Rnd2 was upregulated in human GBM.
Firstly, from public TCGA database, we studied the RND2’s expression profiles in diverse human cancers(http://cancergenome.nih.gov/). The results showed that the RND2 expression was evidently higher in glioblastomas among 14 categories of human cancers (Fig. 1A). Besides, we also analyzed RND2 expression profiles in diversified human tissues based on Human Protein Atlas database (http://www.proteinatlas.org/). We found that RND2 expressed obviously higher in brain and testis than other tissues (Fig.S1A). Furthermore, RND2 significantly upregulated in GBM compared with normal brain tissue according to TCGA database (Fig. 1B). Consistently, we also carried out the western blot and RT- PCR to analysis protein and the mRNA of RND2 from different organs of C57 mouse and found that RND2 were highly expressed in brain (Fig.S1B, C). Further we detected the expression of RND2 in three sub-class glioblastomas and it was found out that, compared with PN or CL GBM subtypes, the expression was at a significantly lower level in MES GBM (Fig. 1C). To determine whether the protein levels of RND2 elevated or not in clinical samples, we analyzed the expression of RND2 in human glioblastoma samples, including 14 normal brain tissue, 31 WHO grade II gliomas, 41 WHO grade III gliomas and 52 glioblastomas, we found that RND2 expressed significantly higher in gliomas compared with normal brain tissues (Fig. 1D, E). All the information of patients was listed in Supplementary TableS1. Furthermore, we found that mRNA level of RND2 was significantly increased in gliomas, no matter low-grade glioma or high-grade glioma (Fig. 1F). All these data suggested that RND2 was upregulated and acts as a potential oncogene in GBM, hence, we furtherly explored the function of RND2 in GBM.
RND2 knockdown induced GBM cell apoptosis in vitro
We conducted function studies to detect apoptosis in U87 and U251 cells in order to track the function of RND2 in GBM. Different shRNAs (shRND2-1 and shRND2-2) targeting RND2 were designed and efficacies of knockdown were ensured by western blot (Fig. 2I and Fig.S2F). Firstly, to evaluate cell death, we performed cleaved-caspase3 immuno-fluorescence and immune-blotting and found that RND2 knockdown groups expressed a significantly higher level of cleaved-caspase3 contrasted to control group both in U87 cells and U251 cells (Fig. 2A, B, Fig. 2I). It was known that one of the hallmarks events of the early stage of apoptosis was the loss of ΔΨm (16). The JC-1 staining manifested that knockdown of RND2 persuaded GBM cells’ loss of ΔΨm (Fig. 2C, D). Furthermore, we performed TUNEL staining and it was demonstrated by the result that the rate of TUNEL-positive cells elevated significantly in RND2 knockdown groups (Fig. 2E, F). Annexin V-PE/7-AAD staining which was detected through flow cytometry showed that RND2 knockdown decreased apoptosis both in U87 and U251 cells (Fig. 2G, H and Fig. S2A, B).
To further support these data, immunoblotting was also used to examine the expression of the BCL-2-associated X protein (BAX) when RND2 was up or down regulated in U87 and U251 cells. The result showed that the expression of BAX was decreased when RND2 was overexpressed in U87 and U251 cells, and reduced levels of RND2 resulted in the opposite effects (Fig. 2I, Fig S2C).
Collectively, all these data suggested the apoptosis of glioblastoma cells was negatively regulated by RND2 expression level.
RND2 overexpression reduced GBM cell autophagy in vitro.
Because autophagy was one of the most important mechanism of cell death, further, we want to explored whether RND2 played a key role in autophagy regulation in glioblastoma cells. First, we found that RND2 was negative with LC3B but positive with p62, which are both the markers of autophagy (17), in the same location of human glioblastoma samples by immunochemistry staining (Fig. 3A and Fig.S3C, D). Furthermore, we detected mRNA level of RND2 was negative with LC3B but positive with p62 by RT-PCR (Fig. S3A, B). In a word, all these data suggested that RND2 had a negative relationship with autophagy.
Further, we detected the effect of RND2 in glioblastoma cells by up or down regulated in U87 and U251 cells, and the LC3B immunofluorescence results showed that the cells overexpressing RND2 had lower fluorescence intensity (Fig. 3B and Fig.S2E). Additionally, GFP-LC3B expressed stably in U87 cells to facilitate the visualization of autophagy, and we found that the overexpression of RND2 inhibited GFP-LC3 puncta formation compared with control cells (Fig. 3C and Fig. S2D).
Besides, it was also showed that the amount of autophagic vacuoles per cell was evidently lowered in the RND2 overexpression group compared with control group (Fig. 3F, G). Moreover, the RND2 overexpression decreased LC3B and Beclin-1 levels but increased p62 levels (Fig. 3D), inversely, the proteins expressed totally different when RND2 knocked down (Fig. 3E).
In a summary, RND2 could inhibit autophagy inhibited cell death, as well as reduce both apoptotic and autophagic cell death.
RND2 Overexpression Reduced Cell apoptosis and Autophagy in an Intracranial Xenograft Model
We established U87 through stable overexpression of RND2 and constructed an intracranial model of xenograft for the purpose of investigating the potential effects of RND2 in vivo. In the beginning, we tested efficacy of the RND2 overexpression and the process was ensured by western blot and RT-PCR (Fig. S7A, B). Next, we created an intracranial xenograft model by implanting U87 cells intracranially. According to Kaplan-Meier curves, mice in the control group survived significantly longer than the RND2 overexpression group mice (Fig. 4A). Within expectation, the mice which were implanted with the RND2 overexpression U87 cells had bigger tumors than the mice with the U87 cells with control group (Fig. 4B, C), and the weight of the tumor in the RND2 overexpression group was significantly higher than that of the control group (Fig. 4D). Immuno-histochemical assays showed that RND2 led to higher expression levels of P62 and Bcl-2 and lower the expression levels of cleaved caspase-3, and LC3B (Fig. 6E). This results also proved that RND2 could weaken autophagy and apoptosis in vivo.
RND2 Interacted with p38 Physically and Inhibited p38 Phosphorylation.
Notch signaling, NF-kb signaling, p53 signaling and Snail1 signaling also played a key role in glioblastoma cells death (18–20), we explored the activity of the signaling by detecting the target gene expression level by RT-PCR, and result showed that there was no significant difference when RND2 was overexpressed (Fig. S4A-G). The interaction between proteins was the basis to active signaling pathway, therefore, we performed co-immunoprecipitation of MDM2, P53 and Snail1 with RND2 in GBM. But the result showed that there was no direct interaction between them (Fig. S4H-G).
It’s also reported in previous studies that the activated p38 MAPK signaling pathway could induce cell death in cancer(5). To determine whether RND2 regulated p38 MAPK signaling pathway, we detected the p-p38 and p38 levels. We found that p38 levels weren’t influenced by RND2, however, p-p38 was induced when RND2 was overexpressed in U87 and U251 cells (Fig. 5C, D and Fig.S5G), meanwhile increased when RND2 was knocked down (Fig. 5C and Fig.S5G). As is known that p38 phosphorylation is an indicator of p38 MAPK signal activation (21), so RND2 could reduce p38 MAPK signal activation in glioblastoma cells.
In order to detect how RND2 decrease p-p38 expression level, co-immunoprecipitation and immunofluorescence assays, which could ensure the association between proteins, was used to explore the potential mechanism. Firstly, the subcellular localization of endogenous RND2 in GBM cells were examined by immunofluorescence and the results showed that the endogenous RND2 was expressed not only mainly in cytoplasm but also in cellular membrane (Supplementary Fig.S5E). Furthermore, the co-localization of RND2 and p38 in GBM patient tissues was observed mainly existing in the cytoplasm by immunofluorescence (Fig. S5A). Consistently, these results of co-immunoprecipitation assays in U87 cells determined the physical interaction between RND2 and p38 in U87 and in U251 cells (Fig. 5B and Fig. S5D). Distribution of p38 were found mainly in cytoplasm and were detected in nucleus as well, but in U87 cells, we found that the p38 co-localized with RND2 was in cytoplasm (Fig. 5A and Fig. S5B, C), which was in accordance with our former results. Notably, the overexpression of RND2 resulted in a huge decrease in the levels of nuclear p-p38 (Fig. S5E). In summary, these results indicated that activity of p38 MAPK signaling pathway was downregulated by RND2 and then p38 regulated its own downstream target genome.
p38 induced autophagy in GBM cells and rescued RND2-mediated autophagy and apoptosis.
p38 MAPK signaling pathway can regulate autophagy and cell death, however, its regulation of autophagy response sometimes has both-side regulation, including positive and negative (11, 22). Therefore, in the next step, we are going to explore the role of p38 MAPK signaling in autophagy of GBM cells. BIRB796, the inhibitor of phosphorylation of p38, was used block the activity of p38 MAPK signaling and the result showed that the cells had lower fluorescence intensity when cells were treated by BIRB796 (Fig. 6A, B). Besides, it was also showed that the amount of autophagic vacuoles per cell was evidently lowered in BIRB796 group compared with DMSO group (Fig. 6C, D). The western blot results showed BIRB796 decreased LC3B and Beclin-1 levels but increased p62 levels (Fig. 6E). All these data demonstrated that p38 MAPK signaling increased autophagy in human glioblastoma cells.
As demonstrated previously, RND2 was able to target p38 MAPK signaling pathway in GBM cells, furtherly, we also speculated that p38 could be involved in RND2-mediated autophagy and apoptosis. The number of autophagic vacuoles in one cell on average was calculated through TEM and it was found that the number soared in U87 cells caused by overexpressed p38, which could rescue the inhibition effects resulted from RND2 (Fig. 7A, B). Furthermore, we detected the autophagy-mediated protein and found that overexpressing p38 could rescue the inhibition of autophagy caused by RND2 (Fig. 7C). And 3-MA, the inhibitor of autophagy, decreased apoptosis caused by knockdown of RND2 in U87-MG (Fig. 7E, F) and the relative apoptosis rate decreased significantly between DMSO and 3-MA group (Fig. 7G). Additionally, western blotting analysis results revealed that 3-MA was significantly decreased the RND2-mediated GBM cells apoptosis (Fig. 7D).
In summary, all of the results demonstrated that RND2 could inhibit autophagy and decrease apoptosis through inhibiting p38 MAPK signaling pathway. Furthermore, RND2 associated autophagy inhibiting was part reason of GBM cells apoptosis resistance.
RND2 Predicted Poor Prognosis of Patients Suffering GBM.
In order to further evaluate RND2’s role in clinical samples, we performed RT-PCR assays to examine levels of the RND2 expression as shown before (Fig. 1F) in tumor tissues sampled from patients suffering primary GBM. In the experiment, we identified low-expression and high-expression of RND2 by RND2 mRNA expression levels, and we carry out Kaplan-Meier analysis (Table 1), which all the information is mentioned in TableS2. Kaplan-Meier curves showed that the patients in low-expression group survived significantly longer than the patients in the RND2 high-expression group (Fig. 8A). Additionally, the expression of RND2 had a positive correlation with tumor volume (Fig. 8B).
In a word, all these data revealed that RND2 could be defined as a biomarker of glioblastomas and would indicate the poor prognosis confronting with GBM patients.