MEK5-ERK5 Axis Promotes Self-renewal and Tumorigenicity of Glioma Stem Cells

Glioma stem cells (GSC) promote the malignancy of glioblastoma (GBM), the most lethal brain tumor. ERK5 belongs to the MAPK family. Here, we demonstrated that MAPK kinase 5 (MEK5)-ERK5-STAT3 pathway plays an essential role in maintaining GSC stemness and tumorigenicity by integrating genetic and pharmacologic manipulation and RNA sequencing analysis of clinical specimens. ERK5 was highly expressed and activated in GSCs. ERK5 silencing by short hairpin RNA in GSCs suppressed the self-renewal potential and GBM malignant growth concomitant with downregulation of STAT3 phosphorylation. Conversely, the activation of the MEK5-ERK5 pathway by introducing ERK5 or MEK5 resulted in increased GSC stemness. The introduction of STAT3 counteracted the GSC phenotypes by ERK5 silencing. Moreover, ERK5 expression and signaling are associated with poor prognosis in patients with GBM with high stem cell properties. Finally, pharmacologic inhibition of ERK5 significantly inhibited GSC self-renewal and GBM growth. Collectively, these findings uncover a crucial role of the MEK5-ERK5-STAT3 pathway in maintaining GSC phenotypes and GBM malignant growth, thereby providing a potential target for GSC-directed therapy. Significance: In this study, we demonstrated that MEK5-ERK5-STAT3 axis plays a critical role in maintaining stemness and tumorigenicity in GSCs by using genetic, pharmacologic, and bioinformatics tools, identifying the MEK5-ERK5-STAT3 axis as a potential target for GSC-directed therapy.


Introduction
Glioblastoma (GBM) is the most malignant and therapeutically intractable primary brain tumor and it accounts for more than 45% of malignant brain tumors (1). GBM contains a subpopulation of cells, glioma stem cells (GSC), which are at the hierarchal apex of differentiation status (2). GSCs display extensive self-renewal and multilineage differentiation. GSCs are strongly associated with enhanced tumorigenesis, metastasis, recurrence, and resistance to treat-oxidative stress, proinflammatory cytokines, and mitogens (11). Previous studies suggested that the MEK5-ERK5 axis is implicated in inflammation and cardiovascular diseases, such as atherosclerosis (12,13). Furthermore, dysregulated the MEK5-ERK5 axis is associated with tumor growth, invasion, metastasis, poor prognosis, and therapeutic resistance in various tumor types such as breast cancer, prostate cancer, colon cancer, and lung cancer (14)(15)(16)(17)(18)(19).
Moreover, recent studies revealed that pharmacologic and genetic inhibition of ERK5 signaling compromises cancer stem cell (CSC) phenotypes in leukemia and colon cancer (20,21), although the effects of ERK5 inhibition were not tested in vivo. These reports raise the possibility that ERK5 also plays a very important role in maintaining the stemness of GSCs. In addition, the functional role of the MEK5-ERK5 axis in GSCs stemness regulation and tumorigenesis remains unknown in vitro and in vivo, regardless of previous reports showing that ERK5 inhibition in differentiated glioma cells decreases tumor cell growth and reduces drug resistance (22).
Here, we aimed to determine whether MEK5-ERK5 axis in GSCs could be a novel therapeutic target by integrating genetic and pharmacological manipulation in GSCs and RNA sequencing (RNA-seq) analysis of clinical specimens.

HEK293T cells and HEK293GP cells were purchased from the RIKEN Cell
Bank and Takara Bio, respectively. These cells were cultured at 37°C in a 5% CO 2 incubator and maintained in DMEM supplemented with FBS. Human patient-derived GBM cell lines TGS-01 and TGS-04 were established as described previously (23). The use of these human materials and protocols were approved by the Ethics Committees of Gifu Pharmaceutical University (Gifu, Japan) and the University of Tokyo (Tokyo, Japan). These cells were confirmed as GSCs and cultured in neurosphere medium containing DMEM/F12 (FUJI-FILM Wako Pure Chemical) supplemented with recombinant human EGF at 20 ng/mL (FUJIFILM Wako Pure Chemical), recombinant human basic FGF at 20 ng/mL (FUJIFILM Wako Pure Chemical), B27 supplement without vitamin A (Gibco), and GlutaMAX (Gibco).

Orthotopic Xenograft Model of GSC-derived GBM and Histology
An orthotopic xenograft model of GSC-derived GBM was generated by transplantation of 5 × 10 4 TGS-01 GSCs into the brain of 4-week-old female nude mice (BALB/cSlc-nu/nu, SLC). Briefly, a small burr hole was drilled in the skull, 0.5 mm anterior and 2.0 mm lateral to the bregma with a microdrill, and dissociated cells were transplanted at a depth of 3 mm below the dura mater. For XMD8-92 pretreatment experiment, TGS-01 GSCs were treated with 5.0 μmol/L XMD8-92 for 24 hours, and XMD8-92 was subsequently washed off; 5 × 10 4 dissociated cells were transplanted into the brain of nude mice following XMD8-92 removal. Mice were sacrificed at the indicated timepoints or upon occurrence of neurologic symptoms. Mouse brains were fixed in a 4% paraformaldehyde solution, embedded in paraffin, and sectioned at a thickness of 5 μmol/L. Sections were stained with hematoxylin and eosin (H&E) and captured using a BZ-X810 fluorescence microscope (Keyence). The tumor area in each section was outlined and calculated using NIH ImageJ. Slide tumor volumes were calculated by multiplying tumor area with slice thickness of 5 μmol/L, and brain tumor volumes were approximated by summation of the slide tumor volumes for each animal. All animal experiments were approved by the Committees on Animal Experimentation of Gifu Pharmaceutical University (Gifu, Japan) and were performed in accordance with the guidelines for the care and use of laboratory animals. The number of animals used per experiment is stated in the figure legends.

Retroviral and Lentiviral Transfection
Vectors were transfected into HEK293GP cells or HEK293T cells using the calcium phosphate method. Virus supernatants were collected 48 hours after transfection and then cells were infected with virus supernatants for 24 hours in the presence of 2 μg/mL polybrene. Cells were then subjected to selection by culture with 1 μg/mL puromycin for 3 days before usage for experiments.

Tumor Sphere Formation Assay and In Vitro Limiting Dilution Assay
For sphere formation assay, cells were dissociated into single cells with StemPro Accutase (Gibco). Cells were then plated in 96-well Costar ultra-low attachment plates at 1 × 10 3 cells per well with neurosphere medium mixed with 1.0% methylcellulose. Tumorsphere number and size were measured on day 7. For in vitro limiting dilution assays, cells were plated in 96-well plate at 1, 5, 10, 20, 40, or 80 cells per well, with 10 replicates for each cell number. The presence of tumorspheres in each well was examined on day 7. Limiting dilution assay analysis was performed using online software (http://bioinf.wehi.edu.au/software/ elda/). Sphere formation was estimated by scoring the number of spheres larger than 50 μm.

Cell Viability Assay, Apoptosis Assay, and Migration Assay
Cells were dissociated into single cells with StemPro Accutase (Gibco). Cell number was then evaluated by trypan blue dye exclusion assay (0.04% in PBS) using an optical microscopy. Apoptosis assay was conducted using PE-Annexin V (BD Biosciences) and 7-Amino-Actinomycin D (BD Biosciences) by flow cytometry on Beckman CytoFLEX S (Beckman Coulter). The migration ability was evaluated by a wound healing assay. Cells were seeded and incubated until they reached 90%-100% confluence. A 100 μL pipette tip was used to make cross lines, and the debris was washed away with PBS. The areas of the wounds were imaged with an optical microscope at 0 and 24 hours and analyzed using ImageJ software (NIH).

Real-time Quantitative PCR
Total RNA was extracted from cells, followed by synthesis of cDNA with reverse transcriptase and oligo-dT primer. The cDNA samples were then used as templates for real-time PCR analysis, which was performed on an MX3005P instrument (Agilent Technologies), by using specific primers for each gene (Supplementary Table). Expression levels of the genes examined were normalized by using the GAPDH expression levels as an internal control for each sample.

Single-cell RNA-seq Data Analysis
We obtained the expression and annotation data (GSE84465) from Gene Expression Omnibus and analyzed them using "Seurat" package on R (ver. 4.0.2; ref. 24). For quality control, cells expressing less than 600 genes, more than 6,000 genes, or more than 100,000 counts were extracted in advance. Next, we performed sctransform normalization with default settings, principal component analysis (PCA), and uniform manifold approximation and projection. Then, we clustered the data and defined EGFR + cells as GBM cells, as per a previous study (24).
We also analyzed another single-cell RNA-seq (scRNA-seq) data reported in 2021 (25). The expression data were normalized by "NormalizeData" function in Seurat. Next, data scaling, PCA, and clustering were performed.
Single-sample gene set enrichment analysis (ssGSEA) and GSEA were performed using "GSVA" and "clusterProfiler" packages, respectively. Wilcoxon rank-sum test was performed for differential expression analysis using "presto" package.

Bulk RNA-seq Data Analysis
For comparing ERK expression between normal and GBM tissues, we analyzed the previous reported data (26). We also analyzed the data of GBM tissues (e.g., grade, MGMT methylation status) in The Cancer Genome Atlas (TCGA), the Chinese Glioma Genome Atlas (CGGA), and Rembrandt databases. Statistical significance was determined using Wilcoxon rank-sum test followed by Bonferroni correction. For correlation analysis, we calculated Pearson correlation coefficient.

Survival Analysis
Clinical data of patients with GBM were obtained from the CGGA database. Survival analysis was performed with log-rank test using survival package, and Kaplan-Meier curves were plotted by "survminer" package.

Statistical Analysis
Unless otherwise specified, Student t test and one-way ANOVA with post hoc Bonferroni test were used to calculate statistical significance. Throughout this study, P < 0.05 was considered statistically significant.

ERK5 is Associated with Stem Cell Phenotypes in GSCs
We first analyzed a scRNA-seq dataset of clinical GBM specimens to determine the properties of GSCs (24). We defined EGFR + cells as GBM cells as per a previous study (24). We then divided the GBM cell population into two clusters: CSC-signature high GBM cells and CSC-signature low GBM cells, based on a ssGSEA (Fig. 1A). We confirmed that gene sets involved in stemness and stem cell were significantly enriched in CSC-signature high GBM cells, allowing us to define these cells as the GSC population (Fig. 1B). CSC-signature high GBM cells significantly upregulated a gene panel involved in the MAPK signaling pathway (Fig. 1C). Moreover, gene expression profiling of the MAPK family showed that the expression level of ERK was significantly upregulated in CSC-signature high GBM cells compared with CSC-signature low GBM cells.
In contrast, expression levels of other MAPK members in the CSC-signature high GBM cells were indistinguishable from those in CSC-signature low GBM cells (Fig. 1D). Similarly, gene set involved in the ERK5 pathway was significantly enriched in CSC-signature high GBM cells, suggesting that ERK5 is related to stem cell properties in GSCs (Fig. 1E). We also analyzed different scRNA-seq datasets of clinical GBM specimens. We confirmed that the GSC pool was characterized by the elevated expression of HAFZ and significant upregulation of the stemness-related genes, and differentiated GBM cells showed high expression of canonical differentiation marker, GFAP in accordance with a previous report (ref. 25; Supplementary Fig. S1A-S1C). In this dataset, we also observed that GSCs significantly upregulated gene panels related to MAPK and ERK5 pathways (Supplementary Fig. S1D and S1E).
Furthermore, we compared the ERK5 expression level between GSCs and differentiated glioma cells. TGS-01 and TGS-04 cells, which are human GBM patient-derived GSCs, were cultured in neurosphere culture condition (for GSCs) or adherent culture condition (for differentiated glioma cells; Fig. 1F).
Consistent with results from our bioinformatics analysis, protein levels of ERK5 were increased in TGS-01 and TGS-04 GSCs, concomitant with higher levels of stem cell transcription factors, SOX2 and c-MYC (Fig. 1G). In addition, the nuclear abundance of ERK5 was also increased in TGS-01 and TGS-04 GSCs (Fig. 1H). Moreover, mRNA levels of SOX and c-MYC were significantly increased in TGS-01 and TGS-04 GSCs, along with a significant reduction of ERK (Fig. 1I). Thus, high expression levels of ERK5 were preferentially present in GSCs, suggesting that ERK5 might have a potential role in regulating the stem cell phenotypes of GSCs.

Disrupting ERK5 Impairs the Self-renewal Potential of GSCs In Vitro
We next proved the functional importance of ERK5 in the maintenance of GSCs in vitro by targeting ERK expression using lentiviral short hairpin RNA (shRNA; shERK) in TGS-01 and TGS-04 GSCs. ERK knockdown exhibited a marked reduction in protein levels of SOX2 and c-MYC in both TGS-01 and TGS-04 GSCs, along with a significant reduction in their mRNA levels ( Fig. 2A and B). ERK silencing remarkably reduced cell proliferation of both TGS-01 and TGS-04 GSCs (Fig. 2C). Annexin V staining showed that targeting ERK did not significantly alter cell apoptosis in both TGS-01 and TGS-04 GSCs  ( Fig. 2D). A tumorsphere formation assay indicated that silencing ERK significantly reduced GSC tumorsphere formation in both TGS-01 and TGS-04 GSCs (Fig. 2E). Furthermore, an in vitro dilution assay demonstrated that the self-renewal potential of GSCs was significantly impaired by ERK silencing in both TGS-01 and TGS-04 GSCs (Fig. 2F). In addition, ERK disruption decreased the migration potential significantly, as demonstrated by the wound healing assay in both TGS-01 and TGS-04 GSCs (Fig. 2G). These data suggest that ERK5 is required for the self-renewal potential and aggressiveness of GSCs in vitro.

Targeting ERK5 Abrogates the Tumorigenicity of GSCs In Vivo
Considering that ERK5 is required for cell growth, aggressiveness, and selfrenewal of GSCs, we next elucidated whether targeting ERK5 could affect the tumorigenic potential of GSCs in an orthotopic xenograft mouse model (Fig. 2H). Equal numbers of TGS-01 GSCs infected with shERK or shControl (shCtrl) were intracranially injected into immunocompromised mice. The mice inoculated with shERK infected TGS-01 GSCs had significantly longer survival than those inoculated with shCtrl infected TGS-01 GSCs (Fig. 2I). In addition, histologic examination demonstrated that the mice inoculated with shERKinfected TGS-01 GSCs displayed a significant reduction in intracranial tumor growth compared with shCtrl-infected TGS-01 GSCs (Fig. 2J). Attenuated tumor growth of GSCs was also observed by ERK knockdown using shRNA targeting different regions of the gene sequence ( Supplementary Fig. S2A and S2B). These results demonstrate that ERK5 contributes to maintenance of the tumorigenic capacity of GSCs in vivo.

The MEK5-ERK5 Axis Promotes the Self-renewal Potential of GSCs In Vitro
We next investigated the effect of the MEK5-ERK5 axis on the cell growth and self-renewal capacity of GSCs. ERK overexpression markedly upregulated protein levels of SOX2, c-MYC, and nuclear ERK5 in both TGS-01 and TGS-04 GSCs, along with a significant increase in their mRNA levels, while DN-ERK (the dominant negative form of ERK5 that lacks two phosphorylation sites for MEK5 and cannot phosphorylate target genes) overexpression markedly decreased SOX2, c-MYC, and nuclear ERK5 protein levels, concomitant with a significant reduction in their mRNA levels ( Fig. 3A and B; Supplementary  Fig. S3A). Contrary to the suppressive effect of ERK knockdown on GSC properties ( Fig. 2C and E), ERK overexpression significantly increased cell proliferation and tumorsphere formation in both TGS-01 and TGS-04 GSCs ( Fig. 3C and D). In contrast, the introduction of DN-ERK decreased cell proliferation and tumorsphere formation in both TGS-01 and TGS-04 GSCs ( Fig. 3C and D), consistent with our prior studies on ERK silencing GSCs ( Fig. 2C and E).
To further confirm the functional significance of the MEK5-ERK5 axis in the regulation of GSC stemness, constitutively active-MEK (CA-MEK) was overexpressed in both TGS-01 and TGS-04 GSCs (Fig. 3E). We confirmed that CA-MEK increased nuclear abundance of ERK5 in both TGS-01 and TGS-04 GSCs (Supplementary Fig. S3B). CA-MEK overexpression significantly increased cell proliferation and tumorsphere formation concomitant with an increased expression of SOX2 and c-MYC at protein and mRNA levels ( Fig. 3E-H). Thus, the MEK5-ERK5 axis could regulate the self-renewal potential of GSCs in vitro.

ERK5 Regulates the Self-renewal Potential of GSCs Partly Through STAT3
We next examined the molecular mechanisms of how ERK5 controls the maintenance of GSCs. We subclustered CSC-signature high GBM cells to identify ERK5-signaling high GSCs and ERK-expression high GSCs, based on a ssGSEA and the expression level of ERK, respectively (Fig. 4A). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway mapping revealed that a gene set involved in the JAK-STAT signaling pathway was significantly enriched in both ERK5-signaling high GSCs and ERK-expression high GSCs ( Fig. 4B and C).
STAT3, which is a transcriptional factor, is associated with poor prognosis in patients with GBM (27). Furthermore, previous studies reported that the STAT3 pathway is constitutively activated in GSCs, and the activation of STAT3 is critical for maintaining the stemness of GSCs (28). Moreover, a recent study showed that ERK deficiency reduced the level of STAT3 phosphorylation at Tyr705 (p-STAT3 Y705 ), which is the active form of STAT3, in macrophages (29). These reports suggest the possibility that STAT3 is a downstream factor of ERK5 regulating GSC stemness. Indeed, STAT3 pathway-related genes were significantly upregulated in both ERK5-signaling high GSCs and ERK-expression high GSCs ( Fig. 4D and E). Likewise, the analysis of different scRNA-seq dataset recapitulated the upregulation of STAT3 pathway-related genes in both ERK5-signaling high GSCs and ERK-expression high GSCs ( Supplementary Fig. S4A-S4C). Moreover, consistent with our bioinformatics analysis, we confirmed the reduction of STAT3 Y705 phosphorylation levels in ERK-silenced TGS-01 GSCs and DN-ERK-overexpressing TGS-01 GSCs (Fig. 4F). Meanwhile, ERK and CA-MEK overexpression increased STAT3 Y705 phosphorylation levels in TGS-01 GSCs ( Fig. 4G; Supplementary  Fig. S4D). STAT3 Y705 phosphorylation levels were also increased in TGS-01 GSCs compared with differentiated TGS-01 GBM cells ( Supplementary   Fig. S4E). Thus, STAT3 target genes were downregulated in ERK silenced TGS-01 GSCs (Fig. 4H). Furthermore, the introduction of CA-STAT restored cell proliferation and rescued the impaired tumorsphere formation induced by silencing of ERK in TGS-01 GSCs (Fig. 4I-K). These results indicate that ERK5 mediates the STAT3 pathway to regulate GSC self-renewal potential.

ERK5 is Associated with Poor Prognosis of Patients with GBM with High Stem Cell Properties
To determine the clinical relevance of our findings, we performed in silico studies on a publicly available clinical datasets in patients with GBM (26).
ERK expression and ERK5 signaling were significantly upregulated in GBM tissues compared with non-tumor brain tissues and were associated with increased glioma grade (Fig. 5A-F). ERK expression does not differ regardless of MGMT methylation status, IDH mutation status, age, and sex ( Supplementary   Fig. S5A-S5D and S5F-S5I). ERK expression was significantly downregulated in the neural subtype in TCGA database while no marked alterations of ERK expression were found among classical, mesenchymal, and proneural subtypes in the CGGA database ( Supplementary Fig. S5E and S5J). The Kaplan-Meier survival analysis demonstrated that patients with GBM with higher ERK expression had significantly shorter survival durations than those with low expression (Fig. 5G), which is consistent with a previous study. Likewise, patients with GBM with elevated ERK5 signaling also displayed a significantly short survival duration (Fig. 5H).
Previous studies indicated that ERK5 expressed by GSCs plays a critical role in maintaining stemness and tumorigenicity in GSCs; thus, we assessed whether AACRJournals.org Cancer Res Commun; 3(1) January 2023  ERK in patients with GBM harboring higher stem cell properties is associated with poor prognosis. We found that high expression of ERK and elevated ERK5 signaling were associated with poor prognosis in GSC signaling high patients (Fig. 5I). Furthermore, high expression of ERK and increased ERK5 signaling are associated with poor prognosis in patients harboring high expression of stemness markers such as SOX, c-MYC, PROM, FUT, and NES ( Fig. 5J and K). In contrast, we found that ERK expression and ERK5 signaling do not correlate with poor prognosis in patients with lower stem cell properties ( Supplementary Fig. S6A-S6C). The human GBM clinical data suggest that ERK5 expressed in GSCs, rather than differentiated GBM cells, could affect the survival of patients with GBM.

The Pharmacologic Inhibition of ERK5 Suppresses the Self-renewal Potential and Tumorigenicity of GSCs
Our studies on patient-derived GSCs in vitro and in vivo and GBM clinical specimens indicated that ERK5 expressed in GSCs could be a promising target for GBM therapy. For clinical application of our findings, we determined whether pharmacologic inhibition of ERK5 by XMD8-92, a small-molecule inhibitor of ERK5 (20,21,30), could suppress the self-renewal potential and tumorigenicity of GSCs; 5.0 μmol/L of XMD8-92 treatment decreased protein levels of SOX2 and c-MYC, concomitant with a reduction in the phosphorylation level of STAT3 Y705 in both TGS-01 and TGS-04 GSCs (Fig. 6A). Moreover, mRNA levels of SOX and c-MYC were significantly downregulated by XMD8-92 treat-ment at 5.0 μmol/L in both TGS-01 and TGS-04 GSCs (Fig. 6). XMD8-92 significantly decreased cell proliferation and tumorsphere number in a dosedependent manner over time in both TGS-01 and TGS-04 GSCs ( Fig. 6C and D). Furthermore, XMD8-92 treatment at 5.0 μmol/L significantly decreased the GSC self-renewal activity (Fig. 6E). No additional decrease in the tumorsphere number was observed following XMD8-92 treatment at 5.0 μmol/L in shERKinfected GSCs (Fig. 6F). XMD8-92 treatment at 5.0 μmol/L suppressed the GSC migratory capacity (Fig. 6G). In addition, treatment with a different ERK5 inhibitor, AX15836 (31) recapitulated the GSC phenotypes observed in XMD8-92 treatment-mediated ERK5 inhibition ( Supplementary Fig. S7A-S7E).
We examined whether XMD8-92 could disrupt tumor growth in GSC-derived orthotopic xenografts. The survival of mice inoculated with TGS-01 GSCs, transiently pretreated with 5.0 μmol/L of XMD8-92 was significantly prolonged, with significant inhibition of GSC-driven tumor growth ( Fig. 6H and I). These data demonstrate that XMD8-92 treatment effectively inhibits the self-renewal potential and tumorigenicity of GSCs, thus, exerting anti-GBM potential.

Discussion
Previous studies reported that STAT3 functions to regulate gene networks involved with maintaining stem cell phenotypes in CSCs (32). Moreover, STAT3 upregulates CSC markers such as CD44, leading to the promotion of CSC   properties (33). In addition, STAT3 plays an important role in controlling selfrenewal, survival, and stem cell transformation in GSCs (34,35). Aberrant STAT3 activation is preferentially present in GSCs relative to differentiated GBM cells, and its inhibition suppresses the self-renewal capacity of GSCs (36), suggesting that inhibition of the STAT3 signaling pathway may have significant therapeutic potential. However, the specific inhibition of STAT3 has been challenging in clinical use because STAT3 is required for fundamental cellular functions in normal cells (37). Thus, a comprehensive understanding of the molecular mechanism of STAT3 activation may provide new GSC-specific therapeutic targets to improve GBM treatment. In this study, we demonstrated that ERK5 mediates phosphorylation of STAT3 Y705 to exert its function in GSCs.
We found that genetic or pharmacologic inhibition of ERK5 decreases the level of STAT3 Y705 phosphorylation in GSCs to control stem cell phenotypes. However, how ERK5 modulates the phosphorylation of STAT3 Y705 in GSCs remains unclear and more mechanistic insights are needed. Because ERK5 is a serine/threonine kinase, ERK5 should indirectly increase the level of STAT3 Y705 phosphorylation through other signaling pathways. A recent study demonstrated that SOCS3, a classic negative regulator of the JAK2-STAT3 pathway, was upregulated by the inactivation of ERK5 (29). Moreover, ERK5 is required for sustained AKT pathway activation upon stimulation of PDGFR (43). These pathways regulate STAT3 Y705 phosphorylation in GSCs to control stem cell renewal potential (44,45). Thus, these pathways may be involved in the mechanism by which ERK5 modulates the phosphorylation of STAT3 Y705 to maintain stemness in GSCs. Meanwhile, the ectopic expression of STAT did not perfectly rescue the impairment of stem cell renewal activity in ERK-silenced GSCs. We could not exclude the possibility that additional pathways may also partly mediate the function of ERK5 in GSCs. Therefore, further studies are required to clarify the mechanisms by which ERK5 regulates the tumorigenesis of GSCs.
Beyond the role in inhibiting tumor initiation and progression, targeting ERK5 has the potential to augment the effectiveness of chemotherapy and radiotherapy. For example, a recent study demonstrated that temozolomide treatment combined with ERK5 inhibition leads to a significant increase in DNA damage in vitro experiments using non-stem GBM cells (22). Pharmacologic inhibition of ERK5 by XMD8-92 promotes the effect of 5-fluorouracil-based chemotherapy in colon cancer cells (17). Moreover, ERK5 inhibition sensitized cancer cells to radiotherapy by suppressing the ability to repair radiationinduced DNA damage in lung and prostate cancer cells (46). Given these reports and our findings presented here, it would be interesting to test whether ERK5 inhibition could synergize with chemotherapy and radiotherapy to disrupt GSC maintenance and overcome the therapeutic resistance of GSCs.
In conclusion, our studies identified the MEK5-ERK5-STAT3 axis as a critical regulator maintaining the self-renewal and tumorigenic potential of GSCs.
Moreover, we confirmed that pharmacological inhibition of ERK5 by XMD8-92 attenuated the self-renewal ability, tumorigenicity, and migratory capacity of GSCs (Fig. 6J). To our knowledge, this is the first preclinical study to ex-amine the functional role of ERK5 in CSC tumorigenesis in vivo. Our findings provide new insights into the molecular mechanism driving CSC proliferation and self-renewal. Our data also suggest that targeting the MEK5-ERK5-STAT3 axis could be an effective therapeutic approach against various cancers whose malignancies are connected to the stemness of CSCs.