The levels of APOE expression in THCA and other cancers
Gene expression analyses using the TIMER database based on TCGA data showed that APOE mRNA levels were significantly higher in breast cancer (BRCA), Esophageal carcinoma (ESCA), Head and Neck squamous cell carcinoma (HNSC), Liver hepatocellular carcinoma (LIHC), Prostate adenocarcinoma (PRAD), Stomach adenocarcinoma (STAD), Thyroid carcinoma (THCA) and Uterine Corpus Endometrial Carcinoma (UCEC) compared with the corresponding normal tissues (Fig.1). To further investigate whether the abnormal expression of APOE affects the occurrence of thyroid cancer, we then evaluated APOE transcription levels from multi-database. As shown in Fig.2a, APOE mRNA expression was significantly upregulated in thyroid cancer than in adjacent normal tissues according to the GEPIA database. Data in the Oncomine database showed that APOE mRNA expression was significantly higher in PTC and ranked within the top 2% based on mRNA expression (Fig.1b). IHC staining according to the the Human Protein Atlas database showed that APOE expression was not detected in normal controls, while the moderate APOE expression was showed in PTC tissue (Fig.2c) .
APOE expression correlated with clinicopathological parameters for PTC patients in TCGA cohort and Shanghai cohort
Next, we used the UALCAN web resource to study the relationship between APOE expression and different clinical characteristics of thyroid cancer. As shown in Fig.3, the expression of APOE was significantly higher in PTC patients than in normal controls with multi-analysis based on gender, age, metastasis status and different stages. In particular, APOE mRNA expression in PTC samples was significantly correlated with Ⅲ-Ⅳstages.
Then, 58 matched tumor and normal samples were immunohistochemically stained to verify whether the expression of APOE protein in the Shanghai cohort was consistent with the TCGA database and The HPA database. We found that in Shanghai cohort, APOE protein expression in PTC samples was significantly increased compared with adjacent tissues: 98.6% of PTC patients had higher APOE expression levels than in the normal tissues (Fig.5). Besides, the APOE expression in PTC of Shanghai cohort was positively correlated with pathological stage (p = 0.001, r = 0.6978) and lymph node metastasis (p = 0.001, r = 0.7) (Table 1). In order to reveal the relationship between APOE and PTC survival outcome, the survival curve analysis based on TCGA data was analyzed by GEPIA. However, the low APOE expression group had significantly shorter overall survival (Logrank, p= 0.027) compared to the high expression group and APOE expression had no association with disease-free survival(Logrank, p = 0.07). To determine the potential diagnostic value of APOE, the ROC curves of APOE were generated in TCGA cohort and Shanghai cohort, respectively (Fig.5). The ROC curve analysis showed that APOE had a satisfactory diagnostic value and the AUC of APOE were 0.7819 and 0.9064 in TCGA cohort and Shanghai cohort, respectively.
Enrichment analysis of APOE in PTC
Two PPI networks analysis of APOE were conducted by using STRING and GeneMANIA to explore the potential interactions among the proteins interacted with APOE. 12 nodes and 778 edges were acquired in the PPI network (Fig.6a). The function of these APOE-related genes was associated with post-translational protein modification, extracellular structure organization, inflammatory response. Results of GeneMANIA suggested that the roles of these APOE-related genes were basically linked to blood microparticle, plasma lipoprotein particle, protein-lipid complex, high-density lipoprotein particle, regulation of plasma lipoprotein particle levels, plasma lipoprotein particle remodeling and protein-lipid complex remodeling (Fig.6b).
To further understand the biological significance of APOE in PTC, LinkedOmics and GSCALite tools were used to explore the APOE co-expression patterns and possible pathways in the TCGA cohort. The results of LinkedOmics platform demonstrated 11534 genes (dark red dots) positively correlated with APOE and 8392 genes (dark green dots) negatively correlated with APOE in PTC (Fig.7a). The top 50 significant genes positively and negatively correlated with APOE in PTC were shown in Fig.7b and Fig.7c , respectively. Moreover, APOC1 (cor=0.9121, p=2.17e-195), APOC1P1 (cor=0.7111, p=2.27e-78), ISYNA1(cor=0.6479, p=5.45e-61) and C7orf61 (cor=0.6777, p=1.21e-68) were the four hub genes most positively correlated with APOE in PTC. Enrichment analysis was also performed. GO items showed that APOE co-expressed genes mainly participated in adaptive immune response, protein-lipid complex subunit organization, artery development, actin cytoskeleton reorganization, cell chemotaxis, amyloid-beta clearance and protein activation cascade, while the activities like ER-nucleus signaling pathway, nucleoside triphosphate metabolic process, Golgi vesicle transport, nucleobase metabolic process, mitochondrial respiratory chain complex assembly, coenzyme metabolic process, mitochondrial gene expression and tricarboxylic acid metabolic process were inhibited (Fig.7d). And KEGG pathway items revealed that enrichment in transcriptional misregulation in cancer, staphylococcus aureus infection, cytokine-cytokine receptor interaction, osteoclast differentiation, neuroactive ligand-receptor interaction, allograft rejection, cholesterol metabolism, leukocyte transendothelial migration, inflammatory bowel disease(IBD), natural killer cell-mediated cytotoxicity, and ECM-receptor interaction(Fig.7e). In order to further explore the potential mechanism of the five key genes (APOC1, APOC1P1, ISYNA1, C7orf61 and APOE) and whether these genes function through common cancer pathways, we analyzed them using the GSCALite platform by pathway activity module. As illustrated in Fig.8, APOE participated in the activation of processes such as Apoptosis, Cell Cycle, DNA Damage Response, EMT and Hormone ER, and the inhibition of Hormone AR, PI3K/AKT, RTK and TSC/mTOR signaling pathways.
Regulators of APOE in PTC
Owing to the significance of APOE in PTC, we further analyzed APOE
networks of kinase, miRNA and transcription factor targets in PTC (Table 2). Only one kinase target of APOE was identified (Kinase_LCK) from the LinkedOmics database. Then, PPI network was constructed to reveal the underlying mechanism of kinase LCK, and showed that the function of these genes in T cell activation, regulation of lymphocyte activation / leukocyte activation, positive regulation of T cell activation/lymphocyte activation/leukocyte activation/cell activation (Additional file 1: Fig. S1). MIR-323 was enriched by GSEA for APOE co-expressed genes (Table 2). Besides, V$AP1_C, V$STAT5B_01, V$NERF_Q2, V$NFKAPPAB65_01, RGAGGAARY_V$PU1_Q6, V$BACH2_01, V$NGFIC_01 and V$LBP1_Q6 were the transcription factor network targets of APOE (Table 2), and the functions of these transcription factors were mostly enriched in JAK-STAT signaling pathway, MAPK signaling pathway, growth hormone receptor signaling pathway and regulation of epithelial cell migration (Additional file 2: Fig. S2).
Association of APOE expression with immune cells and biomarkers
When we analyzed the role of the APOE in PTC using the LinkedOmics platform, we found that the function of the gene was primarily focused on regulating PTC immune response. This suggested that APOE may be involved in the immunoregulatory process of PTC. Then, TIMER platform was used to further clarify the association between APOE expression and immune cell infiltration. For the correlation between APOE expression and immune-related cells, we found a positive association between APOE expression and B cells (Cor=0.228, P=4.39e-07), CD8+T cells(Cor=0.15, P=9.30e-4), Neutrphils (Cor=0.197, P=1.14e-05) and Dendritic cells(Cor=0.229, P=3.58e-07) (Fig.9a). Fig.9b showed that positive correlations were acquired between APOE expression and the expression of CD8A, CD8B. For TAM, biomarkers including CCL2, CD68 and IL10 were positively correlated with APOE expression. Similar results were obtained in M1 and M2 Macrophage (INOS(NOS2), IRF5, COX2(PTGS2), CD163, VSIG4, MS4A4A ). Therefore, these results may indicate that APOE overexpression was related to the immunomodulatory process and APOE may be a potential target of immunotherapy of PTC. And multivariate COX regression analyses of APOE were performed. As shown in Table 3, tumor purity was associated with poor outcome and CD8+T cells (HR=0.000, 95%CI=0.000–0.111), macrophages (HR=0.000, 95%CI= 0.000–0.138) and Dendritic (HR=8478449.036, 95%CI=0.813–8.844267e+13) might be independent favorable prognostic indicators of PTC patients.
In order to reveal in more detail whether there was a correlation between APOE expression and lymphocytes and immunomodulators (immunoinhibitors, immunostimulators, and major histocompatibility complex (MHC) molecules), we analyzed it using TISIDB database. Fig.10a showed correlations between APOE expression and immunoinhibitors. The immune inhibitors showed strong correlations with APOE expression including CD160 (Spearman: ρ = 0.401, P < 2.2e-16), TGFB1 (Spearman: ρ = 0.514, P < 2.2e-16), LGALS9 (Spearman: ρ = 0.338, P = 6.24e-15), and TGFBR1 (Spearman: ρ = 0325, P = 8.14e-16) in PTC (Fig.10b). For immunostimulators, APOE expression was positively correlated with CD40 (Spearman: ρ = 0.444, P < 2.2e-16), KLRK1 (Spearman: ρ = 0.279, P = 1.76e-10), TNFRSF8 (Spearman: ρ = 0.607, P < 2.2e-16), and TNFSF13B (Spearman: ρ = 0.144, P = 0.00115) in PTC (Fig.10c-10d). Fig.10e showed correlations between APOE expression and MHC molecules. And there were positive correlations between APOE expression and HLA-B (Spearman: ρ = 0.153, P = 0.000557), HLA-DOA (Spearman: ρ = 0.144, P = 0.00113), HLA-DPA1 (Spearman: ρ = 0.117, P = 0.00843), and TAP1 (Spearman: ρ = 0.126, P = 0.00437) in PTC (Fig.10f). Besides, the correlation between APOE expression and tumor-infiltrating lymphocytes (TILs) was shown in Fig.10g, and TILs were positively correlated with APOE expression including Act_b (Spearman: ρ = 0.247, P < 1.84e-08), Act_CD8 (Spearman: ρ = 0.254, P = 6.95e-09), Tcm_CD4 (Spearman: ρ = 0.186, P = 2.56e-05), and Tfh (Spearman: ρ = 0359, P = 5.7e-17) in PTC (Fig.10h).
Finally, the EPIC application was used to analyze whether APOE expression was related to PTC immune infiltration. Among 512 PTC samples, samples were divided into 2 groups (top 1/2 and low 1/2 APOE expression groups). As shown in Fig.11, B cells (P = 0.002), CD8 T cells (P = 0.034), Macrophage cells (P = 0.00083) and other cells (P = 0.042) were main immune cells affected by different APOE expression.