198B-overexpressing CL1-5 cells (Supplementary Figure S5, Figure 4D and 4H). 17 We next determined whether N-glycosylation affects FAM198B stability. The de novo 18 FAM198B synthesis was blocked by cycloheximide in CL1-5 cells and FAM198B 19 were detected by immunoblotting (Supplementary Figure S6A). The half-lives of

20 N-glycosylation-defective mutants, T1/2= 4.1, 3.1, 0.1, 3.8 and 0 h for N98Q, N289Q,

21 N322Q, 2NQ and 3NQ, were shorter than that of wild type FAM198B, T1/2= 8.3, 22 (Supplementary Figure S6B). Subsequently, to evaluate whether the protein 23 degradation of N-glycosylation-defective FAM198B mutants was through 26S 24 proteasome pathway. HEK293T cells were transiently cotransfected with wild type 25 FAM198B, FAM198B/3NQ mutant and hemagglutinin (HA)-tagged 26 ubiquitin–expressing vectors in presence of proteasome inhibitor MG132 or not. As 27 shown in Supplementary Figure S7, FAM198B/3NQ exhibited more ubiquitination 28 than that of wild type FAM198B in the presence of MG132. Taken together, these 29 results implied that the N-glycosylation is critical for protein stability but not for 30 intracellular trafficking of FAM198B. 31 32 FAM198B suppresses cancer cell invasion and metastasis through ERK-mediated 33 MMP-1 inhibition 34 To elucidate the underlying mechanism by which FAM198B inhibits invasiveness, 35 the transcriptomic expression microarray was performed. We identified 1,580 genes 36 that were significantly altered in CL1-5/FAM198B compared with CL1-5/mock 37 control cells, with a greater than two-fold change and under a false discovery rate 38 (FDR) < 0.05. The differentially expressed genes were subjected to pathway analysis

1 by the MetaCore analytical suite. Consistent to in vitro and in vivo results, 12 out of 2 Top 20 ranking FAM198B-altered pathways were mainly involved in cell invasion 3 and cell proliferation/apoptosis (Supplementary Table S6). MMP-1 was found in three 4 cell invasion pathways (Supplementary Figure S8-10). We found that FAM198B 5 suppressed both mRNA and protein expressions of MMP-1. It implied that FAM198B 6 might suppress cancer cell invasion through MMP-1 inhibition (Figures 5A and 5B). 7 Based on pathway analysis, the mitogen-activated protein kinase 8 (MAPK)-extracellular signal-regulated kinase (ERK) signaling pathway may be 9 involved in FAM198B-regulated MMP1 expression (Supplementary Figure S9), and 10 ERK/MAPK signaling is known to be required for MMP-1-mediated tumor invasion 11 and metastasis (34, 35). We next examined whether FAM198B downregulates MMP-1 12 expression by inhibiting the ERK pathway. Immunoblotting revealed that the ERK 13 phosphorylation was obviously decreased whereas the protein level of MMP-1 was 14 also decreased in CL1-5/FAM198B cells compared with CL1-5/mock control cells 15 (Figure 5B). 16 To further confirm the relationship between FAM198B and MMP-1, the stably 17 FAM198B-silencing CL1-5 cells were generated by shFAM198B#2 lentivirus 18 infection. The invasion ability of CL1-5/FAM198B/shFAM198B#2 cells was 19 increased compared with CL1-5/FAM198B or CL1-5/FAM198B/shLacZ control cells, 20 and immunoblot assay showed that MMP-1 expression and ERK phosphorylation are 21 up-regulated upon FAM198B knockdown (Figure 5C). Furthermore, MMP-1 22 expression was suppressed by two different shMMP-1 lentiviruses and MMP-1 23 knockdown suppressed cell invasion ability in CL1-5/FAM198B/shFAM198B#2 24 (Figure 5D). Besides, to verify whether FAM198B-induced MMP-1 suppression is 25 through ERK signaling, CL1-5 cells were treated with a specific MEK1/2 inhibitor, 26 AZD6244, for 24 h (36). The immunoblot results revealed that ERK phosphorylation 27 was significantly decreased, and MMP-1 expression was suppressed with a 28 concentration of AZD6244 as low as 0.1 μg/ml (Supplementary Figure S11). We also 29 found that the FAM198B knockdown-induced upregulation of ERK phosphorylation 30 and MMP-1 expression were markedly suppressed by AZD6244 treatment (Figure 31 5E). These results indicated that FAM198B suppressed MMP-1 expression through 32 inactivating ERK signaling, at least partly. Finally, we investigated whether 33 N-glycosylation affects FAM198B-induced invasion suppression. CL1-5 cells were 34 transiently transfected with N-glycosylation-defective FAM198B-expressing vectors 35 and analyzed for invasion activities. We found that N-glycosylated FAM198B 36 significantly inhibits cancer cell invasion, whereas N-glycosylation-defective 37 FAM198B did not and all of the N-glycosylation-defective FAM198B mutants fail to 38 suppress ERK phosphorylation and MMP-1 expression compared with wild type

1 FAM198B (Figure 5F and Supplementary Figure S12). Collectively, our data revealed 2 that complex N-glycosylation of FAM198B plays an important role in the 3 pERK/MMP-1 pathway and regulates cancer metastasis. 4 5 DISCUSSION 6 In this study, we identify FAM198B as a novel potential tumor suppressor in lung 7 adenocarcinoma. The patient survival analysis indicated that low FAM198B 8 expression is associated with shorter overall survival in lung adenocarcinomas. 9 Overexpression of FAM198B inhibited cancer cell invasion, proliferation and 10 tumorigenesis in both in vitro and in vivo assays, whereas knockdown of FAM198B 11 promoted the malignancies. FAM198B inhibits cancer invasion at least partly through 12 downregulating the pERK/MMP-1 signaling pathway and N-glycosylation enhances 13 the stability of FAM198B protein (Figure 5G). 14 Protein glycosylation is an important post-translational modification that occurs 15 in the ER and Golgi apparatus and is regulated by complicated mechanisms, including 16 the localization and expression of glycosyltransferases and the molar ratio of 17 glycosyltransferases to substrates. It is well known that the glycosylation is important 18 for protein functions. For example, the altered glycosylation of membrane receptors 19 interferes the downstream signaling via hindering receptor oligomerization and ligand 20 binding and in turn impacts a wide range of key cellular processes, including cell 21 division, differentiation and localization, and even the progression and malignancy of 22 cancer cells (37-39). First, we showed that both upper and intermediate bands of 23 FAM198B were N-glycosylated based on the results of tunicamycin experiments. 24 However, the Endo H and PNGase F digestion and asparagine site-directed 25 mutagenesis, Asn98 and Asn289, failed to shift FAM198B to expected molecular 26 weight. These facts suggested that FAM198B might have additional glycosylations. 27 Hence, we used LC-MS/MS to characterize N-glycosylation sites and glycan 28 structures of FAM198B and identified another N-linked glycosylation on Asn322 site. 29 The molecular weight of 3NQ was also matched to the expected size of FAM198B. 30 Taken together, these data revealed that the glycosylation of Asn322 might contribute 31 to the small fractions with higher molecular weight presented on N98Q, N289Q, and 32 2NQ of FAM198B. 33 In the immunofluorescence assay, we characterized the subcellular localization 34 of FAM198B by using an anti-V5/FITC antibody in CL1-5 cells instead of using an 35 anti-FAM198B antibody because both commercial available FAM198B antibodies 36 raised by either synthetic peptides or bacterial recombinant proteins failed to detect 37 recombinant FAM198B proteins due to poor quality. In turn, we found that 38 N-glycosylation does not influence the FAM198B trafficking but affects the turnover

1 rate of FAM198B. Elimination of single or all glycosylation sites of FAM198B 2 accelerated the FAM198B protein degradation and resulted in a lower steady-state 3 level of FAM198B. Besides, we also found that the glycosylation of FAM198B 4 inhibits 26S proteasome-mediated protein degradation. A Transwell invasion assay 5 demonstrated that N-glycosylation-defective FAM198B mutants completely lost the 6 invasion suppression activity compared with wild type FAM198B. Given the lack of 7 FAM198-specific antibody, we can not exclude the possibility that certain 8 FAM198B-induced phenotypic alterations are due to overexpression artifact and not 9 reflecting a genuine FAM198B. To minimize this possibility we validated the 10 invasion suppressive activity of enforced FAM198B expression in three cell lines and 11 performed RNA silencing of endogenous FAM198B in two cell lines with three 12 different shRNA constructs. All of these data showed the suppressive activity of 13 FAM198B seems not due to overexpression artifact although there is no available 14 anti-FAM198B antibody. Similarly, the negative correlation of FAM198B with 15 patient survival assayed by qRT-PCR cannot be confirmed by FAM198B 16 immunohistochemistry staining but this clinical finding has provided a new insight of 17 biological function of FAM198B in lung cancer progress. 18 Collagenase-1 (MMP-1) is one of the major proteases and cleaves interstitial 19 collagens with a preference for type I, II and III collagens. MMP-1 has been shown to 20 potently facilitate cancer invasion and metastasis and its expression correlates with 21 poor clinical outcomes and increases recurrence (40-42). It has previously been shown 22 that the ERK and p38 mitogen-activated protein kinase (MAPK) pathways are the 23 major regulators for MMP-1 expression (43, 44). The role of p38 MAPK on MMP-1 24 expression is controversial in certain situations (45). Recent studies showed that 25 EGFR-mediated p38 MAPK signaling pathway augments MMP-1 expression and 26 then leads to promote cancer tumorigenesis and angiogenesis (46). Other studies 27 indicated that p38 MAPK inhibits MMP-1 expression (47, 48). Although certain 28 studies indicated that p38 MAPK is involved in MMP-1 regulation (45-48), but p38 29 MAPK did not play an major role in FAM198B-mediated MMP-1 regulation 30 (Supplementary Figure S13). In contrast, the constitutive activation of the 31 ERK/MAPK pathway not only increases cancer cell proliferation but also enhances 32 MMP-1 expression, which in turn increases tumor invasion and metastasis. Our data 33 also revealed that MMP-1 acts as the major downstream effector of 34 FAM198B-modulated invasion via ERK signaling. However, it remains to further 35 investigate how FAM198B suppresses ERK phosphorylation. Furthermore, we found 36 that N-glycosylation-defective FAM198B mutants, regardless of the degree of 37 deglycosylation, fails to inhibit ERK/MMP-1 pathway. Deglycosylation altered 38 FAM198B stability, shortened the half-life of FAM198B, and reduced FAM198B

1 availability. It may be the reason that FAM198B/2NQ and FAM198B/3NQ mutants 2 did not show synergistic effect on cell invasion, ERK phosphorylation and MMP-1 3 expression. 4 Up to date, there is no available agent could specifically regulate FAM198B 5 glycosylation in vitro or in vivo, but FAM198B might serve as a candidate for drug 6 screening. Further study is needed to identify the glycosyltransferase(s) specific to 7 FAM198B. Selecting drugs that enhance the activity of the FAM198B-specific 8 glycosyltransferase(s) might be another strategy of anti-cancer drug development. 9 Collectively, our findings provide a new insight into the molecular mechanism by 10 which FAM198B inhibits tumor invasion, metastasis, and tumorigenesis through the 11 pERK-mediated MMP-1 signaling pathway, and the complex post-translational 12 N-glycosylation serves a crucial role in stabilizing FAM198B. 13 14 ACKNOWLEDGEMENTS 15 We thank the National RNAi Core Facility, the Integrated Core Facility for Functional 16 Genomics of the National Core Facility Program for Biotechnology, the Microarray 17 Core Facility of the National Taiwan University Center of Genomic Medicine, and 18 Common Mass Spectrometry Facilities of Institute of Biological Chemistry, Academia 19 Sinica for technical support. 20 21 REFERENCES 22 1. Johnson DH, Schiller JH, Bunn PA. Recent Clinical Advances in Lung Cancer 23 Management. Journal of Clinical Oncology. 2014;32:973-82. 24 2. Chen Z, Fillmore CM, Hammerman PS, Kim CF, Wong K-K. Non-small-cell 25 lung cancers: a heterogeneous set of diseases. Nat Rev Cancer. 2014;14:535-46. 26 3. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA: a cancer journal for 27 clinicians. 2015;65:5-29. 28 4. Boroughs LK, DeBerardinis RJ. Metabolic pathways promoting cancer cell 29 survival and growth. Nat Cell Biol. 2015;17:351-9. 30 5. Steeg PS. Targeting metastasis. Nat Rev Cancer. 2016;16:201-18. 31 6. Yuan S, Yu S-L, Chen H-Y, Hsu Y-C, Su K-Y, Chen H-W, et al. Clustered 32 Genomic Alterations in Chromosome 7p Dictate Outcomes and Targeted Treatment 33 Responses of Lung Adenocarcinoma With EGFR-Activating Mutations. J Clin Oncol. 34 2011;29:3435-42. 35 7. Chen W-J, Ho C-C, Chang Y-L, Chen H-Y, Lin C-A, Ling T-Y, et al. 36 Cancer-associated fibroblasts regulate the plasticity of lung cancer stemness via 37 paracrine signalling. Nature communications. 2014;5. 38 8. Su K-Y, Chen H-Y, Li K-C, Kuo M-L, Yang JC-H, Chan W-K, et al.

1 Pretreatment Epidermal Growth Factor Receptor (EGFR) T790M Mutation Predicts 2 Shorter EGFR Tyrosine Kinase Inhibitor Response Duration in Patients With 3 Non–Small-Cell Lung Cancer. J Clin Oncol. 2012;30:433-40. 4 9. Shay G, Lynch CC, Fingleton B. Moving targets: Emerging roles for MMPs in 5 cancer progression and metastasis. Matrix Biol. 2015;44–46:200-6. 6 10. Du X, Lin BC, Wang Q-R, Li H, Ingalla E, Tien J, et al. MMP-1 and 7 Pro-MMP-10 as Potential Urinary Pharmacodynamic Biomarkers of FGFR3-Targeted 8 Therapy in Patients with Bladder Cancer. Clin Cancer Res. 2014;20:6324-35. 9 11. Chao YC, Pan SH, Yang SC, Yu SL, Che TF, Lin CW, et al. Claudin-1 is a 10 metastasis suppressor and correlates with clinical outcome in lung adenocarcinoma. 11 Am J Respir Crit Care Med. 2009;179:123-33. 12 12. Tsai MF, Wang CC, Chang GC, Chen CY, Chen HY, Cheng CL, et al. A new 13 tumor suppressor DnaJ-like heat shock protein, HLJ1, and survival of patients with 14 non-small-cell lung carcinoma. J Natl Cancer Inst. 2006;98:825-38. 15 13. Hsiao Y-J, Su K-Y, Hsu Y-C, Chang G-C, Chen J-S, Chen H-Y, et al. SPANXA 16 suppresses EMT by inhibiting c-JUN/SNAI2 signaling in lung adenocarcinoma; 2016. 17 14. Wu YY, Peck K, Chang YL, Pan SH, Cheng YF, Lin JC, et al. SCUBE3 is an 18 endogenous TGF-[beta] receptor ligand and regulates the epithelial-mesenchymal 19 transition in lung cancer. Oncogene. 2011;30:3682-93. 20 15. Chen C-C, Chen H-Y, Su K-Y, Hong Q-S, Yan B-S, Chen C-H, et al. Shisa3 Is 21 Associated with Prolonged Survival through Promoting β-Catenin Degradation in 22 Lung Cancer. American journal of respiratory and critical care medicine. 23 2014;190:433-44. 24 16. Pan S-H, Chao Y-C, Hung P-F, Chen H-Y, Yang S-C, Chang Y-L, et al. The 25 ability of LCRMP-1 to promote cancer invasion by enhancing filopodia formation is 26 antagonized by CRMP-1. The Journal of Clinical Investigation. 2011;121:3189-205. 27 17. Hung P-F, Hong T-M, Hsu Y-C, Chen H-Y, Chang Y-L, Wu C-T, et al. The Motor 28 Protein KIF14 Inhibits Tumor Growth and Cancer Metastasis in Lung 29 Adenocarcinoma. PLoS One. 2013;8:e61664. 30 18. Chen C-H, Chuang S-M, Yang M-F, Liao J-W, Yu S-L, Chen JJW. A Novel 31 Function of YWHAZ/β-Catenin Axis in Promoting Epithelial–Mesenchymal 32 Transition and Lung Cancer Metastasis. Mol Cancer Res. 2012;10:1319-31. 33 19. Simpson JC, Wellenreuther R, Poustka A, Pepperkok R, Wiemann S. Systematic

34 subcellular localization of novel proteins identified by large‐ scale cDNA sequencing;

35 2000. 36 20. Lindskog C, Korsgren O, Pontén F, Eriksson JW, Johansson L, Danielsson A. 37 Novel pancreatic beta cell-specific proteins: Antibody-based proteomics for

1 identification of new biomarker candidates. J Proteomics. 2012;75:2611-20. 2 21. Meszaros R, Strate I, Pera EM, Durbeej M. Expression of the novel gene Ened 3 during mouse and Xenopus embryonic development. Int J Dev Biol. 4 2008;52:1119-22. 5 22. Meszaros R, Akerlund M, Hjalt T, Durbeej M, Ekblom P. Gene expression 6 profiling of differentiating embryonic stem cells expressing dominant negative 7 fibroblast growth factor receptor 2. Matrix Biol. 2007;26:197-205. 8 23. Helenius A, Aebi, Markus. Intracellular Functions of N-Linked Glycans. Science. 9 2001;291:2364-9. 10 24. Hakomori S. Glycosylation defining cancer malignancy: new wine in an old 11 bottle. Proc Natl Acad Sci U S A. 2002;99:10231-3. 12 25. Häuselmann I, Borsig L. Altered tumor cell glycosylation promotes metastasis. 13 Front Oncol. 2014;4. 14 26. Simizu S, Takagi S, Tamura Y, Osada H. RECK-Mediated Suppression of Tumor 15 Cell Invasion Is Regulated by Glycosylation in Human Tumor Cell Lines. Cancer Res. 16 2005;65:7455-61. 17 27. Tan IL, Dimamay MP, Buerano CC, Alfon JA, Tanig CZ, Matias RR, et al. 18 Development and evaluation of a fluorogenic real-time RT-PCR for the detection of 19 dengue 3 virus. J Med Virol. 2010;82:2053-63. 20 28. Laue T, Emmerich P, Schmitz H. Detection of dengue virus RNA in patients after 21 primary or secondary dengue infection by using the TaqMan automated amplification 22 system. J Clin Microbiol. 1999;37:2543-7. 23 29. Houng HS, Chung-Ming Chen R, Vaughn DW, Kanesa-thasan N. Development 24 of a fluorogenic RT-PCR system for quantitative identification of dengue virus 25 serotypes 1-4 using conserved and serotype-specific 3' noncoding sequences. J Virol 26 Methods. 2001;95:19-32. 27 30. Chu YW, Yang PC, Yang SC, Shyu YC, Hendrix MJ, Wu R, et al. Selection of 28 invasive and metastatic subpopulations from a human lung adenocarcinoma cell line. 29 American journal of respiratory cell and molecular biology. 1997;17:353-60. 30 31. Chen JJW, Peck K, Hong T-M, Yang S-C, Sher Y-P, Shih J-Y, et al. Global 31 Analysis of Gene Expression in Invasion by a Lung Cancer Model. Cancer Res. 32 2001;61:5223-30. 33 32. Shedden K, Taylor JM, Enkemann SA, Tsao MS, Yeatman TJ, Gerald WL, et al. 34 Gene expression-based survival prediction in lung adenocarcinoma: a multi-site, 35 blinded validation study. Nat Med. 2008;14:822-7. 36 33. An HJ, Peavy TR, Hedrick JL, Lebrilla CB. Determination of N-Glycosylation 37 Sites and Site Heterogeneity in Glycoproteins. Anal Chem. 2003;75:5628-37. 38 34. Keld R, Guo B, Downey P, Gulmann C, Ang YS, Sharrocks AD. The ERK MAP

1 kinase-PEA3/ETV4-MMP-1 axis is operative in oesophageal adenocarcinoma. Mol 2 Cancer Res. 2010;9:313. 3 35. Rohani MG, Pilcher BK, Chen P, Parks WC. Cdc42 inhibits ERK-mediated 4 collagenase-1 (MMP-1) expression in collagen-activated human keratinocytes. J 5 Invest Dermatol. 2014;134:1230-7. 6 36. Catalanotti F, Solit DB, Pulitzer MP, Berger MF, Scott SN, Iyriboz T, et al. Phase 7 II Trial of MEK Inhibitor Selumetinib (AZD6244, ARRY-142886) in Patients with 8 BRAFV600E/K-Mutated Melanoma. Clin Cancer Res. 2013;19:2257-64. 9 37. Hu P, Berkowitz P, Madden VJ, Rubenstein DS. Stabilization of plakoglobin and 10 enhanced keratinocyte cell-cell adhesion by intracellular O-glycosylation. J Biol 11 Chem. 2006;281:12786-91. 12 38. Liu YC, Yen HY, Chen CY, Chen CH, Cheng PF, Juan YH, et al. Sialylation and 13 fucosylation of epidermal growth factor receptor suppress its dimerization and 14 activation in lung cancer cells. Proc Natl Acad Sci U S A. 2011;108:11332-7. 15 39. Stowell SR, Ju T, Cummings RD. Protein Glycosylation in Cancer. Annual 16 Review of Pathology: Mechanisms of Disease. 2015;10:473-510. 17 40. Vihinen P, Kahari VM. Matrix metalloproteinases in cancer: prognostic markers 18 and therapeutic targets. Int J Cancer. 2002;99:157-66. 19 41. Bergamaschi A, Tagliabue E, Sorlie T, Naume B, Triulzi T, Orlandi R, et al. 20 Extracellular matrix signature identifies breast cancer subgroups with different 21 clinical outcome. J Pathol. 2008;214:357-67. 22 42. Bonnans C, Chou J, Werb Z. Remodelling the extracellular matrix in 23 development and disease. Nat Rev Mol Cell Biol. 2014;15:786-801. 24 43. Ferguson J, Arozarena I, Ehrhardt M, Wellbrock C. Combination of MEK and 25 SRC inhibition suppresses melanoma cell growth and invasion. Oncogene. 26 2013;32:86-96. 27 44. Xia P, Zhang R, Ge G. C/EBPbeta Mediates TNF-alpha-Induced Cancer Cell 28 Migration by Inducing MMP Expression Dependent on p38 MAPK. J Cell Biochem. 29 2015;116:2766-77. 30 45. Dhillon AS, Hagan S, Rath O, Kolch W. MAP kinase signalling pathways in 31 cancer. Oncogene. 2007;26:3279-90. 32 46. Kim D, Dai J, Park Y-h, Yenwong Fai L, Wang L, Pratheeshkumar P, et al. 33 Activation of EGFR/p38/HIF-1α is pivotal for angiogenesis and tumorigenesis of 34 malignantly transformed cells induced by hexavalent chromium. J Biol Chem. 2016. 35 47. Endo H, Utani A, Shinkai H. Activation of p38 MAPK suppresses matrix 36 metalloproteinase-1 gene expression induced by platelet-derived growth factor. 37 Archives of Dermatological Research. 2003;294:552-8. 38 48. Xiang T, Fei R, Wang Z, Shen Z, Qian J, Chen W. Nicotine enhances invasion

1 and metastasis of human colorectal cancer cells through the nicotinic acetylcholine 2 receptor downstream p38 MAPK signaling pathway. Oncol Rep. 2016;35:205-10. 3 4 TABLES Table 1. Prognostic prediction of FAM198B by multivariate Cox regression analysis. Variable Hazard Ratio (95% C.I.) P value

FAM198B

-Low expression Reference

-High expression 0.41 (0.18-0.91) 0.029

Stage

-I and II Reference

- III and IV 4.16 (1.79-9.70) 0.001

EGFR mutation

-Wild type Reference

-Mutant 0.44 (0.20-1.00) 0.049

5 6 FIGURE LEGENDS 7 Figure 1. High FAM198B expression is associated with favorable outcome in lung 8 adenocarcinoma. (A) Kaplan-Meier analysis of overall survival using 104 9 adenocarcinoma patients from the publicly available MSKCC lung cancer microarray 10 datasets (log rank, P=0.001). Cutoff point for separation of high and low FAM198B 11 expression groups was 25%. (B) Kaplan-Meier analysis of overall survival in 95 12 patients with lung adenocarcinoma according to the expression of FAM198B assayed 13 by qRT-PCR. The patients were divided into high and low FAM198B expression 14 groups using the same cutoff point in (A) (log rank, P=0.0048). 15 16 Figure 2. FAM198B inhibits tumorigenesis and metastasis in vitro. (A) Left and 17 middle panels: ectopic expression of FAM198B in lung cancer cells (CL1-5, H226, 18 and A549) suppresses cell invasion and migration. The three cell lines were mixed 19 stable cells that were transfected with FAM198B-V5-expressing vector or mock 20 control vector, and their cell invasion and migration abilities were examined by

1 Transwell assay (*P<0.05; mean ± SD; n=3). Right panel: the protein expressions of 2 exogenous FAM198B-V5 were analyzed by immunoblotting with anti-V5 antibody. 3 β-actin served as an internal control . (B) Left and middle panels: knockdown of 4 FAM198B using shRNA in lung cancer cells (CL1-0 and EKVX) promotes cells 5 invasion and migration. The two cell lines were transduced with shFAM198B#5 or 6 shLacZ lentivirus and selected for stable clones with puromycin. The invaded cell 7 numbers were determined by Transwell assay (*P<0.05; mean ± SD; n=3). Right 8 panel: The mRNA levels of FAM198B knockdown in lung cancer cells were analyzed 9 by real-time RT-PCR (*P<0.05; mean ± SD; n=3), and GAPDH served as the internal 10 control. (C) Left and middle panels: the single-cell tracking of stable cells was 11 measured by time-lapse video microscopy for 16 h. Migration rate and directionality 12 of individual cells were calculated with 20 cells per condition (*P<0.05; mean ± SD). 13 Right panel: the movement of individual cells was monitored using time-lapse 14 microscopy and is shown as representative trajectories. (D) The cell proliferation 15 effect of FAM198B overexpression and silencing in CL1-5 and CL1-0 cells, 16 respectively. The cell proliferation rate was measured by MTT assays as optical 17 density (O.D.) at 570 nm at the indicated time points (*P<0.05; mean ± SD; n=8). (E) 18 The anchorage-independent growth effect of FAM198B overexpression and silencing 19 in CL1-5 and CL1-0 cells, respectively (*P<0.05; mean ± SD; n=3). 20 21 Figure 3. FAM198B inhibits tumorigenesis and metastasis in vivo. (A) Left panel: 22 in vivo tumor metastasis effect of FAM198B was analyzed by an experimental 23 metastasis assay with mixed stable CL1-5/mock control cells and CL1-5/FAM198B 24 cells, which were injected into the tail vein of SCID mice. The metastatic tumor 25 nodules were calculated (*P<0.05; mean ± SEM; n=8 per group). Right panel: the 26 histology of tumor nodules was confirmed by hematoxylin and eosin (H&E) staining 27 (original magnification, x200). Tumor nodules were indicated by black arrowheads. 28 Scale bar: 5 mm. (B) Left and middle panels: representative photographs of orthotopic 29 and metastatic tumor nodules from null mice. Mice were orthotopically implanted 30 with mixed stable CL1-5/mock control cells and CL1-5/FAM198B cells into the left 31 lung. Scale bar: 2 mm. Right panel: the tissue sections were stained by H&E and the 32 micrometastatic tumor nodules were indicated by black arrowheads. Scale bar: 500 33 μm. (C) Summary of orthotopic implantation assays. (D) Kaplan–Meier 34 metastasis-speciﬁc survival curves of nude mice orthotopically implanted with 35 CL1-5/mock or CL1-5/FAM198B-exoressing cells and the log-rank test for data 36 obtained 60 days after the cancer cells were orthotopically injected. (E) Left panel: 37 The effect of FAM198B on in vivo tumorigenesis was analyzed by a subcutaneous 38 tumor growth assay. CL1-5/mock control cells and CL1-5/FAM198B cells were

1 subcutaneously injected into the dorsal region of nude mice, respectively. Tumor 2 volume was estimated at the indicated time points (*P<0.05; mean ± SEM; n=7 per 3 group). Middle panel: mice were sacrificed, and the weight of subcutaneous tumors 4 was measured at the end of the experiment (*P<0.05; mean ± SEM; n=7 per group). 5 Right panel: final tumor photographs at 30 days post-subcutaneous injection. 6 7 Figure 4. FAM198B is an N-glycosylated protein. (A) Schematic diagram of 8 putative N-glycosylation residues (asparagine 98, 289) and putative O-glycosylation 9 residues (serine 137,149) in the human FAM198B protein. (B) Deglycosylation of 10 exogenous FAM198B by glycosidase in vitro. The cell lysates of mixed stable 11 CL1-5/mock control and CL1-5/FAM198B cells were treated with or without E: Endo 12 H, P: PNGase F, O: O-glycosidase and O/P: combination of O-glycosidase and 13 PNGase F at 370C for 1 h. The samples were electrophoresed and immunoblotted with 14 an anti-V5 antibody. (C) Inhibition of N-glycosylation of FAM198B was performed 15 by treating mixed stable CL1-5/mock control and CL1-5/FAM198B cells with various 16 concentrations of tunicamycin (TM) for 24 h. The cell lysates were immunoblotted 17 with the indicated antibodies. (D) Glycosylation degrees of FAM198B at Asn98, 18 Asn289, Ser137, and Ser149 residues in lung cancer cells. CL1-5 cells were 19 transiently transfected with mock control, wild type FAM198B, 20 N-glycosylation-defective FAM198B mutants (N98Q, N289Q, 2NQ), and 21 O-glycosylation-defective FAM198B mutants (S137A, S149A, 2SA) for 72 h. Then, 22 cell lysates were harvested and subjected to immunoblotting with the indicated 23 antibodies. (E-F) Transient expression of mock control, wild type FAM198B, and 24 N-glycosylation-defective FAM198B mutants in CL1-5 cells, which were then treated 25 with or without 10 μM tunicamycin for 24 h (E) and O-glycosidase at 370C for 1 h (F). 26 Then, cell lysates were immunoblotted with anti-V5 antibody. β-actin served as an 27 internal control. (G) LC-MS/MS-based identiﬁcation of N-glycopeptides 28 corresponding to N-glycosylation site, Asn289. The upper and intermediate gel pieces 29 of purified wild type FAM198B were excised and performed by in-gel digestion. 30 After enriching the glycopeptides, intact glycopeptides were analyzed by 31 LTQ-Orbitrap MS. The collision-induced dissociation (CID) spectrum is shown that 32 the oxonium ions and fragments of the glycopeptides with Man(7)GlcNAc(2) on 33 Asn289 (m/z 1113.485, 2+) were illustrated. Blue square, GlcNAc; green circle, 34 mannose; Y0, whole peptide sequence ILGLNR; Y1, peptide with one GlcNAc; Y2, 35 peptide with two GlcNAcs; b3, peptide fragment ILG. (H) CL1-5 cells were 36 transiently transfected with mock control, wild type FAM198B, N322Q, and 3NQ. 37 Then cell lysates were harvested and subjected to immunoblotting with the indicated 38 antibodies. u, indicate upper band; i, indicate intermediate band; l, indicate lower

1 band. 2 3 Figure 5. N-glycosylated FAM198B inhibits cancer cell invasion through 4 pERK-mediated MMP-1 signaling pathway. (A) The mRNA level of putative 5 downstream target MMP-1 upon FAM198B was measured by real-time RT-PCR 6 (*P<0.05; mean ± SD; n=3), and GAPDH served as the internal control. (B) The cell 7 lysates derived from mixed stable CL1-5/mock and CL1-5/FAM198B cells were 8 immunoblotted with the indicated antibodies. β-actin served as an internal control. (C) 9 Knockdown of FAM198B abolishes the suppression of cell invasiveness in CL1-5 10 cells induced by FAM198B. Left panel: the stably FAM198B-expressing CL1-5 cells 11 were transduced with shLacZ, shFAM198B#2 lentivirus (black bar) and subjected to 12 puromycin selection. The cell invasion ability was examined by Transwell invasion 13 assays. *P value <0.05, Student’s t test, as compared with the mock control (white bar) 14 or FAM198B-expressing CL1-5 cells transduced with shLacZ lentivirus (black bar). 15 All data presented were mean ± SD (n=3). Right panel: Cell lysates were 16 immunoblotted with the indicated antibodies. β-actin served as an internal control. (D) 17 Left panel: CL1-5/FAM198B-expressing cells were infected with 18 lentivirus-expressing shLacZ, shFAM198B and shMMP-1. Two independent sets of 19 shMMP-1 (#1 and #2) targeting distinct regions were used, which to avoid off-target 20 effect. After 96 h puromycin selection, the cell invasion ability was measured by 21 Transwell invasion assays (*P<0.05; mean ± SD; n=3). Right panel: the protein 22 expression of FAM198B-V5 and MMP-1 were analyzed with immunoblotting. β-actin 23 served as an internal control. (E) Cells were treated with or without 0.1μg/ml 24 AZD6244 for 24 h, and the cell lysates were analyzed by immunoblotting with the 25 indicated antibodies. β-actin served as an internal control. (F) Left panel: CL1-5 cells 26 were transiently transfected with mock control, wild type FAM198B, and 27 N-glycosylation-defective FAM198B mutants, and after 24 h, cells were treated with 28 neomycin for stable mixed cell selection. Then, Transwell assays were performed to 29 examine the cell invasion ability (*P<0.05; mean ± SD; n=3). Right panel: The 30 protein expressions of mixed stable CL1-5/FAM198B wild type and 31 N-glycosylation-defective mutant cells were analyzed by immunoblotting and probed 32 with the indicated antibodies. β-actin served as an internal control. (G) Proposed 33 model showing that FAM198B inhibits lung cancer progression. FAM198B is a tumor 34 invasion/migration suppressor while it possesses a vast amount of N-glycosylation 35 and a small amount of O-glycosylation. Glycosylated FAM198B downregulated the 36 pERK-mediated MMP-1 signaling pathway, which resulting in the suppression of 37 tumor tumorigenesis and metastasis.