Expression of FURIN in lung adenocarcinoma
We used Oncomine and TIMER to analyze the expression of FURIN mRNA in different cancer tissues and normal tissues. Among many cancers, FURIN had a higher expression in breast, stomach, head and neck, kidney, ovarian, and lung cancer, but FURIN had a lower expression in esophageal cancer, leukemia, lymphoma, melanoma, and myeloma (Figure 1A). At the same time, we also used Oncomine to analyze tumors in LUAD and normal samples, which showed that FURIN expression in LUAD was higher than normal lung gland tissue in different patient data sets (FURIN expressed in Selamat, Hou data sets) (Figure 1D, E). Analysis in the TIMER database showed that FURIN expression differences in tumor tissue and adjacent normal tissues (Figure 1B): FURIN expression in COAD(colon cancer), KIRC(kidney renal clear cell carcinoma),KIRP(kidney renal papillary cell carcinoma) was significantly lower in adjacent normal tissue; in contrast, FURIN expression in BLCA(road cortical cancer), ESCA(esophageal cancer), HNSC(head and neck squamous cell carcinoma), LUAD(lung adenocarcinoma), STAD(stomach cancer), UCEC(endometrial cancer) were higher than those of neighboring tissues. Data mining was further confirmed using the GEPIA and UALCAN databases. The GEPIA analysis showed a slight increase in expression of FURIN in LUAD tissue compared to normal lung gland tissue (P<0.05) (Figure 1C). On the other hand, UALCAN analyzed the protein expression of FURIN and found that its expression in tumor tissue was slightly higher than that of normal tissue (Figure 1F). In addition, we also used HPA to analyze IHC, and Figure 1G showed that FURIN significantly reduced protein expression in normal tissue and significantly increased protein expression in tumor tissue. According to different analysis results of FURIN above, FURIN was slightly higher in LUAD, indicating that the abnormal expression of FURIN may be closely related to the development, transfer and prognosis of LUAD.
Distribution of FURIN expression in clinical characteristics sub-groups
In order to study whether FURIN expression in lung adenocarcinoma patients associated with clinical factors, we used the UALCAN database to detect the distribution of FURIN in different histological subtypes of lung adenocarcinoma, it was found that there was statistical significance between FURIN expression and the first three tumor stages (P<0.05) and an increase in the distribution of FURIN in the first three tumor stages (Figure 2A). At the same time, FURIN was expressed higher in TP53 in a non-mutant state (Figure 2B). Analysis of gender, age, subtypes of different tissues and different lymph node metastasis showed that the expression level of FURIN in patients with lung adenocarcinoma increased relative to normal samples (Figure 2C-F). In addition, FURIN was significantly higher in women than in men in lung adenocarcinoma (Figure 2C). And smokers had a higher expression of FURIN than non-smokers (Figure 2D).
Prognosis analysis of FURIN expression in LUAD
Next, we used the PrognoScan database to explore the relationship between FURIN expression and LUAD patient prognostics. The study found a significant correlation between prognosis in LUAD patients and expression of FURIN (P<0.05) (Figure 3A, B). The higher FURIN expression was associated with a poorer LUAD prognosis (OS HR=1.26, COX P-VALUE=0.000641; RFS HR=0.69, COX P-VALUE=0.023887). Moreover, there are other types of cancer that showed some correlation between prognostic and FURIN expression, such as breast and ovarian cancer (Figure 3C, D). We then conducted a meta-analysis of the results generated from different data sets in PrognoScan database. Meta-analysis showed that the combined HR and 95%Cl of OS associated with high FURIN expression were 1.08(0.98,1.18), but heterogeneity was observed in 21 data sets (I^2=55.2%, P=0.001) (Figure 3E).
Then, we used the Kaplan-Meier plotter to further validate the prognosis of FURIN expression. Patients were divided into two groups based on the median expression of each group of FURIN. It is found the increased expression of FURIN was associated with the prognosis of LUAD patients (P<0.05), and overall survival (OS) and first progression (FP) were also highly affected by the increased expression of FURIN mRNA (OS:HR=1.67, Log-rank P=1.4e-05; FP:HR=1.62, Log-rank P=0.0024) (Figure 3F, G), suggesting that FURIN may be a reliable biomarker for the prognosis of LUAD.
LUAD patients with FURIN mutation frequency, type and FURIN changed LUAD patient survival analysis
Based on sequencing data from LUAD patients in the TCGA database, we used the cBioPortal database to determine the type and frequency of FURIN changes in LUAD patients. FURIN changed in 30 out of 503 patients (6%) (Figure 4A). These changes were 5 mutations (0.99%), 3 (0.6%) amplification, 1 (0.2%) deep deletion and 20 (3.98%) mRNA high, and 1 multiple change (0.2%). Therefore, mRNA high is the most common type of LUAD. The types, sites and case number of the FURIN genetic alteration are further presented in Figure 4B. We found that missense mutation is the most common type of FURIN change. In addition, we explored the potential link between FURIN gene changes and clinical survival prognostics in LUAD patients. Figure 4C, D, E shows that patients without FURIN change showed better prognosis in Overall survival rate (P=1.636E-3), progression-free survival rate (P=6.679E-4) and disease-specific survival rate (P=2.380E-3) compared to patients with FURIN changes.
Correlation expression and correlation analysis of immune cell penetration in LUAD
Then, we used the TIMER database to determine whether the high expression of FURIN in LUAD tissue was associated with six major permeable immune cells (B cells, CD4+T cells, CD8+T cells, neutrophils, macrophages, and dendritic cells). The results showed that FURIN expression levels were associated with B cells (r=-0.144, p=1.46e-03), CD8+T cells (r=-0.157, p=4.92e-04), CD4+T cells (r=-0.039, p=3. 97e-02), macrophages cells (r=-0.108, p=1.77e-02), neutrophil cells (r=-0.124, p=6.28e-03), dendritic cells (r=-0.243, p=5.64e-08) related. As showed in the scatter chart, the expression level of FURIN was negatively related to most immune cells, indicating that the expression level of FURIN was negatively related to LUAD immune infiltration. To further study the relationship between immune cellular immersion and FURIN expression in LUAD, we used the TIMER database to further map Kalan-Meier to assess the prognostic value of each of the six immune cells (Figure 5B). We found that the expression of B cells (Log-rank P=0) and dendritic cells (Log-rank P=0.048) were associated with the prognosis of LUAD, indicating that FURIN plays an important role in regulating immune cell penetration.
Correlation analysis between FURIN mRNA levels and different subgroup markers of immune cells
Next, based on a set of immune markers in LUAD, we used the TIMER database to further explore the link between FURIN expression and immune cell penetration levels. Specifically, we evaluated the correlation between FURIN expression and Parker levels for a particular subset of cells (including CD8+T cells, B cells, T cells, monocytes, TAM, neutral granulocytes). At the same time, we analyzed different subsets of T cells, namely Th1, Th2, Tfh, Th17, Treg, T cell exhaustion. As a result of tumor purity adjustment results in clinical samples, the results showed that the expression of FURIN in LUAD tissue was significantly related to the expression of most marker genes in immune filtration cells (Table 1).
The study found that the expression of FURIN in LUAD was significantly related to the expression of immune marker genes in CD8+T cells, T cells, monocytes, TAM, and dendritic cells. In particular, CD8A, CD8B of CD8+T cell, CD3D, CD2 of T cells, CD86, CSF1R of Monocyte, CCL2 of TAM cells, ITGAM of Neutrophils, HLA-DPB1, HLA-DQB1, HLA-DRA, HLA-DPA1, CD1C of Dendritic cells are all closely related to LUAD's FURIN level (P<0.0001). In addition, we found that the expression of FURIN is closely related to Th2, Tfh, Th17, T cell exhaustion (e.g. STAT6, ICOS, STAT3, HAVCR2), further confirming the relationship between FURIN expression and LUAD immune infiltration.
Then we used the TIMER 2.0 database and used different algorithms to further explore the relationship between macrophages and FURIN expression, and found that FURIN expression was negatively related to macrophage immersion levels (Figure 6A, B). The expression of FURIN was also related to the polarization of macrophages, and the expression of FURIN was negatively related to the polarization of M1 macrophages and M2 macrophages (Figures 6C, D, E, F). In addition, we found that the expression of FURIN was positively related to markers of M1 macrophages and negatively related to the markers of M2 macrophages (Figure 6G, H). The expression of FURIN in lung adenocarcinoma is associated with macrophage polarization, indicating that FURIN may affect the immune response through polarized macrophages.
FURIN promoter methylation level analysis in LUAD
After analysis, we observed an increase in FURIN expression in LUAD, so we will further study the reasons for the increase in FURIN expression. Because methylation is closely related to disease progression, low methylation can lead to genomic instability and may activate related genes, all of which we use the UALCAN database to verify the methylation levels of FURIN promoters in LUAD. As showed in Figure 7A, the level of methylation of the promoter of FURIN in normal tissues was slightly higher than that of LUAD. In addition, group analysis of promoter methylation was carried out according to the patient's sex, tumor stage, smoking habits, lymph node metastasis, and age. The results showed that the promoter methylation was associated with sex, tumor stage, smoking habits and age in LUAD patients (Figure 7B, C, D, F), but there was no correlation between promoter methylation and LUAD patients with N3's lymph node metastasis. These results showed that lower promoter methylation may lead to high expression levels of FURIN in LUAD.
Construction of gene co-expression network and gene enrichment analysis
To better understand the biological processes of FURIN in LUAD, we used the LinkedOmics database to analyze differences in FURIN co-expression genes. As showed in Figure 8A, 9000 genes (red dots) are positively correlation with FURIN and 10988 genes (green dots) are negatively correlation (P<0.05). As showed in Figure 8B-C, the first 50 positive and negative genes associated with FURIN are shown as heat maps. FURIN expression is presented with genes strong positive correlation such as INSL4 (positive correlation, r=0.618, p=1.13E-55), CPS1 (r=0.597, p=4.37E-51), C6orf176 (r=0.594, p=1.33E-50), but negative correlation with genes such as STK17A (r=0.469, p=1.24E-29), DHDH (r=0.447, p=1.12E-26), CNN3 (r=0.446, p=1,31E-26). What’s more, we used GeneMANIA to construct PPI network of FURIN.As showed Figure8D, NOTCH3, GPC3, PROZ, GDF11, FLNA, GAS6, SERPINB8, LECT1, EDA, F7, PCSK5, PACS1, F10, PCSK1, PROS1, SORL1, MMP7, NGF, PCSK2, F9 interact with FURIN. After that, we used DAVID to perform Gene Ontology (GO) enrichment analysis based on the FURIN co-expression gene network. As showed in Figure 8E, According to GO enrichment analysis, FURIN co-expression genes were mainly concentrated in biological process including the positive regulation of substrate adhesion-dependent cell spreading, blood coagulation, fibrin clot formation, signal transduction, plasminogen activation, positive regulation of heterotypic cell-cell adhesion, CAMP catabolic process, cell cycle, cell protein complex assembly, fibrinolysis, positive regulation of exocytosis, positive regulation of transcription, DNA-templated. At the cell composition level, FURIN co-expression is mainly enrich in extracellular exosomes, and FURIN co-expression in molecular function level were mainly focused on metal ion binding (Figure 8F).
SARS-CoV infection may affect the expression of FURIN
The SARS-CoV-2 S protein has multiple FURIN bits. And FURIN can help the virus enter human cells[12]. In addition, FURIN plays an important role in lung adenocarcinoma. Therefore, it is crucial to analyze the expression of FURIN in LUAD patients who infected with SARS-CoV. We used GSE44274 in the GEO database for research. GSE44274 includes four biological sample types of mouse lung tissue (SARSmus, UV-V, Lung; SARSmus, UV-V+TLR, Lung; SARSmus, PBS, Lung and SAESmus, HKU, Lung). Based on whether the samples infected with SARS-CoV, we divided it into treatment group and control group, which were used to analyze the expression of FURIN in mice lung infected with SARS-CoV-2. As showed in the figure, the FURIN level in the treatment group was significantly lower than that in the control group. The results showed statistical significance, which indicated that SARS-CoV-2 infection may affect the expression of FURIN (Figure 9).