To get a comprehensive analysis of SNAI2 expression patterns, we analyzed the SNAI2 levels in normal tissues and tumor cell lines according to the GTEx and CCLE databases. In the TCGA database, we used the Gene_DE module of TIMER2 (tumor immune estimation resource, version 2) web (http://timer.cistrome.org/) to investigate SNAI2 expression differences between tumors and neighboring normal tissues. For certain tumors without normal or with highly limited normal tissues, we used the GEPIA2 (Gene Expression Profiling Interactive Analysis, version 2) web server (27) to integrate and analyze the two databases (GTEx and TCGA databases).

IHC of SNAI2 was obtained from the THPA (https://www.proteinatlas.org) to assess the expression at the protein level in various types of cancers including pancreatic cancer. The antibody to SNAI2 for immunohistochemistry is CAB011671. Also, the survival analysis in SNAI2 immunohistochemistry was obtained from the aforementioned website. For Immunofluorescence (IF), the cells (HEK293T, Hela, HUVEC, and HCCLM3) were fixed with 4% paraformaldehyde and incubated with 0.1% TritonX-100, then stained with anti-SNAI2 (1:100) and phalloidin (1:1000) overnight at 4°C. After incubating secondary antibodies, the cell nuclei were counterstained with DAPI (1:1000). Finally, take photos with a fluorescence microscope.

We analyzed the genetic alteration characteristics of SNAI2 in Pan-Cancer by logging into the cBioPortal web (28, 29) (https://www.cbioportal.org/). The Cancer Types Summary module showed the frequency of modification, mutation type, and CNA (Copy number alteration) across all TCGA tumors. The mutated site of SNAI2 was observed in the 3D structure. The TCGA database was used to gather data on the expression of MMR genes (MLH1, MSH2, MSH6, EPCAM, and PMS2) as well as DNA methylation surrogates in pan-cancer. For subsequent analysis, the expression data were log2 transformed.

The UCSC Xena database (https://xenabrowser.net/datapages/) was used to assess the prognosis value of SNAI2 in terms of overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (DFI). Then univariate Cox regression and Kaplan-Meier model were used to assess the prognostic role of SNAI2 for a specific prognosis type in each cancer. Moreover, bivariate SNAI2 expression levels were used to perform Kaplan Meier curve analysis, whose cutoff was chosen by the surv-cutpoint function of the survminer R package (0.4.9) The log-rank p-value of the K-M method and hazard ratio (HR) with a 95% confidence interval was computed, and the outcomes were presented as a heatmap.

To analyze the effect of SNAI2 expression on cancers, we conducted the GSEA computational method to explore SNAI2 enrichment. The “gmt” file of the hallmark gene set (h.all.v7.4.symbols.gmt), which contains 50 hallmarks gene sets, was downloaded from the website of Molecular Signatures Database (MSigDB, https://www.gsea-msigdb.org/gsea/index.jsp), and used to calculate the normalized enrichment score (NES) and false discovery rate (FDR) of the DEGs between low- and high-SNAI2 expression cancer group for each biological process in each cancer type. The GSEA was carried out with the help of the R package “clusterProfiler” (30) and the results were summarized in the bubble plot depicted by the R package “ggplot2”.

TIMER is a data source for quantifying immune cell infiltrations across distinct cancers. The immune cell infiltration levels of TCGA cancers were obtained from the TIMER2 database (http://timer.cistrome.org/). We investigated the relationship between SANI2 mRNA expression and 21 immune cell subsets by using the R package “ggplot2”, and generated a heatmap by Spearman correlation analysis.

We using the Tumor Immune Single-cell Hub (TISCH) web tool (http://tisch.comp-genomics.org/documentation/) to explore the single-cell analysis (31). The analysis parameters were as below: SNAI2 (Gene), major-lineage (Cell-type annotation), and all cancers (Cancer type). The expression levels of SNAI2 in each cell type were quantified and visualized by a heatmap.

The Spearman correlation analysis was performed to show the associations between SNAI2 and reported biomarkers of cancer immunotherapy for each cancer type. Tumor mutation burden (TMB) and microsatellite instability (MSI) were the well-known immunotherapy biomarkers. In this study, we analyzed the correlations between SNAI2 and immunotherapy biomarkers in pan-cancer by TCGA database. We first acquired the gene mutation data of all cancer types possessed from the TCGA database, then calculated the TMB and MSI of each cancer sample. Then, we analyzed the association between SNAI2 expression with 47 immune checkpoint-related genes, TMB, and MSI with Spearman’s correlation method. Moreover, the Sangerbox website (http://sangerbox.com) was utilized to investigate the correlation between SNAI2 expression and neoantigens via the “Tool” module and Spearman’s correlation test. All the results were visualized as heatmaps or radar plots. Furthermore, three immune checkpoint blockade (ICB) therapy cohorts were obtained to validate the immunotherapy response prediction ability of SNAI2. The GSE91061 cohort includes 51 melanoma patients receiving nivolumab (anti-PD-1). The Gide2019 cohort includes 32 melanoma patients with anti-PD-1 and anti-CTLA4 therapy. The IMvigor210 cohort contains 348 urological cancer patients treated with atezolizumab (anti-PDL1). Patients were divided into a SNAI2 low-expression group and a high-expression group according to the best cut-off value using the surv-cutpoint function of the “survminer” R package (0.4.9). A Chi-square test was used to evaluate the proportion difference of responders between low- and high-SNAI2 cancer groups.

The Chinese Academy of Sciences Shanghai cell bank provided human pancreatic cancer cells (PANC-1, SW1990, BxPC-1, SW-979) and an H6C7 cell line for this study. The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (Gibco, USA) (FBS) and 100 units/mL of penicillin-streptomycin (Invitrogen, USA). All cells were grown at 37°C with 5% CO₂ in a humidified cell incubator. The antibodies and reagents were described in Supplemental Table 1 . Plasmids encoding shRNAs against SNAI2 were synthesized by Sheweisi Biotechnology Company (Tianjin, China) In addition, the vector encoding SNAI2 was got from Genepharma Company. The cells were transfected with overexpression constructs or shRNA plasmids using Lipofectamine 3000 (Invitrogen, USA) according to the manufacturer’s instructions. Transfected cells were cultured in DMEM medium without FBS and replaced with a complete medium after 6-8h. Subsequent experiments were conducted after transfection for 48 h.

For the western blotting assay, equal amounts of cell lysates were separated using SDS/PAGE, transferred to polyvinylidene fluoride (Millipore, USA) membranes, and then blocked in 5% skim milk. Next, the electrotransferred membranes were incubated overnight at 4°C with the indicated primary antibodies and were incubated with appropriate HRP-conjugated anti-mouse/rabbit secondary antibodies for one hour at room temperature. Lastly, immunoreactive bands were visualized with chemiluminescence kits chemiluminescence after three washes with Tris-buffered saline with 0.1% Tween (TBST). To investigate the expression level of SNAI2 mRNA in tumor tissues and cells, we conducted a qRT-PCR experiment by the conventional method. Briefly, total RNA was isolated from cells or tissues using Trizol reagent (Invitrogen, USA) and qRT-PCR was conducted by using the SYBR Green qPCR Master Mix. The primers used in PCR are as followed: SNAI2, 5’-CGAACTGGACACACATACAGTG-3’ and 5’-CTGAGGATCTCTGGTTGTGGT-3’; GAPDH, 5’-AGAAGGCTGGGGCTCATTTG-3’ and 5’-AGGGGCCATCCACAGTCTTC-3’.

To explore the effect of SNAI2 on cell proliferation, we conducted CCK-8 and 5-ethynyl-2′-deoxyuridine (EdU) staining assays. For the CCK-8 assay, multiple cultures of cells were plated in 96-well plates at a density of 4.0×10³ cells/well. At the indicated time points, viable cells were assessed using the CCK-8 reagent according to the manufacturer’s instructions (KeyGEN, Shanghai, China). For EdU staining assays, the transfected cells were planted in the 96-well plate, incubated for 36 h, and conducted by using a Cell-Light EdU DNA Cell Proliferation Kit (Ribobio, Guangzhou, China), which was calculated as the ratio of the number of EdU-positive cells to the number of total cells. In the colony formation assay, the transfected cells were plated in 6-well plates at a density of 500 cells/well and cultured for 14 days. Then the plates were fixed and stained with 0.5% crystal violet.

Transwell is a common detection method for cell migration, which refers to the movement of cells after receiving a migration signal or feeling the gradient of certain substances. The principle is to plant the cells in the upper chamber. Because the polycarbonate membrane is permeable, the components in the lower layer of the culture medium can affect the cells in the upper chamber. Cell mass can reflect the migration ability of tumor cells. Therefore, we used this experiment to explore the effect of changing SNAI2 on cell invasion and migration. Transwell systems (8-m pore size; BD Biosciences, USA) were used for migration and invasion assays. Briefly, 5×10⁴ cells resuspended in DMEM were seeded into the upper chambers not coated (for migration assay) or coated with Matrigel (for invasion assay; BD Biosciences, Franklin Lakes, USA). The lower chambers were filled with a 15% FBS medium. The chambers were removed after the indicated periods of incubation, and cells on the lower surface of the membrane were fixed, stained with 0.1% crystal violet, and photographed. For each chamber, five random visual fields were manually counted, and migration and invasion experiments were repeated three times independently.

Paired student’s-tests were used to assess gene expression data from the TCGA and GTEx datasets. Spearman’s correlation was used to assess the relationships between SNIA2 expression and immune cell abundance scores. The Kaplan-Meier technique and univariate Cox regression analysis were employed to assess the prognostic value of SNAI2 expression in each malignancy. To compare the proportions of ICI-therapy responders and non-responder between low-SNAI2 and high-SNAI2 cancer subgroups, a chi-square test was used to compute the statistical significance. The experimental data were analyzed with GraphPad Prism 9.0 software. The differences between groups were analyzed by using Student’s t-test or one-way analysis of variance (ANOVA). All experiments were repeated in triplicates. The results were reported as the Mean ± SD. All statistical tests were two-sided, and statistical significance was set at p <0.05. The abbreviations of cancers were represented in Supplemental Table 2 .

In this paper, pan-cancer samples from public databases were included for subsequent analysis including expression differences, the landscape of SNAI2 mutations, the correlation between expression and survival, and immune infiltration. The flow chart of our study design is shown in Figure 1 . Firstly, we use the GTEx database to analyze the SNAI2 expression in normal tissues. The expression of SNAI2 was shown to vary in different organs, with significantly lower expression in blood and bone marrow ( Figure 2A ). In addition, SNAI2 expression was highly expressed in most tumor cells according to the CCLE database ( Figure 2B ).

To determine the SNAI2 expression in cancers and their corresponding adjacent non-cancerous tissues, we employed the TCGA and GTEx datasets to analyze the SNAI2 mRNA levels. The results revealed that high SNAI2 expression was examined in some cancers including ACC, CHOL, ESCA, GBM, HNSC, LAML, LGG, LIHC, LUSC, PAAD, SKCM, STAD, TGCT, and UCS. By contrast, low SNAI2 expression was observed in BRCA, CESC, KICH, KIRP, OV, PRAD, READ, THCA, and UCEC ( Figure 2C ). Furthermore, IHC results also revealed that SNAI2 expression was located in the nuclear, SNAI2 protein was expressed to some extent in both normal pancreatic tissue and pancreatic cancer, and the staining level was significantly increased in pancreatic cancer ( Figure 2D ).

Therefore, we further performed western blot and qRT-PCR assays in clinical PAAD samples to confirm the expression of SNAI2, and the result indicated that the SNAI2 protein and mRNA were upregulated in PAAD samples as expected compared with that in normal tissues ( Figures 2E, F ). Lastly, immunofluorescence (IF) images showed that the SNAI2 protein was mainly localized and distributed in the nucleus in HEK293T, Hela, HUVEC, and HCCLM3 cell lines ( Figure 2G ). Therefore, the above results showed that SNAI2 was differentially expressed in various cancers, which suggested that SNAI2 may play an important role in cancer progression.

In order to explore the mutation rate and mutation type of SNAI2 in different tumors, the genetic alteration of SNAI2 in various cancers from the TCGA cohorts was analyzed by the cBioPortal tool. We found that the highest alteration frequency of SNAI2 (>15%) appears for patients with UCS, in which amplification was the primary alteration type. It is worth noting that the only type of SNAI2 gene alteration in Lymphoid Neoplasm Diffuse Large B-cell Lymphoma (DLBCL) and MESO cases was deep deletion. In addition, we also found that amplification was the most common mutation type, followed by mutation, and finally deep mutation. ( Figure 3A ). Also, the mutation count of SNAI2 in different cancers was shown in Figure 3B . Further, we also presented the types, sites, and case numbers of the SNAI2 genetic alteration. The main type of SNAI2 genetic alteration was missense mutation followed by truncating mutation ( Supplemental Table S3 ). The X209_splice alteration was located in the zf-H2C2_2 domain of SNAI2, which can induce a splice mutation of the SNAI2 gene, which further leads to related functional abnormalities, mainly including the gain in BLCA, diploid in UCEC and KIRP, shallowdel in LUAD ( Figure 3C ). We also presented the 3D structure of the X209_splice mutation in the SNAI2 gene ( Figure 3D ).

The prognostic significance of SNAI2 in various cancers was evaluated ( Figure 4A ). According to the results of Kaplan-Meier OS analysis, SNAI2 is a risk factor for ACC, BLCA, GBM, HNSC, KICH, KIRP, LGG, LUAD, MESO, PAAD, SKCM, SARC, STAD, THYM, and a protective factor in LUSC, and UVM. The results shown in the forest plot ( Figure 4B ) demonstrated that the downregulating of SNAI2 expression was related to the time delay of OS in ACC, BLCA, GBM, KICH, KIRP, LGG, LUAD, MESO, PAAD, and STAD. Furthermore, we analyzed the Kaplan-Meier curve of ACC, BLCA, STAD, PAAD, LGG, KIRP, MESO, and LUAD, and the results suggested that lower expression of SNAI2 was related to a better survival outcome ( Figure 4C ), which indicates that SNAI2 was a prognostic biomarker in these cancers. Therefore, SNAI2 had a prognostic role in predicting the prognosis of cancers, but the roles were complicated and multifaceted across cancers. Further investigation should focus on the function of the SNAI2 protein in cancer cells.

To further explore the tumorigenic role of SNAI2, the differential expression genes (DEGs) between low- and high-SNAI2 subgroups in each cancer were used to perform GSEA to discern the SNAI2-associated cancer hallmarks ( Figure 5 ). It is found that SNAI2 expression was significantly related to immune-related pathways, such as UV response (down), TNFA-signaling-via-NFKB, KRAS signaling (up), IFN-α response, IFN-γ response, inflammatory response, IL-JAK-STAT3 signaling, IL2- STAT5 signaling, complement, coagulation, and allograft rejection, especially in BLCA, COAD, GBM, PAAD, PCPG, and STAD. It is noteworthy that the expression of SNAI2 is significantly negatively correlated with oxidative phosphorylation in the ACC, KIRP, LUAD, PAAD, and THYM. These data uncovered a potential association between SNAI2 expression and immune activation in the tumor microenvironment (TME).

In addition, the epithelial-mesenchymal transition (EMT) hallmark of many kinds of tumors and had a significant positive correlation with ACC, BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LAML, LGG, LUAD, LUSC, MESO, OV, PAAD, PCPG, PRAD, READ, SKCM, STAD, TGCT, THCA, THYM, UCEC, and UCS. In previous studies, EMT has been significantly confirmed to be related to the occurrence, metastasis, and drug resistance of cancers (32), suggesting that SNAI2 might play an indispensable part in the oncogenesis and development of cancers by enrolling in EMT. The above results provided an important direction for further understanding the role of SNAI2 in cancer establishment and development.

The single-cell analysis of SNAI2 in single-cell datasets of cancer samples was analyzed. The heatmap depicted ( Figure 6A ) represents the expression levels of SNAI2 of various cell types including immune cells, stromal cells, malignant cells, and functional cells. And the finding revealed that SNAI2 was mainly expressed in the fibroblasts, especially in BCC, BLCA, CRC, HNSC, PAAD, SKCM, and STAD cell datasets. In the GSE130001 BLCA dataset, the SNAI2 is highly expressed in fibroblasts and myofibroblasts cells in the BLCA microenvironment ( Figure 6B ). In the GSE103322 HNSC dataset, the SNAI2 is highly expressed in fibroblasts and malignant cells in the HNSC microenvironment ( Figure 6C ). In the CRA001160 PAAD dataset, SNAI2 was highly expressed in fibroblast cells and stellate cells in the PAAD microenvironment ( Figure 6D ).

The correlations between SNAI2 expression and immune cell infiltrations were analyzed to further elucidate the relationships between SNAI2 and cancer immunity. Spearman correlation analyses were conducted utilizing pan-cancer immune cell infiltration data from the TIMER2 database. As shown in Figure 7 , SNAI2 was positively associated with the infiltration levels of CAF, Endo, Neutrophil, Monocyte, Macrophage, Gran, and HSC, and negatively associated with the infiltration levels of B cell plasma in most TCGA cancers, especially in BRCA, HNSC, TGCT. In addition, the infiltration levels of NK T cells were also negatively associated with SNAI2 in DLBC, SARC, and UVM. We then integrated ImmuneScore, EstimateScore, StromalScore, and neoantigens to further evaluate the relationship between SNAI2 expression and immune infiltration across cancers. According to the results, SNAI2 expression was positively correlated with the ImmuneScore, EstimateScore, and StromalScore in most cancers ( Figures S1 – S3 ). Also, SNAI2 expression positively correlated with the neoantigens in GBM, while was negative in PRAD ( Figure S4 ). Together, our results indicated that SNAI2 might affect the development, prognosis, and therapy of cancers by associating with immune cells.

The associations between SNAI2 and immune regulators in pan-cancer were displayed ( Figure 8A ). We found that SNAI2 had a strong positive relationship with most immune regulators in ACC, COAD, KICH, READ, and THCA, and a negative relationship with most immune regulators in HNSC, LUSC, SKCM, and TGCT. Besides, there was a robust positive relationship between SNAI2 and NRP1, and CD276 in most of the TCGA cancers. Previous studies find the correlation between PDIA3 and NRP1, CD276 was also very significant in pan-cancer (33). To understand the role of SNAI2 in predicting the efficiency of immune checkpoint inhibitors (ICIs), the correlation between SNAI2 expression and TMB and MIS was further assessed. Positive correlations with TMB were identified in LAML, LGG, OV, SARC, and THYM, and negative correlations were discovered in BLCA, BRCA, ESCA, KIRP, LIHC, and PRAD ( Figure 8B ). Moreover, in the correlation between SNAI2 expression and MSI, positive associations were discovered in COAD, MESO, and TGCT, and negative correlations were discovered in ESCA, PRAD, and STAD ( Figure 8C ). Our results suggested that SNAI2 had the potential to predict the efficiency of ICIs in the corresponding cancers.

The researchers discovered a link between SNAI2 expression and immune cell infiltration, TMB, and MSI, suggesting a degree of correlation with immunotherapy efficacy. Therefore, we assessed the possible correlation between SNAI2 expression levels and immunotherapy response by analyzing previously reported clinical study cohorts. We discovered that patients with low SNAI2 expression had more clinical benefits and therapeutic responses to PD-L1 blockade therapy. In the GSE91061 melanoma cohort ( Figure 8D ), melanoma patients with SNAI2 low-expression had much better survival probability than patients with SNAI2 high-expression patients, and the response rate to anti-PD-1 was 0% in the SNAI2 high-expression subgroup, while 25% of patients responded to the anti-PD1 therapy in the SNAI2 low-expression subgroup ( Figure 8E ). Moreover, in melanoma tumors of the Gide2019 cohort, clinical advantages and therapeutic responses to PD-L1 and anti-CTLA4 blocking therapy were greater in patients with low SNAI2 expression ( Figure 8F ). And the response rate to anti-PD-L1 and anti-CTLA4 therapy was 76.2% in SNAI2 low-expression patients, which is significantly higher than 45.5% in SNAI2 high-expression patients ( Figure 8G ). Furthermore, similar results were found in melanoma patients treated with anti-PD-1 therapy. In urological tumors of the IMvigor210 cohort ( Figure 8H ), the response rate to anti-PD-L1 therapy was 20% in the SNAI2 high-expression patients, which is significantly lower than 45.5% in SNAI2 low-expression patients ( Figure 8I ). These findings supported the potential of SNAI2 for immunotherapy response prediction, indicating that it was a promising biomarker for cancer immunotherapy.

DNA mismatch repair (MMR) is a highly conserved biological mechanism that is critical for preserving genomic integrity (34). The deletion or mutation of key genes in MMR leads to abnormal DNA replication, which also promotes the occurrence and development of tumors (35). To assess the impact of SNAI2 in tumorigenesis, the correlation between SNAI2 expression levels and mutations in five MMR genes including MLH1, MSH2, MSH6, PMS2, and EPCAM was analyzed. According to the findings, the MMR genes were correlated with SNAI2, and this correlation was found in the majority of cancer types ( Figure S5A ). Especially in the UVM, SNAI2 expression was significant positive related to MLH1, MSH2, MSH6, and PMS2 genes. While, a negative correlation with the EPCAM gene was observed in BLCA, BRCA, CESC, COAD, ESCA, HNSC, KICH, LUAD, LUSC, OV, PAAD, PRAD, STAD, and UCEC.

DNA methylation affects a variety of genetic changes, including chromosomal structure, DNA stability, and DNA interaction with specific proteins. The abnormal state of DNA methylation is another important reason for promoting tumorigenesis, which has been heralded as a promising biomarker for diagnosis, treatment, and prognosis (36). The relationship between the expression of four methyltransferase genes (DNMT1, (DNMT2, DNMT3A, DNMT3B) and SNAI2 in pan-cancer was investigated further ( Figure S5B ). The result suggested that SNAI2 expression was closed correlated with at least one methyltransferases gene in SKCM, TGCT, THCA, THYM, UCEC, UVM, BLCA, BRCA, CESC, COAD, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LGG, MESO, PAAD, PRAD, READ, and SARC. Therefore, our study found that SNAI2 might play a potential role in affecting MMR regulation and DNA methylation in various cancers.

Next, we utilized the GeneMANIA online program to create a PPI network for SNAI2, and the SNAI2 expression-correlated genes for a series of pathway enrichment analyses to learn more about the molecular mechanism of the SNAI2 gene in carcinogenesis. SNAI2 demonstrated significant interactions with MTA3, HLF, BBC3, and CNOT9, as illustrated in Figure 9A . In addition, we tried to screen out the targeting SNAI2-binding proteins We combined the TCGA tumor expression data with the GEPIA2 program to find the top 100 genes that are linked with SNAI2 expression. The heatmap data revealed a positive correlation between SNAI2 and the fourteen genes in the majority of detailed cancer types ( Figures 9B, C ). SNAI2 expression levels were positively correlated with BMP1, DUSP7, EXT1, MMP14, TUBB6, MARVELD1, MCC, MICALL1, FSCN1, FRMD6, SERPINH1, PLXNA1, MYH9, and FAM126A.

To reveal the role of SNAI2 in pancreatic cancer cells cell proliferation and invasion, we designed the following assays. qRT-PCR and Western blot analysis suggested that SNAI2 was overexpressed in the four pancreatic cancer cell lines (PANC-1, SW1990, BxPC-1, and SW979) when compared with its expression in the control ( Figure S6A ). Firstly, we used the western blot assay to identify the knockdown status of SNAI2 in the PANC-1 cells. The result showed that SNAI2 shRNA could significantly knock down the protein level of SNAI2 ( Figure 10A ). To uncover the potential ability of SNAI2 in regulating PANC-1 cell proliferation and invasion, we performed the proliferation and transwell assays to reflect the proliferative and invasive ability of PANC-1 cells. CCK-8 ( Figure 10B ), colony formation ( Figure 10C ), and Edu ( Figure 10D ) analysis indicated that the proliferative ability of PANC-1 cells significantly decreased after SNAI2 was knocked down. Besides, transwell assays ( Figure 10E ) also proved that the knockdown of SNAI2 significantly weakened the invasive ability of PANC-1 cells. Conversely, overexpression of SNAI2 in SW1990 cells significantly promoted cell proliferation, migration, and invasion ( Figures S6B – F ). Moreover, EMT has been identified as an important process in tumor invasion and metastasis. We then used Western blot assay to look at the EMT markers (E-cadherin, N-cadherin, and vimentin) to see if SNAI2 could affect EMT in pancreatic cancer cells. The results showed that downregulating SNAI2 reduced the expression of N-cadherin and Vimentin while increasing E-cadherin expression in PANC-1 cells ( Figure 10F ), whereas overexpression of SNAI2 increased the expression of N-cadherin and Vimentin, and decreased E-cadherin expression in SW1990 cells ( Figure S6G ). Taken together, these data indicated that SNAI2 played a crucial role in pancreatic cancer cell proliferation and invasion.