To observe the expression level and the expression difference of FAM107A between tumor and adjacent normal tissues for the different tumors or specific tumor subtypes, FAM107A was firstly investigated via the Oncomine dataset (https://www.oncomine.org/), a platform with the integrated results of high-quality standard tumor tissue microarray in the literature and the microarray database, as well as the analysis of 14 annotation databases. Besides, FAM107A was also input in the “Gene_DE” module of TIMER2 (Tumor immune estimation resource, version 2) website (http://timer.cistrome.org/) which is a website for analysis of gene expression of virous tumor types and tumor infiltrating immune cell components based on the data of the TCGA project (20). Next, the “Expression Analysis-Box Plots” module of the GEPIA2 (Gene expression profiling interactive analysis, version 2) website (http://gepia2.cancer-pku.cn/#analysis) (21), with the settings of P-value cutoff = 0.01, log₂FC (Fold change) cutoff =1, and “Match TCGA normal and GTEx data”, was applied to obtain box plots of the expression difference between the tumor tissues and the corresponding normal tissues of the GTEx (Genotype-tissue expression) database for those certain tumors without normal tissue matched as controls in TIMER2 database. In addition, the “Pathological Stage Plot” module of GEPIA2 was selected to analyze the FAM107A expression in different pathological stages of all TCGA tumors via violin plots. In addition, we conducted FAM107A protein expression analysis of the CPTAC (Clinical proteomic tumor analysis consortium) dataset via the UALCAN portal (http://ualcan.path.uab.edu/analysis-prot.html), a comprehensive website for analysis of cancer OMICS data (22). Moreover, to further analyze the expression level of FAM107A in tumor cells, the Broad Institute Cancer Cell Line Encyclopedia (CCLE) website (https://portals.broadinstitute.org/ccle/about) was applied to visualize the FAM107A expression of diverse cancer cell lines.

According to the median expression level of FAM107A, tumor cases were divided into a high expression group and a low expression group to explore the relationship between FAM107A expression and the prognosis of different tumor patients. The “Survival Map” module of GEPIA2 (21) was used to obtain the overall survival (OS) and disease-free survival (DFS) significance map data of tumor patients with different FAM107A expression in TCGA database. By applying cutoff-high (50%) and cutoff-low (50%) values as the expression thresholds, all tumor cases were separated into the high-expression and low-expression cohorts. The Kaplan-Meier (K-M) survival plots were acquired by performing the log-rank test through the “Survival Analysis” module of GEPIA2 (21).

The cBioPortal (https://www.cbioportal.org/) (23, 24), a website providing visualization, analysis, and download of large-scale cancer genomics data sets, was used for genetic alteration analysis of FAM107A. After logging into the web portal, the “TCGA Pan Cancer Atlas Studies” in the “Quick select” section was chosen. Then, “FAM107A” was input for queries of the genetic alteration characteristics of FAM107A. The “Cancer Types Summary” module was selected to scan the results of the alteration frequency, mutation type, and copy number alteration (CNA) in all TCGA tumors. After selecting the “Mutations” module, the mutational information of FAM107A protein can be displayed. The “Comparison” module was also selected to analyze the survival prognosis differences among the TCGA tumor cases with and without FAM107A genetic alteration. K-M plots of OS, DFS, progression-free survival (PFS), and disease-specific survival (DSS) were generated, respectively, with log-rank P-values.

The GSCALite (http://bioinfo.life.hust.edu.cn/web/GSCALite/) website (25) was applied to analyze the Spearman correlation between FAM107A mRNA expression and copy number variation (CNV), and generate the differential methylation bubble map of FAM107A between cancer tissues and adjacent tissues in 33 types of TCGA cancers. The diagram of correlation between expression and methylation was plotted. Moreover, DiseaseMeth version 2.0 (http://bioinfo.hrbmu.edu.cn/diseasemeth/) was applied to analyze the differential DNA methylation level of FAM107A gene in cancers and normal tissues. The correlation between expression of FAM107A and expression levels of DNA-methyltransferases, tumor mutational burden (TMB), microsatellite instability (MSI), and mutation levels of mismatch repair (MMR) genes was analyzed with SangerBox tool (http://sangerbox.com/).

To explore the association between FAM107A gene expression and immune cell infiltration in TCGA datasets, the “Immune-Gene” module of the TIMER2 was used. To our knowledge, the TIDE (Tumor Immune Dysfunction and Exclusion) database was used to analyze the relationship between FAM107A expression and myeloid-derived suppressor cells (MDSCs). The partial correlation values and P-values were acquired through the purity-adjusted Spearman rank correlation test. Heatmap and scatter plot results were applied for result visualization. The correlation between expression of FAM107A and tumor purity-related factors (Stromal, Immune and ESTIMATE score) and key immune checkpoint genes across various cancers from the TCGA project was investigated with SangerBox tool.

Firstly, the query of a single protein name (“FAM107A”) and organism (“Homo sapiens”) was used and searched in the STRING website (https://string-db.org/). Then, the reliable experimentally determined FAM107A-binding proteins were obtained after setting the following filters: minimum required interaction score [“Low confidence (0.150)”] and active interaction sources (“experiments”). Subsequently, the “Similar Gene Detection” module of GEPIA2 was used to obtain the top 300 FAM107A-correlated genes based on the datasets of all TCGA tumor samples. These genes were considered to have potential similar molecular functions as FAM107A and the “correlation analysis” module of GEPIA2 was applied to conduct a pairwise gene Pearson correlation analysis of FAM107A and FAM107A-correlated genes. The intersection of FAM107A-binding and correlated genes was visualized via a Venn diagram. Moreover, the two sets of genes were combined to perform gene ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis by applying the “clusterProfiler” R package of R language software (https://www.r-project.org/), with two-tailed P <0.05 considered statistically significant. The results were finally visualized with the “tidyr” and “ggplot2” R packages.

The immortalized normal human urothelial cell line SV-HUC1, the human bladder urothelial carcinoma (BLCA) cell lines UM-UC3, T24, 5637 cell lines, the human embryonic kidney cell line 293T, and the human renal cancer(RC) cell lines 786-O, 769-P cell lines were all purchased from the Chinese Academy of Sciences Committee on Culture Collection Cell Bank, Shanghai Institutes for Biological Sciences (Shanghai, China). SV-HUC1 cells were cultured in F-12K Nutrient Mixture (Gibco) supplemented with 10% fetal bovine serum(FBS) and 1% penicillin/treptomycin. UM-UC3 and 293T cells were cultured in DMEM (Gibco, USA) supplemented with 10% FBS and 1% penicillin/treptomycin. T24, 5637, 786-O and 769-P cells were cultured in RPMI 1640 medium(Hyclone) supplemented with 10% FBS and 1% penicillin/treptomycin. All the cells were cultured at 37°Cin a humidified incubator with 5% CO₂. After purchasing the siRNAs against FAM107A and siRNA-control from GenePharma(RiboBio, Shanghai, China), the indicated cells were transfected with siRNAs assisted with the Lipofectamine 2000 according to the manufacturer’s instructions(Invitrogen, USA). The siRNA template sequences in our study were applied as following: siFAM107A#1, sense 5’-GCCAGAAUACAGAGAGUGGTT-3’, antisense 5’-CCACUCUCUGUAUUCUGGCTT-3’; siFAM107A#2, sense 5’-GCUGGAAUAGCAUCUCCUUTT-3’, antisense 5’-AAGGAGAUGCUAUUCCAGCTT-3’; siNC, sense 5’-UUCUCCGAACGUGUCACGUTT-3’, antisense 5’-ACGUGACACGUUCGGAGAATT-3’.

Total RNA of cells was extracted with Trizol reagent(Invitrogen), and then cDNA was synthesized in accordance with manufacturer’s instruction(Accurate Biology, Hunan, China). Quantitative real-time PCR (qRT-PCR) was conducted to determine the mRNA expression using SYBR Green Permix Pro Taq HS qPCR kit (Accurate Biology) on the QS5 PCR machine (Applied Biosystems, USA). Gene expression levels were normalized against human housekeeping gene β-actin and calculated by the 2^-ΔΔCt relative quantification method. The specific primers in our study were applied as following: FAM107A, forward 5’- AGGGAGCGGGCAGACATTGG-3’ and reverse 5’-CACGGGGTTCAGCAGCTTCTTG-3’; GAPDH, forward 5’-GGTGAAGGTCGGAGTCAACG -3’, reverse 5’-CAAAGTTGTCATGGATGACC-3’.

Cell proliferation was assessed by using Cell Counting Kit-8 reagent (Abcam, UK). Cells in the logarithmic growth phase were trypsinized and seeded at a density of 5×10³ cells/well in each well of a 96-well plate. After incubation at 37°Cfor different periods of time, CCK-8 reagent was added 10 μl into each well and the optical density values were measured at 460 nm using a microplate reader (Bio-Tek Instruments, Germany).

We conducted cell colony-formation assay to evaluate the clonogenic ability of FAM107A-interfered BLCA and RC cells And 1,000 tumor cells were plated into 6-well plates and cultured for 10 days. For cell fixation, 4% paraformaldehyde was used and for staining, 2% crystal violet (Beyotime, China) was used. Finally, the cell colonies were counted and imaged.

Cells were seeded in 6-well plates and cultured until the cells were basically 100% confluence. The cell monolayer was scraped by 200μl pipette and rinsed three times with PBS to form a clean wound. After incubation at 37°C for 24h, the closure of scratches was observed under an inverted microscope(Olympus). The corresponding photographs of scratch at 0h and 24h were taken under the inverted microscope. The cell-free area at 0h and 24h were measured by means of the Image J software (National Institutes of Health, Bethesda, MD, USA) and the percentage of wound closure was calculated.

Transwell chambers (pore size, 8.0 μm; Biosciences, Heidelberg, Germany) coated with Matrigel (BD Biosciences, USA) at a concentration of 2 mg/ml and incubated at 4°C for 3h were used for cell invasion assay and 2×10⁴ cells were cultured on upper chamber with 100 μl serum-free medium. As chemo-attractant, 600 medium containing 20% FBS was supplemented into the lower chamber. After incubation for 48h, cells on the inferior surface of the transwell chamber were fixed with 4% paraformaldehyde and stained with 2% crystal violet. Photographs of five visual fields were taken under the inverted microscope (Olympus) and the number of invasive cells was counted.

The total protein was extracted from cells by using Radio Immunoprecipitation Assay (RIPA) buffer with the proteinase inhibitor. After centrifuged for 20min, the supernatants were collected and used for proteic concentration analysis with G250 solution. Twenty micrograms protein lysate were separated on 12% polyacrylamide gels and electrophoretically transferred onto polyvinylidene difluoride(PVDF) membrane. The membrane was blocked with Tris Buffered Saline with Tween(TBST) containing 5% skim milk for 1h at room temperature. Then, the membrane was blotted with the primary antibodies(anti-FAM107A, Absin, anti-β-actin, Cell Signaling Technology, USA) overnight at 4°C. On the second day, the membrane was washed with TBST three times and incubated with horseradish peroxidase (HRP) labeled IgG at room temperature for 1h. The immunoreactive bands were detected by using ECL Western blotting reagents with chemiluminescence detection system.

The data were displayed as means ± standard deviation(SD). SPSS ver. 23.0 software (SPSS, Inc., Chicago, IL, USA) was used for statistical analysis. Pairwise differences between the 2 groups was assessed by Student’s t test. The significance of differences among more than 2 groups was assessed using ANOVA analysis. P <0.05 was considered statistically significantly different.

To identify the differential expression of FAM107A gene between tumor and normal tissues, we firstly applied the TIMER2 approach to analyze the significance of FAM107A expression. The result (Figure 1A) shows lower expression(P<0.05) in BLCA (Bladder urothelial carcinoma), BRCA(Breast invasive carcinoma), CESC (Cervical squamous cell carcinoma and endocervical adenocarcinoma), COAD (Colon adenocarcinoma), ESCA(Esophageal carcinoma), GBM (Glioblastoma multiforme), HNSC(Head and Neck squamous cell carcinoma), KICH(Kidney Chromophobe), KIRC (Kidney renal clear cell carcinoma), KIRP(Kidney renal papillary cell carcinoma), LUAD (Lung adenocarcinoma), LUSC (Lung squamous cell carcinoma), PRAD(Prostate adenocarcinoma), READ (Rectum adenocarcinoma), STAD (Stomach adenocarcinoma), THCA (Thyroid carcinoma), and UCEC(Uterine Corpus Endometrial Carcinoma) than the corresponding normal tissues. Inversely, it was found highly expressed in lymphoma. The down-regulating expression of FAM107A between the normal tissues and tumor tissues was also detected in OV (Ovarian serous cystadenocarcinoma) and THYM(Thymoma) after including the normal tissues of the GTEx dataset as controls (Figure S1, P < 0.01). However, there was an up-regulating expression of FAM107A in TGCT (Testicular Germ Cell Tumors) and UCS (Uterine Carcinosarcoma) (Figure S1, P < 0.01). Besides, no significant difference of FAM107A expression between tumor and normal tissues was obtained significant for ACC (Adrenocortical carcinoma), DLBC (Lymphoid Neoplasm Diffuse Large B-cell Lymphoma), LAML (Acute Myeloid Leukemia) or LGG(Brain Lower Grade Glioma). Subsequently, the expression status of FAM107A across various cancer types of TCGA was analyzed using the Oncomine dataset. As shown in Figure 1B, FAM107A had a relatively lower expression in bladder cancer, brain and central nervous system (CNS) cancer, breast cancer, cervical cancer, colorectal cancer, esophageal cancer, gastric cancer, head and neck cancer, kidney cancer, lung cancer, melanoma, pancreatic cancer, prostate cancer, and sarcoma.

Using the “Pathological Stage Plot” module of GEPIA2, a correlation between FAM107A expression and the pathological stages of cancers was also detected, including KICH, KIRC, LIHC (Liver hepatocellular carcinoma), LUSC, SKCM (Skin Cutaneous Melanoma), STAD (all P<0.05) (Figure 1C). As for the expression of FAM107A protein, the CPTAC dataset was implemented to explore the difference between primary cancers and normal tissues. The results showed lower protein expression of FAM107A in lung adenocarcinoma and UCEC than in normal tissues (P<0.05) (Figure 1D).

To be comprehensive, we also combined the HPA, GTEx, and CCLE datasets to assess the expression level of FAM107A in nontumor tissues and various tumor cell lines. Highest expression of FAM107A mRNA was detected in the central nervous system for normal tissues, similar to the cerebral cortex, cerebellum, hippocampal amygdala, and basal ganglia. Meanwhile, a neglected expression of FAM107A was also observed in endocrine glands (thyroid gland and pituitary gland), lung and urogenital system (kidney, urinary bladder, epididymis, seminal vesicle, prostate, fallopian tube, endometrium, cervix, uterine and placenta) (Figures S2A, B). However, a quite low expression level was found in most tumor cell lines after analyzing the sequencing data from CCLE dataset (Figure S2C).

To explore the prognostic value of FAM107A, the cancer cases were divided into high-expression and low-expression groups according to the median expression value of FAM107A for investigating the correlation of FAM107A expression with prognoses of patients across different tumor types by using the datasets from TCGA and GEPIA2. The results displayed that down-regulated FAM107A expression was associated with poor prognosis in many cancers. A correlation between low FAM107A expression and poor overall survival from the TCGA cases was detected in KIRC (HR=0.59, Logrank p<0.001), LIHC (HR=0.69, Logrank p=0.038), PAAD (Pancreatic adenocarcinoma) (HR=0.62, Logrank p=0.024), PRAD (HR=0.12, Logrank p=0.018), SKCM (HR=0.74, Logrank p=0.027), HNSC (HR=0.72, Logrank p=0.017) (Figure 2A). Besides, low FAM107A expression linked to poor prognosis as for disease-free survival (DFS) was discovered in CHOL (Cholangiocarcinoma) (HR=0.38, Logrank p=0.038), HNSC (HR=0.67, Logrank p=0.019), OV (HR=0.78, Logrank p=0.044), PAAD (HR=0.6, Logrank p=0.024), PRAD (HR=0.52, Logrank p=0.0025) (Figure 2B). Conversely, the high expression level of FAM107A was associated with poor OS for adrenocortical carcinoma (ACC) (HR=3, Logrank p=0.0061) and STAD (HR=1.5, Logrank p=0.0093), and high FAM107A expression was also related to poor DFS for ACC (HR=2.3, Logrank p=0.016) (Figures 2A, B). All these survival data presented a different association of FAM107A expression with the prognosis of different cancer cases.

To study the genetic alteration status of FAM107A in different tumors, the cBioPortal tool was applied to analyze FAM107A genetic alteration with data from TCGA dataset. As shown in Figure S3, four kinds of FAM107A genetic alteration were detected from all TCGA tumor samples with a total frequency of about 0.9%, including deep deletion, amplification, missense mutation, and splice mutation. Deep deletion was the main type of FAM107A genetic alteration. All FAM107A gene mutational sites and sample numbers are presented in Figure 3A. As for the specific tumors, uterine carcinosarcoma appeared the highest alteration frequency of FAM107A(>3%). Meanwhile, an alteration frequency exceeding 2% was determined in skin cutaneous melanoma, esophageal adenocarcinoma, kidney renal clear cell carcinoma, and uterine corpus endometrial carcinoma. The types, sites, and case numbers of the FAM107A genetic alteration are further presented in Figure 3B. Moreover, the potential correlation of FAM107A genetic alteration with the survival prognosis of different tumor cases was explored using the “Comparison” module of cBioPortal tool. The result showed a worse prognosis for LUSC patients with FAM107A genetic alteration with regard to disease-free survival(P = 1.417e-3), disease-specific survival(P = 0.0267) and progression-free survival(P = 5.078e-5), but not overall survival(P = 0.316) (Figure 3C). In addition, the correlation between FAM107A expression and tumor mutational burden (TMB)/microsatellite instability (MSI) was analyzed across all cancers of TCGA. The result showed a negative correlation between FAM107A expression and TMB existed in most cancers including BLCA (P = 0.018), BRCA (P = 0.02), HNSC (P = 0.044), KIRC (P= 0.0043), LGG (Brain Lower Grade Glioma) (P= 0.0013), LIHC (P=0.0028), LUAD (P= 1.5e-07), LUSC (P= 0.0028), PAAD (P= 6e-08), PRAD (P= 5.3e-27), READ (P= 0.0073), SKCM (P= 0.014), STAD (P= 5.7e-15), THCA (P= 0.007) and UCE C (P= 0.00021). However, a positive correlation between FAM107A expression and TMB was found in LAML (P = 0.048) and THYM (Thymoma) (P= 0.0062) (Figure S5A). As for MSI, we observed a negative correlation between FAM107A expression and MSI for CESC (P = 0.0058), ESCA (P = 0.0039), HNSC (P = 0.0012), LUSC (P= 0.043), OV (Ovarian serous cystadenocarcinoma) (P= 0.0041), PAAD (P=0.046), PRAD (P= 0.0014), SKCM (P= 0.0014), STAD(P= 9.8e-11) and UCS (P = 0.016) (Figure S5B). In-depth research should be carried out to determine this result. Interestingly, the expression of FAM107A was detected to have significant correlations with mutation levels of more than one key mismatch repair (MMR) genes (MLH1, MSH2, MSH6, PMS2, and EPCAM) in various cancers like BRCA, KIRC, and THCA (Figure 4D). Moreover, a statistically significant correlation between FAM107A CNV and its mRNA expression was observed in LUAD, BLCA, TGCT, CESC, THYM, ESCA, OV, LUSC and HNSC, which indicated CNV may be one of the important mechanisms leading to the alteration of FAM107A gene expression (Figure 4A).

To identify whether DNA methylation plays an essential role in regulating the gene expression of FAM107A, we firstly applied the GSCALite database to analyze the correlation between DNA methylation of FAM107A and expression in various cancers. As shown in Figure 4B, the expression of FAM107A was found to be negatively correlated with DNA methylation in multiple cancers, mainly including THYM, KIRC, LGG, KIRP, BRCA, STAD, PRAD, COAD, PAAD, MESO(Mesothelioma), PCPG(Pheochromocytoma and Paraganglioma), HNSC, UCEC, SKCM, THCA and LUAD. Subsequently, using the DiseaseMeth(version 2.0) database, we investigated the differential DNA methylation level between tumors and normal tissues in pan-cancers. The results showed a higher DNA methylation level existed in many cancers than in corresponding normal tissues, including COAD, pilocytic astrocytoma tumor, READ, nasopharyngeal carcinoma, CHOL, LUAD, LUSC, LIHC, oral squamous cell carcinoma, KIRC, esophageal squamous cell carcinoma, ESCA, HNSC and PAAD(Figure S4). However, DNA methylation level was detected anomaly higher in normal tissues than LGG. Furthermore, we assessed the correlations between gene expression of four DNA-methyltransferases (DNMT1, DNMT2, DNMT3A, and DNMT3B) and expression of FAM107A mRNA. As shown in Figure 4C, in multiple kinds of cancers including ACC, BRCA, CESC, COAD, ESCA, HNSC, KIRC, LGG, LIHC, LUSC, PAAD, PRAD, SARC, STAD, TGCT, THCA, THYM and UCEC, FAM107A was observed to have an expressive correlation with at least one type of DNA-methyltransferases.

Accumulating evidence indicates that tumor immune microenvironment plays an important role in the occurrence, progression, and metastasis of tumors. Therefore, we used the SangerBox tool to investigate the correlation of FAM107A expression with Stromal, Immune, and ESTIMATE score in pan-cancers, which represented the abundance of stromal components, immune components, and tumor purity to some extent. According to rank of the relationship between the expression level and the score, scatter plots of top nine tumors of Stromal, Immune, and ESTIMATE score were displayed, respectively (Figures S6–S8). Then, we applied the TIMER2.0 database to explore the potential relationship between the FAM107A expression and infiltration level of diverse immune cells in multiple cancers of TCGA. It is worth noting that the immune infiltration of myeloid derived suppressor cells(MDSCs) was negatively correlated with prognosis in many cancers (Figure 5A). Herein, we investigated the correlation between the FAM107A expression and infiltration level of MDSCs, and the results showed a statistically negative correlation between them in many tumors as expected (Figure 5B). Moreover, we further assessed the expression correlations between FAM107A and immune checkpoints in pan-cancers. As shown in Figure 5C, the FAM107A expression was robustly and significantly correlated with recognized immune checkpoints in most cancers including BLCA, BRCA, CHOL, COAD, HNSC, KIRC, PAAD, READ, STAD, and TGCT.

To explore the potential molecular function of FAM107A involved in occurrence, progression, or metastasis of tumors, we combined String and GEPIA2 tools to obtain the FAM107A-binding proteins and expression-correlated genes for enrichment analyses and 10 FAM107A-binding proteins were proved by experimental evidence via STRING tool (Figure 6A). Using the TIMER2.0 database, we performed a heatmap to show the expressive correlation between FAM107A and these 10 corresponding genes in diverse cancers. As shown in Figure 6B, the expression of FAM107A was detected in most cancers to be significantly positively correlated with four of above genes including GPM6A(Glycoprotein M6A), HLF(Hepatic leukemia factor), SLCO1C1(Solute carrier organic anion transporter family, member 1C1), SPARCL1(Secreted protein acidic and rich in cysteines-like protein 1). Afterwards, we used the GEPIA2 tool to obtain top 300 FAM107A-correlated genes by combining all tumor expression data of TCGA. Subsequently, KEGG and GO enrichment analyses were conducted with the union of the 10 FAM107A-targeting genes and 300 expression-correlated genes. The KEGG data indicated that many genes are linked to the pathways of “Cell adhesion molecules” and “cAMP signaling pathway” (Figure 6C). Likewise, the result showed that “developmental cell growth”, “cell-cell adhesion via plasma-membrane adhesion” and “positive regulation of cAMP mediated signaling” were enriched in GO terms and might be in involved in the impact of FAM107A on tumor pathogenesis (Figure 6D).

To further investigate the effect of FAM107A on bladder cancer, we firstly examined the expression of FAM107A in multiple bladder cancer cell lines and renal cancer cell lines. As shown in Figure 7A, a relatively lower expression of FAM107A was detected in UMUC3 and T24 cells compared with SV-HUC cells. Likewise, a lower expression of FAM107A was also noted in renal cancer cell lines (Figure S9A). Subsequently, we downregulated the FAM107A expression in UMUC3, T24, 786-O and 769-P cells via two FAM107A siRNAs. The successful silence of FAM107A expression in these four cell lines was validated at a translational level (Figure 7B, Figure S9B). Next, CCK8 assay was carried out to evaluate cell proliferation. The results suggest a significant promotion of proliferation rates in the tumor cells with FAM107A downregulation (Figure 7C and Figure S9C). Colony-formation assay also confirmed the positive effect of siFAM107A on bladder cancer cell proliferation (Figure 7D, Figure S9D). Finally, wound healing and Transwell assays were performed to investigate the effect of FAM107A on the motility of bladder cancer cells. There was a significant reduction in both migration and invasion in tumor cells after knockdown of FAM107A (Figures 7E, F, Figures S9E, F). In summary, downregulated FAM107A facilitates the proliferation, migration, and invasion in bladder cancer and renal cancer cells.