H3.3K27M Mutation Promotes The Migration and Invasion of Glioma Cells By Activating β-Catenin/USP1 Signaling

H3.3K27M is a newly identied molecular pathology marker in glioma and is especially correlated with the malignancy of diffuse intrinsic pontine glioma (DIPG). In recent years, accumulating research has revealed that other types of glioma also contain the H3.3K27M mutation. However, the role of H3.3K27M in high-grade adult glioma, which is the most malignant glioma, has not been investigated. In this study, we focused on exploring the expression and function of H3.3K27M in high-grade adult glioma patients. We found that H3.3K27M is partly highly expressed in high-grade glioma tissues. Then, we introduced H3.3K27M into H3.3 wild-type glioma cells, U87 cells and LN229 cells. We found that H3.3K27M did not regulate the growth of glioma in vitro and in vivo; however, the survival of mice with transplanted tumors was signicantly reduced. Further investigation revealed that H3.3K27M expression mainly promoted the migration and invasion of glioma cells. Moreover, we certied that H3.3K27M overexpression enhanced the protein levels of (cid:0)-catenin and p-(cid:0)-catenin, the protein and mRNA levels of ubiquitin-specic protease 1 (USP1), and the protein level of enhancer of zeste homolog 2 (EZH2). Importantly, the (cid:0)-catenin inhibitor XAV-939 signicantly attenuated the upregulation of the aforementioned proteins. Overall, the H3.3K27M mutation is present in a certain proportion of high-grade glioma patients and facilitates a poor prognosis by promoting the metastasis of glioma by regulating the (cid:0)-catenin/USP1/EZH2 pathway.


Introduction
Gliomas are a primary type of tumor in the central nervous system. In the 2016 World Health Organization (WHO) classi cation, glioblastoma (GBM) was de ned as grade IV glioma and makes up 45.2% of all malignant central nervous system (CNS) tumors and 80% of all primary malignant CNS tumors [1,2]. GBM and other high-grade gliomas are challenging for neurosurgeons, as no effective treatment exists.
Surgical treatment, radiotherapy and chemotherapy are the main treatment methods, but their curative effect falls short of patients' expectations [3]. Therefore, many neuroscientists put much effort into studying malignant glioma to nd effective ways to cure this disease.
Histones are the core components of the nucleosome subunit, forming an octamer containing four core histones (H3, H4, H2A, and H2B) surrounded by a 147-base pair DNA fragment. Histone tails are in uenced by a wide range of covalent posttranslational modi cations (PTMs) that jointly regulate chromatin status [4]. These PTMs can change the electronic charge and structures of these histone tails, which bind to the DNA, to alter the chromatin status and subsequent gene expression [5]. The PTM of histones is closely related to the occurrence of many tumors [6]. As histone mutations directly affect histone PTMs that are associated with gene activity, it is likely that these mutations contribute to tumor development through either the activation of oncogenes or the repression of essential tumor suppressor genes [7]. Thus, it is necessary to explore the role of H3.3 mutation in the mechanism of tumor development.
In the 2016 WHO classi cation, the H3 K27M mutant was rst added to the list of diffuse glioma for the new classi cation [1]. In recent years, scientists have focused on H3.3K27M-mutant diffuse midline gliomas. However, an increasing number of scholars have reported that some other kinds of intracranial tumors, such as ependymomas and gangliogliomas, exhibit the H3.3K27M mutation [9,10]. Researchers found that K27M-mutant diffuse midline gliomas are associated with signi cantly worse survival across all midline tumor locations [11]. Accordingly, H3.3K27M plays an important role in the occurrence and progression of intracranial tumors; nevertheless, the potential molecular mechanism of H3.3K27M is not very clear and needs further investigation.
Enhancer of Zeste homolog 2 (EZH2) is a signi cant constituent of polycomb repressive complex 2 (PRC2). It plays an oncogenic role in the occurrence and progression of tumors by regulating epigenetic genes [12]. PRC2 is a chromatin-associated methyltransferase catalyzing the mono-, di-, and trimethylation of lysine 27 on histone H3 (H3K27) [13]. EZH2 has been reported as a potential therapeutic target for H3K27M-mutant pediatric gliomas [14]. However, the function of EZH2 in adult H3K27M-mutant glioma is not very clear.
In this study, we rst assessed the expression of H3.3K27M in glioma patients and glioma cell lines by Western blotting. Then, we introduced H3.3K27M into glioma cells and investigated its effect on proliferation, migration and invasion. Finally, we investigated the underlying mechanisms by which the H3.3K27M mutation promotes the progression of glioma.

Glioma specimens
Glioma tissues and normal brain tissues were collected at the A liated Hospital of Xuzhou Medical University (Xuzhou, Jiangsu, China). All human glioma specimens were assessed to con rm the pathological diagnosis. The research was approved by the Research Ethics Committee of Xuzhou Medical University, and written informed consent was obtained from patients who underwent surgery at the A liated Hospital of Xuzhou Medical University.

Cell Counting Kit-8 (CCK-8) assay
Two hundred microliters of medium containing 4000 cells was seeded in a 96-well plate. At the designated time points, the original medium was replaced with medium containing 10% CCK-8 reagent (Victimed, Xuzhou, China) and incubated at 37°C for 2 h. Afterward, the absorbance at 450 nm was detected with a microplate reader. Cell viability was calculated based on absorbance values.

Colony formation assay
Five hundred cells suspended in 5 mL of medium were inoculated into each 60 mm dish and cultured continuously until macroscopic cell colonies were formed. Then, the cells were xed with 100% methanol and stained with 0.1% crystal violet solution for 15 min. After washing with PBS, the plates were photographed using a digital camera. Colonies containing more than 50 cells were counted manually.
Wound healing assay Stable cell lines were inoculated in six-well plates under normal conditions for 24 h. On the second day, the best cell density to perform the scratch assay was when cells were close to con uent. After the scratch was completed by the pipette tip in the middle of the wells, the unattached cells were washed with PBS twice, and the culture medium was replaced with serum-free culture medium. Photographs were taken at 0 h, 24 h, 48 h or 0 h, 12 h, and 24 h by an inverted microscope (IX71; Olympus, Tokyo, Japan).

Transwell invasion and migration assays
Transwell assays were carried out with a polycarbonate lter membrane with a diameter of 6.5 mm and pore size of 8 µm (Corning, Bedford, MA, USA) according to the manufacturer's protocol. Matrigel (BD, San Jose, CA, USA) was used to precoat the lters to analyze cell invasion. Trypsin-treated cells were resuspended in serum-free medium. A total of 100 µl of cell suspension containing 3000 cells was added to the upper compartment, and 500 µl of 3% FBS medium was added to the lower compartment as a chemoattractant. After incubation at 37 ℃ for 36 h, the chamber was washed twice with PBS to remove the noninvasive cells from the upper surface. Then, the lters were xed with methanol for 30 min. After washing with PBS twice, crystal violet was added for 30 min. After drying, photographs of ve randomly selected elds from each well were taken using an inverted microscope. The same experimental design was used for migration experiments except that the lters were not pretreated with Matrigel.

Orthotopic mouse model
Animal experiments were approved by the ethical committee and met the standards required by the guidelines of Xuzhou Medical University (Xuzhou, China). First, we constructed luciferase-mCherry-U87 cells, and then we used lentiviruses to construct cells that are stably expressed the vector Myc-H3F3A and Myc-H3.3K27M in luciferase-mCherry-U87 cells. After narcotizing all the nude mice, we incised the skin of the mouse head for approximately 1 cm and then drilled a small hole 3 mm deep at 1.8 mm right lateral to bregma. Then, the constructed U87 cells (5×10 5 ) were diluted in L15 medium and injected into the striatum of 7-week-old male nude mice by a microinjector. Finally, the incision was sutured, and all mice were rewarmed and put into the cage. A luciferase assay was performed to ensure the existence of glioma on day 7 posttransplantation, and the glioma growth rate and location of the tumors were determined using an IVIS kinetic imaging system. The glioma growth rate was assessed using uorescence imaging analysis.

Statistical analysis
All experiments were performed three times, and data are presented as the mean ± SD. Data were analyzed with GraphPad Prism 7. One-way ANOVA, two-way ANOVA and Tukey's multiple comparisons test were used to analyze differences in each three-group comparison. Overall survival curves were generated using the Kaplan-Meier method and compared using the log-rank test. P < 0.05 was considered to indicate statistical signi cance.

Results
The H3.3K27M mutation is present in a portion of human glioma patients To assess the expression of H3.3K27M in glioma patients, we performed Western blot assays with the total protein from 6 nontumor tissues and 22 glioma tissues by using H3K27M antibody. Interestingly, the H3K27M mutation was found in three glioma patients, all of whom were diagnosed with high-grade glioma (one sample was grade III, and two samples were grade IV) (Fig. 1A). Subsequently, we performed DNA sequencing to con rm the mutation in these three tissues (Fig. 1B). All of them were mutated from the AAG codon at position 27 to the ATG codon. Finally, to con rm whether the mutation was present in the commonly used glioma cell lines, we assessed the H3.3K27M mutation status in six glioma cell lines (U87, U251, LN229, A172, U118 and A172). However, no H3.3K27M mutation was found in these cell lines ( Fig. 1C and 1D).
H3.3K27M has no obvious effect on glioma cell proliferation but promotes glioma cell migration and invasion in vivo Due to the lack of H3.3K27M in these cell lines, we introduced the mutation by constructing cell lines stably expressing H3.3K27M. At the same time, we established a cell line overexpressing wild-type H3.3 as a negative control and a cell line overexpressing empty vector as a blank control. The successful generation of these cells lines was con rmed by Western blotting assay ( Fig. 2A). After constructing the cell lines, we focused on exploring the role of H3.3K27M in tumorigenesis. Colony formation and CCK-8 assays were used to determine the role of H3.3K27M in glioma cell proliferation. The expression of H3.3K27M had no effect on the proliferative ability of glioma cells U87 and LN229 (Fig. 2B, 2C and 2D).
Further detection at the molecular level indicated that overexpression of H3.3K27M did not change the expression of three classical proliferation pathway-related kinases, including the AKT, MEK1/2, and STAT3 pathways (Fig. 2E). Finally, we established a tumor model to con rm the above results by transplanting the above three cell lines expressing luciferase into the right striatum of nude mice. Tumor size was measured by luciferase assay every seven days. The overexpression of H3.3K27M had no obvious effect on tumor growth ( Fig. 3A and 3B). However, H3.3K27M expression signi cantly reduced the survival rate of mice (Fig. 3C). Moreover, by hematoxylin and eosin (HE) staining of mouse brain slices, we found that a large number of invasive tumor cells in ltrated into the surrounding tissues from primary tumors in the H3.3K27M group (Fig. 3D).

H3.3K27M promotes glioma cell migration and invasion in vitro
Based on the above results, we hypothesized that H3.3K27M mainly affects glioma cell motility. Thus, we explored its roles in glioma cell migration and invasion. The wound healing assay showed that the migration ability of U87 and LN229 cells stably expressing H3.3K27M was obviously enhanced compared with that of the control or WT group (Fig. 4A and 4B). The Transwell migration assay (without Matrigel) obtained similar results as the wound healing assay (Fig. 4C). Additionally, the Transwell invasion (with Matrigel) assay showed that the H3.3K27M group had a much stronger invasion ability than the control or WT group (Fig. 4D).
H3.3K27M positively regulates the β-catenin/USP1 signaling pathway In 2019, researchers reported that EZH2 participates in regulating the TGF-β pathway via a novel pathway axis that could potentially be relevant in regulating the metastasis and aggressiveness of GBM [15].
Recent studies also showed that EZH2 is a potential therapeutic target for H3K27M-mutant pediatric gliomas [14]. We thus hypothesized that EZH2 expression is involved in H3K27M mutation-induced glioma cell migration and invasion. We therefore detected the protein expression of EZH2 by Western blotting. As expected, the EZH2 protein was signi cantly upregulated upon overexpression of the H3.3K27M mutant (Fig. 5A). Recent studies have proven that the EZH2 protein is mainly regulated by the ubiquitin-proteasome-dependent degradation pathway, which is directly stabilized by ubiquitin-speci c protease 1 (USP1). Additionally, β-catenin is the main transcription factor that promotes the transcription of USP1 [16]. Based on this knowledge, we assessed whether EZH2 upregulation is attributed to thecatenin-USP1 axis. Then, we further detected the protein and mRNA levels of USP1. Western blotting and qRT-PCR results showed that overexpression of H3.3K27M upregulated USP1 levels at both the mRNA and protein levels in U87 and LN229 cell lines ( Fig. 5B and 5C). Furthermore, USP1 upstream transcription factor -catenin was found to be more signi cantly activated, which was re ected in the signi cant increase in the levels of both total -catenin and phosphorylated -catenin upon H3.3K27M overexpression. More importantly, to con rm that H3.3K27M-induced EZH2 upregulation was dependent on the -catenin/USP1 pathway, the -catenin inhibitor XAV-939 was used to perform a rescue experiment. Overexpression of H3.3K27M in U87 and LN229 cells signi cantly upregulated the levels of USP1 and EZH2, which could be signi cantly blocked by the -catenin inhibitor XAV-939 ( Fig. 5E and 5F). Taken together, these results indicate that H3.3K27M plays an important role in promoting glioma cell migration and invasion by activating the -catenin/USP1/EZH2 signaling pathway.

Discussion
In our study, we assessed the expression of H3K27M in 28 human glioma tissues. We found that three high-grade glioma patients had H3K27M mutations. To con rm the existence of H3.3K27M in these three samples, we performed DNA sequencing and found that all of them mutated from AAG codon at position glioma cell lines U87 and LN229. In the animal experiment, we found that H3.3K27M had no obvious effect on the growth of glioma but promoted the migration and invasion of glioma cells. In parallel, H3.3K27M had no effect on classical kinase-related proliferation pathways, including the AKT, MEK1/2, and STAT3 pathways. This interesting phenomenon motivated us to investigate the role of H3.3K27M in the migration and invasion of cell lines in vitro. Wound healing and Transwell assays demonstrated that H3.3K27M could increase the migration and invasion of U87 and LN229 cells. Finally, we focused on studying the mechanism by which H3.3K27M promotes migration and invasion. We discovered that H3.3K27M signi cantly activated the β-catenin/USP1/EZH2 pathway in U87 and LN229 cells. Our study reveals the molecular mechanism by which the H3.3K27M mutation promotes the malignancy of mutated glioma and provides some direction for research on targeted drugs.
In 2015, it was pointed out that although H3K27M mutations are frequently observed in brainstem and thalamic gliomas, such mutations often lead to a poor prognosis for brainstem gliomas [17]. However, Karremann et al showed that anatomical midline location, histopathological grade and tumor resection range had no effect on survival in H3K27M-mutant glioma [11]. Although anatomic location is controversial for patient outcomes, the poor prognosis of patients with H3K27M mutated glioma and the lack of treatment are issues that need to be addressed. Hence, the speci c molecular mechanism of H3.3 mutation-driven glioma has not been elucidated and should be further studied. The H3.3K27M mutant exists in children's and adults' glioma and may lead to a poor prognosis, so we wanted to investigate whether H3.3K27M has an effect on the malignant behaviors of glioma. We found that the expression of H3.3K27M is related to the migration and invasion of glioma and may explain the poor prognosis of H3.3K27M glioma. The nding of an active pathway may provide some inspiration for precision treatment of H3.3K27M-mutated glioma.
EZH2 is the methyltransferase component of PRC2. H3.3K27M histones bind to the SET domain of EZH2, causing a global reduction in H3K27 dimethylation and trimethylation (H3K27me2/3) [18]. However, researchers reported that the H3.3K27M mutant increased H3K27 methylation at unique loci in SF7761 cells due to the recruitment of EZH2. Their genome-wide sequencing data also suggest that the loss and gain of H3K27me3 by the H3.3K27M mutation may drive tumor formation in pediatric DIPG [19]. Li et al uncovered a SOX4-dependent epithelial-mesenchymal transition (EMT)-inducing mechanism underlying MTA1-driven cancer metastasis and suggested a widespread TGF-β-MTA1-SOX4-EZH2 signaling axis that drives EMT in various cancers [20]. Chen et al identi ed a novel metastasis-promoting lncRNA, MRPL23-AS1, which mediates the transcriptional silencing of E-cadherin by forming an RNA-protein complex with EZH2 [21]. EZH2 is also closely associated with the EMT of other tumors, such as pancreatic cancer, head and neck squamous cell carcinoma and esophageal cancer [22][23][24]. It has been reported that the EZH2 inhibitors MC4040 and MC404 reverse EMT and hamper cell migration and invasion, attenuating the glioma malignant phenotype [25]. Studies have also reported that EZH2 is involved in a novel miR-490-3p/TGIF2/TGFBR1 axis inducing migration and EMT in glioblastomas [15]. EZH2 is also connected with glioma proliferation and metastasis and has been regarded as a potential predictor and therapeutic target in glioma [26,27]. In general, EZH2 could be a marker associated with EMT in glioma. Our study revealed that EZH2 is closely related to the migration and invasion of glioma and supports that EZH2 plays an important role in the EMT of glioma.
Researchers have put much effort into studying the mechanism of H3.3K27M glioma due to the poor prognosis of H3.3K27M-mutant glioma in children, and they have made some advances in clarifying the molecular mechanism of DIPG. Mechanistically, research has revealed that the JNK pathway, NOTCH pathway, RAS/MYC axis and Rb/E2F1 pathway participate in the development of H3.3K27M-positive gliomas [28][29][30][31]. In our research, we wanted to explore the mechanism of enhanced invasion and migration. As previously mentioned, EZH2 participates in the migration and invasion of glioma. Thus, we detected the expression of EZH2 after overexpressing H3.3K27M in U87 and LN229 cells. As a result, the expression of EZH2 increased with H3.3K27M overexpression. It has been reported that abnormal ubiquitination and degradation are the main reasons for EZH2 protein accumulation, which is primarily regulated by the deubiquitinase USP1 [16]. We therefore focused on exploring the expression of USP1 and its upstream pathway. Our results showed that both the protein and mRNA levels of USP1 increased upon H3.3K27M overexpression. It is known that β-catenin acts as both a transcriptional coregulator and an adaptor protein for intracellular adhesion [32]. In the mechanism of glioma occurrence, β-catenin/TCF4 activates the transcription of the deubiquitinase USP1, which then directly interacts with and deubiquitinates EZH2. USP1-mediated stabilization of EZH2 promotes the occurrence and development of glioma [16]. Furthermore, we demonstrated that the upregulation of USP1 is attributed to the activation of -catenin, a classical signal transduction protein related to malignant progression of glioma, and the corresponding signaling pathway. These results indicate that H3.3K27M promotes glioma invasion and migration through the -catenin/USP1 signaling pathway. Nevertheless, -catenin/USP1/EZH2 has no obvious effect on the proliferation of glioma. As reported, the age of the patient and the speci c location of the H3K27M tumor lead to different biological behaviors [33]. H3.3K27M-mediated gliomagenesis was found to be dependent on PDGF signaling in a genetic mouse model [34]. Thus, we speculate that H3.3K27M has no direct effect on the proliferation of glioma, which might be due to the lack of special genetic mutations. On the basis of the above evidence, H3.3K27M promotes glioma malignant behavior by activating the -catenin/USP1/EZH2 pathway.

Conclusions
In conclusion, we provided the rst evidence that H3.3K27M is linked with the -catenin/USP1 pathway. This signaling pathway is also involved in human glioma malignant activity. Our research results may provide some basis for studying the function of H3.3K27M in adult glioma. The -catenin/USP1 pathway may be a potential target for H3.3K27M glioma. However, we should conduct more exhaustive studies to explore the detailed mechanisms of H3.3K27M in future projects, such as whether H3.3K27M directly binds to β-catenin to activate this pathway.