G9a promotes immune suppression by targeting the Fbxw7/Notch pathway in glioma stem cells

Abstract Aim Immunotherapy for glioblastoma multiforme (GBM) is limited because of a strongly immunosuppressive tumor microenvironment (TME). Remodeling the immune TME is an effective strategy to eliminate GBM immunotherapy resistance. Glioma stem cells (GSCs) are inherently resistant to chemotherapy and radiotherapy and involved in immune evasion mechanism. This study aimed to investigate the effects of histone methyltransferases 2 (EHMT2 or G9a) on immunosuppressive TME and whether this effect was related to changes on cell stemness. Methods Tumor‐infiltrating immune cells were analyzed by flow cytometry and immunohistochemistry in orthotopic implanted glioma mice model. The gene expressions were measured by RT‐qPCR, western blot, immunofluorescence, and flow cytometry. Cell viability was detected by CCK‐8, and cell apoptosis and cytotoxicity were detected by flow cytometry. The interaction of G9a and F‐box and WD repeat domain containing 7 (Fbxw7) promotor was verified by dual‐luciferase reporter assay and chromatin immunoprecipitation. Results Downregulation of G9a retarded tumor growth and extended survival in an immunocompetent glioma mouse model, promoted the filtration of IFN‐γ + CD4+ and CD8+ T lymphocytes, and suppressed the filtration of PD‐1+ CD4+ and CD8+ T lymphocytes, myeloid‐derived suppressor cells (MDSCs) and M2‐like macrophages in TME. G9a inhibition decreased PD‐L1 and increased MHC‐I expressions by inactivating Notch pathway companying stemness decrease in GSCs. Mechanistically, G9a bound to Fbxw7, a Notch suppressor, to inhibit gene transcription through H3K9me2 of Fbxw7 promotor. Conclusion G9a promotes stemness characteristics through binding Fbxw7 promotor to inhibit Fbxw7 transcription in GSCs, forming an immunosuppressive TME, which provides novel treatment strategies for targeting GSCs in antitumor immunotherapy.


| INTRODUC TI ON
Glioblastoma multiforme (GBM) is the most lethal primary tumor in the central nervous system. Although precise surgical resection in combination of radiotherapy and chemotherapy, most tumors inevitably recur, and the median survival time is less than 15 months. 1 Glioma stem cells (GSCs), a small subpopulation of GBM cells, have the capability to self-renew, proliferate, and multiple differentiation.
GSCs are inherently resistant to chemotherapy and radiotherapy, causing invasion, and recurrence. 2 Several strategies including in facilitating an inflammatory tumor microenvironment (TME), targeting stemness markers, and changing tumor characteristics by epigenetic modification make GSCs more susceptible to therapy in GBM. 3 Notch signaling pathway plays an important role in promoting tumor dormancy escape, recurrence, and progression after conventional therapy. 4 Notch intracellular domain (NICD), the active intracellular form of Notch after ligand binding, in the nucleus of tumor cells has been demonstrated using immunohistochemistry in human glioma samples. 5 Notch1 expression is higher in high-grade glioma compared to low-grade glioma. 6 Previous study has shown that activated Notch1 maintained a stemness phenotype in GSCs, and its downstream targets were highly upregulated in GSCs. 7 Notch inhibitor decreased population growth, clonogenicity, and expressions of GSCs markers in GBM-derived neurospheres, 8 facilitating conventional therapy.
A large number of immunosuppressive cells are involved in GBM microenvironment, and GBM cells generate immunosuppressive factors and highly express immune checkpoint ligand to inhibit local immune response. 9 Activated Notch signal in GSCs regulates molecular adaptation to the local immunosuppressive TME. 10 Single-cell RNA sequencing found that Notch signaling regulated tumor immune microenvironment, and Notch blockade delayed tumor recurrence. 11 Targeting Notch pathway, making tumor in an inflammatory TME, may be an effective strategy to sensitize immunotherapy.
The epigenetic modifiers have important functions on regulating gene expression and remodeling chromatin through DNA methylation or histone modifications. 12 Histone methyltransferases 2 (EHMT2 or G9a) as a permanent epigenetic marker predominantly governs histone H3 lysine 9 methylation (H3K9me). Studies have identified that G9a was correlated with malignancies, and G9a inhibitor decreased glioma cell viability. 13,14 G9a upregulation and concomitant H3K9me modification promoted transcriptional suppression of multiple genes. Lower H3K9me2 levels at the promoters of autophagy and differentiation-related genes were identified in differentiated cells than in GSCs. G9a inhibitors BIX01294 treatment upregulated the expressions of these genes in tumor spheres, 15 suggesting that G9a regulates cell stemness in GSCs. However, whether G9a is involved in the epigenetic regulation of Notch gene transcription in GSCs remains unclear.
Herein, we investigated the mechanism in which G9a remodeled tumor immune microenvironment by regulating stem cell signaling pathway. Our data showed that G9a promoted the expression of Notch signaling, and further demonstrated that G9a-mediated epigenetic silencing of F-box and WD repeat domain containing 7 (Fbxw7), a known Notch suppressor, elevated the expression of Notch1, which influenced the expressions of stem cell markers and immune-associated molecules in GBM progression.

| Cell culture
Human GSCs line 51A was from GBM patient as described previously. 16 GSCs were cultured in serum-free DMEM/F12 medium supplied with 2% B27, 20 ng/mL epidermal growth factor (EGF), and 20 ng/mL basic fibroblast growth factor (bFGF). 51A GSCs were cultured in high-glucose DMEM with 10% FBS to produce non-GSCs for 1 month. Human glioma cell line SHG140 was from primary culture of patient's glioma tissue in the First Affiliated Hospital of Soochow University. 17 Human glioma cell lines U251 and U87 were purchased from the Shanghai Institutes for Biological Sciences. Glioma cells were cultured in the medium of GSCs for 1 month, then CD133+ cells were sorted using CD133 MicroBead (Miltenyi Biotec) according to producer's direction, and defined as GSCs. The murine glioma cell line GL261 was obtained from American Type Culture Collection (ATCC), and cultured in high-glucose DMEM with 10% FBS. All cells were used after identification as previously described. 16,17

| Animal experiments in vivo
Male C57BL/6J and nude mice 18-20 g were fed in specific pathogenfree facilities at the Soochow University Animal Center. Experimental protocol was approved by the Medical Ethics Committee of the First Affiliated Hospital of Soochow University.
For tumor challenge experiments, C57BL/6J mice or nude mice were intracranially injected with 1 × 10 5 GL261 cells in 5 μL into the frontal lobe of cerebrum to establish intracranial allografts. All mice were sacrificed when they experienced suffering symptoms including inactivity, feeding interfere, and severe weight loss, and survival was recorded. In vivo imaging system (IVIS) 50 system was used to quantify the size of grafts at day 28 after C57BL/6J mice were implanted with GL261 cells carrying luciferase lentivirus. Bioluminescence imaging signal was reported as average flux. Intracranial grafts in mice were surgically resected to perform IHC or flow cytometry on the same day ( Figure 1B). For CD4 or CD8 T cell deletion, 20 μg of IgG, CD4, or CD8 antibody in 100 μL was i.p. administered three times per week.

| Cell transfection and stable cell lines generation
Cells were transfected with G9a-shRNA and Fbxw7-shRNA for gene knockdown. Plasmids carrying human G9a wild-type cDNA or SET site

| Immunohistochemical (IHC) staining
The tissues were fixed in formalin, embedded in paraffin, and sectioned at 4 μm. The slides were blocked with 5% BSA, incubated with primary antibodies including anti-CD4 and anti-CD8 (Abcam) at 4°C

| Flow cytometry
Tumor tissues from C57BL/6J mice were collected and dissociated into single-cell suspension by filtering with a 70μm mesh cell strainer. Red blood cells were lysed in ACK lysing buffer. Cells were resuspended in Percoll gradients and centrifuged at 400 g for 20 min to isolate lymphocytes. PMA, ionophore, and protein transport inhibitor were added into cell suspension for 5 h incubation.

| Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
Total RNA was isolated by TRIzol reagent (Invitrogen) and reversetranscribed to cDNA with a 1st Strand cDNA Synthesis Kit (Thermo Scientific). PCR was performed with SYBR Green PCR Master Mix

| Western blot
Total protein was extracted from lysis buffer (Beyotime Biotechnology) containing phosphatase inhibitor cocktail on ice.
The protein content was determined, then 20 μg of total protein was separated by SDS-PAGE gel and transferred onto PVDF membrane (Millipore). The membrane was blocked with 5% BSA for 2 h, then incubated with primary antibodies overnight at 4°C. After incubated with HRP-coupled secondary antibodies for 2 h at room temperature, the protein was analyzed by an enhanced chemiluminescence kit (Beyotime) and detected by chemiluminescence system (Bio-Rad).

Primary antibodies including anti-SOX2 (Cell Signaling) and anti-Oct4
(Cell Signaling) were used for immunostaining at 4°C overnight, then Alexa Fluor 488 or Alexa Fluor 555-conjugated secondary IgG antibody (Invitrogen) was added for reaction. The nuclei were stained with DAPI, and Images were captured by a fluorescence microscope.

| Dual-luciferase reporter assay
The binding of G9a and Fbxw7 promotor was validated using a dualluciferase reporter assay. The Fbxw7 promotor was synthesized and inserted into pGL3 luciferase vectors. The wild-type and SET deleted sequences for G9a were constructed by genepharma Co., Ltd. Fbxw7 promotor reporter plasmids were transfected into GSCs with G9a shRNA, wild-type or SET-deleted plasmid. Luciferase activity was measured 48 h after transfection using a dual-luciferase reporter gene assay system (Promega).

| Chromatin immunoprecipitation (ChIP)
ChIP experiment was performed using Chromatin IP Kit (Cell Signaling) according to manufacturer's protocol. In brief, GSCs were added with 37% formaldehyde for 10 min, then DNA was processed to the length of 150-900 bp by nuclease digestion and sonication.  Tumor tissue from GL261 cell-bearing mice was removed and dissociated into single-cell suspension (schedule shown in Figure 1B), then CD4+ and CD8+ T cells were analyzed using flow cytometry. n = 5. (B) IFNγ releasing and PD-1-expressing immune cells were showed. (C) The percentage of G-MDSCs and M-MDSCs was analyzed using flow cytometry. (D) M2-like macrophages were in all macrophage subpopulations were showed. (E) CD8+ T cells were purified from C57BL/6J mice spleen, then added into G9a-shRNA or G9a cDNAtransfected GL261 cells at indicated ratio of effector: target cells for 24 h co-culture. GL261 cells were labeled by CFSE, and double-positive cells of CFSE and PI staining were evaluated as dead tumor cells. (F) The expressions of MHC-I and PD-L1 were detected in GL 261 cells transfected with G9a-shRNA or G9a cDNA. (G) GL261 cells were transfected with vector or G9a cDNA, then co-cultured with CD8+ T cells at effector: target of 10: 1 in existence of 10 μg/mL anti-PD-L1 antibody or not. Cytotoxicity was measured using flow cytometry. *p < 0.05, **p < 0.01 vs. Scr or Vector, #p < 0.05 vs. G9a cDNA.

| Cell viability analysis
Cell viability was assessed using a CCK-8 assay according to the manufacturer's instructions. Briefly, 5000 cells were seeded in a 96well plate and treated with UNC0642 at the indicated concentration. CCK-8 was added and optical density at 492 nm was read using a microplate reader.

| Spheroid formation assay
The primary spheres were dissociated mechanically into single cells and resuspended in DMEM/F12 medium with EGF, bFGF, and B27.
Tumorspheres were observed and counted under an inverted microscope 5 days later.

| T-cell cytotoxicity assay
Human peripheral blood was collected in lithium heparin tubes, and peripheral blood mononuclear cells (PBMCs) were separated by density-gradient centrifugation using Lymphocyte Separation Medium
Murine CD8+ T cells were isolated from spleen of C57BL/6J mice.
Tumor cells were co-cultured with CD8+ T cells at indicated ratio, then cytotoxicity was detected using flow cytometry. Tumor cells were prelabeled by CFSE, dead cells were stained using PI, and the doublepositive cells of CFSE and PI were considered as dead tumor cells.

| Statistical analysis
Statistical analyses were performed using the Graphpad Prism 8.0 software. All data are presented as the mean ± standard deviation.
Unpaired two-tailed Student's t-test was performed to compare two groups, and one-way analysis of variance (ANOVA) with Tukey's post hoc test was performed among multiple groups. Prior to statistical analyses, the datasets for each group were tested for normality of distribution using the Kolmogorov-Smirnov test. Results with p values of <0.05 or <0.01 were set to be significant.

| G9a function in tumor growth promotion depends on the immune system
To analyze the role of G9a in tumor growth and survival, we stably transfected lentivirus vector carrying shRNA to knockdown G9a or cDNA to overexpress G9a in mouse GL261 cells ( Figure 1A).

F I G U R E 3
The effect of G9a downregulation on immune activation and cell apoptosis in GSCs. (A) GSCs were transfected with scramble, G9a-shRNA, and G9a cDNA, and the expressions of G9a were measured using western blot. (B) The expressions of MHC-I and PD-L1 were detected using flow cytometry. (C) CD8+ T cells were isolated from healthy human PBMCs, then co-cultured with GSCs. Cytotoxicity was measured using flow cytometry. (D) GSCs with vector or G9a cDNA co-cultured with CD8+ T cells in the existence of anti-PD-L1 antibody or not, then cytotoxicity was assayed. (E) CCK-8 assay was performed after GSCs were transfected with G9a-shRNA or treated with 20 μM UNC0642 for 48 h. (F) Cell apoptosis was detected using flow cytometry after G9a-shRNA transfection or 20 μM UNC0642 treatment for 24 h. *p < 0.05, **p < 0.01 vs. Scr or Vector, #p < 0.05 vs. G9a.

| G9a inhibited local antitumor immunity and modulated immune cell subpopulations
To explore the effect of G9a on local anti-tumor immunity in tumor tissue, we analyzed subtypes of immune cells in the TME of GL261 transplanting C57BL/6J mice using flow cytometry. G9a knockdown remarkably increased the percentage of CD8+ and CD4+ T cells in tumor tissues. In contrast, G9a overexpression reduced the percentage of these subtypes (Figure 2A) Figure 2G). These results suggested that G9a promoted the formation of immunosuppressive TME through PD-L1 upregulation.

| G9a downregulation suppressed immune activation and promoted cell apoptosis in GSCs
The interaction of GSCs and immune system promotes immune evasion and tumor growth by establishing an immunosuppressive TME. 19 We investigated the effect of G9a on immune molecules To investigate the effect of G9a on cell proliferation and apoptosis in GSCs, G9a-shRNA and UNC0642, a G9a inhibitor, were used.
UNC0642 treatment showed a dose-and time-dependent decrease in cell proliferation ( Figure S1). After G9a knockdown or inhibition in GSCs, CCK8 assays showed that cell proliferation was significantly decreased ( Figure 3E). Flow cytometry showed an increased percentage of apoptotic cell ( Figure 3F), suggesting the role of G9a on promoting proliferation and inhibiting apoptosis.

| G9a downregulation decreased stem characteristics in GSCs
To investigate the role of G9a in keeping stemness, we first detected the relative levels of G9a in non-GSCs and GSCs. Western blot ( Figure 4A and Figure S2A), immunofluorescence ( Figure 4B and Figure S2B), and RT-qPCR ( Figure 4C and Figure S2C) demonstrated that G9a expressions were increased in identified 4 GSC lines than in corresponding non-GSC lines. G9a knockdown or inhibition dramatically decreased the expression of pluripotent transcription factor SOX2 by immunofluorescence assay ( Figure 4D) and spheroid formation ability ( Figure 4E) in GSCs.

| G9a promoted stemness characteristics of GSCs and suppressed immune response via Notch signaling pathway
Notch1, an oncogene in GBM, is required for proliferation and survival of stem cells, and Notch signaling plays an important role in immune responses in tumors. 20 To investigate whether G9a regulates immune response through Notch1 signaling pathway, we observed the expressions of Notch1 pathway-associated genes after G9a knockdown or inhibition in GSCs. The expressions of Notch 1 signaling and its downstream target genes HES1 were significantly increased in GSCs than in non-GSCs ( Figure S3). Compared with control group, G9a knockdown or inhibition decreased expressions of Notch1, HES1, and c-myc ( Figure 5A). Furthermore, we transfected G9a cDNA to observe the effect of G9a on cell stemness and immune molecules regulation when Notch1 signaling pathway was blocked.
Notch1 was activated to maintain stemness in GSCs ( Figure S3), and Notch1 pathway blockade abrogated the increase in spheroid formation ability ( Figure 5B) and the expressions of stemness markers ( Figure 5C) induced by G9a overexpression. Moreover, the expressions of MHC-I and PD-L1 were similarly reverted by Notch1 inhibitor ( Figure 5D). In addition, Notch1 inhibitor increased cytotoxicity of CD8+ T cells in co-culture with G9a overexpressing GSCs ( Figure 5E).
These data suggested that G9a maintained stemness characteristics of GSCs through Notch1 signaling pathway.

| G9a activated Notch signaling via binding to Fbxw7 promotor
Fbxw7 suppresses cancer niche proliferation through ubiquitination and degradation of Notch pathway. 21 To investigate the mechanism of Fbxw7 on regulating stemness in GSCs, we speculated that Fbxw7 might be a potential G9a interaction partner. Both western blot ( Figure 6A) and RT-qPCR ( Figure 6B) results showed that the expressions of Fbxw7 were increased by G9a knockdown or inhibition, suggesting that G9a regulated Fbxw7 expression in transcriptional level. Furthermore, recombinant pGL3 plasmid carrying Fbxw7 promotor was constructed and transfected into GSCs. Dual-luciferase assays showed that luciferase expression was notably increased by  Figure 6I). Taken together, these data indicated that G9a upregulated Notch signaling by suppressing Fbxw7 promotor H3K9me2.