FoxM1 inhibition sensitizes resistant glioblastoma cells to temozolomide by downregulating the expression of DNA-repair gene Rad51

Abstract

Purpose: Recurrent glioblastoma multiforme (GBM) is characterized by resistance to radiotherapy and chemotherapy and a poor clinical prognosis. In this study, we investigated the role of the oncogenic transcription factor FoxM1 in GBM cells' resistance to alkylator temozolomide (TMZ) and its potential molecular mechanism.

Experimental design: FoxM1 expression levels were measured by immunohistochemical analysis in 38 pairs of primary and recurrent GBM tumor samples. Expression levels were also measured in primary recurrent GBM cell lines, and their responses to TMZ were characterized. In a mechanistic study, an siRNA array was used to identify downstream genes, and a chromatin immunoprecipitation assay was used to confirm transcriptional regulation.

Results: Recurrent tumors that were TMZ resistant expressed higher levels of FoxM1 than did primary tumors. Recurrent GBM cell lines expressed higher levels of FoxM1 and the DNA damage repair gene Rad51 and were resistant to TMZ. TMZ treatment led to increased FoxM1 and Rad51 expression. FoxM1 knockdown inhibited Rad51 expression and sensitized recurrent GBM cells to TMZ cytotoxicity. FoxM1 directly regulated Rad51 expression through 2 FoxM1-specific binding sites in its promoter. Rad51 reexpression partially rescued TMZ resistance in FoxM1-knockdown recurrent GBM cells. A direct correlation between FoxM1 expression and Rad51 expression was evident in recurrent GBM tumor samples.

Conclusion: Targeting the FoxM1-Rad51 axis may be an effective method to reverse TMZ resistance in recurrent GBM.

Figure 1. FoxM1 and DNA repair genes are upregulated in recurrent GBMs

A, Immunohistochemical staining of FoxM1 in 38 pairs of GBM samples. Left panel, representative images of FoxM1 expression in primary and recurrent tumors. Right panel, scoring of FoxM1 expression levels in 38 pairs of primary and recurrent tumors. B and C, Two GBM sample pairs, including primary and recurrent tumors, were subjected to Western blot (B) and real-time RT-PCR analyses (C). The real-time PCR data are from three independent experiments. *p<0.05 and **p<0.001. D, Western blot analysis shows the expression levels of FoxM1, Rad51, BRCA2, Chk2, and MGMT in primary cultured cell lines from the paired primary and recurrent GBM1 samples.

Figure 2. Recurrent GBM cells are resistant to TMZ treatment by upregulating FoxM1 and Rad51

A, Cell viability assay shows the cell growth of primary and recurrent GBM1 cells after being treated with 40 μM TMZ for 1 to 5 days. DMSO was used as a treatment control. B, Colony formation assay of the primary and recurrent GBM1 cells after being treated with different concentrations of TMZ for 14 days. C, Western blot analysis shows the expression of FoxM1 and DNA repair genes after being treated with 40 μM TMZ for 1 to 5 days in primary and recurrent GBM1 cells. β-actin was used as a loading control. D, Immunofluorescence staining of Rad51 foci in primary and recurrent GBM1 cells. Percentage of cells with more than five Rad51 foci in 10 random microscopic fields was calculated after TMZ treatment for 5 days.

Figure 3. FoxM1 transcriptionally regulates Rad51 expression by directly binding to the promoter region

A, Western blot analysis reveals Rad51 protein levels in the two recurrent GBM cells stably expressing sh-FoxM1. ShRNA, with a sequence that targets no genes, was used as a negative control. B, Real-time PCR assay shows Rad51 mRNA levels in recurrent GBM1 and GBM2 cells expressing sh-FoxM1. Data were from three independent assays. **P<0.001. C, Dual luciferase assay shows the relative promoter activities of constructs harboring different fragments of the human Rad51 promoter in sh-FoxM1 or control recurrent GBM1 cells. The values were normalized to a Renilla transfection control. Three independent assays were performed. **P<0.001. D, ChIP assay shows the direct binding of FoxM1 to Rad51 promoter. Left panel, schematic illustration of four PCR-amplified fragments of Rad51 promoter; right panel, ChIP assays were performed using FoxM1 antibody in recurrent GBM1 and GBM2, and in U87 cells. IgG was used as a negative control.

Figure 4. FoxM1 binding sites in the Rad51 promoter are critical for Rad51 transactivation

A, Western blotting showing the knockdown of FoxM1 using specific siRNAs in recurrent GBM1 and GBM2 cells. SiRNA, with a sequence that does not target genes, was used as a negative control. B, Relative promoter activity of Rad51 after transfecting FoxM1 siRNAs in GBM1 and GBM2 cells. C, Schematic structure of the putative FoxM1 binding sites in Rad51 promoter. FoxM1-binding site sequences are shown in both wild-type (WT) and mutant (Mut) forms. D, Rad51 promoter activities of the promoter, with or without mutations, in the predicted FoxM1 binding sites. All experiments were repeated in triplicate, **P<0.001.

Figure 5. Silencing FoxM1 sensitized TMZ-induced cytotoxicity in recurrent GBM cells, and Rad51 overexpression rescued chemoresistance in FoxM1-silenced recurrent GBM cells

A, Left, Western blot analysis reveals different responses of DNA repair genes to TMZ in sh-FoxM1 and control recurrent GBM1 cells. Right, Western blot analysis reveals the expression of different DNA repair genes in recurrent GBM1 cells expressing sh-FoxM1, sh-Rad51, or sh-FoxM1 plus Rad51. B, Immunofluorescence staining of Rad51 foci in recurrent GBM1 cells expressing sh-FoxM1, sh-Rad51, or sh-FoxM1 plus Rad51. Percentage of cells with more than five Rad51 foci in 10 random microscopic fields was calculated. C, Flow cytometry analysis shows the percentage of G2 arrest in recurrent GBM1 stable cells expressing sh-FoxM1, sh-Rad51, or sh-FoxM1 plus Rad51 after TMZ treatment. D, Colony formation assay shows cell survival in recurrent GBM1 cells expressing sh-FoxM1, sh-Rad51, or sh-FoxM1 plus Rad51 treated by TMZ. All experiments were performed in triplicate, **P<0.001.

Figure 6. FoxM1 and Rad51 expression levels were highly correlated with each other and were predictive of poor prognosis in recurrent GBM

A, Immunofluorescence assay showing the co-localization of FoxM1 with Rad51 in recurrent glioblastoma specimens. Images shown are representative of 10 frozen glioblastoma specimens. Scale bar, 200 μM. B, Co-expression of FoxM1 and Rad51 in recurrent GBM surgical specimens. Left panel, representative image of immunohistochemical staining of FoxM1 and Rad51 in two recurrent GBM surgical specimens. Right panel, statistical analysis of FoxM1 and Rad51 expression correlation in 38 recurrent GBM specimens (r=0.77, P<0.001). C, Survival curves of recurrent GBM patients with high and low FoxM1 expression levels (P <0.01, log-rank test). D. Survival curves of recurrent GBM patients with high and low Rad51 expression levels (P <0.05, log-rank test).