3.1. The ANLN mRNA expression levels in HCC and other human cancers
The ANLN mRNA levels in different tumors and in the normal tissues of multiple cancer types were analyzed using the Oncomine (Fig. 1A) and TIMER (Fig. 1B) databases, with an aim to reveal the aberrant expression of ANLN from a holistic perspective. To determine the differences in ANLN expression between HCC and normal samples, the differential scatter plot (Fig. 1C) and paired plot (Fig. 1D) were drawn. Also, ANLN expression in different GEO datasets was also analyzed via the HCCDB database (Fig. 1E and Table 2). Our results suggested that, ANLN expression was upregulated in HCC among different datasets, as well as in bladder, breast, biliary, colorectal, head and neck, esophageal, gastric, kidney, lung, prostatic and rectal cancers compared with that in normal tissues.
3.2. Prognostic value of ANLN in HCC
As displayed in Fig. 2A, results of survival analysis showed that HCC with high ANLN expression was associated with worse prognosis than that with low ANLN expression (p < 0.001). Similar results were also obtained from the analysis of HCC samples in ICGC-LIRI-JP (Fig. 2B). In addition, the relationships among ANLN expression, clinical characteristics and survival prognosis for HCC patients were also investigated in the Kaplan-Meier plotter databases (Table 3). According to the results, ANLN over-expression was related to the worse OS in male and female patients, white and Asian patients, as well as patients with different clinical stages and T stages according to the American Joint Committee on Cancer (AJCC), none or micro-vascular invasion status, alcohol consumption, and hepatitis virus infection (P < 0.05). Specifically, the effects of high ANLN expression on OS were particularly significant in Asian (hazard ratio [HR] =5.66, P<0.001), male (HR=2.43, P<0.001) and G1 (HR=6.52, P<0.001) patients. Besides, the values of area under the receiver operating characteristic (ROC) curve (AUC) for 1-, 3- and 5-year OS were 0.679, 0.587 and 0.602, respectively (Fig. 2C).
Moreover, univariate analysis revealed that high ANLN expression was remarkably correlated with poor OS (HR: 1.305; 95% confidence interval [CI]: 1.095−1.556; p < 0.01) (Table 4). Additionally, race was another clinicopathological variable associated with poor survival. Meanwhile, ANLN (HR: 1.584; CI: 1.158−2.167, p < 0.01) and race (HR: 4.651; CI: 1.778−12.152, p < 0.01) remained the factors associated with OS upon multivariate analysis.
3.3. Associations of ANLN expression with clinicopathological variables
Altogether 371 HCC samples with available ANLN expression data stratified based on multiple patient characteristics were analyzed from TCGA. As shown in Figure 3A–3G, the transcription levels of ANLN were notably higher in HCC patients than in healthy people upon subgroup analyses stratified according to gender, age, weight and obese, ethnicity, disease stage, tumor grade and nodal metastasis status. In addition, the logistic regression univariate analysis demonstrated that ANLN expression, which served as a categorical dependent variable, was linked with poor prognostic clinicopathological characteristics (Table 5). Besides, the up-regulated ANLN expression in HCC was evidently associated with high grade (4.82 for G3 vs. G1, 4.88 for G4 vs. G1), stage (OR = 1.68 for II vs. I, 2.14 for III vs. I), and T stage (OR = 1.73 for T2 vs. T1, 1.95 for T3 vs. T1). These findings demonstrated that HCC with high ANLN expression was more inclined to progress to a more advanced and malignant stage than that with low ANLN expression.
3.4. GSEA identified the ANLN-related signaling pathways and Gene Ontology (GO) terms
GSEA was carried out to further investigate the differences between the high and low ANLN expression groups. As revealed by our findings, the GO terms, including mRNA binding, G1/S phase transition in the cell cycle, and Atpase activity, were the top 3 with the highest ES scores, and they were differentially enriched in the high ANLN expression phenotype (FDR=0, NOM=0) (Fig. 4A -4C); in the meantime, the high density lipoprotein particle, protein activation cascade and protein lipid complex were closely correlated with the low ANLN expression phenotype (FDR<0.01, NOM=0) (Fig. 4D – 4F). Furthermore, results of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis revealed that the genes in high ANLN expression group were mainly enriched in the cell cycle, ubiquitin-ediated proteolysis and oocyte meiosis (FDR=0.01, NOM=0) (Fig. 5A -5C); in addition, the complement and coagulation cascades, primary bile acid biosynthesis, and fatty acid metabolism were primarily enriched in low ANLN expression group (FDR<0.05, NOM=0) (Fig. 5D -5F). Fig.4G and Fig.5G exhibit the comprehensive diagrams showing the above items. More details on the above pathways and GO terms are summarized in Supplementary Table 1-4.
3.5. Co-expression analysis and ANLN networks of kinases, miRNAs and TF targets in HCC
The Function module of LinkedOmics was utilized to analyze the mRNA sequencing data of 371 HCC patients from TCGA database. As shown in the volcano plot (Fig. 6A), 129,55 genes (dark red dots) were significantly positively correlated with ANLN, whereas 6,965 genes (dark green dots) were exhibited significant negative correlations (FDR < 0.01). Thereafter, those 50 genes that showed significant positive and negative correlations with ANLN were exhibited in the heat map (Fig. 6B and 6C, respectively). Supplementary Fig. 1A–1F display the statistical scatter plots for individual genes. To further explore the ANLN targets in HCC, the kinases, miRNAs and TFs target networks of the positively correlated gene sets generated by GSEA were analyzed, respectively. Of them, the top 5 most significant target networks were the kinase-target networks mainly related to the polo-like kinase 1 (PLK1), cyclin-dependent kinase 1 (CDK1), aurora kinase B (AURKB), ATM serine/threonine kinase, and cyclin-dependent kinase 2 (CDK2) (P = 0, FDR = 0) (Table 6). Besides, the miRNA-target network was associated with (AGTCTTA) miR-499, (CTGTTAC) miR-194, (GTAGGCA) miR-189, miR-380-3P and MIR-452 (P = 0, FDR < 0.1). The TF-target network was mainly related to the E2F (E2F) TF family, including E2F1_Q6, E2F4DP1_01, E2F_Q6, E2F1DP1_01, and E2F1DP2_01 (P = 0, FDR = 0). Besides, the protein-protein interaction (PPI) network constructed by GeneMANIA revealed correlations among genes for kinases PLK1 (Fig. 7), MIR499 (Supplementary Fig. 2), and TF E2F1_Q6 (Supplementary Fig. 3). Typically, the gene set enriched for kinases PLK was mainly responsible for mitosis, nuclear division and organelle fission, while that enriched for TF E2F1_Q6 was related to DNA replication.
3.6. ANLN expression was correlated with the immune infiltration levels in HCC
The correlations of ANLN expression with immune infiltration levels in HCC samples collected from TIMER were analyzed, and the results suggested that ANLN expression was significantly correlated with the infiltration levels of B cells (r = 0.429, P<0.001), CD8+ T cells (r = 0.273, P<0.001), CD4+ T cells (r = 0.342, P<0.001), macrophages (r = 0.42 P<0.001), neutrophils (r = 0.374, P<0.001) and dendritic cells (r = 0.422, P<0.001) (Fig. 8A). These findings suggested that ANLN might play a specific role in immune infiltration within the HCC microenvironment, especially for those of B cells, macrophages and dendritic cells.
3.7. ANLN expression in Hep3B cells was dramatically upregulated compared with that in LO2 cells
The cell culture results are presented in Supplementary Fig. 4. As observed, ANLN expression was significantly upregulated in Hep3B cells relative to that in normal liver LO2 cells (Fig. 8B) (P <0.05). The amplification plot and melt curve were also displayed in Fig. 8C and Fig. 8D, respectively.
3.8. The protein levels corresponding to ANLN mRNA levels in cell lines detected by Western blotting
Fig. 9 shows the protein expression levels corresponding to ANLN mRNA levels in Hep3B and L-O2 cells. Clearly, the protein level corresponding to ANLN mRNA level was lower in LO2 cells than in Hep3B cells (P< 0.05).
3.9. IHC results regarding ANLN expression in HCC tissues
IHC was carried out to examine the ANLN protein expression in HCC tissues and their counterparts, as well as ANLN expression in HCC. It was discovered that, the ANLN protein expression was upregulated in HCC tissues (Fig. 10A and 10B) relative to that in normal tissues (Fig. 10C and 10D).