Loss of the novel mitochondrial protein FAM210B promotes metastasis via PDK4-dependent metabolic reprogramming

Recent advances in tumor metabolism have revealed that metabolic reprogramming could dramatically promote caner metastasis. However, the relation and mechanism between metastasis and metabolic reprogramming are not thoroughly explored. Cell proliferation, colony formation, and invasion analysis were performed to evaluate the role of FAM210B in human cancer cells. Human ovarian cancer xenograft model was used to determine the effects of inhibiting FAM210B by shRNA on tumor metastasis. Microarray analysis was used to determine the target genes of FAM210B. FAM210B cellular localization was performed by mitochondria isolation and mitochondria protein extraction. To detect FAM210B-mediated metabolic reprogramming, oxygen consumption rate and extracellular acidification rate were measured. Our previous study screened a novel cancer progression-suppressor gene, FAM210B, which encodes an outer mitochondrial membrane protein, by the suppression of mortality by antisense rescue technique (SMART). Here we demonstrated that FAM210B loss was significantly associated with cancer metastasis and decreased survival in a clinical setting. Additionally, it was found that low expression of FAM210B was significantly correlated with decreased survival and enhanced metastasis in vivo and in vitro, and the loss of FAM210B led to an increased mitochondrial respiratory capacity and reduced glycolysis through the downregulation of pyruvate dehydrogenase kinase 4 (PDK4), which activated the EMT program and enhanced migratory and invasive properties. Collectively, our data unveil a potential metabolic target and mechanism of cancer metastasis.

Ovarian cancer is the most lethal gynecological malignancy. 1 Although ongoing efforts have improved the early diagnosis, metastases decrease the cure rate substantially. 2,3 An emerging hallmark of cancer is the reprogramming of energy metabolism to fuel tumor cell growth, division, and survival. [4][5][6] Recent studies have revealed that enhanced mitochondrial function have important roles in tumor proliferation and metastasis. [7][8][9][10] The metabolic status malignant tumors exploit to regulate growth and proliferation is becoming increasingly clear. 11 However, how cancer cells alter the metabolic program and affect metastasis is not well understood.
Our previous study screened the novel gene FAM210B from the library involved in the tumor-selective killing of trichostatin A (TSA) by suppression of mortality by antisense rescue technique (SMART). 12 In this study, further analysis indicated that FAM210B was a novel mitochondria outer membrane protein and low expression of FAM210B obviously promoted malignant metastasis in vitro and in vivo. Depletion of FAM210B in ovarian cancer cells displayed elevated mitochondrial respiration and decreased glycolysis relative to control cells via the suppression of its downstream target, pyruvate dehydrogenase kinase 4 (PDK4). Thus our results suggested that low expression of FAM210B promoted metastasis through metabolic reprogramming.

Results
Low expression of FAM210B was associated with metastasis and decreased survival. We previously screened a new gene, FAM210B, from the library involved in the tumor-selective killing of trichostatin A (TSA) by SMART. 12 In this study, we also found that the expression of FAM210B was selectively upregulated by TSA in tumor cells (Supplemenetary Figures S1A and B). As in our previous study, the role of FAM210B in TSA-induced apoptosis was tested. Cells were stably infected with lentivirus expressing either a small hairpin RNA (shRNA) non-targeting control or two specific shRNAs that knock down FAM210B (Supplemenetary Figure S1C). TSA resulted in profoundly reduced proliferation and colony formation in shScramble cells but not in shFAM210B cells (Supplemenetary Figures S1D and E). Decreased expression of FAM210B also moderately inhibited TSA-induced apoptosis, suggesting that FAM210B was essential for apoptosis mediated by TSA (Supplem-  Figure S1F). On the other hand, we also found that low expression of FAM210B was significantly associated with metastasis and decreased survival ( Figure 1). Because FAM210B can be regulated at the mRNA level (Supplemenetary Figure S1A), the available public gene expression databases were surveyed. In the ovarian cancer data set GSE14407, FAM210B expression was significantly reduced in ovarian cancer epithelia (CEPI) samples compared with normal human ovarian surface epithelia (HOSE) (Figure 1a). Furthermore, TCGA data showed that FAM210B mRNA tended to be lower in many types of cancer than in normal cells (Supplemenetary Figure S2). Based on the above results, we further elucidated the clinical outcome relevance of FAM210B in various cancer patients using Kaplan-Meier (KM) plots (http://www.kmplot.com). In the ovarian cancer data sets GSE9891, GSE19829 and GSE27651, analysis showed that low expression of FAM210B was significantly correlated with decreased overall survival (Figures 1b-d). All of the breast cancer and lung adenocarcinoma data sets in KM plotter also revealed that patients with low FAM210B had a significantly poor prognosis (Figures 1e and f). Thus it supposed that FAM210B may have a crucial role in cancer progress and metastasis. Consistent with the above results, in the breast cancer metastatic model of GSE9187 and GSE37344, FAM210B expression was significantly low in the metastatic phenotype (Figures 1g and h). GSE26511 revealed that cervical cancer patients with positive lymph nodes had a significantly lower FAM210B expression (Figure 1i). To investigate the potential role of FAM210B in ovarian cancer metastasis, we used orthotopic mouse models of ovarian cancer. Immunodeficient mice were injected with parental SKOV3-luc-GFP cells under the bursal membrane of the ovary, and peritoneal cavity metastases were developed after 4 weeks (Figure 1j). mRNA expression and immunohistochemistry (IHC) scores of FAM210B were higher in the metastatic site than in the primary ovarian cancer (Figures 1k and l). These findings showed the clinical significance and metastasis potential of FAM210B in cancer.
To further study the role of FAM210B in human metastatic ovarian cancer, we assessed its expression in primary tumor sections from ovarian cancer patients and their matched metastatic tumors by IHC ( Figure 2). Analysis of these tumor sections revealed diminished FAM210B staining (either signal intensity or percentage of positivity of FAM210B staining) in metastatic tumor sections compared with paired primary tumors in all eight patients examined.
Inhibition of FAM210B Promoted Metastasis in vitro and in vivo. Given that FAM210B loss was associated with metastasis and decreased survival, we used lentiviral shRNA to suppress FAM210B expression in SKOV3 and A549 cells and measured its effect on cell metastasis in vitro and in vivo.  Figure S1E). We also observed overexpression FMA210B in SKOV3 cells using lentivirus myc-FAM210B, and the invasion cells were less than control cells after 48 h (Supplemenetary Figures S3A  and B). The migration assay was also dependent on the  expression level of FAM210B ( Figure 3e). SKOV3-Luc-GFP-shFAM210B orthotopic ovarian cancers showed similar tumor weight to control shScrbl tumors ( Figure 3f). The number of pelvic cavity metastatic nodules was significantly increased in mice with SKOV3-Luc-GFP shFAM210B tumors compared with mice with control shScrbl tumors, as assessed by the number of GFP+ nodules (Figures 3g and h). Thus these results indicated that the loss of FAM210B increases the invasive and migratory properties of cancer cells.
FAM210B was transported into and localized in mitochondria. FAM210B is a novel protein, and its cellular location and function are unclear. First, the protein expression of FAM210B was detected in various human cell lines (Supplemenetary Figure S3C). Based on its expression profile, we found that it was highly conserved among metazoa but not in yeast (Supplemenetary Figure S4A). Protein sequence analysis predicted a cleaved sequence located in the N-terminal region in human FAM210B that may serve as a mitochondrial targeting sequence ( Figure 4d). Immunoblotting of endogenous FAM210B expression also revealed that FAM210B was located in the mitochondria ( Figure 4e). Furthermore, when mitochondria were sonicated for membrane disruption and then were washed, FAM210B was associated with disrupted mitochondrial pellets, suggesting that FAM210B was a mitochondrial membrane protein ( Figure 4f). To further identify the submitochondrial localization, intact mitochondria isolated from HeLa cells were treated with proteinase K. FAM210B is sensitive to protease digestion similar to Tom20 and in contrast to NDUFS1 and MnSOD, which are resistant to degradation, indicating that endogenous FAM210B is localized on the outer membrane of mitochondria ( Figure 4g). Taken together, these results showed that FAM210B resided on the outer membrane of mitochondria, and the N-terminal 47 aa MTS of FAM210B are necessary for its proper mitochondrial localization.
Loss of FAM210B mediated metabolic reprogramming in cancer cells. A recent study reported on the important roles of the reprogramming of cellular metabolism in tumor metastasis. 9 Based on the mitochondrial localization of FAM210B and loss of the FAM210B effect on cancer metastasis, we measured its oxygen consumption rates (OCRs) and extracellular acidification rates (ECARs) in the presence or absence of distinct inhibitors. FAM210B was transiently silenced (gene expression knockdown; Stealth siFAM210B RNA pool) in SKOV3 and A549 cells. The baseline and maximum OCR were significantly increased in siFAM210B cells compared with the negative control in SKOV3 cells (Figures 5a and b) and A549 (Supplementary Figures S5A and B), while baseline ECAR levels were decreased, and the maximum ECAR levels were unchanged (Figures 5c, d, Supplemenetary Figures S5C and D). Consistently, siFAM210B cells showed reduced lactate production ( Figure 5e) and higher glucose uptake ( Figure 5f) and were also less sensitive to glucose deprivation ( Figure 5g). However, total reactive oxygen species (ROS; Supplemenetary Figure S6A), mitochondria-derived ROS (Supplemenetary Figure S6B), mitochondrial respiratory activity (Supplemenetary Figure S6D), and mitochondrial structure (Supplemenetary Figure S6E) were unchanged in FAM210B-deficient cells while mitochondrial mass (Supplemenetary Figure S6C) was slightly increased. Thus siFAM210B cells exhibit metabolic reprogramming with higher Krebs cycle but with reduced baseline glycolysis compared with control cells.
FAM210B knockdown decreased the expression of PDK4. To further define the mechanism of FAM210B metabolic remodeling, the transcriptome of siFAM210B A549 cells and control cells were assessed by gene expression microarray. A heat map selected the top 25 differentially expressed genes in siFAM210B cells versus control, and PDK4 was dramatically downregulated (Figure 6a). Microarray data were verified by qRT-PCR in SKOV3 cells (Po0.001) (Figure 6b). Reduced expression of PDK4 in siFAM210B cells was partially restored via overexpressing FAM210B and was further validated by western blotting (Figure 6c). Nevertheless, we did not find a change in the expression of the other three pyruvate dehydrogenase (PDH) kinase genes (PDK1, PDK2, and PDK3) upon FAM210B knockdown (Supplemenetary Figure S7A). PDK4, a regulatory kinase, phosphorylates PDH to inhibit its activity (depicted schematically in Figure 6d). Therefore, the deficiency of PDK4 decreases glycolysis and activates mitochondrial energy metabolism. 13 In line with this, the phosphorylation levels of PDH-E1α were attenuated after the reduced expression of FAM210B or PDK4 (Figure 6e). The suppression of PDK4 also showed a higher Krebs cycle but reduced baseline glycolysis compared with control cells (Figures 6f and g). Collectively, the data suggested that the metabolic and metastatic phenotype due to the knockdown of FAM210B may be PDK4 dependent.
PDK4 decrease was essential for metabolic remodeling and metastasis due to FAM210B knockdown. Next, we wondered whether the loss of FAM210B-mediated metabolic features occurred through the decreased expression of PDK4. We overexpressed and validated PDK4 in siFAM210B cancer cells (Supplemenetary Figure S7B). As shown in Figures 7a and b, the basal and maximum OCR were partially reversed by restoring PDK4 gene expression in siFAM210B pooled cancer cells. Similarly, the basal ECAR was also partially restored (Figures 7c and d). To address the effects of the alteration of oxidative phosphorylation (OXPHOS) activity on the metastasis potential, we tested  (Figures 7e and f). To make clear the mechanism of cell invasion promoted by low expression of FAM210B, bioinformatics analysis of gene expression profiling revealed that TGF-β signaling differentially modulated the canonical pathway in siFAM210B compared with the control cells (Figure 7g). Mesenchymal genes (Vimentin and N-cadherin) were significantly upregulated while epithelial gene (Ecadherin) was downregulated in siFAM210B and siPDK4 cells compared with the control (Figures 7h and i). Together, these findings indicated that the downregulation of FAM210B assumed a characteristic EMT phenotype and remodeled metabolic features of mitochondrial OXPHOS.

Discussion
Epithelial ovarian cancer (EOC) is the leading cause of death from gynecological malignancy, with a high mortality attributable to widespread intraperitoneal metastases. 14 Therefore, we applied the suppression of mortality by SMART to screen genes involved in tumor progression. We showed that low expression of the new gene FAM210B attenuated the TSA effects in cancer cells and promoted malignant metastasis inhibition activated TGF-β signaling and subsequent EMTassociated gene expression. Accumulating evidence has indicated that mitochondria had a central and multifunctional role in malignant tumor progression, and targeting mitochondria provides therapeutic opportunities. 15 In our screen model, we demonstrated that the loss of a novel mitochondrial protein, FAM210B, was significantly associated with metastasis and decreased survival in a clinical setting. Furthermore, the reduced level of FAM210B was significantly correlated with decreased  Figure 3) in EOC, a finding that was not reported previously. FAM210B and its homologs belong to the domain of unknown function 1279 (DUF1279) family of proteins, and their functions in cells are poorly understood. A recent new study identified it as a putative mitochondrial protein using transiently transfected His/Myc-tagged FAM210B and to be involved in erythroid differentiation. 16 Our observation revealed that different cell lines express FAM210B and FAM210B was an outer membrane protein of the mitochondria with an N-terminal 47 aa MTS (Figure 4). These findings suggest that the novel mitochondrial protein FAM210B may connect mitochondrial function to cancer progression.
A common feature of cancer cells is to utilize the nutrients to satisfy their needs. Thus targeting metabolic processes is attractive for therapy. Aerobic glycolysis is common in cancer cells and correlates with poor tumor prognosis. 17,18 However, glycolytic inhibitors have little effects on solid tumor growth. 19 Another group assumed that altered metabolic and signaling cues in response to glycolysis inhibition in cancer cells promotes metabolic reprogramming and eventually the escape from glycolysis dependency. 20 Recent studies have indicated that the migratory and invasive properties of cancer cells are dependent on mitochondrial respiration. 8,9,21 Based on the mitochondrial location of FAM210B, we tried to explore whether its effect on metastasis is associated with mitochondrial apoptosis, biogenesis, respiratory chain activity, or mitochondrial ROS; however, the results are negative (Supplemenetary Figure S6). Next, we wondered whether mitochondrial metabolism contributes to the metastasis feature in FAM210B knockdown cells. We measured the OCR and ECAR in the presence or absence of distinct inhibitors and found that the loss of FAM210B cells showed the metabolic signature of increased mitochondrial basal OCR and respiratory capacity but reduced glycolysis ( Figure 5). The phenomenon is in agreement with recent work reflecting the metabolic plasticity of tumor cells. 20,22 Knockdown of FAM210B cancer cells with high OXPHOS but relatively low basal ECAR showed some differences from the others. Microarray revealed that the reduced expression of FAM210B in cancer cells also showed a lower expression of PDK4 ( Figure 6). This may explain the negative results of mitochondrial apoptosis, biogenesis, and respiratory chain activity and only effects on OCAR in FAM210B knockdown cells. However, the contribution of the mechanism of metabolism reprogramming to metastasis is not thoroughly clarified. Settleman and colleagues described that PDK4 downregulation changes pyruvate entry into the TCA cycle and is a critical regulator of EMT. 23 KEGG pathway analysis of the microarray data demonstrated that the loss of FAM210B activates TGF-β signaling and subsequent EMT-associated gene expression, and we also confirmed the phenomenon in FAM210B or PDK4 knockdown cells (Figures 7g-i). Focal hypoxia and ROS accumulation may induce an EMT program. 24,25 However, the decreased expression of FAM210B or PDK4 demonstrated little to no effect on mitochondrial-derived ROS. On the other hand, PDK4 restoration or blockade of mitochondrial respiration in FAM210B knockdown cancer cells partially reversed their invasive potential (Figures 7a-f). Collectively, we propose that the EMT program and enhanced migratory and invasive properties may coexist with mitochondrial reprogramming in FAM210B or PDK4 knockdown cancer cells, while the mechanism needs further clarification.
In this study, we successfully exploited the SMART technique to identify a novel cancer progression-suppressor gene, FAM210B, and demonstrated that it encodes an outer mitochondrial membrane protein and is related to cancer metabolism reprogramming and metastasis. Our results offer a mechanistic rationale for the further investigation of FAM210B as a potential therapy to prevent or treat cancer metastasis. Mitochondrial isolation, extraction of mitochondrial proteins, proteolysis of mitochondria fraction and western blotting analysis. Mitochondria isolation was performed using a mitochondria isolation kit (Thermo Scientific, Rockford, IL, USA) according to the manufacturer's instructions. The extraction of mitochondrial proteins was performed as previously described. 26 The proteolysis of mitochondria was performed as previously described with some modifications. 27 Briefly, purified mitochondrial pellets were treated as follows: proteinase K (0.1 mg/ml), 1% SDS (membrane solubilization) at 37°C and centrifuged at 12 000 × g for precipitation. Western blotting analysis was determined according to standard protocols with commercially available antibodies: anti-GFP (Abcam, ab6556), anti-α-Tubulin (Proteintech Group, Wuhan, Hubei, P.R.C, 11224-1-AP), anti-Tom20 (Proteintech Group, Wuhan, Hubei, P.R.C, 11802-1-AP), anti-Fam210b (Abcam, Cambridge, UK, ab150889 and Santa Cruz,  Microarray data analysis and mining. Cells were homogenized with TRIzol (Invitrogen) and submitted to Genminix Informatic Ltd. (Shanghai, China). Microarray analysis was performed using Affymetrix Human Transcriptome Array 2.0 (HTA2.0, Affymetrix, CA, USA). The data were analyzed with Robust Multichip Analysis algorithm using Affymetrix default analysis settings and global scaling as normalization method. Gene expression validation by real-time PCR was performed using the primers listed in Supplementary Table S1. The heat map was drawn using Cluster 3.0. Gene expression data and survival analyses of ovarian, breast, and lung cancer patients with annotated clinical outcomes were downloaded from the Gene Expression Omnibus database (GSE9187, GSE37344, GSE14407, GSE19829, GSE9891, and GSE11969). For survival analyses, the median gene expression value was used to define the high and low Fam210b expression groups, and KM analyses were performed. Significance for these plots was evaluated using the log-rank test.
In vivo animal studies. For the orthotopic mouse model of ovarian cancer studies, 4-week-old female NOD-SCID mice were purchased from and were housed at the Experimental Animal Center, Tongji Medical College, HUST, Wuhan, China. The orthotopic mouse model of ovarian cancer was performed as previously described. 28 For the development of the pelvic cavity metastasis study, mice were surveyed by in vivo imaging for the tumor growth curve or by stereomicroscopy for GFP+ nodules in tumor implantation after 1-4 weeks.
Invasion and migration assays. For the invasion assay, the polycarbonate membranes (transwell; 8-μm pore) were coated on both sides with Matrigel (BD Biosciences, USA), and the cells were seeded in the upper chamber, and the migrated cells on the basal side of the membrane were fixed in 4% paraformaldehyde and were stained with crystal violet before microscopic evaluation. For the scratch/migration assay, the cell-free area was measured 24 h after scratching the dish, and 0.1 μM rotenone was used.
Clonogenic assay and MTT. After treatments, 2 × 10 3 cells were plated in 6cm culture dishes for 14-20 days. Colonies were fixed with 4% paraformaldehyde, stained with 2% crystal violet, and counted. For the cell viability MTT assay, treated cells were seeded in 96-well plate for 24-72 h with or without the indicated concentrations of TSA (Sigma, St. Louis, MO, USA). The MTT assay was performed as described previously. 29 Measurement of the OCR and ECAR. The OCR and ECAR were measured using the Seahorse XF24 and XF96 instruments (Seahorse Bioscience, USA). Next, 2.5 × 10 4 cells were plated in XF24 Cell Culture Microplates and 8000 cells in XF96. For the OCR test, the cell medium was changed by the assay medium (Seahorse Bioscience) supplemented with 2 mM glutamine, 10 mM glucose, and 1 mM pyruvate for 1 h in a 37°C incubator prior to the measurements using the XF Cell Mito Stress Kit (Seahorse Bioscience). The concentrations of oligomycin and FCCP were 1.0 and 0.5 μM FCCP for SKOV3 and A549 cells, respectively. For the glycolytic metabolism test, the cells were incubated in the basal medium (Seahorse Bioscience) for 1 h in a 37°C incubator prior to measurement using the Glycolytic Stress Test Kit (Seahorse Bioscience). The OCR and ECAR results were adjusted to the cell numbers. Finally, the cells that were prepared but not undergone any treatment were stained with CCK8 following spectrophotometric measurements at 450 nm (Molecular Device, USA).
Analysis of lactate production. Cell culture supernatants were tested following the manufacturer's instructions (Lactate Colorimetric/Fluorometric Assay Kit, BioVision, CA, USA), and the data were normalized to the total protein.
Mitochondrial DNA (mtDNA) content determination. mtDNA content measurement was performed as previously described. 30 Briefly, the relative mtDNA content was measured by quantitative real-time PCR analysis of total DNA extracted Transmission electron microscopy. The cells were collected and fixed with 2.5% glutaraldehyde in 0.1 mol/l sodium cacodylate buffer and then were postfixed with 2% osmium tetroxide in 0.1 mol/l sodium cacodylate buffer plus 0.3% potassium ferrocyanide. The cells were stained with 4% aqueous uranyl acetate, dehydrated, infiltrated, and embedded in epoxyresin. Ultrathin sections (80 nm) were cut and imaged with an aFEI Tecnai G 2 electron microscope (Tongji Hospital, Wuhan, PRC).
Immunohistochemistry. Tissues were subjected to standard heat-induced epitope retrieval in 10 mM citrate buffer (pH 6.0) using a steamer. The slides were incubated overnight at 4°C with a rabbit anti-Fam210b (Abcam, ab150889). Following incubation with the primary antibody, the secondary biotin-conjugated antibody was applied for 1 h at 37°C. All of the slides were counterstained using Harris hematoxylin (Abcam).
Statistical analysis. The statistical significance of the treatment groups was determined using the independent samples t-test or one-way ANOVA. All of the analyses were conducted using SPSS19.0 (USA).

Conflict of Interest
The authors declare no conflict of interest.