Differential expression of IRFs in PC patients
We firstly detected the level of IRFs in PC in Oncomine database. The results were shown in Fig. 1 and Table S1. We found that the level of IRF3, IRF6, IRF7, IRF8 and IRF9 were upregulated in tumor tissues in PC (Fig. 1, P < 0.05). In addition, we also noticed that no difference was found between tumor tissues and normal tissues about the level of IRF1/3/4/5/6 in PC (Fig. 1). To be more specific, Malte’s dataset revealed that IRF2 expression was increased in Pancreatic Ductal Adenocarcinoma with a fold change(FC) of 2.051 [11]. According to the data of Huadong’s study, IRF6 was upregulated in Pancreatic Carcinoma tissues and the FC is 2.43 [12]. A total of two datasets demonstrated the upregulation of IRF7 in PC [11, 13]. Moreover, three datasets suggested that IRF8 expression was increased in PC [14–16]. We also found that the level of IRF9 was elevated in PC with the FC of 2.205 and 2,095 [12, 16]. This is followed by the verification of the expression of IRFs in PC using the TCGA dataset. We found that the mRNA level of IRF1, IRF2, IRF3, IRF5, IRF6, IRF7, IRF8 and IRF9 (Fig. 2A-I) were upregulated in PC (All p < 0.05). Therefore, we suggested that the level of IRF3, IRF6, IRF7, IRF8 and IRF9 were upregulated in tumor tissues of PC.
The association between the level of IRFs and patient’s pathology stage in PC were also detected. Interestingly, a significant association was obtained between IRF7 expression and patient’s pathology stage in PC (Fig. 3G, p < 0.00908). However, there was no association between IRF1/2/3/4/5/6/8/9 expression and patient’s pathology stage in PC (Fig. 3, p > 0.05).
Prognostic value of IRFs in PC patients
The prognostic value of IRFs in PC was explored using TCGA dataset. The data showed that PC patients with high IRF2 (HR = 1.8, p = 0.0069) and low IRF3 expression (HR = 1.6, p = 0.031) were associated with poor overall survival (Fig. 4A). Particularly, PC patients with high IRF6 expression had both poor overall survival (HR = 1.6, p = 0.03) (Fig. 4A) and poor disease-free survival (HR = 1.6, p = 0.028) (Fig. 4B).
Co-expression, genetic alteration, and drug sensitivity analyses of IRFs in PC patients
Comprehensive analyses were performed to explore the molecular character of IRFs in PC using cBioportal. There was a low to moderate correlation among the mRNA level of each IRFs member in patients with PC (Fig. 5A). Moreover, the genetic alterations analysis revealed that IRF1, IRF2, IRF3, IRF4, IRF5, IRF6, IRF7, IRF8 and IRF9 were altered in 6%, 8%, 8%, 2.7%, 6%, 6%, 4%, 4%, and 4% of the queried PC samples, respectively (Fig. 5B). High mRNA expression, amplification and deep deletion were the three most common type of genetic alterations in these samples (Fig. 5B). To clarify whether these genetic alterations could affect the prognosis of PC patients. Kaplan-Meier method was drawn and revealed that genetic alterations of IRFs could not affect the overall survival and disease-free survival of PC patients (Fig. 5C, p > 0.05). Drug sensitivity analysis was also performed. And the results suggested that low expression of IRF2/4/5/8 were resistant to most of the drugs or small molecules from GDSC (Fig S1).
Immune cell infiltration analysis of IRFs in PC patients
Tumor-infiltrating lymphocytes could serve as a biomarker for predicting sentinel lymph node status and cancer patients’ survival [17, 18]. The previous study has revealed close correlation between immune infiltration analysis and IRFs in cancers[19]. In our study, a comprehensive detection of the correlation between IRFs and immune cell infiltration in PC was conducted using TIMER. As shown in Fig. 6, the level of IRF7 was positively associated with the infiltration abundance of B cells (Cor = 0.436, P = 2.40e-09), CD8 + T cells (Cor = 0.401, P = 5.32e-08) macrophages (Cor = 0.227, P = 2.84e-3), Neutrophils (Cor = 0.471, P = 8.03e-11) and Dendritic cells (Cor = 0.566, P = 6.71e-16) (Fig. 6A). Interestingly, the expression of IRF2 and IRF6 also showed a positive association with the infiltration abundance of these five immune cells in PC (Fig. 6B and 6F, all p < 0.05). As for IRF3, a positive correlation was obtained between IRF3 expression and the infiltration abundance of B cells, CD8 + T cells and CD4 + T cells (Fig. 6C). Moreover, the expression of IRF4 (Fig. 6D), IRF5(Fig. 6E), IRF8(Fig. 6H) and IRF9(Fig. 6I) was positively associated with all these six immune cells, including B cells, CD8 + T cells, CD4 + T cells, macrophages, Neutrophils and Dendritic cells (all p < 0.05).We also found that IRF7 expression was associated with the infiltration abundance of CD8 + T cells (Cor=-0.209, P = 6.07e-083), CD4 + T cells (Cor = 0.389, P = 1.77e-7), Neutrophils (Cor = 0.252, P = 8.72e-4) (Fig. 6G). We also explored the effect of copy number alteration of IRF on the immune cell infiltration in PC. As a result, copy number alteration of IRF could suppress the infiltration level of immune cells to some extent (Fig S2).
IRFs-associated biologic functions in PC
DAVID 6.8 and Metascape were utilized to explore the biological functions of IRFs and their neighboring genes (Table S2) in PC. As we could see in Fig. 7 the results of functional analysis obtained from DAVID 6.8. The item of GO enrichment analysis revealed that IRFs and their neighboring genes were mainly involved in defense response to virus, T cell receptor signaling pathway, immune response, regulatory region DNA binding, protein binding, sequence-specific DNA binding, transcription factor activity, sequence-specific DNA binding, cadherin binding involved in cell-cell adhesion and type I interferon signaling pathway (Fig. 7A). The item of KEGG pathway revealed that IRFs and their neighboring genes were mainly linked to RIG-I-like receptor signaling pathway, T cell receptor signaling pathway, Toll-like receptor signaling pathway, Cell adhesion molecules (CAMs) and Cytosolic DNA-sensing pathway (Fig. 7B). PPI network showed that IRFs were mainly involved in immune response, sequence-specific DNA binding, response to Type I interferon (Fig S3).
To further detect IRFs-associated functions in patients with PC, Metascape was further used to perform enrichment analysis. Interestingly, the result suggested that IRFs and their neighboring genes were mainly linked to regulation of cytokine production, immune response-activating signal transduction in GO function analysis and type I interferon signaling pathway (Fig S4A and S4B, Table S3). The data of KEGG pathways analyses were shown in Fig S4C, S4D, and Table S4. As expected, IRFs and their neighboring genes were involved in T cell receptor signaling pathway, Cell adhesion molecules (CAMs), Antigen processing (presentation) and Hippo signaling pathway. Moreover, PPI network and Molecular Complex Detection (MCODE) components were isolated to identify the correlation between IRFs and their neighboring genes. The result indicated the involvement of IRFs in T cell receptor signaling pathway and Pertussis (Fig S4E and S4F).