3.1 The mRNA and protein expression of AJUBA and YAP1 in Glioma (Bioinformatics analysis )
We first used Oncomine database to analyze the transcriptional levels of AJUBA in various cancer types and corresponding normal tissues(Fig 1). The result shows that there are a total of 300, 436 unique analyses for AJUBA and YAP1 respectively. In tumor tissues,AJUBA was significantly increased in 30 datasets and 7 datasets showed an reduced level.Additionally, higher expression of AJUBA were observed in in tumor tissues, especially in brain and CNS, colorectal,gastric,cervical,leukemia, esophageal,head and neck cancers. Compared to normal tissues,YAP1 showed elevated mRNA levels in tumor tissues, especially in brain and CNS, colorectal,gastric,lymphoma,pancreatic cancers.YAP1 was significantly increased in 22 datasets and 7 datasets showed an reduced level.
Oncomine analysis results showed that AJUBA mRNA level was significantly elevated in glioma.The transcription levels of AJUBA were significantly higher in Glioma in five datasets. AJUBA was upregulated in Glioblastoma in Sun brain42, Murat Brain43 and Liang Brain dataset44 and upregulated in LGG in Sun brain42,French Brain45 and TCGA Brain 2 dataset. The stypes of LGG are mainly included in Anaplastic Oligodendroglioma,Oligodendroglioma,Anaplastic Astrocytoma,Diffuse Astrocytoma. However,The transcription levels of YAP1 also were significantly higher in Glioma in five datasets.YAP mRNA level was elevated in glioblastoma in Sun brain42, Murat Brain43,Bredel Brain 246 and TCGA Brain datasets.
Further, Meanhile,Bredel Brain 2 and TCGA Brain 2 datasets showed elevated levels in Anaplastic Oligoastrocytoma,Oligoastrocytoma and Astrocytoma in LGG. According to TCGA Brain datasets, YAP1 was upregulated in Oligodendroglioma ,Astrocytoma and Glioblastoma as compared with normal brain tissues. Conversely, YAP1 was greatly downregulated in two analyses of French Brain dataset45 and one analyses of Liang brain dataset.All the results with corresponding p-values for statistical significance are summarized in Fig 1 and Table 1.
We also compared the transcription levels of AJUBA and YAP1 between glioma and normal tissues by using GEPIA(Fig. 2 A-B). We found that AJUBA was upregulated in tumor tissues, while AJUBA mRNA levels was significantly upregulated in GBM. The YAP1 was upregulated in tumor tissues, while YAP1 mRNA levels also was significantly upregulated in GBM.In addition, the relationship between the mRNA levels of AJUBA and YAP1 and the tumor suvival and stages of glioma were analyzed. The results revealed that AJUBA and YAP1 expressions were associated with the tumor suvival and stage (Fig. 2 C-D)
To further explore the AJUBA protein expression in glioma, we analyzed Immunohistochemistry staining images from the HPA(Fig. 3). The result showed that not expressions of AJUBA were observed in LGG tissues, while high protein expressions of them were observed in HGG. YAP1 proteins were not expressed or low expressed in LGG , while medium protein expressions of YAP1 were expressed in HGG. In conclusion, we found that the proteins expression of AJUBA and YAP1 in HGG were higher than that in LGG.
KEGG pathway enrichment analysis was employed to obtain a comprehensive understanding of the target proteins in the Hippo signaling pathway. The results showed that AJUBA promoted the translocation of YAP1 to the nucleus by dephosphorylation, where it induced the expression of TEAD and Smad1/4 activated (Fig. 4). Therefore,AJUBA and YAP1 were vital proteins of the Hippo signaling pathway in glioma according to the bioinformatics analysis.
We analyzed the relationship of AJUBA and YAP1 at the gene level by using GeneMANIA (Fig. 5a).Physical interactions were found between AJUBA and Hippo signaling pathway, as well as between AJUBA and YAP1. Relationships were noticed in co-localization between AJUBA and YAP1. In addition, the relationships were found between AJUBA and YAP1 in genetic interactions.We identified interactions of AJUBA and YAP1 at the protein expression level by using STRING (Fig. 5b). AJUBA was shown to interact with YAP1 and Lats in gene co-occurrence, text-mining, and protein homology.Therefore, AJUBA and YAP1 were vital proteins of the Hippo signaling pathway in glioma according to the bioinformatics analysis.
3.2 Clinicopathological characteristics
The characteristics of 217 patients with glioma are summarized in Table 2. This cohort consisted predominantly of LGG (65.4 %; 142/217) and partly of HGG (34.6 %; 75/217). The mutations at codon 117 of IDH1 gene were heterozygous in 142 LGG paraffin specimens with rate of 82.6 % (Table 2). The total cohort mainly included cases with tumor size ≥3 cm (82.49 %; 179/217), tumor total resection (66.4 %; 144/217), and no local recurrence (65.9 %; 143/217). Differences in age (p < 0.000), tumor location (p < 0.000), local recurrence (p < 0.000), tumor resection (p < 0.000) ,ki-67 expression (p < 0.001) and p53 expression(p <0.02) between patients with LGG and HGG reached statistical significance.
3.3.1 Association between clinicopathological variables and protein expression of AJUBA and YAP1.
The expression of AJUBA, YAP1 and their relationship with clinicopathological parameters in all cohorts (LGG and HGG) were detected by immunohistochemistry to investigate the possible role of AJUBA in the development of glioma (Table 3) (Fig. 6). In 217 cases of glioma, AJUBA expression was low in 133 tumors (61.3%) and high in 84 tumors (38.7%). The high expression of AJUBAwas prominent in GBM (55.95 %), suggesting its preferential activation in the most aggressive tumors. Statistical analysis revealed that AJUBA expression was related to WHO grade (p = 0.000), recurrence (p =0.018) ,Tumor resection(p =0.012) and p53 expression (p = 0.000) but not to age (p = 0.950), gender (p = 0.672),Ethnic(p=0.068), tumor location (p = 0.052), tumor size (p = 0.856), postoperative radio-chemotherapy (p = 0.07), and Ki-67 expression (p = 0.163) in all cohorts. Meanwhile, there was no statistical significance for the association between AJUBA and IDH 1mut in LGG (p = 0.451). However, Low expression of AJUBA was associated with 1p19q codeletion in LGG (r= -0.22,p = 0.009) (Table 4). The results suggested that AJUBA expression is related to glioma molecular subtypes. Meanwhile, YAP1 expression was low in 94 tumors (43.3 %) and high in 123 tumors (56.7 %). Statistical analysis indicated that YAP1 expression was related to tumor resection (p=0.000),Local recurrence (p = 0.021),WHO grade (p =0.027). but not to age (p = 0.784),tumor size (p =0.05), gender (p =0.073),Ethnic(p =0.679), tumor location (p =0.494), postoperative radiochemotherapy (p =0.13), P53 expression (p=0.05), Ki-67 expression (p = 0.133) in glioma. An increasing trend of YAP1 was found in WHO grade II, III, and IV glioma tissues. (Table 3). Low expression of YAP1 was associated with IDH 1mut in LGG (r = 0.197,p = 0.019). However, there was no statistical significance for the association between YAP1 and 1p19q codeletion (p = 0.412) (Table 4). The results suggested that YAP1 expression is norelated to glioma molecular subtypes(Table 4).
3.3.2. Association among AJUBA and YAP1 protein markers
Spearman correlation analysis was performed to clarify the associations among AJUBA and YAP1 protein markers (Table 5).AJUBA expression was associated with YAP1 (p = 0.001; r = 0.219) .When AJUBA was highly expressed in HGG, YAP1 was also highly expressed in HGG. The high expression of YAP1 is associated with the overexpression of AJUBA in glioma. Therefore, AJUBA and YAP1 are connected to each other and related to the development of glioma.
3.4 Survival analysis
Survival analysis was conducted in total glioma cohort (Table 6).Among the 217 cases, two cases had missed visits. Univariate survival analysis through Kaplan–Meier method revealed that the following prognostic factors affected the OS in glioma (Fig. 7 A–I): age (≤50 vs. > 50 years old; p < 0.0001),tumour location(Frontal vs. Other;p=0.0114), local recurrence (no vs. yes; p < 0.0001), tumor resection (total vs. partial; p < 0.0001), WHO grade (LGG vs. HGG; p < 0.0001), ki-67 expression (≤5% vs. > 5%; p < 0.0001), AJUBA expression (low vs. high; p <0.0001), and YAP1 expression (low vs. high; p < 0.0001),YAP1 and AJUBA co-expression(YAP1 low+AJUBA low vs.YAP1 low+AJUBA High vs.YAP1 High+AJUBA Low vs. YAP1 High+AJUBA High; p <0.0001). Multivariate analysis (Cox’s proportional hazards regression model) on OS showed that the possible independent prognostic factors for OS were as follows: age (HR: 2.867; 95 % CI: 1.987–4.135; p < 0.0001), Tumour location (HR: 1.551; 95 % CI: 1.093–2.201; p = 0.0024), Local recurrence(HR: 3.48; 95 % CI: 2.312–5.238; p < 0.0001) ,WHO grade (HR: 6.28; 95 % CI: 3.920–10.06; p < 0.0001), ki-67 expression (HR: 2.541; 95 % CI: 1.789–3.609; p < 0.0001), and AJUBA expression (HR: 2.065; 95 % CI: 1.418–3.007; p=0.045), YAP1 expression (HR: 2.377; 95 % CI: 1.666–3.392; p < 0.0001),. In univariate survival analysis and multivariate analysis of PFS (Fig. 7J–R), the prognostic factors were the same as those of OS,
and AJUBA expresion(HR: 2.044, 95 % CI:1.383 -3.021;p<0.0001) ,also YAP1 expression (HR: 1.857; 95 % CI: 1.299–2.657; p = 0.002) were independent prognostic factors for PFS. These results showed that the independent factors that can predict short OS and PFS in the total glioma cohort were >50 years old, Frontal location,recurrence,tumor total resection,WHO grade IV and Ki-67 expression >5%. AJUBA high expression and YAP1 high expression predicted short OS in glioma. Therefore, AJUBA and YAP1 may be biomarkers of poor prognosis in glioma.