Positive TAZ-TDG feedback loop drives proneural-to-mesenchymal transition to promote glioblastoma progression

Abstract

Background

Glioblastoma multiforme (GBM) is the most prevalent malignant tumor of the central nervous system, featuring high heterogeneity and invasiveness. Proneural-mesenchymal transition (PMT) has garnered extensive attention as a crucial driving mechanism for glioblastoma invasion and progression; however, its molecular regulatory network remains ill-defined.

Methods

The expression and prognostic significance of TAZ were analyzed using bioinformatics tools. The function of TAZ was evaluated using in vitro and in vivo experiments. MeDIP and McrBC analyses were used to investigate the effect of TDG on methylation levels in the WWTR1 promoter region. ChIP-qPCR and Luciferase reporter assays were used to evaluate the binding affinity of TEAD4 to the TDG promoter.

Results

Aberrantly expressed TAZ promotes PMT in GBM cells. Mechanistically, the high expression of TAZ is caused by thymine DNA glycosylase (TDG), which maintains the low methylation state of its promoter region. TDG is transcriptionally regulated by TAZ-TEAD4, forming a TAZ-TDG positive feedback loop that promotes PMT. Importantly, we confirmed the significant inhibitory effect of dual targeting of TAZ and TDG, both in vitro and in vivo.

Conclusion

Our data strongly indicate that the self-amplifying TDG-TAZ loop represents a previously unrecognized key participant in the progression of glioblastoma (GBM) via PMT, suggesting that TAZ and TDG may serve as potential therapeutic targets for GBM.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12967-026-08504-4.

Introduction

Glioblastoma (GBM) is classified into three distinct molecular subtypes: proneural (PN), mesenchymal (MES), and classical (CL) [1]. Proneural-mesenchymal transition (PMT) makes tumor cells more invasive, promotes tumor infiltration into the surrounding brain tissue, and increases the risk of recurrence [2, 3]. As GBM cells with a mesenchymal phenotype exhibit enhanced cell adhesion and invasion capabilities, patients with GBM of the mesenchymal subtype have a worse prognosis [4]. Therefore, PMT is a key factor in the malignant progression and recurrence of GBM [5]. Nevertheless, current research on the molecular regulatory network of PMT remains imperfect, and its key drivers and regulatory mechanisms require further investigation.

As a transcriptional coactivator, TAZ does not possess intrinsic ability to bind directly to DNA. TEAD, a transcription factor, interacts with MCAT elements (such as GGAATG) located within the promoter regions of specific genes via its DNA-binding domains. This interaction facilitates the formation of the TAZ-TEAD complex, which targets and regulates the transcription of downstream genes. This cooperative mechanism is fundamental to the role of TAZ in transcriptional regulation [6–8]. TAZ directly drives the PMT process in glioblastoma by activating TEAD-dependent transcriptional programs and integrating multiple signaling pathways (such as BMP, MAPK, and WNT), leading to enhanced invasiveness and treatment resistance [9–12].

Thymine DNA glycosylase (TDG) is a base excision repair enzyme that participates in DNA damage repair and plays a critical role in the epigenetic regulation of DNA demethylation [13]. Specifically, TDG maintains the unmethylated status of CpG islands downstream of the ten-eleven translocation (TET) family [14]. TDG initiates gene expression by replacing the two oxidized forms of 5mC, namely 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC), with unmodified cytosine [13, 15, 16]. In recent years, the role of TDG in cancer has gradually been uncovered, as it can regulate the expression of multiple genes related to tumorigenesis through demethylation [17, 18]. TDG is overexpressed in gliomas, and its expression level is significantly correlated with poor patient prognosis [19]. TDG directly promotes the malignant mesenchymal phenotype in tumors, such as gliomas, through epigenetic regulation and SUMOylation modification [20].

In this study, we examined the aberrant expression of TAZ in GBM cell lines and tissue samples. TAZ exhibited a significant correlation with the proliferation, migration, and invasion of GBM cells and PMT both in vitro and in vivo. Mechanistically, TAZ is regulated by TDG-mediated demethylation. Simultaneously, TDG is positively transcriptionally regulated by TAZ-TEAD, forming a self-amplifying loop that promotes PMT progression. Importantly, the best tumor suppression was observed in vitro and in vivo when both TAZ and TDG were simultaneously knocked out or inhibited (TAZ inhibitor: Verteporfin; TDG inhibitor: Juglone). In conclusion, our findings indicate that TAZ and TDG are potential therapeutic targets for clinical intervention in patients with GBM.

Materials and methods

Clinical samples acquisition

GBM tissues were collected from patients at the Department of Neurosurgery at The Affiliated Drum Tower Hospital of Nanjing University Medical School (Nanjing, Jiangsu, China). The research protocol was approved by the Institutional Review Board and the Ethics Committee of Affiliated Drum Tower Hospital of Nanjing University Medical School, and the collection and processing of patient samples were conducted in accordance with the approved guidelines. All patients provided informed consent.

Generation and culture of primary GBM cells

Primary GBM cell lines were isolated from post-surgical specimens of patients with GBM. Briefly, fresh tumor tissues obtained from patients with GBM were meticulously dissected into small pieces and subjected to five consecutive washes with cold phosphate-buffered saline (PBS). Subsequently, the tissues were enzymatically dissociated and incubated with trypsin (Gibco, USA) at 37 °C for 15 min. Repeated pipetting was performed to ensure thorough dissociation, and the resulting single-cell suspension was sequentially filtered through sterile 70 μm and 40 μm strainers (BD Falcon, USA) and recognized as primary GBM cells. Transcriptome sequencing (mRNA-seq), single-sample gene set enrichment analysis (ssGSEA), and western blotting were used to definitively classify the primary cell subtypes. Moreover, the cells were incubated with CD44 or SOX2 antibodies at 4 °C for 30–60 min followed by fluorescence-activated cell sorting (FACS) to isolate PN cells (PN20 and PN24) and MES cells (MES50 and MES52). The methods were performed as previously described [21]. The cells were authenticated using STR assays. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) or DMEM/F12 (Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, USA) and 1% penicillin-streptomycin solution (Solarbio, China) at 37 °C with 5% CO2 and were negative for mycoplasma contamination.

Plasmid construction and cell transfection

Overexpression and knockdown plasmids (TAZ and TDG, respectively) were designed and constructed by GeneChem (Shanghai, China). Small interfering RNA (siRNAs) and the corresponding negative control (NC) siRNAs were procured from Sangon Biotech (Shanghai, China). siRNAs (siLATS1/2 and siMST1/2) were transfected into the cells using the Lipofectamine 3000 transfection reagent, following the manufacturer’s protocols. The siRNAs were validated using western blotting. Polybrene was used to enhance the transfection efficiency. Stably transfected cell lines were established by adding puromycin (2 µg/mL) for screening. The corresponding sequence information is provided in Supplementary Table S1.

Cell counting kit-8 (CCK8) and colony formation assay

For the CCK8 assay, cells transfected with either the knockdown or overexpression plasmid were seeded into 96-well plates at appropriate densities and incubated for 24, 48, 72, or 96 h. Subsequently, 10 µL of CCK-8 solution (Glpbio, USA) was added to each well containing 100 µL of fresh medium, and the plates were incubated at 37 °C with 5% CO2 for 1 h. Absorbance values were subsequently measured at 450 nm using an enzyme-labeling device, and the data were normalized to that the of blank control.

For colony formation assays, 800 cells were seeded in 6-well plates and cultured for 1–2 weeks until visible colonies formed. The colonies were then fixed with 4% paraformaldehyde for 30 min and stained with 2.5% crystal violet solution for 15 min. Finally, the number of colonies was quantified using the ImageJ software.

Wound healing and transwell assays

For the transwell migration assay, cells (1 × 10⁴) were seeded in the upper chamber of a 24-well plate (Corning, USA) with 200 µL of serum-free DMEM per well. Next, 500 µL of DMEM supplemented with 10% FBS was added to the lower chamber. Following a co-culture period of 24–48 h, non-migrated cells on the lower surface of the upper chamber were gently removed using a cotton swab. The remaining cells were fixed with 4% paraformaldehyde for 15 min and stained with 2.5% crystal violet solution for another 15 min. Finally, the migrated cells that adhered to the lower surface of the membrane were imaged and counted using a microscope.

For the wound-healing assay, cells were seeded in a 6-well plate. Once confluence exceeded 90%, a straight-line scratch was made across the cell monolayer using a 200 µL pipette tip. The cells were subsequently washed with phosphate-buffered saline (PBS) to remove detached cell debris and cultured in serum-free DMEM. Images of the wound areas were captured under a microscope at 12- and 24-hours post-scratch. The percentage of wound closure was calculated as (wound area at 0 h-wound area at 48–72 h)/wound area at 0 h × 100.

Total RNA isolation and quantitative real-time PCR (RT-qPCR)

Total RNA was extracted using TRIzol reagent (Invitrogen, USA) according to the manufacturer instructions. RNA concentration was quantified using a Nanodrop 60-Touch spectrophotometer (Implen, Germany). Subsequently, total RNA (1 µg) was used as a template for cDNA synthesis using the M5 Sprint qPCR RT kit and gDNA remover (Mei5bio, China). Quantitative RT-PCR was conducted on the qtower3g PCR system (Analytik Jena, Germany), with GAPDH serving as the reference gene for mRNA expression analysis. The primer sequences used in this study are listed in Supplemental Table S2.

Western blot analysis (WB)

To extract total protein, RIPA lysis buffer (Solarbio, China) was used, and the protein concentration was quantified using a BCA protein assay kit (Beyotime, China). A total of 4 µg of protein lysate was loaded onto an SDS-PAGE gel for separation, followed by transfer to a PVDF membrane. The membrane was blocked with 5% skim milk at room temperature for 2 h and incubated with primary antibodies overnight at 4 °C. The following day, the membrane was incubated with horseradish peroxidase-conjugated secondary antibody (Proteintech, Wuhan, China). Finally, the protein bands on the membrane were visualized using a GV6000M imaging system (GelView6000pro). The details of the primary antibodies used in this study are provided in Supplementary Table S3.

Immunofluorescence (IF)

The cells were seeded on glass coverslips in a 6-well plate. After 24 h, the cells were fixed with 4% paraformaldehyde for 15 min, followed by permeabilization with 0.3% Triton X-100 for 10 min. The cells were then blocked with 5% bovine serum albumin (BSA) at room temperature for 2 h. Subsequently, the cells were incubated overnight at 4 °C with primary antibodies diluted in 5% goat serum. The following day, the cells were incubated with Alexa Fluor 488-or Alexa Fluor 647-conjugated secondary antibodies diluted in 5% goat serum in PBS for 2 h in the dark. Finally, the nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI, Solarbio). Immunofluorescence images were acquired using a laser-scanning confocal microscope (Leica TCS SPE).

Immunohistochemical assay (IHC)

Briefly, paraffin-embedded tissue specimens were sliced continuously to a thickness of 4 μm. Formalin-fixed, paraffin-embedded tissue sections were incubated at 80 °C for 15 min. Subsequently, the sections were dewaxed in xylene, rinsed with graded ethanol, and rehydrated with double-distilled water. For antigen retrieval, the slides were pretreated by steaming in sodium citrate buffer at 95 °C for 15 min. The samples were incubated with primary antibodies at 4 °C overnight. After washing with PBS, the slices were incubated with the corresponding secondary antibody. A DAB staining kit (Soleibao, China) was used for the analysis.

Dual-luciferase reporter assay

The binding sites of TEAD4 and TDG gene promoters were predicted using the UCSC (http://www.genome.ucsc.edu/) and JASPAR (http://jaspar.genereg. net) online platforms. We selected Site#1–3 with the highest credibility to design Region#1 (644–851), Region#2 (1097–1341), and Region#3 (1740–1924) using Snapgene software. To verify the direct interaction between TAZ-TEAD4 and the predicted sequences on the TDG promoter, partial promoter sequences of TDG (Region#1, Region#2, and Region#3), as well as the wild-type full-length sequence, were cloned into the pGL3-Basic (Promega) reporter plasmids. Wild-type or mutant plasmids were transfected into cells that were pre-seeded in 24-well plates using Lipofectamine 3000, according to the manufacturer’s protocol. Luciferase activity in each well was measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA), and the relative luciferase activity was normalized to Renilla luciferase activity.

DNA methylation analysis

The UALCAN tool (https://ualcan.path.uab.edu/cgi-bin/ualcan-res.pl) was used to analyze the methylation levels of the TAZ promoter region. Detailed usage methods can be found in previous reports [22]. The CpG island of the WWTR1 gene was predicted, and primers were designed using MethPrimer (http://www.urogene.org/methprimer/). Genomic DNA was incubated with McrBC enzyme (New England Biolabs) at 37 °C for 2 h to cleave methylated DNA. Subsequently, RT-qPCR was performed using specific primers. Finally, the PCR products were subjected to semi-quantitative analysis using 1.5% agarose gel electrophoresis.

Methylated DNA immunoprecipitation (MeDIP)

MeDIP analysis was conducted according to the manufacturer’s protocol (Abcam, Cambridge, USA). Total genomic DNA was subjected to sonication and denatured by heating to yield single-stranded DNA. Subsequently, 1 µL of anti-5caC or 5mC antibody was used for immunoprecipitation at room temperature for 2 h, with normal rabbit IgG serving as the negative control. DNA was liberated from the DNA/antibody complex by incubating with proteinase K at 65 °C for 1 h. The target DNA was captured, eluted, and used for real-time quantitative PCR detection. The sequences of the PCR primers are listed in Supplementary Table S4.

Methylation-specific PCR (MSP)

Genomic DNA was treated with sodium bisulfite and DNA methylation was tested by MSP using EZ DNA Methylation-Direct kit (Zymo Research). Two primer groups were used to amplify the promoter region of WWTR1 containing multiple CpG sites, and the primer sequences were designed using MethPrimer (http://www.urogene.org/methprimer/).

Chromatin immunoprecipitation (ChIP) assay

ChIP assays were performed using the SimpleChIP^® Enzymatic Chromatin Immunoprecipitation Kit (Cell Signaling Technology). GBM cells were cross-linked with 1% formaldehyde at room temperature for 10 min. Subsequently, cross-linked chromatin DNA was disrupted by sonication. Chromatin immunoprecipitation was executed using 4 µg of antibody and protein G magnetic beads, and the samples were incubated overnight at 4 °C with rotation. Normal rabbit immunoglobulin G (IgG) was used as a negative control. Finally, the protein G magnetic beads were washed, chromatin was eluted from the antibody/protein G magnetic bead complex, and the cross-links were reversed. The precipitated DNA was purified and analyzed using qPCR. The primer sequences used for ChIP-qPCR are listed in Supplementary Table S5.

Intracranial injection mouse model

Thymus-deficient nude mice (BALB/c) and NSG mice aged 6–8 weeks were procured from the Animal Center of SPF (Beijing) Biotechnology Co., Ltd. and reared in a specific pathogen-free environment at 24 °C. The experimental animals were randomly grouped into 10 mice per group. Tumor cells expressing luciferase (3 × 10^5 cells suspended in 8µL PBS) were injected into the right striatum using a digital stereotactic apparatus. On the 7th, 14th, 21st, and 28th days post-injection, bioluminescence imaging of intracranial tumors was conducted via intraperitoneal injection of D-luciferin potassium salt, and the intensity of the fluorescence signal was quantitatively analyzed. Mice were euthanized when they exhibited neurological symptoms. Subsequently, brain samples were collected, fixed in 4% paraformaldehyde, paraffin-embedded, and prepared for IHC. The survival rates of the mice were recorded, and Kaplan-Meier survival curves were constructed to analyze their survival status. All experiments were performed in strict compliance with the stipulations of the “Animal Welfare Law” and were approved by the Animal Ethics Committee.

Statistical analysis

All data were analyzed using GraphPad Prism 9.0 and are presented as the mean ± SEM. All experiments were independently replicated at least thrice. Student’s t-test was used for comparisons between two groups, whereas one-way analysis of variance (ANOVA) was used for comparisons between at least three groups. Overall survival curves were used to illustrate the survival distributions, and the log-rank test was used to evaluate the statistical significance between the various groups. Statistical significance was defined as P < 0.05 (ns indicates non-significance; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Results

TAZ is upregulated in GBM and correlates with poor prognosis

TAZ (also known as WWTR1), a core transcriptional coactivator of the Hippo pathway, is regulated by multiple modifications (phosphorylation, acetylation, and methylation) and non-coding RNAs, making it a potential target for glioma therapy [23]. Pan-cancer analysis revealed that TAZ was highly overexpressed in various malignant tumors, including GBM (Figure S1A). Transcriptome sequencing suggested that TAZ was highly expressed in the MES subtype in the TCGA and CGGA databases (Fig. 1A). Similarly, single-cell transcriptome sequencing suggested that TAZ was highly enriched in the MES subtype (Fig. 1B). These data suggest that TAZ is closely associated with the MES phenotype. In addition, TAZ expression levels were positively correlated with glioma grade (Figures S1B-C). To verify these findings, we performed TAZ protein detection and immunohistochemical analyses on glioma tissue specimens of varying grades. The results indicated that TAZ protein expression was significantly upregulated compared to that in adjacent tissues and exhibited a positive correlation with tumor grade (Fig. 1C-E). In previous studies, we isolated primary GBM cells (PN20, PN24, MES50, and MES52) from post-surgical GBM tissues [21]. The results showed that TAZ was expressed at higher levels in MES50 and MES52 cells than in PN20 and PN24 cells (Fig. 1F). Survival analysis indicated that elevated TAZ expression was associated with a reduced overall survival rate in patients with GBM (Fig. 1G). Taken together, these findings indicate that TAZ is upregulated in glioma tissues and correlates with malignant clinicopathological features, suggesting its pivotal role in glioma progression.

Fig. 1.

TAZ is upregulated in GBM and correlates with poor prognosis. (A) TCGA and CGGA databases revealed that the expression of WWTR1 varied across different molecular subtypes of glioblastoma (GBM). (B) Scatter plots of the single-cell dataset (GSE131928) illustrating the distribution of WWTR1 expression across various cell types in GBM. (C-D) IHC staining of TAZ in gliomas of different grades and adjacent normal tissues (C). The IHC scores of TAZ were quantified (n = 6). Scale bars, 50 μm. Data are mean ± SEM. NS, non-significant; ***P < 0.001. Two-tailed unpaired Student’s t-test (D). (E) Protein expression of TAZ in different grades of glioma tissues (grade I, II, III, and IV). (F) Western blotting was used to detect the protein expression of TAZ in primary cells (PN20, PN24, MES50, and MES52). (G) Kaplan-Meier curve demonstrating a significant association between elevated TAZ expression levels and worsened overall survival (OS) in the TCGA and CGGA databases (http://gliovis.bioinfo.cnio.es/)

TAZ enhances the growth and motility of GBM both in vivo and in vitro

Previous studies have shown that abnormal TAZ expression is involved in the carcinogenesis of many types of solid tumors [24–26]. Therefore, we further explored the tumor-promoting effects of TAZ in glioma. We established stable TAZ knockdown and overexpression cell lines using patient-derived GBM cells (MES50, MES52, PN20, and PN24) (Figure S2A-B). The results of the clone formation and CCK-8 assays showed that TAZ knockdown significantly weakened the proliferation of GBM cells, whereas TAZ overexpression produced the opposite effect (Fig. 2A-B and Figure S2E-F). Transwell and wound-healing assays further demonstrated that downregulation of TAZ expression significantly inhibited the invasion and migration abilities of GBM cells, whereas overexpression of TAZ effectively promoted these processes (Fig. 2C-D and Figure S2G-H). Next, an orthotopic intracranial xenograft tumor model was established in BALB/c-nu mice (Fig. 2E), and the results showed that TAZ overexpression accelerated tumor growth and significantly shortened survival in the nude mice. In contrast, TAZ knockdown restricted tumor growth and prolonged survival (Fig. 2F-K). In summary, our results elucidate the core regulatory role of TAZ in the malignant biological behavior of GBM in vivo and in vitro.

Fig. 2.

TAZ enhances the growth and motility of GBM both in vivo and in vitro. (A-B) Effect of TAZ changes on the colony-forming ability of MES and PN cells. (C-D) The invasive ability of TAZ-overexpressing and defective MES/PN cells was evaluated using the Transwell assay. (E) Formation of an intracranial orthotopic xenograft mouse model. (F-I) Bioluminescent imaging analysis and representative H&E staining of tumor growth in GBM xenograft mice derived from MES50 (F) or PN20 (G) cells. Representative images are shown at 7, 14, and 21 days after transplantation. (G and J) The recorded total flux at each time point reflects the dynamic growth of the xenograft tumor size in each group of nude mice. ***P < 0.001. (H and K) Kaplan-Meier survival curves of mice implanted with MES50 (H) or PN20 (K) cells are shown. Survival curves were analyzed using a two-tailed log-rank test (n = 10 per group). Representative images and quantitative histograms are shown. Data are mean ± SEM. **P < 0.01 and ***P < 0.001. Two-tailed unpaired Student’s t-test (A, B, C, D, G, H, J, and K)

TAZ expression was positively correlated with mesenchymal features in GBM

In 2021, Hara et al. systematically defined three molecular subtypes of GBM [27]: proneural (PN), classical (CL), and mesenchymal (MES). The MES subtype exhibited the strongest invasiveness and treatment resistance. In this context, the phenotypic transition (PMT) from proneural to mesenchymal is a key mechanism of malignant progression of GBM and is closely related to enhanced tumor heterogeneity and increased invasiveness [28, 29]. Our previous studies confirmed that TAZ is a key transcription factor that promotes PMT [21]. Importantly, GSEA enrichment analysis showed that TAZ expression was significantly positively correlated with the signature genes of the MES subtype but negatively correlated with the PN subtype (Fig. 3A). In addition, immunofluorescence analysis confirmed that knockdown of TAZ significantly inhibited the mesenchymal marker CD44 while increasing the proneuron factor OLIG2, whereas overexpression of TAZ showed a completely opposite regulatory pattern (Fig. 3B-C and Figure S3A-B), thus inferring that TAZ is involved in PMT regulation. Considering that TAZ serves as the ultimate effector molecule of the Hippo pathway, we introduced the specific inhibitors TRULI [30] (an LATS1/2 inhibitor) and XMU-MP-1 [31] (an MST1/2 inhibitor) to develop a model for assessing the impact of Hippo pathway inactivation on PMT (Figure S3C-F). Western blotting and RT-qPCR showed that Hippo pathway inactivation (TAZ overexpression) upregulated the expression of mesenchymal markers (CD44 and YKL-40) and downregulated the expression of proneuronal markers (OLIG2 and ASCL1) (Fig. 3D-G). These findings systematically demonstrate the important role of TAZ in driving PMT progression in GBM.

Fig. 3.

TAZ expression was positively correlated with mesenchymal features in GBM. (A) GSEA enrichment analysis showed that TAZ was associated with the MES or PN subtypes in the TCGA database. (B-C) Immunofluorescence analysis of the mesenchymal marker CD44 (red) and pro-neural marker OLIG2 (green) in MES50 cells transfected with Control, shTAZ#1, and shTAZ#2 (B). Scale bar, 10 μm. Relative fluorescence intensities of the indicated proteins (C). (D-E) Western blot analysis was performed to detect the protein levels of PMT-related markers in PN20 and PN24 cells supplemented with TRULI, XMU-MP-1, or transfected with oe-TAZ. (F-G) RT-qPCR analysis was performed to detect the mRNA levels of PMT-related markers in PN20 and PN24 cells supplemented with TRULI, XMU-MP-1, or transfected with oe-TAZ. Data are mean ± SEM. NS non-significant, *P < 0.05, **P < 0.01, ***P < 0.001. Two-tailed unpaired Student’s t-test (C, F, and G)

Thymine DNA glycosyltransferase (TDG) may demethylate WWTR1 promoter region

Previous studies have confirmed that high DNA methylation levels tend to inhibit TAZ expression (through epigenetic inhibition mechanisms) [32]. Therefore, we speculated that the high expression of WWTR1 in GBM is related to methylation modification. Our analysis revealed that the methylation level of the WWTR1 promoter region in the MES subtype was significantly downregulated compared to that in the PN subtype (Fig. 4A), suggesting that DNA demethylation may maintain the abnormally high expression of WWTR1 to promote PMT. The TDG/TET family is closely related to the DNA demethylation of regulatory genes [13]. We systematically screened for DNA demethylases (TDG/TET1-3) and found that WWTR1 expression showed the most significant positive correlation with TDG compared to that with TET-3 (Fig. 4B). We found that knockdown of TDG significantly reduced TAZ protein and mRNA levels, whereas overexpression of TDG significantly increased TAZ expression (Fig. 4C-F and Figure S4A-F). TDG primarily mediates site-specific active demethylation by synergizing with TET enzymes to recognize and excise intermediates (such as 5fC and 5caC) generated by TET oxidation of 5-methylcytosine (5mC). TDG-mediated 5fC/5caC excision occurs primarily in CpG island-rich promoter regions [33, 34]. Notably, MethPrimer analysis revealed that the WWTR1 promoter region was rich in CpG islands, and two primer pairs were designed (Fig. 4G). McrBC, a methylation-dependent restriction endonuclease, specifically recognizes and cleaves DNA containing methylated cytosine sites, such as 5mC, 5hmC, and 4mC [35]. Highly methylated regions are fragmented by McrBC, resulting in PCR amplification failure or significantly reduced amplification efficiency, which indirectly reflects the DNA methylation levels. We found that TDG knockdown led to a decrease in the PCR product of the WWTR1 promoter region, whereas TDG overexpression promoted PCR amplification (Fig. 4H and Figure S4G). To confirm that the downregulation of WWTR1 methylation levels in gliomas is facilitated by the active demethylation activity of TDG, we employed ChIP-qPCR and confirmed that TDG binds to the promoter region of WWTR1 (Fig. 4I). Subsequently, DNA immunoprecipitation (DIP) assays demonstrated that TDG significantly reduced the abundance of 5caC and 5mC modifications in the WWTR1 promoter region (Fig. 4J-K and Figure S4H-I), suggesting that TDG can reduce the abundance of 5caC and 5mC in the WWTR1 promoter region and thereby regulate its expression. Importantly, Methylation-specific PCR (MS-PCR) confirmed that the methylation level of the WWTR1 promoter region is regulated by TDG (Figure S4J-K). These findings elucidate the molecular mechanism by which TDG may maintains its high expression through the regulation of methylation modification of the WWTR1 promoter region.

Fig. 4.

Thymine DNA glycosylase-induced demethylation of the WWTR1 promoter region. (A) Relative methylation levels of the WWTR1 promoter region in PN, MES, and CL were evaluated using the TCGA dataset. (B) Correlation analysis of WWTR1 and demethylases (TDG and TET1-3) in GBM in the CGGA dataset. (C-D) Alterations in WWTR1 protein and mRNA levels in MES50 and MES52 cells following TDG knockdown. (E-F) Immunofluorescence analysis was conducted to assess the alterations in the expression levels of TDG (red) and TAZ (green) in MES50 cells following the knockdown of TDG (E). Scale bar, 10 μm. Relative fluorescence intensities of the indicated proteins (F). (G) Prediction of CpG islands and primers in the WWTR1 promoter region. (H) McrBC-PCR analysis detected changes in WWTR1 promoter methylation levels following TDG knockdown in MES50 and MES52 cells. (I) ChIP assays were performed on TDG and WWTR1 binding regions. (J) The relative abundance of 5caC in the WWTR1 promoter after TDG knockdown was detected using a DIP assay with a 5caC antibody, followed by qPCR with specific primers in MES50 cells. (K) The relative abundance of 5mC in the WWTR1 promoter after TDG overexpression was detected using a DIP assay with a 5mC antibody, followed by qPCR with specific primers in PN20 cells. Data are mean ± SEM. NS non-significant, *P < 0.05, **P < 0.01, ***P < 0.001. Two-tailed unpaired Student’s t-test (D, F, H, I, J, and K)

TAZ-TEAD4 positively regulated TDG expression at the transcriptional level

To elucidate the regulatory mechanism of TAZ in GBM, we identified its downstream targets. Unexpectedly, we observed that TAZ knockdown downregulated TDG expression (Fig. 5A), prompting us to investigate whether TDG is regulated by TAZ in GBM cells. TDG knockdown promoted the upregulation of PN markers and suppressed the expression of MES markers (Fig. 5B). Subsequently, we found that TAZ overexpression significantly increased the protein and mRNA expression levels of TDG in GBM cells, whereas both TDG protein and mRNA expression were significantly reduced after TAZ knockdown (Fig. 5C and Figure S5A). TDG overexpression led to a poor prognosis in GBM (Figure S5B). As TAZ is negatively regulated by LATS1/2 and MST1/2 [36], we used small interfering RNA (siRNA) to inhibit LATS1/2 and MST1/2. The results showed that TAZ upregulation induced by Hippo pathway inactivation enhanced the expression levels of TDG at both the mRNA and protein levels (Fig. 5D-G and Figure S5C-F), suggesting that TDG may be transcriptionally regulated by TAZ. Notably, TAZ, which does not possess a DNA-binding domain, must be directed to specific genomic locations by transcription factors, such as TEAD1-4, to function effectively as a transcriptional coactivator [37, 38]. Analysis of the TEAD family in the Rembrandt database revealed that TEAD4 was closely associated with TDG (Fig. 5H). KM survival analysis indicated that TEAD4 was correlated with poor prognosis (Fig. 5I). ChIP-seq data from the Cistrome database showed that TEAD4 binds to the promoter region of TDG (Fig. 5J). Therefore, we used JASPAR (http://jaspardev.genereg.net/) to predict and identify three conserved binding sites for TEAD4 in the TDG promoter region (Fig. 5K). Next, we incorporated the promoter sequence (Region 1–3) into the pGL3-Basic reporter plasmid and transfected it into GBM cells. Dual-luciferase reporter gene assays demonstrated that TAZ overexpression selectively enhanced the luciferase activity of the Region 1 promoter, but had no significant effect on the Region 2–3 promoter (Fig. 5L and Figure S5G). Furthermore, ChIP-qPCR assays revealed that TEAD4 has significant chromatin-binding activity at the TDG promoter (Region 1) (Fig. 5M and Figure S5H). Collectively, these data indicate that the TDG promoter is a direct target of TAZ-TEAD4.

Fig. 5.

TAZ-TEAD4 positively regulates TDG expression at the transcriptional level. (A) Genes altered upon WWTR1 knockdown in the GEO dataset. The results are presented as a volcano plot. (B) Heatmap illustrating the mRNA levels of PN and MES markers in MES50 cells following TDG knockdown, as determined by RT-qPCR. (C) Changes in TDG protein and mRNA levels in PN20 and N24 cells after TAZ overexpression. (D-E) Variations in the protein levels of TAZ and TDG in PN20 cells following LATS1/2 (D) or MST1/2 (E) depletion. (F-G) Variations in mRNA levels of TAZ and TDG in PN20 cells following the depletion of LATS1/2 (F) or MST1/2 (G). (H) Correlation analysis of TDG and TEAD1-4 in GBM within the Rembrandt dataset. (I) The Kaplan-Meier curve demonstrates a significant association between elevated TEAD4 expression and worsened overall survival (OS) in the Rembrandt database (http://gliovis.bioinfo.cnio.es/). (J) Schematic representation of the binding motif of the transcription factor TEAD4. (K) Predicted sites of the TEAD4 binding motif in the 100–2000 TDG promoter. (L) Dual luciferase reporter gene assays were performed to detect the binding site of TAZ-TEAD within the TDG promoter region in PN20 cells. (M) ChIP-qPCR analyses‌‌ of the binding of TEAD4 to the TDG promoter Region 1–3 in PN20 cells were conducted. CTGF was used as the positive control. Data are mean ± SEM. **P < 0.01, ***P < 0.001. Two-tailed unpaired Student’s t-test (C, F, G, L, and M)

The positive feedback loop of TAZ-TDG-mediated PMT promotes carcinogenesis

TDG-mediated DNA demethylation upregulated TAZ expression. Conversely, TAZ-TEAD4 positively regulated TDG transcription. Therefore, we speculated that TAZ and TDG may form a positive feedback loop and play a pro-cancer role in GBM development. To test this hypothesis, we constructed a cellular model of a positive feedback loop using cross-intervention experiments. The transfection efficiencies of oeTAZ+shTDG and oeTDG+shTAZ in GBM cells were validated using western blotting and qPCR (Figure S6A-B). As expected, knockdown of TDG significantly impaired cell proliferation in the context of TAZ overexpression, whereas inhibition of TAZ in the context of TDG overexpression similarly reversed this pro-proliferative effect (Figs. 6A-F and S6C-H). Invasion and migration assays further demonstrated that the TAZ-TDG regulatory loop bidirectionally regulated cell motility (Figs. 6G-N and S6I-P). TAZ is a key transcription factor that sustains MES differentiation [39]. In colorectal cancer, glioma, and hepatocellular carcinoma, TDG has been reported to promote the transformation of tumor cells to a mesenchymal-like phenotype [19, 40, 41]. Importantly, we discovered that these PN markers were suppressed by the activated TAZ-TDG loop, thereby facilitating PMT progression (Fig. 6O and Figure S6Q). Therefore, TAZ and TDG form a positive feedback loop that jointly promotes PMT in GBM cells and exerts a pro-cancer effect.

Fig. 6.

Positive feedback loop of TAZ-TDG-mediated PMT that promotes carcinogenesis. (A-B) CCK-8 assay was performed to detect cell viability. (C-F) Cells were transfected with the negative control vector and lentiviruses for the overexpression or knockdown of TAZ (C-D) and TDG (E-F). Subsequently, the colony-forming abilities of MES50 cells transfected with various vectors were assessed. (G-J) Wound healing assay was performed to assess the migration capacity of MES50 cells transfected with different vectors. (K-N) Transwell assay was performed to verify the alterations in MES50 cell invasiveness. (O) Western blotting was used to evaluate the expression levels of PMT markers in MES50 cells. Representative images and quantitative histograms are shown. Data are mean ± SEM. **P < 0.01, ***P < 0.001. Two-tailed unpaired Student’s t-test (A, B, D, F, H, L, N, and J)

The combination of verteporfin and juglone synergistically inhibited tumorigenesis

To evaluate the therapeutic potential of the positive feedback loop of TAZ-TDG, we performed in vitro and in vivo studies using patient-derived glioblastoma cells (MES50). Different treatment conditions were set for MES50 cells (control group, sh-Con; TAZ knockdown group, sh-TAZ; TDG knockdown group, sh-TDG; and double gene knockdown group, sh-TAZ + sh-TDG). The results showed that combined knockdown of TAZ and TDG almost completely inhibited tumor cell proliferation (Figures S7A-C). Transwell assays further validated this finding (Figures S7D-E), demonstrating that dual-gene knockdown significantly enhanced the inhibitory effect of single-gene knockdown on GBM cell migration. Western blot analysis showed that knockdown of either TAZ or TDG alone delayed proneural-mesenchymal transition (PMT) in GBM cells, whereas double-gene knockdown significantly enhanced this inhibitory effect (Figure S7F). We then validated the therapeutic effects of dual knockdown of TAZ and TDG in an animal model. MES50 cells treated under different conditions were orthotopically injected into the brains of the NSG mice (Figure S7G). Bioluminescence imaging analysis showed that mice injected with sh-TAZ or sh-TDG had significantly reduced tumor volumes and prolonged survival compared to the control group. Notably, mice injected with combined TAZ and TDG knockout GBM cells exhibited smaller tumor volumes and higher survival rates (Figures S7H-I). These results indicate that the combined knockdown of TAZ and TDG has a significant synergistic effect on inhibiting the occurrence, development, and PMT process of GBM.

Based on these findings, we further evaluated the potential efficacy of TAZ and TDG inhibitors in GBM treatment. Verteporfin is a small-molecule compound that specifically targets YAP/TAZ transcriptional activity [42], whereas juglone targets TDG and significantly increases intracellular 5caC levels [43]. We established an orthotopic intracranial PDX xenograft mouse model using MES50 cells to compare the blood-brain barrier (BBB) permeability of the two drugs. Following intravenous injection, both drugs reached effective concentrations in both tumor and non-tumor areas of the brain, with no significant difference in distribution between the two regions, demonstrating their good BBB penetration (Figure S7J). Western blot analysis showed that verteporfin significantly reduced the expression of MYC, a classic target gene of YAP/TAZ-TEAD, in MES50 cells, whereas juglone successfully inhibited the expression of TDG (Figure S7K). Furthermore, continuous monitoring of cell viability showed that both verteporfin and juglone effectively reduced the proliferation of GBM cells, and the corresponding half-maximal inhibitory concentrations (IC50) were determined (Figure S7L). Verteporfin and juglone attenuated the expression of PMT progression markers in GBM cells, and their combination enhanced this inhibitory effect (Fig. 7A). Importantly, the combination of these two drugs exhibited a significant synergistic antitumor effect (Fig. 7B-C and Figure S7M-N). Subsequently, we validated the combined efficacy of verteporfin and juglone in an NSG (PDX) mouse model (Fig. 7D). The results showed that in an orthotopic NSG mouse model, the combination of verteporfin and juglone significantly reduced tumor volume and prolonged survival (Figs. 7E-G). Furthermore, mice treated with either verteporfin or juglone showed no significant weight loss compared with the control group (Fig. 7H). Pathological section analysis further confirmed that dual inhibition of TAZ and TDG exerted the most potent effect in reversing PMT (Figs. 7I-J). In summary, these results strongly demonstrate the potential of verteporfin and juglone as candidate small-molecule inhibitors for GBM treatment, and their safety and efficacy warrant further investigations. These findings highlight the potential of TAZ and TDG as therapeutic targets for GBM, and the significant synergistic effects of combined targeted therapy.

Fig. 7.

Combination of verteporfin and juglone synergistically inhibits tumorigenesis. (A) Western blotting was used to evaluate the expression levels of PMT markers in MES50 cells. (B-C) Heatmap illustrating the synergy index determined by synergyFinder for the combined use of Verteporfin and Juglone. (D) Schematic diagram of the Verteporfin and Juglone treatment schedule for intracranial orthotopic xenograft mice. (E-F) Bioluminescent imaging analysis of tumor growth in NSG mice engrafted with GBM xenografts derived from MES50 cells with the addition of Verteporfin, Juglone, or Verteporfin + Juglone. Representative images are displayed at 7, 14, and 28 days after transplantation (E). Quantification map of fluorescence intensity (F). (G) Kaplan-Meier survival curves of NSG mice implanted with MES50 cells with the addition of Verteporfin, Juglone, or Verteporfin + Juglone. Survival curves were analyzed using the two-tailed log-rank test. (n = 10 for each group of Verteporfin, Juglone, or Verteporfin + Juglone). (H) Body weights in the different treatment groups were measured. (I-J) IHC assays for Ki67, CD44, and OLIG2 in tumors (I). IHC scores for TAZ were quantified (J). Representative images and quantitative histograms are shown. Data are mean ± SEM. **P < 0.01, ***P < 0.001, ****P < 0.0001. Two-tailed unpaired Student’s t-test (F, G, and J)

Discussion

Glioblastoma (GBM) is divided into three molecular subtypes: proneural (PN), mesenchymal (MES), and classical (CL). PMT refers to the process of tumor transformation from the PN subtype to the more aggressive MES subtype, similar to the epithelial-mesenchymal transition (EMT) [44, 45]. In primary or recurrent GBMs, studies have reported factors that lead to PMT, such as (1) cell-intrinsic factors (e.g., transcription factors), (2) cell-extrinsic factors: tumor immune microenvironment (e.g., macrophages) [46, 47], and (3) cytotoxic therapy (e.g., radiotherapy/chemotherapy) [48]. PMTs are the core manifestation of the malignant progression of GBM, directly leading to enhanced tumor heterogeneity, increased invasiveness, and significant resistance to radiotherapy and chemotherapy (such as TMZ). During recurrence, the pressure of previous treatments may make PMTs more prevalent and prominent, becoming a key pathological basis for the refractory nature of recurrent GBM [49, 50]. Targeting crucial regulatory molecules of PMT (such as STAT3 and TAZ) or exploiting subtype-specific metabolic weaknesses is expected to become a new strategy to overcome the bottleneck of glioma treatment.

In this study, we found that TAZ expression was upregulated in glioma tissues and correlated with malignant clinicopathological features. Experiments have shown that inhibition of TAZ function significantly inhibits cell colony formation, migration, and tumor formation, which may be related to PMT. Importantly, we revealed a new aspect of TAZ regulation: methylation levels in the TAZ promoter region were significantly reduced in GBM tissue samples compared to non-tumor tissue samples, as TDG augmented 5caC abundance in the TAZ promoter region, facilitating the demethylation process and thereby enhancing TAZ expression in tumors. Interestingly, TDG was identified as a downstream target of TAZ in the GEO database. These findings indicate that TDG is transcriptionally regulated by TAZ. TAZ has been unequivocally identified as the primary driver of PMT in gliomas [51, 52]. As a transcriptional coactivator, TAZ activates downstream gene expression by forming a complex with TEAD (TAZ-TEAD), which directly drives the PMT process [10]. TDG mediates DNA demethylation in the tenascin-C (TNC) promoter region, significantly reducing its methylation level and upregulating TNC expression [19]. TNC is a key marker for MES subtype gliomas, and its high expression directly promotes tumor invasiveness and PMT progression [53]. Increased CD44 and decreased OLIG2 are typical molecular characteristics of PMT [21, 54, 55]. We found that the TAZ-TDG loop synergistically activates MES-type signaling, promoting PMT and malignant progression of glioma, leading to poor prognosis.

In addition, we found that in the TAZ-TDG positive feedback loop system, single-target inhibition of TAZ or TDG showed limited antitumor effects, and the inhibition efficiency of the two was similar, which may be due to the characteristic differences in the tumorigenesis mechanisms of the two. TAZ, the core effector molecule of the Hippo pathway, usually co-localizes with TEAD transcription factors in the promoter region of mesenchymal phenotype-related genes, directly regulating the PMT process [56, 57]. In addition to participating in DNA demethylation and epigenetic regulation in the TAZ promoter region, TDG-mediated base excision repair dysfunction may promote PMT phenotypic transformation by inducing genomic instability [13, 58]. This complementary mechanism prompted us to explore whether dual-target inhibition of TAZ and TDG could exert more powerful antitumor effects.

Considering the importance of TAZ and TDG as potential therapeutic targets, we administered verteporfin (TAZ inhibitor) and juglone (TDG inhibitor) to confirm the inhibition efficiency of dual targets in the NSG (PDX) mouse model. Combined treatment exhibited a pronounced tumor-suppressive effect in mouse xenograft models and significantly prolonged survival in orthotopic mice. Hence, TAZ and TDG may serve as effective therapeutic targets for GBM and have significant clinical application potential in treating GBM. In the future, our research will focus on the survival analysis of these markers in patients with GBM and their potential as targeted drugs.

Conclusions

In this study, we confirmed that TAZ is activated through demethylation and functions as an oncogene in GBM cells. We further confirmed that TDG, as a demethylase, can lower the methylation level of the TAZ promoter region, thereby facilitating the expression in tumors. Additionally, our research indicated that the TAZ/TEAD complex positively regulates the expression of TDG at the transcriptional level, suggesting a positive feedback loop between TDG and TAZ that jointly promotes the progression of GBM. The dual knockout of TAZ and TDG exerted an optimal tumor-suppressive effect, providing novel molecular targets for GBM treatment (Fig. 8).

Fig. 8.

Schematic diagram of the molecular mechanism of this study

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

TDG

Thymine DNA glycosylase

PMT

Proneural-to-mesenchymal transition

GBM

Glioblastoma

PN

Proneural

MES

Mesenchymal

CL

Classical

TET

Ten-eleven translocation

5fC

5-formylcytosine

5caC

5-carboxylcytosine

5mC

5-methylcytosine

PBS

Phosphate-buffered saline

WB

Western blot analysis

IHC

Immunohistochemical assay

ChIP

Chromatin immunoprecipitation

EMT

Epithelial-mesenchymal transition

Author contributions

All authors contributed to the data analysis in this study. Performing the laboratory experiments: J Xie and WY; Writing, review, and/or revision of the manuscript: J Xie and WY; Acquisition, analysis, and interpretation of data: LP, SC, SG, and DW; Administrative, technical, or material support: ZT; Study supervision: J Xiao, ZC, and HL. Conception and design: J Xiao, ZC, and HL. All authors contributed to the study and approved the submitted version.

Funding

This study was supported by the National Natural Science Foundation of China (No. 82203876), Natural Science Foundation of Basic Research Program of Jiangsu Province, Youth Program (BK20220184), China Postdoctoral Science Foundation-Special Funding (2022TQ0143), and China Postdoctoral Science Foundation-General Support (2022M711587) to Z.T. The Chinese Academy of Medical Sciences (ZZ15-WT-04) and the National Natural Science Foundation of China (no. 82460526) to Z.C.; the National Natural Science Foundation of China (no. 82360475) and the Jiangxi Provincial Natural Science Foundation (20242BAB20398) to H.L.; and the Graduate Innovation Special Fund Project (YC2024-B063) to W.Y.

Data availability

TCGA transcriptome, DNA methylation, and clinical data were obtained from the website (http://cancergenome.nih.gov/). The CGGA transcriptome and clinical data were collected from the Chinese Glioma Genomic Atlas (CGGA; http://www.cgga.org.cn/). The TAZ knockdown (GSE261673, GSE219088, and GSE188738) microarray datasets were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/gds) database.

Declarations

Ethical approval and consent to participate

The human GBM tissues was approved by the Ethics Committee of Affiliated Drum Tower Hospital of Nanjing University Medical School, and informed consent was obtained from the patients or their guardians according to the Declaration of Helsinki. All animal experiments were approved by the Animal Experiment Ethics Committee of the Second Affiliated Hospital of Nanchang University (Approval No. NCULAE-20221031035).

Consent for publication

Not applicable.

Competing interests

The authors declare no potential conflicts of interest.