FAM198B Is Associated with Prolonged Survival and Inhibits Metastasis in Lung Adenocarcinoma via Blockage of ERK-Mediated MMP-1 Expression

Lung cancer is the leading cause of cancer-related deaths worldwide, and lung adenocarcinoma is the predominant histologic subtype of lung cancer in woman, never-smokers, and younger adults (1–3). Cancer development is a multiphase process resulting from genomic instability, transcriptional alterations, cancer stemness, abnormal metabolic pathways, epigenetic alteration, tumor-promoting inflammation, and evasion of the immune system, which drive the progression of cancer (4–8). Proteolytic degradation of the extracellular matrix (ECM) and the basement membranes surrounding the primary tumor by the matrix metalloproteinase (MMP) is a critical step for tumor angiogenesis, invasion, and metastasis (9). The MMPs have served as potential prognostic markers and therapeutic targets in cancer (10). Previously, we established an isogenic lung cancer metastasis cell model and discovered several metastasis-related gene (11–18). However, the comprehensive understanding of mechanisms involved in the tumor metastasis remains to be explored further. Thus, the discovery and understanding of novel metastasis-related genes are crucial for developing more effective diagnoses and treatments for lung cancer.

The human family with sequence similarity 198, member B (FAM198B), is a novel gene with unknown function and predicted to be a membrane-bound glycoprotein localized on Golgi apparatus (19, 20). FAM198B might play a role in the regulation of mouse and Xenopus embryonic development and is a downstream target of the FGF receptor signaling pathway (21, 22). Nearly one half of all known eukaryotic proteins are N-glycosylated, which is a ubiquitous posttranslational modification (23), and the alteration of glycosylation has been reported to be associated with tumor proliferation, invasion, metastasis, angiogenesis, receptor activation, and intracellular or cell–matrix interactions (24–26).

In this study, we identified a novel potential tumor suppressor, FAM198B, from an isogenic lung cancer metastasis cell model by expression microarrays. The clinical relevance of FAM198B was analyzed in both public databases and Taiwanese lung adenocarcinoma patients. The tumor suppression activity of FAM198B was characterized by in vitro and in vivo metastasis and tumorigenesis assays. Microarray and pathway analyses were used to investigate the molecular signaling of FAM198B, and the effect of glycosylation on FAM198B stability was also investigated.

A total of 95 frozen tissues were collected from lung adenocarcinoma who underwent complete surgical resection at the Taichung Veterans General Hospital (Taichung, Taiwan) between May 2000 and June 2009. After surgical resection, half tumor specimen was immediately frozen within 30 minutes, and half was formalin-fixed, paraffin-embedded. The average tumor content and necrosis were 53.09 ± 21.10 and 10.35 ± 13.19, respectively (mean ± SD, n = 92 for tumor content, n = 90 for necrosis). Total RNA was extracted from the frozen tumor tissue specimens by using the TRIzol Plus RNA Purification Kit (Thermo Fisher Scientific). The qRT-PCR of TATA-box binding protein (TBP) was used to assess the quality of RNAs. Ct 40 was used as the cutoff value to define undetectable or detectable, and specimens with Ct less than 40 were enrolled in this study (27–29). This study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. Institutional Review Boards approved all aspects of the study. All participants provided written-informed consent.

For the in vivo metastasis assay, 1 × 10⁵ cells were suspended in 0.1 mL of PBS and intravenously injected into the lateral tail vein of 6-week-old SCID mice. All mice were sacrificed 4 weeks after injection, and the mouse lungs were removed and fixed in 10% formalin. The lung surface tumor foci were counted under a dissecting microscope. The mouse lungs were further embedded in paraffin, sectioned into 4 μm layers, and stained with hematoxylin and eosin (H&E) for histologic analysis. For the subcutaneous tumor growth assay, 1 × 10⁶ cells were suspended in 0.1 mL of PBS and implanted subcutaneously into the dorsal region of 6-week-old null mice. Tumor growth was examined thrice a week, and tumor volume was estimated by the formula LW²/2, where L is the length and W is the width of the tumor. Tumor weight was measured at the end of the study. For the orthotopic tumor implantation assay, 5 × 10⁴ cells were suspended in 20 μL of PBS containing 10 ng Matrigel (R&D Systems) and inoculated into left lung of 6-week-old null mice. The mice were sacrificed 45 days after implantation. The mouse lungs were removed and fixed in 10% formalin, and the size and number of lung tumor colonies were measured by microscopic examination. All animal procedures were performed with the approval of the Institutional Animal Care and Use Committee of the National Taiwan University.

Overall survival curves were calculated by the Kaplan–Meier analysis, and the log-rank test was performed to test the difference between survival curves. Each cutoff point for overall survival and disease-free survival for definition of the high/low-FAM198B expression groups is listed in Supplementary Tables S1, S3, and S4. Cox proportional hazards regression analysis with stepwise selection method was used to evaluate independent prognostic factors. Covariates of the regression model were FAM198B, age, gender, stage, EGFR status, KRAS status, smoking history, and histologic subtypes. For in vitro or in vivo studies, the Student t test was used to compare the difference between two groups. All tests were two-tailed, and P values < 0.05 were considered significant.

Detailed methods on cell lines and culture conditions, plasmids and transfection, viruses and transduction, qRT-PCR, immunoprecipitation and immunoblot, in vitro invasion and migration assays, single-cell tracking migration assay, cell proliferation assay and anchorage-independent growth assay, microarray and pathway analysis, glycosidase assay, identification of intact N-glycopeptides by LC-MS/MS, protein stability assay, and immunofluorescence analysis are described in the Supplementary Methods.

To investigate which novel candidate genes contribute to metastasis of lung cancer, the genome-wide RNA expression of a well-established isogenic metastasis cell line model, including low metastatic CL1-0 cells and high metastatic CL1-5 cells, was profiled by microarray analyses (30, 31). Through comparison of the differentially expressed genes between CL1-0 and CL1-5 cells, we found a gene, FAM198B, whose mRNA expression of CL1-5 cells was 130-fold lower than that of in the parental CL1-0 cells. The differential expression of FAM198B was confirmed by qRT-PCR in CL1-0 compared with CL1-5 cells (86.15 ± 14.08 and 1.00 ± 0.27, P < 0.05; Supplementary Fig. S1).

Next, we assessed whether the FAM198B expression is associated with survival of lung adenocarcinoma patients. We first conducted a survival analysis using the publicly available Memorial Sloan Kettering Cancer Center (MSKCC) microarray dataset (32). There was a significant association between FAM198B expression and overall survival in the dataset of 104 lung adenocarcinoma patients. Patients with low FAM198B expression had a worse overall survival compared with those with high FAM198B expression (Fig. 1A and Supplementary Table S1).

To further validate this correlation in Asia population, 95 lung adenocarcinoma patients were enrolled in Taiwan to detect the FAM198B expression in tumor specimens by qRT-PCR. The clinical characteristics of these patients are summarized in Supplementary Table S2. Patients with low FAM198B expression had worse overall survival than those with high FAM198B expression (Fig. 1B and Supplementary Table S3), but no significance was found between FAM198B expression and progression-free survival at any defined cutoff of FAM198B expression (Supplementary Table S4). Result of the multivariate Cox proportional hazard regression analysis with stepwise selection method showed that that high FAM198B expression [HR, 0.41; 95% confidence interval (CI), 0.18–0.91; P = 0.029], late stage (HR, 4.16; 95% CI, 1.79–9.70; P = 0.001), and EGFR mutation (HR, 0.44; 95% CI, 0.20–1.00; P = 0.049) were independent prognostic factors of the overall survival (Table 1). It suggested that FAM198B may function as a tumor suppressor in lung adenocarcinoma.

Based on the clinical finding, we then explored the potential role of FAM198B in the invasiveness and mobility of lung cancer cells. First, the endogenous FAM198B expressions of 11 lung cancer cell lines were determined by qRT-PCR (Supplementary Fig. S2), and the high and low FAM198B-expressing cells were identified for further functional experiments. To determine whether FAM198B can modulate cell invasion and migration, FAM198B was ectopically expressed in three endogenously low FAM198B-expressing cell lines, CL1–5, H226, and A549. We found that FAM198B significantly inhibits the invasion and migration abilities of these lung cancer cell lines (Fig. 2A).

Five, shFAM198B#2, #5, #7, #8, and #9, out of nine shFAM198B vectors purchased from The RNAi Consortium effectively inhibited FAM198B expression assayed by HEK293T cell transfection and immunoblotting assay (Supplementary Fig. S3A). The knockdown of FAM198B in two endogenously high FAM198B-expressing cell lines, CL1-0 and EKVX, by transduction of shFAM198B#5 lentivirus significantly enhanced the activities of invasion and migration (Fig. 2B). The FAM198B-suppressive activity was also confirmed in CL1-0 cells by the other two constructs, shFAM198B#2 and shFAM198B#7 lentiviruses (Supplementary Fig. S3B, left). Next, we performed a single-cell tracking assay to assess the effect of FAM198B on cell mobility directly. The migration rate and directionality of CL1-5/FAM198B cells were greatly reduced from 4.49 ± 3.05 to 1.57 ± 0.16 and from 0.55 ± 0.14 to 0.29 ± 0.07, respectively, compared with the CL1-5/mock control cells. In contrast, CL1-0/shFAM198B#5 cells showed an increase in cell-migratory rate and directionality (6.06 ± 0.78 and 0.38 ± 0.10, respectively) compared with CL1-0/shLacZ cells (1.85 ± 0.29 and 0.15 ± 0.06, respectively; Fig. 2C).

To explore whether FAM198B modulates other cancer phenotypic functions, we performed 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and anchorage-independent growth assays to assess the role of FAM198B on cell proliferation and colonogenesis. We found that the cell proliferation rate and colony-forming ability of stably FAM198B-expressing CL1-5 cells were significantly reduced compared with the mock control cells (Fig. 2D and E, left). Conversely, knockdown of FAM198B increased the cell proliferation of CL1-0 cells compared with the shLacZ control group by three different shFAM198B lentiviruses (Fig. 2D, right; Supplementary Fig. S3B, middle). Similarly, inhibitory activity of FAMB198B on colony formation was found (Fig. 2E, right). These data implied that FAM198B might act as a tumor suppressor to inhibit the growth and invasion of lung cancer cells in vitro.

To further confirm the metastasis suppression activity of FAM198B in vivo, a mouse metastasis assay was performed, in which the stably FAM198B-expressing CL1-5 cells or mock control cells were intravenously injected into the lateral tail vein of SCID mice. Mice were sacrificed after 4 weeks, and lung metastases were counted by a dissection microscope. The number of metastatic nodules in the mouse lungs was significantly reduced in the FAM198B-expressing group compared with the mock control group (170.13 ± 50.24 and 604.50 ± 58.77, n = 8 for each group, Fig. 3A). Furthermore, to mimic the entire lung metastasis process, an orthotopic lung implantation experiment was performed (16). Tumor cells were orthotopically inoculated into the left lobe of mice lung, and we examined the metastatic nodules formed on the mice right lung. FAM198B group not only formed smaller orthotopic tumor growth (0.73 ± 0.31 and 2.84 ± 0.69, respectively) but also decreased the number of lung metastatic tumors (2.16 ± 1.32 and 4.66 ± 1.03, respectively) compared with the mock control group (Fig. 3B and C). The effect of FAM198B on mouse survival was evaluated using another orthotopic implantation experiment. As shown in Fig. 3D, FAM198B-expressing group had a longer survival compared with the mock control group (medium survival, 39 vs. 27 days; P = 0.003). Next, we used the subcutaneous xenograft mouse models to assess the impact of FAM198B in tumor growth in vivo. The stably FAM198B-expressing CL1-5 cells or mock control cells were subcutaneously injected into the flanks of nude mice, and the tumor volumes were measured thrice a week. At 30 days after implantation, the mice were sacrificed. The FAM198B-expressing group exhibited a significant reduction in both tumor volume and weight compared with those of the mock control group (Fig. 3E). These data implicated that FAM198B inhibits tumorigenesis and metastasis of lung cancer in vivo.

Occasionally, we noticed two bands in the immunoblot of overexpressed FAM198B-V5. The position of the minor lower band is close to the predicted molecular weight, 64 kDa. The presence of multiple protein bands is often due to posttranslational modification such as glycosylation (33). Thus, we used the publicly available neXtProt database (http://www.nextprot.org/) and NetNGlyc server (http://www.cbs.dtu.dk/services/NetNGlyc/) to predict the possible posttranslational modifications of FAM198B, particularly for glycosylation sites. We found two potential N-glycosylation sites at asparagine 98 and 289 and two potential O-glycosylation sites at serine 137 and 149. All of these glycosylation sites are located in the predicted extracellular domain of the Golgi apparatus membrane (Fig. 4A; ref. 20). After Endo H digestion, the intermediate band converted to a lower band, suggesting that the intermediate band is a high mannose-type glycoprotein. The molecular weight of both upper and intermediate bands was simultaneously decreased after PNGase F treatment (Fig. 4B, left). These results indicated that FAM198B might contain hybrid or complex N-linked glycosylation. O-glycosidase treatment only resulted in a slight molecular shift of both the upper and intermediate bands compared with the combined treatment of O-glycosidase and PNGase F (Fig. 4B, right). Furthermore, cells were cultured in the presence of an N-glycosylation inhibitor, tunicamycin, for 24 hours, and the FAM198B glycosylation was severely impaired and similar to that observed in PNGase F treatment (Fig. 4C).

To examine which predicted site is responsible for FAM198B glycosylation, first, two potential N-glycosylation sites of FAM198B were mutated from asparagine to glutamine, individually or simultaneously (N98Q, N289Q, and 2NQ), and two potential O-glycosylation sites were mutated from serine to alanine (S137A, S149A, and 2SA). The mutant proteins were transiently expressed in CL1-5 cells and subjected to immunoblotting assays. We found that the glycosylation was strongly impeded in FAM198B/N289Q and FAM198B/2NQ and slightly in FAM198B/N98Q. Conversely, the molecular shift of FAM198B/S137A, FAM198B/S149A, and FAM198B/2SA was not obvious (Fig. 4D). To clarify whether Asn98 and Asn289 are the major N-glycosylation sites, the transiently FAMB198-expressing CL1-5 cells were treated with tunicamycin and analyzed by immunoblotting. Figure 4E indicated that the N-glycosylation inhibitor treatment resulted in an obvious molecular weight shift in wild-type FAM198B and FAM198B/N98Q but not in FAM198B/N289Q and FAM198B/2NQ. By contrast, the O-glycosidase treatment resulted in a similar and minor shift pattern in wild-type, N98Q, N289Q, and 2NQ mutants of FAM198B (Fig. 4F).

Furthermore, to identify the glycan structure on Asn98 and Asn289, V5-tagged FAM198B proteins were purified from CL1-5/FAM198B cells and analyzed by nanoscale LC-MS/MS. After identifying by Byonic software and manually confirming each MS/MS spectrum, we found that high mannose types, Man(7–10)GlcNAc(2) presented on Asn289 (Fig. 4G). In addition, we identified hybrid and complex-type of tri-antennary glycoforms with sialic acid and core- or terminal fucose on Asn98 while another complex poly-LacNAc with core-fucosylation located on Asn289 (Supplementary Fig. S4B; Supplementary Table S5). Most interestingly, we also identified another unpredicted N-linked glycosylation site, Asn322, which had high fucosylation (Supplementary Fig. S4A; Supplementary Table S5). To examine whether the Asn322 was occupied by a glycosyl residue, we mutated Asn322 from asparagine to glutamine (N322Q) and generated a triple-mutant (3NQ) containing N98Q, N289Q, and N322Q. Then FAM198B/N322Q or FAM198B/3NQ proteins were transiently expressed in CL1-5 cells and immunoblotted by anti-V5 antibody. Figure 4H indicated that the glycosylation was slightly impeded in FAM198B/N322Q and markedly destroyed in FAM198B/3NQ. These results supported that Asn98, Asn289, and Asn322 are three major N-glycosylation sites of human FAM198B.

To determine the biological function of N-glycosylation, we examined the subcellular localization and expression of FAM198B by immunofluorescence and immunoblot assays. N-glycosylation–defective mutations did not influence cellular localization, Golgi apparatus, but the expression of N-glycosylation–defective FAM198B mutants was much lower than wild-type FAM198B in transiently FAM198B-overexpressing CL1-5 cells (Supplementary Fig. S5; Fig. 4D–H). We next determined whether N-glycosylation affects FAM198B stability. The de novo FAM198B synthesis was blocked by cycloheximide in CL1-5 cells, and FAM198B were detected by immunoblotting (Supplementary Fig. S6A). The half-lives of N-glycosylation–defective mutants, T_1/2 = 4.1, 3.1, 0.1, 3.8, and 0 hours for N98Q, N289Q, N322Q, 2NQ, and 3NQ, were shorter than that of wild-type FAM198B, T_1/2 = 8.3 (Supplementary Fig. S6B). Subsequently, to evaluate whether the protein degradation of N-glycosylation–defective FAM198B mutants was through 26S proteasome pathway, HEK293T cells were transiently cotransfected with wild-type FAM198B, FAM198B/3NQ mutant, and hemagglutinin (HA)-tagged ubiquitin–expressing vectors in the presence of proteasome inhibitor MG132 or not. As shown in Supplementary Fig. S7, FAM198B/3NQ exhibited more ubiquitination than that of wild-type FAM198B in the presence of MG132. Taken together, these results implied that the N-glycosylation is critical for protein stability but not for intracellular trafficking of FAM198B.

To elucidate the underlying mechanism by which FAM198B inhibits invasiveness, the transcriptomic expression microarray was performed. We identified 1,580 genes that were significantly altered in CL1-5/FAM198B compared with CL1-5/mock control cells, with a greater than 2-fold change and under an FDR < 0.05. The differentially expressed genes were subjected to pathway analysis by the MetaCore analytical suite. Consistent with in vitro and in vivo results, 12 of top 20 ranking FAM198B-altered pathways were mainly involved in cell invasion and cell proliferation/apoptosis (Supplementary Table S6). MMP-1 was found in three cell invasion pathways (Supplementary Figs. S8–S10). We found that FAM198B suppressed both mRNA and protein expressions of MMP-1. It implied that FAM198B might suppress cancer cell invasion through MMP-1 inhibition (Fig. 5A and B). Based on pathway analysis, the MAPK–ERK signaling pathway may be involved in FAM198B-regulated MMP1 expression (Supplementary Fig. S9), and ERK/MAPK signaling is known to be required for MMP-1–mediated tumor invasion and metastasis (34, 35). We next examined whether FAM198B downregulates MMP-1 expression by inhibiting the ERK pathway. Immunoblotting revealed that the ERK phosphorylation was obviously decreased, whereas the protein level of MMP-1 was also decreased in CL1-5/FAM198B cells compared with CL1-5/mock control cells (Fig. 5B).

To further confirm the relationship between FAM198B and MMP-1, the stably FAM198B-silencing CL1-5 cells were generated by shFAM198B#2 lentivirus infection. The invasion ability of CL1-5/FAM198B/shFAM198B#2 cells was increased compared with CL1-5/FAM198B or CL1-5/FAM198B/shLacZ control cells, and immunoblot assay showed that MMP-1 expression and ERK phosphorylation are upregulated upon FAM198B knockdown (Fig. 5C). Furthermore, MMP-1 expression was suppressed by two different shMMP-1 lentiviruses, and MMP-1 knockdown suppressed cell invasion ability in CL1-5/FAM198B/shFAM198B#2 (Fig. 5D). Besides, to verify whether FAM198B-induced MMP-1 suppression is through ERK signaling, CL1-5 cells were treated with a specific MEK1/2 inhibitor, AZD6244, for 24 hours (36). The immunoblot results revealed that ERK phosphorylation was significantly decreased, and MMP-1 expression was suppressed with a concentration of AZD6244 as low as 0.1 μg/mL (Supplementary Fig. S11). We also found that the FAM198B knockdown–induced upregulation of ERK phosphorylation and MMP-1 expression were markedly suppressed by AZD6244 treatment (Fig. 5E). These results indicated that FAM198B suppressed MMP-1 expression through inactivating ERK signaling, at least partly. Finally, we investigated whether N-glycosylation affects FAM198B-induced invasion suppression. CL1-5 cells were transiently transfected with N-glycosylation–defective FAM198B-expressing vectors and analyzed for invasion activities. We found that N-glycosylated FAM198B significantly inhibits cancer cell invasion, whereas N-glycosylation–defective FAM198B did not and all of the N-glycosylation–defective FAM198B mutants fail to suppress ERK phosphorylation and MMP-1 expression compared with wild-type FAM198B (Fig. 5F and Supplementary Fig. S12). Collectively, our data revealed that complex N-glycosylation of FAM198B plays an important role in the pERK/MMP-1 pathway and regulates cancer metastasis.

In this study, we identify FAM198B as a novel potential tumor suppressor in lung adenocarcinoma. The patient survival analysis indicated that low FAM198B expression is associated with shorter overall survival in lung adenocarcinomas. Overexpression of FAM198B inhibited cancer cell invasion, proliferation, and tumorigenesis in both in vitro and in vivo assays, whereas knockdown of FAM198B promoted the malignancies. FAM198B inhibits cancer invasion at least partly through downregulating the pERK/MMP-1 signaling pathway, and N-glycosylation enhances the stability of FAM198B protein (Fig. 5G).

Protein glycosylation is an important posttranslational modification that occurs in the ER and Golgi apparatus and is regulated by complicated mechanisms, including the localization and expression of glycosyltransferases and the molar ratio of glycosyltransferases to substrates. It is well known that the glycosylation is important for protein functions. For example, the altered glycosylation of membrane receptors interferes the downstream signaling via hindering receptor oligomerization and ligand binding and in turn affects a wide range of key cellular processes, including cell division, differentiation and localization, and even the progression and malignancy of cancer cells (37–39). First, we showed that both upper and intermediate bands of FAM198B were N-glycosylated based on the results of tunicamycin experiments. However, the Endo H and PNGase F digestion and asparagine site–directed mutagenesis, Asn98 and Asn289, failed to shift FAM198B to expected molecular weight. These facts suggested that FAM198B might have additional glycosylations. Hence, we used LC-MS/MS to characterize N-glycosylation sites and glycan structures of FAM198B and identified another N-linked glycosylation on Asn322 site. The molecular weight of 3NQ was also matched to the expected size of FAM198B. Taken together, these data revealed that the glycosylation of Asn322 might contribute to the small fractions with higher molecular weight presented on N98Q, N289Q, and 2NQ of FAM198B.

In the immunofluorescence assay, we characterized the subcellular localization of FAM198B by using an anti-V5/FITC antibody in CL1-5 cells instead of using an anti-FAM198B antibody because both commercially available FAM198B antibodies raised by either synthetic peptides or bacterial recombinant proteins failed to detect recombinant FAM198B proteins due to poor quality. In turn, we found that N-glycosylation does not influence the FAM198B trafficking but affect the turnover rate of FAM198B. Elimination of single or all glycosylation sites of FAM198B accelerated the FAM198B protein degradation and resulted in a lower steady-state level of FAM198B. Besides, we also found that the glycosylation of FAM198B inhibits 26S proteasome-mediated protein degradation. A Transwell invasion assay demonstrated that N-glycosylation–defective FAM198B mutants completely lost the invasion suppression activity compared with wild-type FAM198B. Given the lack of FAM198-specific antibody, we cannot exclude the possibility that certain FAM198B-induced phenotypic alterations are due to overexpression artifact and not reflecting a genuine FAM198B. To minimize this possibility, we validated the invasion-suppressive activity of enforced FAM198B expression in three cell lines and performed RNA silencing of endogenous FAM198B in two cell lines with three different shRNA constructs. All of these data showed that the suppressive activity of FAM198B seems not due to overexpression artifact, although there is no available anti-FAM198B antibody. Similarly, the negative correlation of FAM198B with patient survival assayed by qRT-PCR cannot be confirmed by FAM198B immunohistochemistry staining, but this clinical finding has provided a new insight of biological function of FAM198B in lung cancer progress.

Collagenase-1 (MMP-1) is one of the major proteases and cleaves interstitial collagens with a preference for type I, II, and III collagens. MMP-1 has been shown to potently facilitate cancer invasion and metastasis, and its expression correlates with poor clinical outcomes and increases recurrence (40–42). It has previously been shown that the ERK and p38 MAPK pathways are the major regulators for MMP-1 expression (43, 44). The role of p38 MAPK on MMP-1 expression is controversial in certain situations (45). Recent studies showed that EGFR-mediated p38 MAPK signaling pathway augments MMP-1 expression and then leads to promote cancer tumorigenesis and angiogenesis (46). Other studies indicated that p38 MAPK inhibits MMP-1 expression (47, 48). Although certain studies indicated that p38 MAPK is involved in MMP-1 regulation (45–48), p38 MAPK did not play a major role in FAM198B-mediated MMP-1 regulation (Supplementary Fig. S13). In contrast, the constitutive activation of the ERK/MAPK pathway not only increases cancer cell proliferation but also enhances MMP-1 expression, which in turn increases tumor invasion and metastasis. Our data also revealed that MMP-1 acts as the major downstream effector of FAM198B-modulated invasion via ERK signaling. However, it remains to further investigate how FAM198B suppresses ERK phosphorylation. Furthermore, we found that N-glycosylation–defective FAM198B mutants, regardless of the degree of deglycosylation, fail to inhibit the ERK/MMP-1 pathway. Deglycosylation altered FAM198B stability, shortened the half-life of FAM198B, and reduced FAM198B availability. It may be the reason that FAM198B/2NQ and FAM198B/3NQ mutants did not show synergistic effect on cell invasion, ERK phosphorylation, and MMP-1 expression.

Up to date, there is no available agent that could specifically regulate FAM198B glycosylation in vitro or in vivo, but FAM198B might serve as a candidate for drug screening. Further study is needed to identify the glycosyltransferase(s) specific to FAM198B. Selecting drugs that enhance the activity of the FAM198B-specific glycosyltransferase(s) might be another strategy of anticancer drug development. Collectively, our findings provide a new insight into the molecular mechanism by which FAM198B inhibits tumor invasion, metastasis, and tumorigenesis through the pERK-mediated MMP-1 signaling pathway, and the complex posttranslational N-glycosylation serves a crucial role in stabilizing FAM198B.

No potential conflicts of interest were disclosed.

Conception and design: C.-Y. Hsu, C.-C. Chiang, P.-C. Yang, S.-L. Yu

Development of methodology: C.-Y. Hsu, Y.-J. Chen, P.-C. Yang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.-Y. Hsu, G.-C. Chang, K.-Y. Su, J.-S. Chen, W.-H. Ku, C.-Y. Wu, S.-L. Yu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.-Y. Hsu, G.-C. Chang, Y.-C. Hsu, K.-Y. Su, H.-Y. Chen, C.-Y. Lin, W.-H. Ku, B.-C. Ho, C.-C. Chiang, P.-C. Yang, S.-L. Yu

Writing, review, and/or revision of the manuscript: C.-Y. Hsu, Y.-C. Hsu, Y.-J. Hsiao, K.-Y. Su, H.-Y. Chen, P.-C. Yang, S.-L. Yu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.-Y. Hsu, G.-C. Chang, Y.-C. Hsu, Y.-J. Hsiao, Q.-S. Hong, W.-H. Ku, P.-C. Yang

Study supervision: Y.-J. Chen, P.-C. Yang, S.-L. Yu

Other (mass spectrometry and glycosylation analysis): Y.-J. Chen

We thank the National RNAi Core Facility, the Integrated Core Facility for Functional Genomics of the National Core Facility Program for Biotechnology, the Microarray Core Facility of the National Taiwan University Center of Genomic Medicine, and Common Mass Spectrometry Facilities of Institute of Biological Chemistry, Academia Sinica for technical support.

This study was supported by grants from the Ministry of Science and Technology, Taiwan (NSC 98-2314-B-002-120-MY3, NSC 102-2911-I-002-303, MOST 103-2911-I-002-303, MOST 104-2911-I-002-302, 104R8400). Mathematics in Biology Group of the Institute of Statistical Science Academia Sinica and Taiwan Biosignature Project of Lung Cancer supported data analysis work.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.