The “Gene_DE” module in the Tumor Immune Estimation Resource (TIMER; http://timer.cistrome.org/) database (49) was accessed to compare the expression of OAS1 in different tumors. Gene transcriptome data in TPM format of normal samples from the Genotype-Tissue Expression (GTEx) project and pan-cancer samples from The Cancer Genome Atlas (TCGA) database were downloaded based on the UCSC Xena browser (https://xenabrowser.net/datapages/).The differences of OAS1 expression between tumor tissues and normal tissues were analyzed by wilcoxon rank sum test in R software, and differences in expression levels were visualized using ggplot2 (Version 3.4.2).The Clinical Proteomic Tumor Analysis Consortium (UPTAC, http://ualcan.path.uab.edu/analysis-prot.html) database, obtained from UALCAN (50) was utilized to compare differences in OAS1 protein expression among multiple tumor types at the proteomic level.The Gene Expression Profiling Interactive Analysis (GEPIA; http://gepia2.cancer-pku.cn/) database (46) was used to compare OAS1 expression in different tumor stages. Human Protein Atlas (HPA; www.proteinatlas.org) was a comprehensive human proteomics database that provides information on the expression and distribution of human proteins in different tissues, organs, and cell types. Tissue microarrays from HPA database showed differential expression of OAS1 in tumor tissue versus normal tissue by immunohistochemistry. Bubble maps of OAS1 related diseases were obtained from the Open Target Platform (https://platform.opentargets.org/) to identify potential disease associations with OAS1 expression.

GEPIA is an online gene expression data analysis tool that can be used for survival analysis in different tumors. The “Survival Analysis” module was used to analyze the relationship between OAS1 expression and patient survival. median OAS1 expression values were selected as cut-off values and used to classify patients. the GEPIA database was used to obtain Kaplan–Meier plots for patient overall survival (OS) and disease-free survival (DFS). Hazard ratio (HR) was defined as the risk in the high OAS1 expression group divided by the risk in the low expression group, which was used to assess the effect of OAS1 expression on patient survival.

To conduct gene mutation analysis, we downloaded the Simple Nucleotide Variation dataset of level 4 processed by MuTect2 software from the Genomic Data Commons (GDC; https://portal.gdc.cancer.gov/). The mutation data was integrated, and structural domain information for the proteins was obtained using the R package maftools (version 2.2.10). The Tumor Mutation Burden (TMB) was calculated using the “tmb”function of the R package maftools for each tumor sample. The TMB and gene expression data were then integrated, and a log2 (x+0.001) transformation was performed for each expression value to calculate the correlation between OAS1 expression and TMB scores. Microsatellite instability (MSI) scores for each tumor were obtained from a previous study (47). We integrated the MSI and gene expression data of the samples to calculate the correlation between OAS1 expression and MSI scores. Finally, neoantigen data for each tumor were obtained from a previous study (51). The neoantigen and gene expression data were integrated to calculate the correlation between OAS1 expression and immune neoantigen scores.

To identify clinically relevant alternative splicing (AS) events, we used the ClinicalAS module of the OncoSplicing website (http://www.oncosplicing.com/) (52). We searched for AS events of OAS1 that were included in both the SplAdder and SpliceSeq projects. We selected OAS1_alt_3prime_56568 for the further study. The PanDiff plots were presented to compare percent spliced-in(PSI)differences in AS events between cancers and GTEx normal tissues. For further analysis, we plotted Kaplan-Meier curves to investigate the prognostic significance of the OAS1_alt_3prime_56568 in pan-cancer.

To assess stromal, immune, and overall tumor purity (ESTIMATE) scores for each patient in each tumor, we utilized the R package ESTIMATE (version 1.0.13) (53). Additionally, we employed the deconvo_tme function in the R package IOBR (version 0.99.9) (54) with the xCell (55) algorithm to re-evaluate major immune cell infiltration scores based on gene expression. To further evaluate the infiltration score of M2 subtype macrophages, we used both CIBERSORT (56) and QUANTISEQ (57) algorithms. Finally, we investigated the correlation between OAS1 expression and immune cell infiltration scores using Pearson correlation analysis to calculate the correlation coefficient between these two factors. We utilized the Tumor Immune Dysfunction and Exclusion (TIDE, http://tide.dfci.harvard.edu) database (58) to assess the prognostic significance of OAS1 expression and cytotoxic T lymphocyte (CTL) infiltration in patients. The T cell dysfunction score for each gene is determined by calculating the Z-score of the Wald test, which is obtained by dividing the coefficient by its standard error. The utilization of Z-Score standardization allows for the transformation of data with diverse metrics into a standardized metric, thus enhancing the comparability of the data. Additionally, we compared OAS1 with other indicators for predicting the efficacy of immunotherapy using the TIDE database. The Tracking Tumor Immunophenotype database (TIP, http://biocc.hrbmu.edu.cn/TIP/) (59) was used to evaluate the anticancer immune status in seven different stages of the tumor immune cycle, and the Gene Set Cancer Analysis database (GSCA, http://bioinfo.life.hust.edu.cn/GSCA/) was used to analyze the relationship between OAS1 and exhausted T cell scores (60).

The pan-cancer samples were segregated into two groups using the median expression of OAS1 across all samples as a threshold. We then compared the differences in gene expression between the two groups and ranked them based on the fold change values. We applied the Gene Set Enrichment Analysis (GSEA) method to study all samples by analyzing gene sets h.all.v7.4.symbols.gmt which were downloaded from the Molecular Signatures Database (http://www.gsea-msigdb.org/gsea/downloads.jsp) (61) using the R package clusterProfiler (version 4.0.5) (62). The primary statistic for examining gene set enrichment results was the normalized enrichment score (NES). Moreover, the false discovery rate (FDR) was estimated as the probability that a gene set with a given NES represents a false positive finding. The indicators of enrichment were NES and FDR, and we considered gene sets with |NES|>1 and FDR < 0.25 to be significantly enriched. Finally, we visualized the outcomes using ggplot2 (version 3.4.2) package. Based on the expression levels of OAS1, we divided the TCGA_PAAD cohort into high OAS1 expression group and low OAS1 expression group. The R package DESeq2(version 1.38.3) was used to analyze the differentially expressed genes (with criteria of |log2FC| > 1.5 and adj.pvalue < 0.05). The differentially expressed genes were then input into the STRING database (https://cn.string-db.org/) with a minimum required interaction score module set at 0.7 to obtain the PPI network graph. The R package igraph(version 1.5.0) was employed for constructing the network modules, while the clusterProfiler package was utilized for conducting enrichment analysis.

Tumor Immune Single-cell Hub (TISH, http://tisch.comp-genomics.org/home/) database (63) was used to download single-cell transcriptome data ( Table S4 ). We utilized the R package Seurat (version 4.3.0) for data processing and analysis. The “Read10X_h5” function was used to read the data, while “CreateSeuratObject” was used to build the SeuratObject. Data normalization was performed using LogNormalize with a scale factor of 10000. Highly variable genes were selected using the “FindVariableFeatures” function with the parameter ‘nfeatures’ set as 2000.Principal component analysis (PCA) was performed on the dataset using the “RunPCA” function. The top 10 principal components were used to construct a k nearest neighbor graph using the “FindNeighbors” function. Finally, we reduced the dimensions of the data using the uniform manifold approximation and projection (UMAP) algorithm via the “RunUMAP” function for visualization purposes.To visualize the gene expression patterns within the dataset, we used the “FeaturePlot” function to display the co-expression relationships between two different genes. Additionally, we utilized the “getScatterplot” function to calculate and visualize the correlation between the expression of two different genes at the single-cell level.

Nine pancreatic cancer patients’ samples were selected in the pathology department of the First Affiliated Hospital of Xi ‘an Jiaotong University. All experiments on tumor samples from patients performed were under the supervision of the Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University (XITU1AF2022LSK-339). They were all confirmed as pancreatic ductal carcinoma through tissue biopsy. Samples after surgical resection were fixed in formalin, embedded in paraffin and stored. Formalin-fixed, paraffin-embedded sections were cut into 4-μm sections. The sections were deparaffinized and rehydrated using xylene and graded alcohol solutions. They were blocked with 2% bovine serum albumin, incubated with specific primary antibodies for 12 hours at 4°C followed by incubation with a biotinylated secondary antibody. Subsequently, diaminobenzidine (DAB) was added dropwise over a 3 to 5 minutes period and sections were then counterstained with hematoxylin. The primary antibodies included rabbit monoclonal antibody of recombinant Anti-OAS1 (dilution 1:500, cat. No: ab232862, Abcam).

Immunohistochemical staining results were scored according to the following criteria. According to the cell staining intensity, the score is grade 4, zero for no positive staining (negative), one for light yellow (weak positive), two for tan (positive) and three for tan (strong positive). According to the percentage of positive cells, it is rated as grade 4, with ≤25% as 1 point, 25%-50% as 2 points, 51%-75% as 3 points, and > 75% as 4 points. The final score can be obtained by multiplying the two scores. 3 views of cancer tissues and 3 of para cancerous tissues were randomly selected for each sample by pathologist. And then all of views were scored by three independent researchers or pathologists in a double-blind condition.

Panc-1 and Bxpc-3 cells were maintained in Dulbecco’s modified Eagle medium (DMEM) containing 10% FBS in an incubation at 37°C, with 5% CO₂. Petri dish size is 10 cm. The passage ratios of Panc-1 and Bxpc-3 were 1: 3 and 1: 4, respectively.

The cells were reseeded at a density of 5×10⁴ cells per well in a 6-well plate, 24 hours prior to transfection. Subsequently, the cells were incubated in a 5% CO2 incubator at 37°C.The cells were transfected when the cell density reached 70%–80%. The transfection complex was prepared: 4μL of each siRNA (10 μM) was added to 100μL of serum-free medium, and the mixture was allowed to stand for 5 minutes at room temperature. 12μL of Lipofectamine™ RNAiMAX (Invitrogen, the US) transfection reagent was added. Allow to stand at room temperature for 15 minutes. The transfection complex was uniformly added into the 6-well plate, and returned to the incubator for continuous culture for 6 h. The medium containing the transfection complex was discarded and washed once with 2 mL PBS. Then 2mL cell culture medium containing 10% serum was added and the culture was continued for 18h in a 5%CO₂ incubator at 37°C.

The cells were collected 24h after transfection. Total RNA was extracted using RNA extraction kit following rigorously the manufacture’s instruction. Then prepare a reaction mixture before reverse transcription: gDNA remover 1.0ul, 10×gDNA remover Buffer 1.0ul, Total RNA 1000ng, RNase Free dH₂O up to 10.0. Incubate the mixture for 2 min at 42°C, 5 min at 60°C and cool rapidly on ice. Add the following reagents 10.0 ul to give a final total of 20 μl: 5×Goldenstar™ Buffer 4.0 ul, dNTP Mix 1.0 ul, Goldenstar™ Oligo (dT)17 1.0 ul, Randomer 1.0 ul, DTT 1.0 ul, Goldenstar™ RT6 1.0 ul, RNase Free ddH₂O up to 20.0 ul. The reverse transcription reaction conditions were as follows: 25°C for 10 minutes, 55°C for 50 minutes, 85°C for 5 minutes and 4°C for +∞. Reverse transcription polymerase chain reaction (RT-PCR) primer sequences in Supplement Table 1 . RT-PCR reaction system was as follows (formulated on ice):2 ×T5 Fast qPCR Mix 10.0 ul, 10 μM Primer F 0.8 ul, 10 μM Primer R 0.8 ul, 50 ×ROX Reference Dye II 0.4 ul, Template DNA 0.5 ul, ddH₂O up to 20.0 ul. The two-step quantitative polymerase chain reaction procedure included Holding stage and Cycling Stage. the former keeps 95°C for 30 s. The latter include 95°C for 5 s, 55°C for 30 s and 72°C for 30 s, 40 cycles total.All the above reagents are from Tsingke Biotechnology Co.,Ltd.

Bxpc-3 and Panc-1 cells were seeded at a density of 3000 cells per well in 96-well plates and cultured in an incubator at 37°C for about 4 hours to attachment, followed by treatment with siRNA for 24 hours. After 24h, the 96-well plate was taken out and the medium were replaced by DMEM 10% FBS containing 10% CCK8 (Bimake). After 2 hours of incubation, the absorbance at 450nm was determined using a micro-UV-visible spectrophotometer (NanoDrop One/One C Thermo, USA). Cell viability was calculated strictly following the instructions.

Cell migration chambers consisting of 24 well plates were used for invasion assays (Corning, USA). Transwell plates were coated with Matrigel^® (diluted with serum-free DMEM 1:5; Thermo Fisher Scientific) and 500 μL of DMEM containing 20% FBS was added to the lower well. Cells were transfected by siRNA and incubated for 24 hours. Then a total of 5 × 10⁴ knocked down or not cells were added to the Matrigel^®-coated chamber without FBS. After 48 hours of culture, the cells transferred to the other side were fixed with 4% paraformaldehyde for 20 minutes, washed 3 times with PBS, and stained with 0.1% crystal violet for 20 minutes. The migrated cells were photographed with a microscope and camera and then counted in three different areas.

The cells were cultured at a density of 5×10⁴ cells per well in 6-well plates. After their adhesion, cells were transfected by siRNA and incubated for 24 hours. Cells were collected using trypsin (Gibco) and centrifuged at 1000g for 5 minutes. Apoptosis was assessed using apoptosis assay (Biolegend). Briefly, after the cells were re-suspended with 100 μl binding solution, 5 μl AnnexinV-APC and 10 μl PI were added and mixed. After an incubation protected from light at room temperature (20–25 °C) in the dark for 15 minutes, 400 μl of binding solution was added to each sample. Flow cytometry (CantoII, BD, the US) tests were performed immediately and data were treated and plotted by FlowJo (version 10).

All bioinformatics analysis were conducted through R software (version 4.2.2). The above visualization was performed with R software and GraphPad Prism 9. The comparison of difference between two groups was analyzed using Wilcoxon rank-sum test. The comparison of difference between three groups or more groups was analyzed using the Kruskal–Wallis test. The paired t-test was used to study the differences in OAS1 expression between pancreatic cancer tissues and adjacent normal tissues in the GSE28735 dataset. The log-rank test was used to analyze patient survival significance. Pearson’s or Spearman’s correlation coefficients were used to quantify the correlations. p-Values less than 0.05 (p < 0.05) were considered significant.

We initially investigated the expression of OAS1 in the CCLE database and observed high expression levels in organs of the digestive system such as the pancreas and intestine cancer cells ( Figure S1A ). In addition, exploring diseases associated with OAS1 using OpenTargets revealed that it is primarily linked to COVID-19 and chronic lymphocytic leukemia ( Figure S1B ). To better understand the role of OAS1 in cancer, we analyzed its expression changes in The Cancer Genome Atlas (TCGA) database, which contains descriptions of 33 types of cancer ( Table S3 ). Through analysis of the GTEx database, we found that OAS1 exhibits higher expression levels in bladder tissues ( Figure 2A ). Furthermore, our analysis revealed significant overexpression of OAS1 mRNA in several tumor tissues, including BLCA, BRCA, CESC, CHOL, COAD, ESCA, GBM, HNSC, KIRC, KIRP, LUAD, PCPG, THCA, and UCEC ( Figure 2B ). Moreover, we examined OAS1 protein expression through analysis of the CPTAC database and discovered significant increases in Clear cell RCC, HNSC, UCEC, and PAAD ( Figure 2C ). Additionally, we evaluated the correlation between OAS1 expression and tumor pathological stage and found significant associations in BLCA, PAAD, LUAD, and SKCM ( Figure 2D ). It is worth noting that in the GSE28735 dataset, we analyzed the expression levels of OAS1 in pancreatic cancer tissues and adjacent normal tissues from 45 patients. The results indicated a significantly higher expression of OAS1 in pancreatic cancer tissues compared to adjacent normal tissues ( Figure 2E ). To confirm these results, we performed immunohistochemistry using a XJTU PAAD cohort. Results demonstrated expression levels of OAS1 protein greatly increased in PAAD ( Figures 2F, G ). Finally, we examined immunohistochemical data from the HPA database and observed significantly elevated expression levels of OAS1 in LUAD, STAD, PAAD, KIRC, PRAD, and OV ( Figure S1C ). Overall, our results suggest that OAS1 may play an oncogenic role in various tumors, and its clinical application value is worth exploring at the pan-cancer level.

To evaluate the potential prognostic significance of OAS1 in various human tumors, we analyzed its correlation with overall survival (OS) using data from the TCGA database. Forest plot revealed that OAS1 expression was associated with the prognosis of several tumor types ( Figure 3A ). Specifically, we observed that high OAS1 expression levels were linked to shorter OS compared to low OAS1 expression levels in LUAD (HR=1.32, p=0.008), LAML (HR=1.43, p=0.05), PAAD (HR=1.66, p=0.047), LGG (HR=1.76, p<0.001), ACC (HR=1.79, p=0.007) ( Figures 3B-F ), and in MESO (HR=0.91,p = 0.018), SKCM (HR=0.93, p < 0.001), BLCA (HR=0.88,p = 0.009) ( Figures 3G-I ), the opposite results were observed. Furthermore, we found that higher expression of OAS1 was linked to shorter disease-free survival (DFS) among patients with LUAD (HR=1.4, p =0.019), LGG (HR=1.6, p =0.005), PRAD (HR=1.6, p =0.028), UVM (HR=3.4, p =0.009) ( Figures S2B-E ). Our findings highlight the crucial role of OAS1 as a novel biomarker for predicting the prognosis of patients with various tumors.

Our study revealed that OAS1 is highly expressed in tumor tissues compared to normal tissues and is associated with the prognosis of patients with various types of tumors, including pancreatic cancer. We aimed to further explore the role played by OAS1 in pancreatic cancer development. Initially, we analyzed differential genes between patients in the OAS1 high expression group and those in the low expression group and conducted Protein-protein interaction analysis. Our PPI network plots showed that differential gene interactions were associated with epithelial differentiation, which has been linked to malignant tumor progression ( Figure 4A ). Subsequently, To investigate the implication of OAS1 in tumor cell progression, we performed the knock-down test using siRNAs in two pancreatic cancer cell lines. Three siRNA were designed and their knock-down efficiency verified using RT-PCR. Results indicated that siRNA3 represented the best knockdown efficiency (up to 70%) ( Figure S2A ), making it the ideal choice for further research. To evaluate the impact of OAS1 on the invasion ability of pancreas cancer, trans-well assays were performed using cells Bxpc-3 and Panc-1after OAS1 knock-down. Our results demonstrated that in both Bxpc-3 (siRNC: 402 ± 3.05, siRNA3: 145± 11.67) and Panc-1 (siRNC: 450 ± 6.51, siRNA3: 140± 1.00) cell lines, the invasion ability of pancreatic cancer cells was significantly reduced after OAS1 knockdown ( Figure 4C , Figure S2B ). Additionally, to evaluate its involvement in apoptosis, apoptosis assay by flow cytometry was performed, and showed that knockdown of OAS1 increased apoptosis in both Bxpc-3 (siRNC: 11.93 ± 6.77, siRNA3: 27.29 ± 5.20) and Panc-1 (siRNC: 12.72± 1.01, siRNA3: 21.97± 3.34) cell lines ( Figure 4D ). However, cell viability tests after knocking down OAS1 showed that it had a moderate effect (less than 50%) on the viability of Bxpc-3 (siRNC: 1± 0.07, siRNA3: 0.82± 0.06) and did not affect the viability of Panc-1 cells (siRNC: 1± 0.03, siRNA3: 0.95± 0.03) ( Figure 4B ).

Furthermore, AUC plots showed that the expression of OAS1 in patients with pancreatic cancer can predict overall survival at 1 year (AUC=0.646), 2 years (AUC=0.723) or 3 years (AUC=0.734) ( Figure 4E ). Based on the expression of OAS1 and pathological staging, we developed a survival prediction model for pancreatic cancer patients using the Cox model and visualized it using Nomogram plots ( Figures 4F, G ). In conclusion, our findings suggest that OAS1 could be a potential biomarker for pancreatic cancer prognosis and a target for therapeutic intervention.

To evaluate OAS1 genetic alteration in cancers, we conducted a comprehensive analysis of the OAS1 gene mutation status in various tumor types. Our findings indicate that different tumors exhibit distinct genetic alterations of the OAS1 gene, including frameshift insertion, missense mutation, nonsense mutation, and frameshift deletion ( Figure 5A ). Furthermore, we investigated the impact of copy number variations (CNVs) in the OAS1 gene on the prognosis of cancer patients. We stratified cancer patients based on their OAS1 CNV levels and found that high OAS1 CNV levels were associated with shorter survival rates in COAD (HR=2.24, p = 0.001) and LUAD (HR=1.92, p=0.037) ( Figure 5B ). To assess the relationship between OAS1 expression and other genomic alterations commonly observed in cancer that impact patient prognosis and treatment response, we compared TMB, MSI, and neoantigens at the pan-cancer level. Radar plots revealed that OAS1 expression was positively correlated with TMB in four cancers (KIRC, LGG, PAAD, STAD) and with MSI only in KIRC. By contrast, it was negatively correlated with TMB in two cancers (THCA, TGCT) and with MSI in three cancers (TGCT, SKCM, OV) ( Figures 5C, D ). We also examined the effect of OAS1 expression on the formation of neoantigens in different tumors. Results showed that the expression level of OAS1 was positively correlated with the score of neoantigen in KIRP, SKCM, LGG, OV, PRAD, and negatively correlated with READ ( Figure 5E ). Finally, we explored the impact of OAS1 gene expression on the Mismatch Repair System (MMRs) in tumors. MMRs play a critical role in maintaining DNA replication accuracy and integrity. Our analysis revealed a significant association between OAS1 expression in various tumors and the expression of genes related to MMR systems, such as MLH1, MSH2, MSH6, PMS2, and EPCAM ( Figure 5F ).

To investigate the implication of OAS1 in epigenetics, we preformed a gene methylation association evaluation. Our study revealed that the OAS1 gene is associated with DNA methyltransferases in multiple types of tumors ( Figure 6A ). Furthermore, we analyzed the impact of OAS1 promoter methylation on prognosis and found that higher levels of OAS1 promoter methylation were associated with better prognosis in tumor types such as BCRA (p = 0.016), LGG (p < 0.001), LIHC (p = 0.017), and LUSC (p = 0.097) ( Figure 6B ). We also investigated the alternative splicing events of OAS1 gene and discovered a significant correlation between tumorigenesis and OAS1_alt_3prime_56568 ( Figure 6C ) in various cancers such as LUAD, STAD, UCS, BRCA, CESC, and SKCM ( Figure 6D ). Finally, we found that this specific alternative splicing event was associated with prognosis in patients with multiple tumors: high PSI values were linked to poorer prognosis in LUSC (p = 0.001), LIHC (p < 0.001), GBM (p = 0.019), CESC (p = 0.022), SARC (p = 0.026), and OV (p = 0.049) ( Figure 6E ). In summary, our findings suggest that OAS1 promoter methylation and alternative splicing events may have important implications for diagnosis, prognosis, and treatment of different types of tumors.

The infiltration of cytotoxic T lymphocytes is significantly associated with response to immunotherapy and tumor prognosis. As previously identified, greater infiltration of these cells when the OAS1 gene is expressed at low levels indicates a better prognosis. However, this situation changes dramatically when OAS1 expression is higher. In several cancers, such as LIHC (p = 0.020), COAD (p = 0.022), UCEC (p < 0.001), BLCA (p = 0.035), BRCA (p = 0.024), and LUAD (p = 0.034) ( Figure 7A ), high expression of OAS1 diminishes or even reverses the benefits of CTL infiltration levels on patient survival. Further investigation revealed that OAS1 can upregulate the expression of immune checkpoint markers associated with T cell dysfunction, including LAG3, CTLA4, PDCD1 (PD-1), IDO1, and CD274 (PD-L1) ( Figure 7B ). At the single-cell level in PAAD, we observed that OAS1 gene expression coincides with EPCAM, a label of malignant cells (R = 0.35, p < 0.001) ( Figure 7C ). Pan-cancer gene set enrichment analysis showed that high expression of OAS1 is linked to the activation of multiple signaling pathways, such as the interferon α (NES = -2.4, FDR <0.001) and γ signaling pathways (NES = -2.4, FDR <0.001), as well as the IL6-JAK-STAT3 signaling pathway (NES =-1.98, FDR = 0.01) ( Figure 7D ). Furthermore, almost all cancers demonstrated a positive correlation between OAS1 expression and GSVA scores of the interferon-γ signaling pathway ( Figure 7E ) and IL6-JAK-STAT3 signaling pathway ( Figure 7F ). We also found a positive correlation between the expression of OAS1 and STAT1 (R = 0.51, p < 0.001) ( Figure 7G ). Finally, we discovered that infection by certain pathogens significantly increases OAS1 expression, especially mycobacterium tuberculosis (log FC = 2.6, p = 0.01) and Zika virus (log FC = 5.8, p < 0.001) ( Figure 7H ), which were prone to immune escape.

The correlation between OAS1 expression and tumor immune microenvironment was finally investigated across various tumor types. We found that OAS1 expression was positively correlated with tumor immune score, stromal score, and ESTIMATE score in most tumors ( Figures S4A-C ), indicating a significant association with tumor immune infiltration. We further analyzed the relationship between OAS1 expression and specific immune cell types using the xCell algorithm. Results showed a significant positive correlation between OAS1 expression and macrophage infiltration ( Figure 8A ). We also used the QUANTISEQ and CIBERSORT algorithms to analyze the correlation between OAS1 expression and M2 subtype macrophage infiltration. Our findings revealed a positive correlation between OAS1 expression and M2 macrophage infiltration in several tumor types using both algorithms ( Figures 8B, C ). At the single-cell level, we investigated the correlation between OAS1 expression and M2 macrophage biomarkers CD68 and CD163. Our results showed a positive correlation between OAS1 expression and CD163 in PAAD (R=0.66, p < 0.001) ( Figure 8D ), LIHC (R=0.54, p < 0.001) ( Figure 8E ), HNSC (R=0.41, p < 0.001) ( Figure 7F ), and BRCA (R=0.61, p < 0.001) ( Figure 8G ). As expected, OAS1 was also co-expressed with CD68 in these tumors ( Figure S4D ). Finally, we compared the predictive accuracy of OAS1 with that of standard immunotherapy prediction models. We found that OAS1 was more accurate in predicting the efficacy of the Gide2019_PD1_Melanoma cohort (AUC=0.75) compared to the standard immunotherapy prediction model ( Figure S4E ). Our findings suggest that OAS1 could serve as a promising biomarker for predicting tumor immune infiltration and response to immunotherapy, especially M2 macrophage polarization.

We conducted a comprehensive investigation into the correlation between OAS1 expression and the immune microenvironment in pancreatic cancer. Firstly, we employed seven algorithms (TIMER, QUANTISEQ, CIBERSORT_ABS, MCPCOUNTER, CIBERSORT, XCELL, EPIC) to evaluate the infiltration of distinct immune cells in various samples obtained from the TCGA_PAAD cohort. Subsequently, based on the median value of OAS1 expression, we categorized the samples into two groups: the “OAS1 high expression group” and the “OAS1 low expression group.” We performed an analysis to compare the infiltration of immune cells between these groups ( Figure 9A ). Interestingly, using algorithms such as TIMER (p=0.029), QUANTISEQ (p=0.017), CIBERSORT_ABS (p=0.016), and CIBERSORT (p=0.006), we observed that the group with higher OAS1 expression tended to have lower CD8⁺ T cell scores compared to the group with lower OAS1 expression ( Figures S5A-G ). Furthermore, we delved into the association between OAS1 and exhausted T cells by analyzing the GSCA database. Our investigation revealed a positive correlation between OAS1 expression and T cell exhaustion scores across various tumor types, including BRCA, CESC, HNSC, KIRC, LGG, LUAD, LUSC, OV, SARC, TGCT, THCA, THYM, UVM, etc. ( Figure 9B ). In addition, we explored the relationship between OAS1 and immune checkpoints by analyzing multiple pancreatic cancer bulk transcriptome datasets (GSE71729, GSE21501, ICGC_array, E_MTAB_6134, ICGC_CA_seq, GSE79668, GSE62452, GSE78229, GSE28735, TCGA_PAAD, GSE57495). Notably, our findings revealed a positive correlation between OAS1 expression and LGALS9, IDO1, and CD274 in several datasets ( Figure 9C ). Moreover, we extended our investigation to assess the association between OAS1 expression and these three immune checkpoints at the single-cell level, utilizing the PAAD single-cell dataset CRA001160. Intriguingly, at the single-cell level, we observed a positive correlation between OAS1 expression and LGALS9 (R=0.54, p<0.001) ( Figure 9D ), IDO1 (R=0.44, p<0.001) ( Figure 9E ), as well as CD274 (R=0.57, p<0.001) ( Figure 9F ). These compelling results suggest that OAS1 expression is positively linked with the aforementioned immune checkpoints, both at the single-cell and bulk transcriptome levels.Finally, employing the TIP database, we assessed the anticancer immune status across seven distinct stages of the tumor immune cycle, including the release of cancer cell antigens (Step 1), cancer antigen presentation (Step 2), priming and activation (Step 3), trafficking of immune cells to tumors (Step 4), infiltration of immune cells into tumors (Step 5), recognition of cancer cells by T cells (Step 6), and killing of cancer cells (Step 7). Our analysis revealed that the lower OAS1 expression group exhibited a notable increase in the infiltration of CD8+ T cells and displayed enhanced cytotoxic effects on cancer cells, compared to the group with higher OAS1 expression ( Figure 9G ). These findings suggest that reduced OAS1 expression is associated with improved immune cell infiltration and heightened cytotoxicity against cancer cells within the various stages of the tumor immune cycle.