GPBAR1 expression in HGG tissues and cell lines.
First of all, data were retrieved in TCGA database to show GPBAR1 expression in high-throughput assay. The TPM (transcripts per million) of GPBAR1 in 163 HGGs was significantly higher than in 207 normal brain tissues (Figure 1A). In our collection of 10 fresh HGGs and their corresponding normal tissues, GPBAR1 mRNA was detected with qRT-PCR. The results also showed the up-regulation of GPBAR1 in HGGs (Figure 1B). In our cohort with 149 HGG patients, GPBAR1 expression was detected with IHC, and evaluated with IHC score which divided the cohort into subsets with low and high GPBAR1 expression (Figure 1C). The percentages of patients with low and high GPBAR1 account for 45.6% (68/149) and 54.4% (81/149), respectively. In addition, GPBAR1 expression in GBM cell lines U251, U118, U87 and A172 cells were detected with Western blot.
The clinical significance of GPBAR1 expression
The clinical significance of GPBAR1 was first evaluated by analyzing its correlation between clinicopathological factors with chi-square test (Table 1). The clinical information including the sex and age of patients, the tumor size, Karnofsky Performance Scale (KPS) score, and adjuvant therapy. Intriguingly, high expression of GPBAR1 was significantly associated with large tumor size (P=0.004). The correlation between GPBAR1 and other factors had no obvious statistical significance.
Furthermore, the prognostic value of GPBAR1 and other clinicopathological factors were analyzed with univariate and multivariate analyses (Table 2). The Kaplan-Meier method was applied to analyze the correlations between these factors and the overall survival rates. In our study, GPBAR1 expression was substantially associated with the overall survival(OS) rate (Figure 2A). High GPBAR1 predicted the poor outcome of patients with high-grade glioma. Moreover, high KPS and adjuvant therapy were also associated with the OS rate of patients (Figure 2B and 2C).
In addition, the independent prognostic factors were identified with multivariate analysis (Table 2). In the Cox-regression Hazard model, GPBAR1 was an independent prognostic factor of high-grade glioma (P=0.016), with a hazard ratio of high GPBAR1 as 1.64. In addition to GPBAR1, higher KPS could independently indicated the favorable prognosis (P=0.008).
GPBAR1 expression was correlated with MAFB
In the clinical analysis, we found that GPBAR1 was associated with larger tumor size, indicating that GPBAR1 may be involved in the tumor proliferation, so we further screened the potential target proteins of GPBAR1 which may participate in the GPBAR1-invovled proliferation. MAFB was previously reported to promote the cancer progression such as tumorigenesis, proliferation and stemness in several cancer types including osteosarcoma and colon cancer6; 29. In TCGA database, MAFB expression was substantially correlated with GPBAR1 expression (Figure 3A). In our study, we also investigated the mRNA correlation between GPBAR1 and MAFB, and demonstrated that GPABR1 was also positively associated with MAFB (Figure 3B). In the cohort with 149 patients, the average IHC score of MAFB in patients with low GPBAR1 was significantly lower than that in patients with high GPBAR1 (Figure 3C). Chi-square test also validated the significant correlation between GPBAR1 and MAFB (Table 1). In U118 and U251 cells, we silenced or overexpressed GPBAR1, and found that MAFB expression was correspondingly changed (Figure 3D and 3E).
GPBAR1 activation promoted the expression of MAFB.
GPBAR1 was a GPCR receptor which could be stimulated by bile acid, so we further investigated the molecular signaling between GPBAR1 and MAFB. The cAMP stimulator forskolin and inhibitor NF449 were used to incubate U118 cells for 24 hours, and we showed that MAFB expression was enhanced when GPBAR1 was overexpressed, or when Forskolin was used (Figure 4A). On the contrary, NF449 could inhibit GPBAR1-induced MAFB expression. These results suggested that GPBAR1 regulated MAFB expression in a cAMP-independent pathway. U118 cell proliferation was also detected after GPBAR1 overexpression, in the presence of forskolin or NF449 with CCK8 assay. GPBAR1 overexpression and forskolin stimulation accelerated U118 proliferation, while NF449 attenuated the proliferation (Figure 4B). GPBAR1 can stimulate and activate MAPK-ERK pathway by phosphorylating ERK23, so we further verified that in GBM cells. Int777 is a well-accepted specific stimulator of GPBAR131, which was used to activate GPBAR1 signaling in U251. In our study, Int777 promoted the phosphorylation of ERK and the expression of MAFB. However, GPBAR1 kncokdown and Ulixertinib decreased ERK activation and MAFB expression (Figure 4C), suggesting that ERK activation was essential in GPBAR1-induced MAFB expression. U251 cell proliferation had the similar tendency with MAFB expression (Figure 4D). To show the role of MAFB in GPBAR1-induced proliferation, we silenced MAFB when overexpressing GPBAR1 and/or stimulating U118 with Int777 (Figure 4E), and we detected the influence of these factors to U118 proliferation. MAFB knockdown significantly impaired cell proliferation, which was increased by Int777 or GPBAR1. These results showed an essential role of MAFB in GPBAR1-induced proliferation (Figure 4F).
Co-expression of GPBAR1 and MAFB was a more sensitive prognostic biomarker
The clinical significance of MAFB has never been elucidated in glioma, so we further evaluated the prognostic significance of MAFB. The cohort was divided into low and high MAFB according to MAFB IHC score (Figure 5A). The number of patients with low and high MAFB was 87 and 62, respectively, accounting for 58.39% and 41.61%. With univariate analysis, we showed that patients with low MAFB expression had more favorable prognosis than those with high MAFB (P=0.002), with the 3-year OS rate as 32.7% and 19.1% respectively (Figure 5B). Moreover, we divided the patients into subsets with those with co-expression of GPBAR1 and MAFB, and those with other expression patterns, which accounted for 42 and 107 patients respectively. Co-expression of GPBAR1 and MAFB can predict the poor prognosis of high-grade glioma more effectively and sensitively (P<0.001), showing that detecting GPBAR1 and MAFB may be a possible method for individual treatment.