Identication of IFITM1 with prognostic value in pancreatic cancer using network module-based analysis

Background: Despite improvements reported in diagnosis and treatments in recent decades, pancreatic cancer still characterized with poor prognosis and low survival rate among solid tumors. Intensive interests have grown in exploring novel predictive biomarkers, aiming to enhance the eciency in early detection and treatment prognosis. Methods: In this study, we identied the differentially expressed genes (DEGs) in pancreatic cancer through analyzed ve gene expression proles, and established the functional modules according to the functional interaction (FI) network between DEGs. Results: Signicant upregulation of the selected DEG, IFITM1, was evaluated in several bioinformatics online tools and veried with immunohistochemistry staining from 90 pancreatic cancer patients’ samples. Prognostic data showed that high expression of IFITM1 associated with poor survival and multivariate Cox regression analysis showed that IFITM1 was one of independent prognostic factor for the overall survival. Meanwhile, signicant correlations of the expression of IFITM1 and the inltration of immune cells were found by TIMER. Furthermore, higher level of IFITM1 was assessed in pancreatic cancer cell lines compared to normal human pancreatic duct epithelial cell and silencing IFITM1 in tumor cells remarkedly inhibited the cancer tumorigenicity. Conclusion: Collectively, our ndings suggested that IFITM1 might have promising clinical utility for pancreatic cancer. Three human pancreatic cancer cell lines, PanC-1, BxPc-3, SW1990, and one normal pancreatic cell line, HPDE6-C7, were used in the study. The panC-1 and BxPc-3 cells were purchased from the Cell Bank of Shanghai of Biochemistry and Cell Biology (SH, China), and the others were purchased from the American Type Culture Collection (ATCC, VA, United States). The panC-1 and HPDE6-C7 cells were cultured in DMEM (Gibco, NY, United States) and the BxPc-3 and SW1990 cells were cultured in RPMI 1640 (Gibco, NY, United States) with both cultures supplemented with 10% FBS (Gibco, NY, United States) and 100 U penicillin/streptomycin (Gibco, NY, United States) in a humid atmosphere containing 5% CO 2 at 37 °C. Short-tandem repeat (STR) proles were authenticated to ensure the quality of the cell lines. All of the cells were passaged by standard cell culture techniques.


Background
Pancreatic cancer is expected to be the second leading cause of cancer related deaths in developed countries by 2030, as characterized with highly invasive and metastatic as well as extremely resistant to chemo-radio-therapy [1]. Despite of some efforts made on adjuvant therapeutics recent years, the clinical mortality ratio remains elevated to date [2]. 5-year survival rate was reported less than 10% with an estimated 432,242 deaths occurring worldwide in 2018 [3]. Because of this poor prognosis, better identi cation of clinical biomarkers useful for making therapeutic decisions and developing targets for innovative drugs is urgently needed in this complex and heterogeneous mutational landscape.
The human interferon (IFN)-induced transmembrane protein 1 (IFITM1), also named Leu13 or CD225, is a 17-kDa cell-surface membrane protein in the IFN-stimulated genes (ISGs) protein family along with IFITM2 and IFITM3. Since it was rst discovered in neuroblastoma cells in 1984, it has been identi ed participating in various biological processes including cell proliferation and adhesion, and being important in immunity and antiviral activities [4,5]. Particularly, increasing studies have con rmed that IFITM1 is overexpressed in numerous human cancers during the last decade, such as lung, gastric, colorectal and ovarian cancers [6][7][8][9]. Furthermore, positive IFITM1 expression was illustrated in correlations with poorer prognosis in diverse tumors [10,11]. Hence, it is very likely that IFITM1 may also hold hitherto undiscovered value in pancreatic cancer and might be further associated with prognosis of pancreatic cancer patients.
In this study, we investigated gene expression pro les from Gene Expression Omnibus (GEO) repository to screen out common DEGs between pancreatic cancer and normal pancreas tissue, and analyzed them through establishing a protein functional interaction (FI) network. Functional network-based modules were constructed and biological pathways were analyzed afterward. It was con rmed that IFITM1 playing a vital role in pancreatic cancer by 90 pancreas cancer patients with clinical and pathological characteristics and survival time as well as was veri ed through several bioinformatics online tools. Our study suggested that IFITM1 may have promising clinical utility for prognostic strati cation, and combining IFITM1 expression pro les systematically with patients' clinical characteristics is promising to be effective for developing treatment.

Patient samples preparation
A total of 90 human pancreatic cancer specimens were collected, 60 of which with paired non-malignant tissue, between 2004 and 2008 at Taizhou People's Hospital of China for the tissue microarray detection(TMA). All the specimens were stored at the Biobank Center of National Engineering Center for Biochips at Alenabio and Shanghai Outdo Biotech Company, China. All the patients received no preoperative anticancer treatment or post-operative adjuvant chemo-radio-therapy. Histological differentiation grade and disease stages were classi ed according to the American Joint Committee on Cancer (AJCC) TNM Classi cation (7th edition). The follow-up time was recorded from the date of the surgery to the date of death or last visit to the clinic. The use of pancreatic cancer samples and clinical data were reviewed and approved by the Ethical Committees of the National Engineering Center for Biochips at Shanghai along with the Taizhou People's Hospitals. Informed consent was written in accordance with the Declaration of Helsinki from all participants.

Microarray data
To investigate biologic characteristics of genes differently expressed between pancreatic cancer and normal pancreas tissue, ve datasets retrieved from National Center for Biotechnology Information (NCBI)-GEO (http://www.ncbi.nlm. nih.gov/geo/) were obtained for the analysis, including GSE101448 (18 with pancreatic tumor and 13 non-tumor pancreatic tissue samples), GSE62165 (118 pancreatic tumor samples and 13 control samples), GSE62452 (69 pancreatic tumor and adjacent non-tumor tissues), GSE41372 (matched pancreatic ductal adenocarcinoma samples from 15 patients ) and GSE28735 (45 matching pairs of pancreatic tumor and adjacent non-tumor tissues). NCBI-GEO is a public repository containing an initial database of 1,325 biologically de ned gene pro les, and was developed to help analysis and interpretation of the long lists of genes produced from microarray-based experiments. Details of the series data are listed in Table I.

Initial identi cation of DEGs
As microarray batch effects appear to be the major contributions to the differential expression, SangerBox online tool (http://sangerbox.com/) was used to eliminate them before taken into account [15]. Next, data were imported into BRB-Array Tools to identify the DEGs between pancreatic tumor and normal pancreas tissues. The affymetrix platform was applied for dataset and gene annotation. The probes were ltered when some gene expression data was missing. The Bioinformatics & Evolutionary Genomics webtool (http://bioinformatics.psb.ugent.be/webtools/Venn/) was used to depicted graphically in the form of Venn to detect the commonly DEGs among the ve datasets.

Establishment of FI network
The Reactome pathway database (version 7.2.0; http://www.reactome.org) was applied to analyze the pathway enrichment for the sets of genes, investigate the functional relationships of the genes and visualize the outcomes with diagrams that can be manually generated [16]. Over 60% human proteins were covered to help construct the Reactome FI network. The pairwise interaction information was uploaded into Cytoscape software to establish the FI network [17]. A false discovery rate (FDR) < 0.05 was accepted as the cut-off criterion.
Construction of the FI network-based functional modules Page 5/48 To construct the FI network-based module analysis, the Microarray Data Analysis bioinformatics tool installed in the ReactomeFIViz was applied [18]. After uploaded the txt le containing gene expression data, correlations between genes in the FI-network were analyzed. The Markov cluster algorithm (MCL) was used to achieve the functional modules with a selected module size. In order to supervise the size of the modules, we inputted 5.0 for the in ation coe cient [19]. MCL modules in size of three or more was applied as well as the average Pearson correlation coe cient de nition ≥ 0.25. Six different colors were used to label the nodes in six different network modules. Module size ≥ 7 and FDG < 0.05 were used as cutoffs for the KEGG pathway and GO term enrichment analysis of the modules in a separate manner Preliminary validation of the selected gene The Human Protein Atlas (www.proteinatlas.org) presented an expression map of human protein across 20 cancer types and 44 normal human tissues. It measured the RNA level as well as used antibody pro ling to precisely localize the corresponding proteins [20]. With the Human Protein Atlas (www.proteinatlas.org), we detected the IFITM1 protein expression in both pancreatic cancer and normal pancreas tissues. Oncomine microarray database (www.oncomine.org) was a cancer microarray database as well as an online data-mining platform which contained 65 gene expression datasets comprising nearly 48 million gene expression measurements form over 4700 microarray experiments [21]. With the Oncomine online tool we explore the expression of IFITM1 in pancreatic cancer and normal pancreas tissues across eight different datasets.
Validation of IFITM1 with the Cancer Genome Atlas (TCGA) GEPIA is an online server, containing more than 8,000 normal and 9,000 tumor samples, providing interactive and customizable functions based on TCGA and Genotype-Tissue Expression (GTEx) database [22]. Boxplot with jitter was applied to compare the expression level of IFITM1 mRNA in pancreas cancer and paired normal pancreas tissue. To validate the expression of IFITM1 correlates with pancreatic cancer patients' survivals and pathological stages, it was searched into the database to obtain Kaplan-Meier survival curves and violin plots. The requested mRNA expressions above or below the median were classi ed the patients into both high and low expression groups. The log rank P value and hazard ratio (HR) with 95% con dence intervals were calculated and displayed on the website. A P-value < 0.05 was considered statistically signi cant.

Immunohistochemical (IHC) staining
To analyze IFITM1 expression by IHC staining experiment, 90 pancreatic cancer patients were included in the TMA analysis. Sample slides were dried for adherence, depara nized in xylene decreasing ethanol concentrations. Tissue sections were marked and incubated with rabbit anti-IFITM1 diluted 1:1,500 (ab224063, Abcam, Cambridge, United Kingdom) at 4℃ overnight. The percentage of the IHC staining and the staining intensity were assessed and recorded by two pathologists from the department of pathology, who were both blinded to the patients' features. The H-scores ranged from 0 to 100, with 100 represented 100% of regions with strong intensity. The H-scores were dichotomized by the cutoff, the median value, for the statistical analysis.
Cell lines culture and short hairpin RNA (shRNA) lentiviral transduction Three human pancreatic cancer cell lines, PanC-1, BxPc-3, SW1990, and one normal pancreatic cell line, HPDE6-C7, were used in the study. The panC-1 and BxPc-3 cells were purchased from the Cell Bank of Shanghai Institute of Biochemistry and Cell Biology (SH, China), and the others were purchased from the American Type Culture Collection (ATCC, VA, United States). The panC-1 and HPDE6-C7 cells were cultured in DMEM (Gibco, NY, United States) and the BxPc-3 and SW1990 cells were cultured in RPMI 1640 (Gibco, NY, United States) with both cultures supplemented with 10% FBS (Gibco, NY, United States) and 100 U penicillin/streptomycin (Gibco, NY, United States) in a humid atmosphere containing 5% CO 2 at 37 °C. Short-tandem repeat (STR) pro les were authenticated to ensure the quality of the cell lines. All of the cells were passaged by standard cell culture techniques.
Two lentivirus-derived shRNAs against IFITM1 were designed by Biolink Biotechnology Co. Ltd (SH, China) to silence IFITM1 mRNA expression (the targeted sequences as follows: IFITM1 shRNA-1, 5'-CCTCATGACCATTGGATTCAT-3'; IFITM1 shRNA-2, 5'-CCTGTTCAACACCCTCTTCTT-3'. The procedure of transducing pancreatic cancer lines with shRNA lentiviral transduction was described in detail previously [23]. Cell viability assay Cells in logarithmic growth phase were seeded in 96-well Falcon Petri dishes (3000 cells/well) the day before assays were performed. Cell viability was assessed with a cell counting kit (Dojindo, Kumamoto, Japan) based on sensitive colorimetric assays with the absorbance measured at 450 nm (OD 450 nm), according to the manufacturer's instructions. The results were con rmed with three independent experiments. Wound healing assay Cells were seeded in 48-well Falcon Petri dishes. Once reached 80-90% con uence, a 200µL sterile plastic pipette tip was used to made a single wound by gently scratching in every well. Cell debris were removed with phosphate buffered saline and refreshed with serum-free medium immediately. Cells that extended the borders of the wound were taken photos and quanti ed in three randomly selected regions per well.
Colony formation assay Cells in logarithmic growth phase were seeded in 6-well Falcon Petri dishes (500 cells/well) and incubated for 7-14 days. The medium was refreshed every three days. The colonies were xed with 70% ethanol and stained with crystal violet (0.5% w/v). Cell colonies containing at least 50 cells were taken photos and quanti ed. Results were presented as means ± SD of three independent experiments.

Statistical analysis
Statistical analyses were performed using SPSS 23.0 (SPSS Inc.; Chicago, IL, United States) and Prism 8 (GraphPad Software, San Diego, CA, United States). Correlations between protein expression and clinicopathologic features in pancreatic cancer patients were analyzed by the χ2 test. Overall survival was evaluated using the Kaplan-Meier method and the log-rank test. A Cox proportional hazards regression model was used to assess the univariate and multivariate analyses. The Enter method was used to select the independent variables in the multivariate analysis.

Screening of DEGs
Five datasets from GEO containing gene expression pro les of pancreatic cancer as well as normal pancreas samples were downloaded for microarray analysis. Random variance mode method along with the paired t-test was applied for the differential expression calculation. Considering the criteria of |log2 (Fold change) |>1 adjusted P value < 0.05, 483 DEGs were obtained ( Fig. 2A). Of those, 231 DEGs were identi ed up-regulated and 252 down-regulated (Table S1).

Establishment of pancreatic cancer-related FI network
FIs data derived from the Human Protein Reference Database [24], BioGrid [25], I2D [26], IntACT [27], MINT [28], the Database of Interacting Proteins [29] as well as multiple high-throughput assays [30]. We established the FI network through mapping the pancreatic cancer-related DEGs to the FIs data. 245 isolated nodes and 227 in eight clusters were identi ed with an effective mean degree of 2.20 (Fig. 2B). The connectivity of a node represents the number of its neighbors and the neighborhood connectivity represents the average connectivity of all neighbors of the selected node. Under the criteria of neighborhood connectivity ≥ 20 and the shortest path length ≥ 3.5 for nonsingle modules, eight genes (EDIL3, RSAD2, AHR, GBP2, IFITM1, AOX1, KIF11, and CEP170) among all the nodes in the FI network were ltered out (Table S2).

Functions analysis of the FI network
To gain insights into the role of the putative targets involved in various biological processes and molecular functions, preliminary GO function and KEGG pathways analysis were performed with OmicShare online database. The most signi cant KEGG pathway was the extracellular matrix (ECM)receptor interaction (Table S3) and the pathways in cancer had the most numbers of the genes involved (Fig. 2C). Furthermore, the GO outcomes showed that the FI network mainly related to cell part (GO:0044464; cellular component), cellular process (GO:0009987; biological process), and binding (GO:0005488; molecular function) ( Fig. 2D; details see in Tables S4-S6).
Identi cation of FI network-based functional modules MCL was applied for the network-based module-clustering algorithm, and the absolute value of the Pearson correlation coe cient of the interacting genes was used to weigh each correlation edge. A total of six modules involving 63 genes were obtained, ranging from 7 to 19 genes per module ( Table 2, Fig. 3A). Next, GO function and pathway analyses were performed on the six network-based functional modules. Through enrichment analysis (Table S7), we identi ed Module 0 in relation with classical protein adhesion and absorption in KEGG and integrins in Panther Pathways. Module 1 was related to JAK-STAT and JAK-STAT signaling pathway in KEGG and IFN-gamma pathway in NCI-PID. Module 2 was found related to ERBB signaling in KEGG and Module 3 to immunological activations in KEGG. Interleukin families in Reactome were enriched in Module 4. Meanwhile, the enrichment of GO term was also analyzed. The calculation results showed that GO functions of six modules complied with the pathway annotations (Table S8-10).  Five genes common to both Module 1, with an average correlation equal to 1.00, and the set of previously selected eight genes were selected for the volcano plot analysis of the ve GEO datasets (Fig. 3B, IFITM1 marker in red triangle). Combining the results from literature mining, we were interested in IFITM1 could be a potential biomarker in pancreatic cancer.

Aberrant Expression of IFITM1 in pancreatic cancer
First, we used the Oncomine database to explore the transcriptional level of IFITM1 in pancreatic cancer and normal pancreas tissues (Fig. 4D). Eight studies reported from 2003 to 2009 were ltered out with the screening condition of "IFITM1; Cancer vs. Normal Analysis; Cancer Type: Pancreatic Cancer", and assessed with a heat map analysis (Fig. 4A). Segara [38]. Meanwhile, we checked the protein expression of IFITM1 with the help of the Human Protein Atlas and the positively strong level was also found in pancreatic cancer specimens compared with normal tissues (Fig. 4B, C). Furthermore, the GEPIA online tool is also used to validate the differential mRNA expression analysis of IFITM1 in pancreatic cancer and normal tissues (Fig. 4E).
IHC experiments were applied to validate IFITM1 expression in 90 pancreatic cancer patients. Representative IHC stained images showed that IFITM1 mainly located in the cell membrane of the pancreatic cancer tissues. A signi cantly up-regulated expression of IFITM1 was found in 76.7% of the patients (69/90), which were consistent with the results analyzed by the microarray data (P < 0.001, see in Fig. 5A-B).

Correlation of IFITM1 expression with pancreatic cancer patients' clinical features
The χ2 test was applied to assess the relations between the expressions of IFITM1 to the clinical and pathologic features of the pancreatic cancer patients, including patients' gender, age at diagnosis, tumor size, histologic differentiation, tumor stage, nodal status, distant metastasis as well as TNM stage.
Histologic differentiation (P = 0.032) and tumor stage (P = 0.044) were found have close correlations with IFITM1 expression ( Table 3). The signi cant correlation between IFITM1 and tumor stage was also veri ed from the GEPIA (Fig. 5C). TCGA-based analysis validates the prognostic IFITM1 signature A total of 184 pancreatic cancer patients in the TCGA database, with ve of them had no documented T or N stage, leaving 179 patients for the con rmation of the relationship between IFITM1 and human pancreatic cancer. Kaplan-Meier analysis and the logrank test were used to analysis. As shown in Fig. 6A, patients with low level of IFITM1 had better disease-free survival than those with high IFITM1(Logrank p = 0.036), and was also true for a better overall survival (Fig. 6B, Logrank p = 0.014).

Veri cation of IFITM1 with prognostic signature
For the survival analysis, the median IHC value was used as an objective cutoff to stratify 90 pancreatic cancer patients with low and high expression. We used the Kaplan-Meier analysis and the log-rank test to explore the correlation. The follow-up time was 81 months and high IFITM1 level was found signi cantly related to decreased overall survival (P = 0.032) (Fig. 6C). The median overall survival of the pancreatic cancer patients with the low and high IFITM1 expression was 10.0 vs. 43.0 months, respectively As for the univariable Cox regression analysis, positive nodal status (P = 0.008), TNM stage (P = 0.012), histologic differentiation (P = 0.004) and IFITM1 expression (P = 0.040) were signi cantly associated with worse outcome (Table 4). In the multivariable analysis, IFITM1 expression (HR = 2.134; 95% CI 1.073, 4.245; P = 0.031) remained an independent prognostic marker for overall survival of pancreatic cancer patients (Table 4). Immune cell in ltration of IFITM1 in pancreatic cancer patients IFITM1 was found involving in in ammatory processes and immune cells in ltration, which might further in uence the clinical outcomes of pancreatic cancer patients. We used the TIMER database to explore a comprehensive relation between IFITM1 levels and Immune in ltrations. Positive correlations were assessed between IFITM1 and B cell (Cor = 0.171, P = 2.51E-2), CD4 + T cell (Cor = 0.307, P = 4.90E-5), macrophage (Cor = 0.204, P = 7.34E-3), neutrophil (Cor = 0.475, P = 5.20E-11) and dendritic cell (Cor = 0.421, P = 1.01E-8). IFITM1 level was in negative association with purity (Cor=-0.165, P = 3.1E-2). Only CD8 + T cell was found having no signi cant relation to IFITM1 expression (Fig. 7).

IFITM1 silencing inhibits tumorigenicity of pancreatic cancer
Three types of pancreatic cancer cell lines and normal human pancreatic duct epithelial cell line HPDE6-C7 were used to examined the expressions of IFITM1. As shown in Fig. 8B-C, higher mRNA and protein levels of IFITM1 were found in BxPc-3, PanC-1 and SW1990, compared to HPDE6-C7. Two shRNA targeting IFITM1 were transfected into PanC-1 and SW1990 to detect the IFITM1 silencing effects, and the e ciency was examined both by qRT-PCR and western blot analysis ( Figure D-E). We performed cell counting kit-8 assay to assess the cancer cell proliferation. Results showed that downregulation of IFITM1 signi cantly decreased the proliferation of PanC-1 and SW1990 cell lines (Fig. 8A). Moreover, wound-healing assay indicated that IFITM1 silencing inhibited cell migration capability and colonyformation assay showed it greatly reduced the cancer stem cell-like properties of pancreatic cancer (Fig. 8F-G).

Discussion
With the highest incidence-to-mortality ratio of solid tumors, pancreatic cancer is an elusive disease seeking for explorations to elicit the redundant mechanisms involving pathogenesis, progression and metastasis. To improve the poor clinical expectations, we identi ed and veri ed a potential molecular biomarker through network-based analytic methods of multiple datasets and etiology-based malignancies characterizing [30].
Systematically, through ve microarray gene expression datasets, the FI network-based approach was used to build the functional modules for pancreatic cancer. On the basis of the diverse function of each module, the degree value and the connectivity of the nodes, IFITM1 was identi ed as a promising biomarker. Consistent with several bioinformatics online tools' outcomes, IFITM1 was found highly expressed in pancreatic cancer compared to normal pancreas tissues, and was con rmed through IHC experiment afterwards. Similarly, previous studies have shown high expressed IFITM1 in a variety of cancers. Through the human cancer pro ling array containing 10 lung carcinoma as well as the corresponding normal samples, IFITM1 was detected increasing signi cantly in lung carcinoma [39]. The difference was also observed in two other studies demonstrated IFITM1 overexpression through patients' tissues and lung cancer cell lines [6,40]. Immunohistochemistry on TMA of a consecutive cohort including 174 gastroesophageal adenocarcinoma patients also detected a signi cant elevation of IFITM1 in primary tumors along with lymph node metastases, compared to the adjacent normal epithelium and intestinal metaplasia [41]. Notably, histologic differentiation (P = 0.032) and tumor stage (P = 0.044) of the pancreatic cancer patients were signi cantly correlated with IFITM1 expression in the clinical and pathological analyses. To In order to explore the underlining mechanisms, we found the role of IFITM1 have been shown in the tumorigenesis involving tumor cells unlimited proliferation, invasion, angiogenesis, metastasis and treatment resistance. Overexpression of IFITM1 was rst time found enhancing cancer migration and invasion in 2005, with the cell model of gastric cancer [42]. The phenomenon was then discovered in HNSCC, glioma and gastric cells in vitro experiments and meanwhile downregulation of IFITM1 signi cantly decreased the activity of matrix metalloproteinase 9 (MM9) [44,46]. Two studies revealed that depletion of IFITM1 signi cantly inhibited aromatase inhibitor (AI)-resistant breast cancer growth and invasion and the observation reversed when the ectopic level of IFITM1 [47,48]. Recently, Y.G. Yang et al. demonstrated IFITM1 was essentially required for the progression of non-small cell lung cancer (NSCLC) both in vitro and in vivo, and silenced expression of IFITM1 signi cantly reduced the capacity of sphere-formation, invasion, and migration of NSCLC [6].
To further validate the prognostic signature of IFITM1 in pancreatic cancer patients, both univariable and multivariable Cox regression analysis were assessed and the data suggested IFITM1 an independent prognostic marker. This similar prognostic values were also seen in other type of tumors [41,43,49].
Chemotherapy, radiation therapy and endocrine therapy are three critical and effective strategies in the cancer treatment. Accumulating evidence has indicated that IFITM1 involved in radio-and chemoresistance as well as endocrine therapy resistance. In our previous study, we explored the radiosensitive possibility of adipose-derived mesenchymal stem cells and discovered IFITM1 could be important target involving in radioresistance [23] Mechanically, upregulation level of IFITM1 was correlation to high expression of STAT3 and MMP family, also to downregulation of P53 and caspase family. The association between IFITM1 and signal transducer STAT at the molecular level was also detected and evaluated by knockdown of IFITM1 in oral tumors [50,51]. While the interaction between IFITM1 and P53 signaling cascade, the apoptotic effects of other IFITM proteins was mainly P53-independent [52]. Yet, it is still unclear whether IFITM1 is of prime importance in the responsiveness of pancreatic cancer patients to clinical therapies.
In addition to the multiple cell functions involving anti-proliferation and adhesion, the IFITM1 is predominantly important in immune responses [53]. As we discovered above, immune cell in ltration was in signi cant correlations with IFITM1 in pancreatic cancer. Escape of natural killer cell recognition was reported from gastric cancer cells with increased level of IFITM1 [42]. This indicated that tumors expressing high IFITM1 were insensitive to the antiproliferative therapies, might be attributed to escape from the immune surveillance.
Although we discovered the potential role of IFITM1 in pancreas cancer, limitations to this study were acknowledged. Tumor xenograft model is needed to verify the outcomes in vivo experiments, and a larger cohort of pancreatic cancer patients are expected to validate the data.

Conclusion
Taken together, our study constructed an effective FI network-based module analysis to discover the genes differently expressed between pancreatic cancer and normal pancreas tissue, and suggested that      Depletion of IFITM1 decreased the tumorigenicity of pancreatic cancer cells. A. Control (shNC) or IFITM1depleted (shIFITM1-1 and shIFITM1-2) of PanC-1 and SW1990 cell lines were incubated for 72 hours to determine the cell proliferation rate by cell counting kit-8 assay. Data represent the means±SD of three independent experiments ( * * * P < 0.001). B-C. Detection of IFITM1 mRNA expression in three different pancreatic cancer cell lines as well as HPDE6-C7 by qRT-PCR and western blot analysis. GAPDH and β-Actin were used as controls, respectively. The results were expressed as the means±SD of three independent experiments ( * * * P < 0.001). D-E. Pancreatic cancer cell lines were transfected with shRNAs (shIFITM1-1 and shIFITM1-2), and the expression level of IFITM1 was assessed by qRT-PCR and western blot analysis using GAPDH and β-Actin as controls, respectively. The results are expressed as means±SD of three independent experiments. F. Colony-formation assay was performed to show decreased cancer stem cell-like properties of PanC-1 and SW1990 cells transfected with shIFITM1-1 or shIFITM1-2 compared with control in vitro. G. The wound-healing assay was performed to assess the effect of IFITM1 silencing on the wound closure ability of PanC-1 cells. Representative images were shown (magnification 100×).