Abnormal overexpression of TROAP in glioma
The GEPIA database contains gene expression data for a variety of tumors. Analysis GEPIA data revealed that TROAP expression was increased in glioblastoma multiforme (GBM), lower-grade glioma (LGG), colon adenocarcinoma (COAD) and stomach adenocarcinoma (STAD) tissues compared with non-tumor tissues (Figure 1a). In addition, we further downloaded the GSE50161 and GSE116520 glioma datasets from the GEO database. In the GSE50161 dataset, 34 glioma tissue specimens were compared with 13 normal brain tissue specimens (Figure 1b). In the GSE116520 dataset, 34 glioma tissue specimens were compared with 8 normal brain tissue specimens (Figure 1c). Through the above research on glioma and normal tissue data from the GEPIA and GEO databases, we found that the expression level of TROAP in glioma was indeed higher than that in the corresponding non-tumor tissues. To verify these analysis results, we further detected the expression level of TROAP in four glioma cell lines (T98, LN229, A172 and U251) and human astrocyte (HA) by RT-qPCR. The result demonstrated that TROAP was highly overexpressed in glioma cell lines compared with HA (Figure 1d).
Increased expression of TROAP is associated with a reduction in the overall survival(OS) of glioma patients
Additionally, we retrieved three groups of data, namely, CGGA RNA-seq, CGGA microarray and TCGA RNA-seq data, to further explore the role of TROAP in the pathological processes of glioma and then plotted the survival curves of the three groups of patients via Cox regression and Kaplan-Meier analyses. High expression of TROAP in the CGGA-RNA sequencing, CGGA microarray and TCGA-RNA sequencing datasets was associated with a significant decrease in patient OS (Figure. 2a, b, and c). These results suggested that high TROAP expression may be associated with the OS of glioma, but whether this gene can serve as an independent risk factor for glioma remains to be further verified.
TROAP is an independent risk factor for glioma patients.
Next, prognostic factors for OS were analysed by the Cox regression model. In the CGGA RNA-seq data cohort, enhanced TROAP expression in glioma was considerably linked with poor prognosis (hazard ratio [HR]= 1.605), old age (HR= 1.624), advanced histology (HR= 4.487), PRS type (HR= 2.123), high grade (HR= 2.883), chemotherapy (HR= 1.647), 1p19q co-deletion (HR= 0.231), and IDH mutation (HR= 0.317) (Figure 3a). In the CGGA microarray dataset, increased TROAP expression in glioma was significantly associated with poor prognosis (HR= 1.929), TCGA subtype (HR= 0.632), high grade (HR= 2.567), old age (HR= 1.736), chemotherapy (HR= 1.530), advanced histology (HR= 4.437), radiotherapy (HR= 0.459), PRS type (HR= 2.042), high grade (HR= 2.567) and IDH mutation (HR=0.423) (Figure 3c). In the TCGA RNA-seq dataset, overexpression of TROAP in glioma was significantly related to poor prognosis (HR=1.128), old age (HR=1.072) and high grade (HR=4.634) (Figure 3e).
Subsequently, we performed multivariate analysis with the Cox regression model. In the CGGA RNA-seq dataset, enhanced TROAP expression in glioma was closely linked with poor prognosis (HR = 1.217), chemotherapy (HR= 0.680), PRS type (HR = 1.994), old age (HR = 1.269), IDH mutation (HR=0.600), high grade (HR = 2.378), and 1p19q co-deletion (HR = 0.424) (Figure 3b). In the CGGA microarray, overexpression of TROAP in glioma was significantly linked with poor prognosis (HR = 1.420), radiotherapy (HR= 0.583), PRS type (HR=1.501) and high grade (HR = 2.666) (Figure 3d). In TCGA RNA-seq dataset, enhanced TROAP expression in glioma was importantly related to poor prognosis (HR = 1.060), high grade (HR = 2.888), and old age (HR = 1.048) (Figure 3f). These results reveal that high expression of TROAP might be a significant factor leading to unfavorable clinical outcomes.
Clinical diagnostic value of TROAP
Cox regression and the Kaplan-Meier method were used to plot ROC curves for the three datasets mentioned above to validate the clinical prognostic value of overexpression of TROAP in glioma. Overexpression of TROAP was confirmed to have diagnostic value (Figure 4a, b, c). The area under the curve was larger than 0.7 at 1,3,5 years, which indicated that high TROAP expression had modest prognostic value.
Correlations between TROAP expression and clinical characteristics in glioma patients
Software was used to carry out Wilcox or Kruskal tests to analyse the relationship between TROAP expression and different clinical characteristics from glioma samples from the three datasets (Figure 5a–g). As depicted in Figure 5a and c, the expression level of TROAP was significantly positively correlated with WHO grade and age in the CGGA RNA-seq, CGGA microarray and TCGA RNA-seq cohorts (p <0.005). As shown in Figure 5b and d, the expression level of TROAP was significantly correlated with 1p19q codeletion status and chemotherapy status in the CGGA RNA-seq cohort (p <0.001). In addition, Figure 5e, f and g demonstrate that the expression level of TROAP was significantly correlated with IDH mutation status, PRS-type and histology in the CGGA RNA-seq and CGGA microarray cohorts. These studies indicated that the expression level of TROAP was significantly associated with multiple clinical features related to the prognosis of glioma.
Identification of the signalling pathways affected by TROAP in glioma by GSEA.
Table 2. The gene set enriches the high TROAP expression phenotype.
|
CGGA RNA-seq
|
CGGA microarray
|
TCGA RNA-seq
|
Gene set name
|
NES
|
NOM p-value
|
FDR q-value
|
NES
|
NOM p-value
|
FDR q-value
|
NES
|
NOM p-value
|
FDR q-value
|
KEGG_CELL_CYCLE
|
2.033
|
0
|
0.018
|
1.955
|
0
|
0.056
|
2.157
|
0
|
0.007
|
KEGG_HOMOLOGOUS_RECOMBINATION
|
1.891
|
0
|
0.061
|
1.922
|
0
|
0.044
|
1.972
|
0
|
0.016
|
KEGG_P53_SIGNALING_PATHWAY
|
1.986
|
0
|
0.021
|
1.736
|
0.014
|
0.228
|
2.098
|
0
|
0.008
|
|
|
|
|
|
|
|
|
|
|
|
|
NES: normalized enrichment score; NOM: nominal; FDR: false discovery rate. Gene sets with NOM P-
value <0.05 and FDR q-value <0.25 were considered as significantly enriched.
By gene set enrichment analysis using low and high expression datasets, the signaling pathways affected by TROAP in glioma were identified, with significantly enriched pathways subjected to analysis via MSigDB (c2.cp.biocarta and h.all. v6.1. symbols). As shown in Table 2 and Figure 6a, b and c, the homologous recombination, cell cycle and p53 signaling pathways were identified in the high TROAP expression group. Therefore, TROAP may play a potential role in the carcinogenesis and progression of glioma.
TROAP protein expression was increased in glioma tissues.
We downloaded six immunohistochemistry (IHC) datasets (2 normal tissues, 2 low grade tissues, and 2 high grade tissues) generated with HPA044102 from the HPA online website and divided them into a male group (Figure 7a, b and c) and a female group (Figure 7d, e and f) to verify the differences in TROAP protein expression between normal tissues, low-grade tissues and high-grade tissues. The results showed that TROAP expression was upregulated in glioma tissues compared with non-tumor tissues.
The higher the grade of the tissue, the more protein was expressed.
Table 3. Four small molecule compounds identified as potential drugs for glioma treatment in CMap analysis.
CMap name
|
Mean
|
N
|
Enrichment
|
P-value
|
Bezafibrate
|
-0.599
|
4
|
-0.833
|
0.00143
|
Clobetasol
|
-0.668
|
3
|
-0.903
|
0.00172
|
CMap: connectivity map
Identification of four potential therapeutic drugs for the treatment of glioma based on CMap analysis.
To identify small molecule drugs that can inhibit the expression of TROAP to improve the OS of glioma patients, we screened the identified differential genes using R language, selected the top 10 positive and 10 negative genes according to their correlation values, and made a Venn diagram of the relationship between them (Figure 8a and b). Next, we uploaded these genes to CMap, which predicted potential drugs. Bezafibrate, clobetasol, scriptaid and thioguanosine were identified, as shown in Table 3. In addition, the two-dimensional and three-dimensional structures of these drugs were obtained from PubChem (Figure 9 a, b, c and d).