3.1 TMEM59L mRNA expression was elevated after irradiation.
Student's t-test was performed on the CGGA database and matched data to identify differentially expressed genes after radiotherapy (Figure-1 A and B). TMEM59L expression was significantly elevated both in recurrent GBM, which have received RT (P<0.001) and in the matched patients (P=0.04). The result was validated in virto experiments that TMEM59L was significantly up-regulated in the three irradiated cell lines(U87 2.0-fold at 8 hours P=0.04, LN229 2.9-fold p=0.019 and U251 1.7-fold P<0.01) compared with their parent cells. The expression level gradually increased after irradiation between 0 to 8 hours (Figure-1 C, D and E).
3.2.1 The high expression of TMEM59L was associated with a better prognosis.
Kaplan–Meier survival curves were constructed to determine the prognosis value to evaluate if the TMEM59L expression is associated with survival in glioma patients. TMEM59L expression level was divided into low/high groups according to the median value of TMEM59L mRNA expression in 325 patients. The results showed that high expression of TMEM59L was related to longer overall survival (OS) than low expression in the CGGA dataset (Figure-2). Multivariate Cox analyses showed that TMEM59L expression (HR: 0.667, 95% CI: 0.56-0.79, P <0.001) was an independent predictor of longer survival(Figure-2 A, C). Similar results were validated in 699 patients from the TCGA dataset (Figure-2 B, D). The results suggested that TMEM59L might be a novel independent prognostic biomarker for glioma patients.
3.2.2 TMEM59L expression level shows a subtype preference
We further invested the clinical prognostic significance of the TMEM59L in glioma.
The high-expression of TMEM59L was enriched in IDH mutant and MGMT methylated glioma (Table-1, Figure-3). The TMEM59L expression level was negatively correlated with tumor grade (Supplemental Figure-1)and significantly up-regulated in the favorable neural subtype.7 On the contrary, TMEM59L expression was down-regulated in the mesenchymal type with poor prognosis8. The ROC curves showed that the area under the curve (AUC) was up to 80.2% and 89.6% in the CGGA and TCGA sequencing dataset, respectively(Supplemental Figure-2).
Table 1 Clinical features of patients with glioma in CGGA and TCGA stratified by TMEM59L level
|
TCGA(699)
|
CGGA(325)
|
Characteristics
|
Low(349)
|
High(350)
|
Low(162)
|
High(163)
|
Gender
|
|
|
|
|
Male
|
192
|
176
|
106
|
97
|
Female
|
135
|
133
|
56
|
66
|
NA
|
22
|
41
|
0
|
0
|
Age(years)
|
|
|
|
|
≤40
|
101
|
145
|
57
|
86
|
>40
|
226
|
164
|
105
|
77
|
Na
|
22
|
41
|
0
|
0
|
Grade
|
|
|
|
|
2
|
76
|
147
|
27
|
76
|
3
|
110
|
135
|
36
|
43
|
4
|
141
|
27
|
99
|
44
|
Na
|
22
|
41
|
0
|
0
|
IDH mutation
|
|
|
|
|
Mutant
|
158
|
285
|
62
|
114
|
Wild type
|
184
|
62
|
100
|
49
|
NA
|
7
|
3
|
0
|
0
|
MGMT-status
|
|
|
|
|
Methylated
|
209
|
283
|
66
|
73
|
Unmethylated
|
109
|
59
|
75
|
42
|
NA
|
31
|
8
|
21
|
48
|
3.3 TMEM59L related biological process
Pearson correlation analysis was performed to investigate the biological process tightly associated with TMEM59L expression in the CGGA and TCGA sequencing datasets. Those genes tightly correlated with TMEM59L expression (Pearson |R| > 0.5,P < 0.05) in the CGGA and (Pearson |R| > 0.4, P < 0.05) in the TCGA were used for Gene ontology analysis with DAVID. Pearson correlation analysis was also applied to several classical immune checkpoints and GSCs related genes in CGGA and TCGA datasets.
It is shown that the negative correlation genes with TMEM59L expression are highly enriched in the immune and inflammatory response, cell proliferation and migration, apoptosis process, response to drug and DNA damage. The positive correlation genes tended to be enriched in biological processes that are normal and indispensable, such as neurotransmitter secretion and nervous system development. In addition, TMEM59L expression was negatively related to CD44, STAT3, IL6 and FUT4 and positively related to L1CAM(Supplemental Figure-3). TMEM59L expression was tightly related to the PD1 family, B7 family, LAG3 , TIM3, CTIL4 and IDO regarding the immune system.
KEGG pathway analysis revealed that the negatively related genes with TMEM59L expression were enriched in the PI3K‐AKT signaling pathway, the NF-kappa B, P53 signaling pathway and the positively related genes were enriched in the GABA ergic-synapas, Glutamatergic synapse. The two datasets shared all the results mentioned above (Figure-4).