SUV39H1 regulates progression of pediatric diffuse high-grade gliomas through modulation of β-catenin/TCF4 levels

Abstract

Abstract

Pediatric diffuse high-grade gliomas (pHGGs) are heterogeneous and infiltrative tumors with poor prognosis. Recent molecular characterization of pHGGs has revealed an altered epigenetic profile with aberrant post-translational modifications and increased histone 3 lysine 9 trimethylation (H3K9) to be implicated in their pathology. Herein, we investigate the expression and biological role of the methyltransferase suppressor of variegation 3–9 homolog 1 (SUV39H1) in regulating key cancer hallmarks, including cell proliferation, motility, and epithelial-mesenchymal transition (EMT) gene regulation of pHGG. Bioinformatic analysis of SUV39H1 mRNA levels was performed in a public database of pediatric gliomas with respective controls. SUV39H1 protein expression was further investigated in a cohort of 24 pHGG tissues and controls by immunohistochemistry and western immunoblotting, followed by correlation analysis with patients’ clinicopathological features. Gene silencing of SUV39H1 was performed in patient-derived pHGG cell lines (SJ-GBM2 and CHLA-200) to investigate their functional role in cell proliferation, migration, cell adhesion, and EMT markers. SUV39H1 mRNA was found enriched in pHGG tissues compared to normal brain, with pathway analyses (GO and KEGG) revealing a significant correlation with cell adhesion and focal adhesion pathways in SUV39H1-high pHGG. Immunohistochemical analysis of the pHGG cohort detected significantly increased SUV39H1 protein expression in pHGG tissues, followed by elevated H3K9me3 expression compared to normal brain tissues. Moreover, SUV39H1 and H3K9me3 levels were higher in pHGG with H3K27M mutation compared to H3-WT tumors and correlated with worst patients’ survival. Gene silencing of SUV39H1 in SJ-GBM2 and CHLA-200 cells induced a significant decrease in cell viability and cell migration, followed by reduced expression of vimentin, β-catenin, and TCF4 protein levels. Furthermore, SUV39H1 silencing reduced the mRNA levels of EMT marker genes CDH2, SNAI1, and MARCKS. Our findings demonstrate that SUV39H1 regulates cell proliferation and adhesion in pHGG, contributing to epithelial-mesenchymal transition through regulation of β-catenin/TCF4 axis, presenting a promising therapeutic target to be investigated further.

Key messages

• SUV39H1 gene expression is enriched in pHGG compared to normal brain.

• SUV39H1 expression is increased in pHGG tissues along with H3K9me3 expression.

• Gene silencing of SUV39H1 reduces cell viability and migration.

• SUV39H1 silencing decreased expression of vimentin, β-catenin and TCF4.

• SUV39H1 silencing reduced EMT marker genes CDH2, SNAI1 and MARCKS.

• SUV39H1 has an oncogenic role in pHGG and presents a promising therapeutic target.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00109-025-02592-3.

Introduction

The high-grade gliomas of children are infiltrative and aggressive brain tumors with a poor prognosis and a major cause of cancer-related mortality [1]. A large amount of studies indicate that pediatric gliomas have unique molecular characteristics and genetic changes that confer a completely different biological behavior from the respective adult tumors [2]. The World Health Organization (WHO) classification 2021 of central nervous system (CNS) tumors has divided pediatric-type gliomas into low-grade (pLGG) and high-grade (pHGG) tumors, which are further characterized by distinct histological and clinical features, as well as an extended range of molecular hallmarks.

Histone H3 wild-type (H3-WT) pediatric gliomas represent a heterogeneous group of tumors that can be of low or high grade, lacking mutations in H3F3A or HIST1H3 genes but often presenting alterations in other oncogenic pathways depending on grade. The BRAF V600E mutations are often present in low-grade gliomas, while TP53, ATRX, or IDH1 mutations are more rare, often seen in high-grade tumors, and MYCN or EGFR amplification is frequent in high-grade, aggressive cases [3]. The H3-mutant gliomas exhibit fully clonal H3K27M somatic mutations in midline pHGG tumors and H3G34V mutations in cerebral hemispheric tumors [4], regarded as the founding driver events in HGGs, crucial for their initiation and maintenance [5–7].

Widespread changes in the epigenomic landscape are often characteristic of pediatric gliomas, inducing global deregulation and effects in gene expression [8–10]. Several histone marks, such as the transcriptional repressor H3K27 trimethylation (H3K27me3) and H3K9me3, participate actively in the epigenomic alterations and deregulated gene activity observed in pediatric gliomas [11, 12]. A recent study of histone mark profiling in pediatric astrocytomas from our group revealed higher expression and prognostic significance of H3K9me3 in grade 4 astrocytomas [11]. The presence of H3K9me3 histone mark promotes the interaction of the modified histones with Heterochromatin protein 1 (HP1) and is mainly involved in transcriptional repression [13]. Defective pericentric heterochromatin and abnormal transcription of pericentric repeats have been tightly involved in cancer-related genomic instability [14]. It is therefore apparent that deregulation of H3K9 methylation machinery may represent a pathogenic feature of pediatric astrocytomas, suggesting a potential implication of its writer, suppressor of variegation 3–9 homolog 1 (SUV39H1) in pHGG pathobiology.

Previous studies have shown that SUV39H1 is involved in several biological processes such as DNA damage repair, telomere maintenance, cell differentiation, and aging [15]. SUV39H1-mediated H3K9 trimethylation has been linked to the silencing of tumor suppressor genes in acute myeloid leukemia (AML), whereas its default function is to maintain genome stability by limiting the acute activation of oncogenes [16]. Moreover, deregulation of SUV39H1 has been observed in several human malignancies, including adult gliomas [17, 18]. A differential expression pattern of SUV39H1 has been observed in adult astrocytomas, with cytoplasmic SUV39H1 levels increasing progressively from low-grade to high-grade glioma tumors, while nuclear SUV39H1 expression has been inversely correlated with grade [18]. Additionally, SUV39H1 has been demonstrated to play a critical role in glioma stem cells (GSC) maintenance and glioblastoma progression, while its inhibition through chaetocin was shown to sensitize cells to temozolomide [19]. Some studies have also reported a tumor suppressive role of SUV39H1 through the stabilization and silencing of heterochromatin [15].

In respect to pediatric gliomas, previous studies have shown that SUV39H1 expression is an independent prognostic factor of adverse significance in pediatric astrocytomas [11] while treatment of diffuse intrinsic pontine gliomas (DIPG/DMG) with SUV39H1’s inhibitor chaetocin was demonstrated to significantly reduce growth, followed by downregulation of stemness and oncogenic networks [20].

Given the pivotal role of SUV39H1 in regulating gene expression and the limited information of its function in pHGG, we proceeded to explore its implication in vital cellular processes such as cell proliferation, migration, and adhesion as well as EMT regulation in diffuse pHGG tissues and respective patient-derived cell lines.

Material and methods

Patients and tissue sample description

The study involved 24 diffuse high-grade astrocytoma cases (1–15 years old) diagnosed based on WHO criteria at the Neurosurgery Department of Mitera Children Hospital between 2017 and 2019. In all cases, the diagnosis, grading, and genetic profiling of tumors (H3K27M, G34V/R) were peer-reviewed according to the principles laid down in the 2021 WHO Classification [3, 21], following the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. All cases with available tumor tissue were included. Their clinicopathological features and genetic mutations are presented in Table 1. Five (5) archival normal brain (cerebellum) tissues, either frozen or in paraffin blocks from children (1–3 years old), were also used. The study was approved by the University of Athens Medical School Ethics Committee (27/06/2017, 1,617,031,069), and written informed consent was obtained from the parents of all patients. When this study was conducted, 15 patients (62.5%) had died of disease after 24 (12–48) months, whereas 9 patients (37.5%) were alive after 48 (24–60) months follow-up.

Table 1.

Clinicopathological data of patients’ cohort

Variable	Patients’ cohort (n = 24) median (range)
Age	8 (1–15) years
	Number of cases (%)
Gender
Male	18 (75)
Female	6 (25)
Tumor location
Cerebellum	19 (79.2)
Brain stem	5 (20.8)
P53 expression (%)	24 (100)
H3K27M positivity	12 (50)
BRAFV600E mutation	0 (0)
H3G34V mutation	2 (8.3)
IDH1 R132H expression	0
Surgery	24 (100)
Partial resection	24 (100)
Complete resection	0 (0)
Radiotherapy
Yes	15 (62.5)
No	9 (37.5)
Chemotherapy (TMZ)
Yes	16 (66.7)
No	8 (33.3)

Cell lines and culture conditions

Two patient-derived H3-WT, IDH-WT pHGG cell lines, CHLA-200 and SJ-GBM2, were obtained from the Texas Tech University Health Sciences Center, School of Medicine Cancer Center, COG Cell Line & Xenograft Repository (www. cccells.org). The cell lines were tested and authenticated for antibiotics and mycoplasma as well as validated by short tandem repeat (STR) genotyping. Culturing of the cells was performed in Iscove’s Modified Dulbecco’s Medium (IMDM, Biosera) supplemented with 20% fetal bovine serum (FBS) (Gibco, Life Technologies), 1 × ITS (Millipore, MA) (5 μg/mL insulin, 5 μg/mL transferrin, 5 ng/mL selenous acid), 1% penicillin–streptomycin (Gibco, Life Technologies), and 4 mM L-Glutamine (Gibco, Life Technologies). Cell cultures were incubated in a humidified atmosphere containing 5% CO₂–95% air at 37 °C.

Gene knockdown by small interfering RNA (siRNA)

siRNAs used for SUV39H1 silencing (sc-38463) and respective negative control (NC, non-target-siRNA, sc-37007) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). According to the manufacturer, the SUV39H1 siRNA (h) is a pool of 3 target-specific 19–25 nt siRNAs designed to knock down gene expression. CHLA-200 and SJ-GBM2 cells were seeded on 24-well plates and allowed to reach ~ 70–90% confluency. They were transfected using the Lipofectamine 3000 kit (Invitrogen) with 20 nM and 40 nM of siRNA for SUV39H1, respectively. Briefly, three different solutions were prepared. The first one consisted of the Lipofectamine reagent diluted in Opti-MEM medium, while the other two were master mixes containing either the corresponding siRNA sequence or the non-target siRNA sequence diluted in Opti-MEM. Afterwards, equal portions (1:1) of each Master mix and the diluted Lipofectamine reagent were incubated together to form the siRNA-lipid complex. Cells were then treated with the complex solution and full growth medium was added to each well after 5–6 h. In the case of cells used as controls, only Opti-MEM medium was added to the wells. After 24 h, cells underwent trypsinization, and the effect of siRNA transfection was confirmed using western immunoblotting.

Real-time PCR analysis

Total RNA extraction was performed from SJ-GBM2 and CHLA-200 cells previously transfected with siSUV39H1 or ΝC, using the RNeasy Mini Kit (Qiagen, Hilden, Germany), according to manufacturer’s recommendations. The total RNA quality and quantity were validated by UV spectroscopy, by calculating the ratio of absorbance at 260 nm and 280 nm. For cDNA synthesis, the PrimeScript RT Reagent Kit-Perfect Real Time (Takara Bio, Japan) for RT-PCR was used. The real-time PCR was performed with SYBR Green using the KAPA SYBR® FAST qPCR Master Mix (2X) Universal Kit (Sigma-Aldrich, KK4602) and specific primer pairs to assess the expression levels of the following genes: GAPDH, CDH2, MARCKS, SNAI1 (Table 2). The PCR reaction conditions were 3 min at 95 °C followed by two-step PCR for 40 cycles, each consisting of 3 s at 95 °C and 40 s at 60 °C. A melting curve analysis was performed to confirm the specificity of quantitative polymerase chain reaction (qPCR) products. The fold value, which depicts the differential expression of each selected gene, was calculated from the processing of the PCR data according to the 2^−ΔΔct method, using the reference gene GAPDH for normalization.

Table 2.

Primer pairs used in real-time PCR analysis

Gene	Forward primer (5′ → 3′)	Reverse primer (5′ → 3′)
GAPDH	AGGGCTGCTTTTAACTCTGGT	CCCCACTTGATTTTGGAGGGA
CDH2	CTCCTATGAGTGGAACAGGAACG	TTGGATCAATGTCATAATCAAGTGCTGTA
MARCKS	CTGCTCTTTGCCACCGATA	AGGCACTGTACTTGGCCATC
SNAI1	GAGGCGGTGGCAGACTAG	GACACATCGGTCAGACCAG

Western blot analysis

Protein extraction was performed using ice-cold RIPA buffer from tissue samples and cell lines (Thermo Fisher Scientific #89,901) or Cell Lysis buffer (Cell Signaling Technology #9803) supplemented with PMSF (Cell Signaling Technology #8553), respectively. Proteins were resolved by electrophoresis in SDS–polyacrylamide gels with several densities (6–15%) depending on the molecular weight of each protein and then transferred to a nitrocellulose membrane (Porablot NCP, Macherey–Nagel, Germany, #12,807,411). Blocking of membranes was performed in Tris-buffered saline Tween-20 (TBS-T 0.1%) with 5% non-fat milk for 1 h at room temperature, followed by incubation with primary antibodies overnight at 4 °C. Primary and secondary antibodies were diluted as indicated in Table 3. Upon incubation with HRP-conjugated secondary antibodies, the detection of the immunoreactive bands was performed with the SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific, Cat no.34580). Relative protein amounts were evaluated by densitometry using ImageJ software (La Jolla, CA, USA) and normalized to the corresponding actin levels. All experiments have been performed at least 3 times, and representative results are shown.

Table 3.

Characteristics of primary and secondary antibodies used in western immunoblotting and in immunohistochemical analysis

Primary antibody	Company	Catalog no.	Clone	Host	WB dilution	IHC dilution, incubation	Antigen retrieval	Positive control
SUV39H1 (KMT1A)	Novus Biologicals	NBP1-21367	Polyclonal	Rabbit	1:2000	1:50	Citrate buffer, pH = 6	Human breast cancer
H3K9me3	Millipore, MA	#05–1250	Monoclonal	Mouse	1:1000	1:250	Tris/EDTA, pH = 9	Human breast cancer
Vimentin	SantaCruz Biotechnology	sc-73259	Monoclonal	Mouse	1:100	-	-	-
β-catenin	SantaCruz Biotechnology	sc-59737	Monoclonal	Mouse	1:200	-	-	-
pGSK3α/β	SantaCruz Biotechnology	sc-81496	Monoclonal	Mouse	1:500	-	-	-
TCF4	SantaCruz Biotechnology	sc-166699	Monoclonal	Mouse	1:300	-	-	-
Actin, clone C4	Millipore, MA	MAB1501	Monoclonal	Mouse	1:20,000	-	-	-

Immunohistochemical staining

Immunostaining of formalin-fixed paraffin-embedded 4 μm sections (FFPE) was performed using the VECTASTAIN® Elite® ABC Universal Kit, PK-6200, Vector Laboratories) as previously described [11]. The primary antibodies were diluted as shown in Table 3. Normal human brain paraffin tissues and breast cancer paraffin tissues were used as positive controls, available at the First Department of Pathology, Medical School, National and Kapodistrian University of Athens. Sections in which the primary antibody was substituted with non-immune serum were used as negative controls and stained in each run. Evaluation of immunohistochemical staining was performed by a pathologist (PK) without knowledge of clinical information. The Histo score (H-score) was calculated based on the percentage of stained neoplastic cells (labeling index—LI) multiplied by the staining intensity.

Cell viability assay

The XTT Cell Proliferation Assay Kit (10010200, Cayman Chemical, USA) was used for the assessment of CHLA-200 and SJ-GBM2 cell viability. Seeding of cells in 96-well plates at a density of 10³–10⁵ cells/well was followed by transfection with siSUV39H1. After 48 h, the medium was replaced with XTT reagent diluted in serum-free, phenol red (PR)–free medium and mixed gently on an orbital shaker for 1 min. The cells were incubated for 2 h at 37 °C in a CO₂ incubator, and the absorbance was measured using a microplate reader at 450 nm. Each experiment was conducted in triplicate.

Cell migration assay

The 12-well culture plates were used for the culture of CHLA-200 and SJ-GBM2 cells (at a density of 10 × 10⁵ cells per well) at 4 sites. Upon the siRNA transfection, cells were coated, the monolayer was stained with a sterile 200-µL pipette tip (denoting zero migration), and media were added. After the 24 h incubation, the samples were washed twice with PBS, pH 7.2, and each site was photographed on a computer-connected microscope at × 4 and × 20 magnifications. Phase-contrast images were taken at the start (0 h) and 24 h of incubation for the same region. Images were analyzed using WimScratch software (Wimasis image analysis platform), and the results were expressed as percentages of the cell etched and covered area.

Statistical analyses

All graphical and statistical analyses were performed by using the SPSS 25.0 software (SPSS Inc., Chicago, IL) and GraphPad Prism 6.0. The statistical significance threshold was set at p < 0.05. Evaluation of the differences in the distribution of numerical variables among groups was performed with Student’s t-test/ANOVA, as appropriate. The immunohistochemical expression and intensity (H-score) of SUV39H1 and H3K9me3 was compared to the differential status of different parameters using the chi-square test and Fisher’s exact test. The log-rank test was applied for univariate survival analysis, and cut-offs were determined according to the mean H-score.

Bioinformatic analysis

Data acquisition and preprocessing

RNA-seq data from GEO accession GSE243682 [22], which includes pediatric normal (n = 8), low- (n = 24), and high-grade (n = 25) gliomas were analyzed. Raw expression values (fragments per kilobase per million (FPKM)) were log₂-transformed as log₂(FPKM + 1) for all downstream analyses. Pediatric tumors were classified as pHGG or pLGG based on histopathology and molecular profiling. For heatmap visualization, gene expression values were further standardized using z-score ($Z=\frac{\chi -\mu }{\sigma }$) across samples.

Differential expression analysis

Differential expression between groups (normal vs. pLGG vs. pHGG) was assessed using the limma-voom pipeline [23]. Briefly, voom variance-stabilizing weights were computed on read counts, and linear models were fitted using limma to obtain log₂ fold-changes and moderated t-statistics [23]. Comparisons of interest included normal vs. pHGG and pLGG vs. pHGG. Genes with Benjamini–Hochberg–adjusted p-value < 0.05 were considered significant. In parallel, nonparametric Wilcoxon rank-sum tests were performed for pairwise group comparisons (e.g., SUV39H1 in pHGG vs. normal) with FDR correction.

Gene-set enrichment analysis

Functional enrichment of GO biological processes and KEGG pathways was conducted using the clusterProfiler package [24] (with enrichGO and enrichKEGG functions) in R (v4.5.1). Genes were ranked by differential expression (log₂FC from limma) for each contrast. Significant GO terms were visualized using the enrichplot routines (mountain/ridge plots).

Data visualization and heatmaps

Heatmaps of gene expression were generated using the ComplexHeatmap package [25]. Expression values (log₂(FPKM + 1)) were scaled to z-scores per gene. Rows (genes) were annotated for function (e.g., “Chromatin regulator”), and columns (samples) were annotated by group (normal, pLGG, pHGG). Color scales were selected to highlight relative over- and under-expression across groups. All analyses were performed in R (v4.5.1) using Bioconductor packages. Key libraries included limma/voom, clusterProfile (and enrichplot), ComplexHeatmap, ggplot2 for plots, and dplyr for data manipulation. Statistical tests (Wilcoxon rank-sum) and multiple testing corrections followed standard implementations in R. Figure panels were composed with ggplot2 and ComplexHeatmap, and final formatting was done in Inkscape.

EMT gene signature correlation

Fifteen canonical epithelial–mesenchymal transition (EMT) or mesenchymal‐signature genes were selected on the basis of seminal glioma subtype publications [26–28]. The list included transcription factors (ZEB1, ZEB2, SNAI1/2, TWIST1), structural markers (CDH2, CHI3L1), signaling receptors (MET), and regulators such as MARCKS and MYC. Lastly, gene expression values were correlated with SUV39H1 transcript abundance using two-tailed Spearman’s rank correlation coefficient (ρ). Resulting ρ coefficients were adjusted for multiple testing (Benjamini–Hochberg); correlations with |ρ|≥ 0.25 and FDR < 0.05 were considered significant.

Results

SUV39H1 is upregulated in pediatric high-grade gliomas and functionally linked to cell adhesion and focal adhesion pathways

Analysis of RNA-seq data from pediatric gliomas GSE243682 [22] revealed that SUV39H1 mRNA is markedly upregulated in pHGG relative to pLGG and normal brain (Fig. 1A). In particular, the median log₂(FPKM + 1) SUV39H1 expression in pHGG was found to exceed that in controls (Wilcoxon p < 0.01, FDR < 0.05). Differential expression analysis confirmed SUV39H1 among the highly upregulated genes in pHGG (adjusted p < 0.001) [23]. The heatmap of chromatin regulator genes including SUV39H1 (Fig. 1B) further illustrates the coordinated overexpression of SUV39H1 and related methyltransferases in pHGG samples (pHGG vs. Normal). More specifically, SUV39H1 upregulation occurs alongside coordinated increases in PRC2 components (EZH2, EED, SUZ12) and DNMTs, consistent with the widespread chromatin remodelling previously observed in aggressive tumors.

Fig. 1.

SUV39H1 overexpression shows epigenetic rewiring in paediatric gliomas. Box plot of SUV39H1 mRNA abundance (log₂ FPKM + 1) in normal pediatric brain (n = 8), pLGG (n = 24), and pHGG (n = 25) from the GSE243682 cohort. p values shown are two-sided Wilcoxon rank-sum tests with Benjamini–Hochberg correction: p = 7.7 × 10⁻⁴ (Normal vs. pLGG), p = 6.7 × 10⁻³ (pLGG vs. pHGG), and p = 3.1 × 10⁻¹ (normal vs. pHGG) (A). Z score Heatmap of 12 chromatin-modifying enzymes grouped by functional complex (left color bar: DNMT, H3K9, PRC2). Columns represent individual samples ordered by diagnosis; annotation tracks above indicate WHO grade, histone H3 mutation status, and BRAF fusion status. Right-hand bars display log₂-fold change for the pHGG vs. normal comparison and associated FDR (***, padj < 0.001; **, padj < 0.01; *, padj < 0.05) (B).

To explore biological pathways associated with high SUV39H1, we performed Gene‐Set Enrichment Analysis (GSEA) using clusterProfiler [24] on all genes ranked by pHGG vs. pLGG log‐fold change. The GSEA “mountain/ridge” plot in Supplementary Fig. 1 highlights multiple Gene Ontology (GO) terms related to cell adhesion that are significantly enriched (FDR < 0.05). Notably, terms such as “extracellular matrix organization,” “cell junction organization,” and “regulation of cell adhesion” showed positive enrichment scores, indicating upregulation in pHGG (Supplementary Fig. 1). These adhesion-related GO enrichments suggest that SUV39H1 overexpression is functionally linked to adhesion processes in pHGG.

Consistent with the GO results, KEGG pathway enrichment identified focal adhesion and cell adhesion molecule (CAM) pathways in pHGG (Table 4). Here, we include “focal adhesion” and “cell adhesion molecules” among the enriched pathways (gene ratio and adjusted p-value annotated). This provides strong support that SUV39H1 expression correlates with cell adhesion pathways.

Table 4.

Enriched KEGG pathways in pediatric high-grade gliomas versus normal brain

Pathway	Category	NES	p-value	Gene-set size	Regulation
Base excision repair	Other	1.952	1.23 × 10⁻4	33	↑
Cell adhesion molecules	Adhesion	−1.370	2.32 × 10⁻3	132	↓
Cell cycle	Other	2.216	1.05 × 10⁻4	124	↑
DNA replication	Other	2.224	1.22 × 10⁻4	36	↑
Focal adhesion	Adhesion	1.405	2.14 × 10⁻3	199	↑
Homologous recombination	Other	1.921	1.26 × 10⁻4	28	↑
Nucleotide excision repair	Other	1.815	1.19 × 10⁻4	44	↑
Pyrimidine metabolism	Other	1.875	1.08 × 10⁻4	97	↑
Ribosome	Other	2.086	1.09 × 10⁻4	87	↑
RNA degradation	Other	1.818	1.15 × 10⁻4	57	↑

Each pathway is shown with its category, normalized enrichment score (NES), adjusted p-value, gene set size, and overall regulation in pHGG (↑ upregulated, ↓ downregulated). Adhesion-related pathways (“cell adhesion molecules” and “focal adhesion”) are highlighted among the significant enrichments, indicating coordinated dysregulation of cell–cell and cell–matrix adhesion processes in SUV39H1-high tumors

Lastly, we compared expression of 15 epithelial–mesenchymal transition (EMT) markers that typify the mesenchymal glioma subtype [26–28]. Differential analysis revealed robust upregulation of key mesenchymal transcripts in pHGG (Supplementary Fig. 2). CHI3L1 (log₂FC =  + 3.6, FDR = 2.1 × 10⁻⁶), CD68 (+ 3.1, FDR = 4.8 × 10⁻⁵), and MYC (+ 2.7, FDR = 6.2 × 10⁻⁴) ranked highest, while TWIST1 and MET changed only modestly (≤ 0.3 log₂ units, FDR > 0.1). Spearman correlation analysis demonstrated that SUV39H1 expression tracks closely with several mesenchymal drivers (Supplementary Fig. 2): MYC (ρ = 0.46, padj = 0.001), ZEB1 (ρ = 0.38), and CDH2 (ρ = 0.34) showed the strongest positive associations. Collectively, we observed that SUV39H1-overexpressing pHGGs adopt a mesenchymal-like transcriptional profile characterized by elevated extracellular-matrix interaction (CHI3L1), immune-infiltration markers (CD68), and EMT regulators (ZEB1, CDH2), supporting a role for SUV39H1 in promoting an invasive phenotype.

SUV39H1 protein expression is increased in pHGG tissues compared to normal brain (NB) tissues

To experimentally investigate SUV39H1 expression in pediatric HGG tumors and explore associations with clinicopathological features, we collected tumor samples of 24 pHGG from young patients (1 to 15 years) and detected SUV39H1 expression by immunohistochemistry (IHC) (Table 1). Five archival normal pediatric brain tissues were also stained for SUV39H1 and used as controls.

SUV39H1 exhibited increased nuclear expression in pHGG tissues compared to NB. SUV39H1 nuclear expression was detected in 15/24 cases with mean H-score = 72 (ranging from 2–270), while it was absent in normal brain astrocytes (p < 0.001). Nuclear immunoreactivity of H3K9me3 was also observed in 23/24 cases and was significantly elevated in pHGG (H-score = 232, ranging from 45 to 300) compared to NB (mean H-score = 126, ranging from 50 to 300) (p < 0.05) (Fig. 2A, B).

Fig. 2.

Immunohistochemical expression of SUV39H1 and repressive histone mark H3K9me3 in normal brain (NB) tissues and pHGGs (Magnification 20×) (A). SUV39H1 and H3K9me3 expression (H-score) were significantly increased in pHGG compared to NB (*p < 0.05, **p < 0.001) (B). Detection of SUV39H1 and H3K9me3 protein levels and quantification by western blot in normal brain tissues and pHGGs (C, D). All the experiments were repeated at least three times, and representative data are shown

Further validation of SUV39H1 protein expression was performed in five samples of pHGG by western blot analysis and compared to NB tissues. Nuclear H3K9me3 expression was significantly increased in pHGG tissues compared to NB (p < 0.05, Fig. 2C, D).

Overall, increased protein levels of SUV39H1 were detected in pHGG, followed by elevated H3K9me3 expression compared to NB tissues.

In turn, we assessed the correlations between the clinicopathological features and SUV39H1 expression levels to explore the potential clinical significance of the methyltransferase in pHGG. Among several variables, age was significantly correlated with high SUV39H1 levels (> 8 yrs, p = 0.012).

However, a significantly higher expression of SUV39H1 was observed in pHGG tumors bearing the H3K27M mutation compared to pHGG H3-WT (p < 0.0001). A similar trend was observed for H3K9me3 levels, being higher in pHGG-H3K27M tumors compared to H3-WT, without reaching statistical significance (p = 0.074). (Fig. 3A). Notably, the increased SUV39H1 and H3K9me3 expression in H3K27M pHGG was significantly correlated with worse patients’ survival (p = 0.014 and p = 0.006, respectively, Fig. 3B, C).

Fig. 3.

SUV39H1 expression (H-score) was significantly increased in H3K27M-mutant pHGG compared to H3-WT (***p < 0.001). In accordance, H3K9me3 H-score was higher in H3K27M-mutant pHGG compared to H3-WT (p = 0.074) (A). Correlation of SUV39H1 expression in H3K27M-mutant pHGG with patients’ survival (p = 0.014) (B). Correlation of H3K9me3 levels in H3K27M-mutant pHGG with patients’ survival (p = 0.006) (C)

SUV39H1 suppression affects viability of pHGG cells

To determine the functional relevance of SUV39H1 in pHGG, we used the two pHGG cell lines, CHLA-200 and SJ-GBM2, expressing both SUV39H1 and H3K9me3 as previously described [14]. The potential impact of SUV39H1 on cell growth was investigated upon transient transfection of SJ-GBM2 and CHLA-200 cells with SUV39H1-siRNA or non-target siRNA (nt-siRNA), respectively. Twenty-four hours post-siRNA transfection, gene silencing was verified by Western immunoblot analysis. SUV39H1 expression was reduced to 75.7% in CHLA-200 and 84.8% in SJ-GBM2 SUV39H1-siRNA-transfected cells in comparison to nt-control. In accordance, H3K9me3 levels were markedly reduced in CHLA-200 and SJ-GBM2 SUV39H1-siRNA-transfected cells compared to nt-control (Fig. 4A, B).

Fig. 4.

SUV39H1 protein expression and H3K9me3 levels were significantly reduced after the siSUV39H1 transfection in CHLA-200 (A) and SJ-GBM2 (B) cell lines compared to nt-siRNA transfected cells. XTT viability assays in CHLA-200 (C) and SJ-GBM2 (D) cells were performed at 48 h following siSUV39H1 transfection, showing decreased cell proliferation compared to nt-siRNA-transfected cells (***p < 0.001, respectively)

After SUV39H1 gene silencing in pHGG cell lines, we investigated their potential impact on cell viability by performing XTT assays after 48 h of siRNA treatment. In turn, after the knockdown of SUV39H1 expression, cell viability of siSUV39H1-transfected cells was reduced by 42% in CHLA-200 cell line (p < 0.001) and by 47.5% in SJ-GBM2 cells (p < 0.001) (Fig. 4C, D).

Altogether, these data demonstrate that SUV39H1 suppression affects cell proliferation of pHGG cell lines.

SUV39H1 suppression affects cell migration in pHGG cells

The potential implication of SUV39H1 in cell motility and the progression of pHGG was investigated by the monolayer scratch migration assays at 24 h after siSUV39H1 transfection.

Cell migration in the scratch area decreased by 32.5% in CHLA-200 and by 10.4% in SJ-GBM2 siSUV39H1-transfected cells (Fig. 5A, B).

Fig. 5.

Monolayer scratch migration assay of CHLA-200 (A) and SJ-GBM2 (B) cells after siSUV39H1 transfection. Cell migration in the scratch area was calculated for siSUV39H1 transfected CHLA-200 and SJ-GBM2 cells and nt-siRNA-transfected cells. The Student’s t-test results are expressed as percentages of the cell etched and covered area. Wound recovery was estimated for siSUV39H1-transfected cells compared to nt-siRNA-transfected cells (C, D). All experiments were repeated at least three times, and mean values of the three experiments are shown (**p < 0.01, *** p < 0.001)

The wound recovery in CHLA-200 was 97.4% in nt-control cells, and 64.9% in siSUV39H1-transfected cells (p < 0.001); whereas wound recovery in SJ-GBM2 was 97.8% in nt-control cells and 87.4% in siSUV39H1-transfected cells (p < 0.01) (Fig. 5C, D).

Overall, these data indicate that SUV39H1 suppression has an impact on the motility of glioma cells.

Effect of SUV39H1 on cell adhesion and EMT markers

After the effects of SUV39H1 in cell proliferation and migration, we proceeded to investigate its potential involvement in cell adhesion using western blotting. Reduced protein expression of vimentin, β-catenin, and TCF4 levels was observed following siSUV39H1 transfection of SJ-GBM2 cells compared with nt-siRNA controls (fold change = 0.6, 0.6, 0.09, p < 0.05, respectively) (Fig. 6).

Fig. 6.

Western blotting of vimentin, β-catenin, p-GSK3α/3β, and TCF4 expression in SJ-GBM2 and CHLA-200 cell lines after siSUV39H1 transfection and protein quantification relative to actin (A, B). Graphs of fold change in CDH2, SNAI1, and MARCKS mRNA levels post siSUV39H1 in SJ-GBM2 and CHLA-200 cells (C). All experiments were repeated at least three times, and mean values of the three experiments are shown (*p < 0.05, **p < 0.01, ***p < 0.001)

In accordance, reduced vimentin, β-catenin, and TCF4 levels were detected in siSUV39H1-transfected CHLA-200 cells compared to nt-treated cells (fold change = 0.5, 0.6, 0.3, p < 0.05, respectively). Upstream regulators of β-catenin, p-GSK3α/3β were not significantly changed by SUV39H1 silencing (Fig. 6).

In turn, mRNA analysis of EMT markers showed significantly reduced CDH2, SNAI1, and MARCKS levels following siSUV39H1 transfection of SJ-GBM2 cells compared with nt-siRNA controls (fold = 0.4, 0.5, 0.6, p < 0.05, respectively). In a similar way, CDH2, SNAI1, and MARCKS levels were significantly reduced following siSUV39H1 transfection of CHLA-200 cells compared with nt-siRNA controls (Fold change = 0.7, p < 0.05, respectively).

Altogether, SUV39H1 plays a prominent role in the establishment of the mesenchymal phenotype of pHGG through regulation of β-catenin/TCF4 levels and EMT markers.

Discussion

Over the last decade, significant advances in understanding the genetic basis of pediatric brain tumors have enabled the identification of distinct tumor subtypes and facilitated the development of more targeted therapeutic schemes. The recent WHO classification highlights the distinct molecular profile of pediatric gliomas from adult counterparts and the important role of post-translational modifications and histone 3 mutations in the pathogenesis of pHGG [3].

Following our previous study of histone marks profiling in pediatric astrocytomas, revealing a significant role of H3K9me3 mark in pediatric gliomagenesis [11], we investigated in more detail the biological role of its writer SUV39H1 in fundamental cancer cell properties such as proliferation, migration, and adhesion as well as EMT regulation.

Bioinformatic analysis of a pediatric astrocytoma dataset [22] showed increased mRNA expression of SUV39H1 methyltransferase in pHGG compared to normal brain tissue and associations with cell adhesion processes as well as a mesenchymal transcriptional profile. In agreement with this hypothesis, investigation of the nuclear SUV39H1 expression in our cohort of pHGG tissues was found elevated compared to normal brain, along with increased H3K9me3 levels compared to normal brain.

Further investigation of the functional role of SUV39H1 in two pHGG patient-derived cell lines by gene silencing revealed that SUV39H1 knockdown significantly decreased cell viability, indicating its involvement in cell proliferation and growth processes. Accordingly, previous studies have shown that treatment of pediatric glioma cells with the H3K9 methylation inhibitor chaetocin significantly decreases cell proliferation, possibly through downregulation of SUV39H1 activity [11].

Furthermore, our data show that SUV39H1 suppression significantly reduces cell migration in both pHGG cell lines, suggesting the implication of this methyltransferase in glioma progression by modulating cell adhesion and migration. To elucidate the underlying mechanism of SUV39H1 in cell adhesion, we investigated the potential effects of SUV39H1 silencing on the expression levels of vimentin, a critical organizer of cell attachment, migration, and cell signaling. Vimentin is known to be highly expressed in glioma cells and implicated in tumor progression, maintaining their resistance to inhibition of oxidative phosphorylation [29]. Upon SUV39H1 silencing, vimentin expression levels were significantly reduced, indicating the potential interaction of these two molecules.

Following this reduction in vimentin levels, we proceeded to investigate the potential effects of SUV39H1 in β-catenin and its downstream transcription factor TCF4 in the two pHGGs cell lines. It is known that pediatric astrocytomas are characterized by loss of E-cadherin (CDH1) and induction of N-cadherin (CDH2) expression, triggering a series of events that allow cancer cells to become less dependent on their microenvironment, thereby promoting cell migration, invasion, and metastasis [30–33]. Since the intracellular domain of E-cadherin binds β-catenin (CTNNB1), loss of E-cadherin expression causes β-catenin to locate to the nucleus where it acts as a transcription factor for various genes associated with the mesenchymal phenotype [32, 33].

In agreement, SUV39H1 silencing in pHGG cell lines was shown to significantly reduce β-catenin and TCF4 levels, indicating the potential regulation of mesenchymal phenotype through this route. Moreover, p-GSK3α/3β levels remained unaffected upon SUV39H1 silencing, suggesting that SUV39H1 does not possibly employ the canonical Wnt/β-catenin axis to mediate its effects.

Although the direct interaction of SUV39H1 and β-catenin has not been previously demonstrated, a regulatory crosstalk has been proposed in BMSCs and different cancer types via H3K9me3 silencing of β-catenin target gene promoters [34, 35]. In colon cancer, SUV39H1 has been shown to repress c-Myc, Axin2, or Cyclin D1 genes, which are typically activated by β-catenin/TCF complexes [36], while in breast cancer, it can repress EMT-related genes via H3K9me3, indirectly suppressing Wnt/β-catenin-induced EMT [37]. The study of Song et al. has also shown that epigenetic silencing of the Wnt antagonist DKK1 through SUV39H1-mediated H3K9me3 may enhance β-catenin signaling [38]. Recently, two different studies have shown the regulation of glioma cell stemness by β-catenin and SUV39H1, promoting therapy resistance [39] and conferring to GBM progression [18], suggesting a possible interaction of SUV39H1 and β-catenin in GSCs.

In addition, high β-catenin/TCF4 activity has been previously involved in glioma progression through regulation of genes involved in cell proliferation and metastasis. Although the β-catenin/TCF4 axis is essential for normal brain development and homeostasis, there is evidence that it is often hijacked in high-grade gliomas, promoting tumorigenesis, self-renewal, and therapy resistance [40]. To this end, inhibition of β-catenin/TCF4 activity has been shown to block glioma growth in vivo, suggesting an important role of this axis in malignant progression [41].

Since the ΕMT process is known to be subjected to dynamic epigenetic regulation and reprogramming of gene expression, mediated by histone modifying enzymes [42], we investigated any potential effects of SUV39H1 in EMT markers (CDH2, SNAI1, MARCKS) known to be regulated by β-catenin/TCF4. SUV39H1 silencing was found to significantly reduce CDH2 in both cell lines, indicating a positive correlation between these molecules along with a reduction in SNAI1. The interaction of SUV39H1 with CDH2 was also revealed by our bioinformatic analysis in a different pHGG cohort, and there is evidence of interplay with SNAI1, which can recruit SUV39H1 to promote DNA methylation of the CDH1 promoter in breast cancer cells [43, 44].

In addition, the EMT regulator MARCKS, which is often overexpressed in HGG, was also reduced upon SUV39H1 silencing in both cell lines. Previous investigations have shown a correlation of MARCKS expression with short survival time and patients’ prognosis, while MARCKS inhibition in vitro reduced cell migration and invasion, leading to decreased expression of its downstream EMT-associated genes [45]. A positive correlation of SUV39H1 expression and MARCKS levels was also revealed by our bioinformatic analysis, strengthening their interaction in pHGG.

Although our study was performed in a limited number of cell lines, our findings demonstrate for the first time the distinct functional role of SUV39H1 in pediatric diffuse HGG contributing to their progression through modulation of β-catenin/TCF4 axis and EMT gene regulation, presenting a promising therapeutic target that needs further investigation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

Conceptualization: Christina Piperi. Methodology: Alexia Klonou, Penelope Korkolopoulou, Angeliki-Ioanna Giannopoulou, Antonios N. Gargalionis. Formal analysis and investigation: Dimitrios S. Kanakoglou, Andromachi Pampalou, Christos Adamopoulos, Panagiotis Sarantis, Andreas Mitsios, Spyros Sgouros. Writing—original draft preparation: Christina Piperi, Alexia Klonou. Writing—review and editing: Christina Piperi, Christos Adamopoulos, Penelope Korkolopoulou, Athanasios G. Papavassiliou. Funding acquisition: Athanasios G. Papavassiliou. Resources: Penelope Korkolopoulou, Spyros Sgouros, Andreas Mitsios. Supervision: Christina Piperi, Athanasios G. Papavassiliou.

Funding

Open access funding provided by HEAL-Link Greece.

Data Availability

The data that support the findings of this study are openly available in GSE243682 [23].

Declarations

Ethical approval and consent to participate

The study has been approved by the Ethics Committee of the University of Athens Medical School Ethics Committee (27/06/2017, 1617031069). All study procedures adhere to the tenets of the Declaration of Helsinki. Informed consent was obtained from the parents of all individual participants included in the study.

Consent for publication

The parents of patients have signed informed consent regarding publication of their data and photographs.

Competing interests

The authors declare no competing interests.