Synapsin III promotes neuronal-like transdifferentiation of glioblastoma stem cells by disrupting JAG1-Notch1 interaction

Abstract

Background

Glioblastoma (GBM), a formidable and highly aggressive form of brain cancer, is predominantly driven by GBM stem cells (GSCs), which are characterized by their ability for self-renewal and aberrant differentiation. Targeting the terminal differentiation of GSCs represents a promising therapeutic strategy. This study aimed to elucidate the role of synapsin III (SYN3) in driving the differentiation of GSCs into neuron-like cells and its effect on the tumor-suppressive pathways in GBM.

Methods

Proliferation assays, limited dilution assays, immunocytochemistry, western blot, RT-qPCR, and GSC tumor models were used to determine gene function and assess the role of γ-secretase inhibitors. Co-immunoprecipitation and microscale thermophoresis were conducted to explore the underlying regulatory mechanisms. Intracranial orthotopic injection of adeno-associated virus (AAV) was performed to evaluate therapeutic potential.

Results

We demonstrate that SYN3, uniquely within the synapsin family, acts as a tumor suppressor by steering GSCs toward neuronal-like transdifferentiation. Mechanistically, SYN3 enhances the expression of Neuregulin 3 (NRG3), which serves as a non-canonical antagonist of Notch signaling by competitively binding to specific epitopes within the EGF-like domain of JAG1, a critical site for the canonical engagement of Notch receptors. This critical interaction disrupts the JAG1-Notch1 signaling pathway, a key mechanism driving GSCs toward neuronal-like transdifferentiation, thereby reducing their stemness. Furthermore, SYN3 demonstrated significant antineoplastic activity in a mouse model harboring GSCs. AAV-mediated overexpression of SYN3 markedly impeded GBM progression.

Conclusions

Our research reveals the therapeutic potential of SYN3 in regulating GSC fate and offers a novel differentiation-based approach for GBM therapy.

Graphical Abstract

Graphical Abstract.

Key Points.

1. SYN3 promotes the transdifferentiation of GSCs into neuron-like cells, reducing their stemness and tumorigenic potential.

2. SYN3 enhances NRG3 expression, disrupting the JAG1-Notch1 pathway, crucial for GSC stemness.

3. AAV-mediated SYN3 overexpression significantly inhibited GBM progression in a GSC-derived mouse model, highlighting its therapeutic promise.

Importance of the Study.

The synapsin family plays critical roles at presynaptic terminals, yet the functionality and mechanisms of synapsins in GBM have remained elusive. We recently identified SYN3 as the only member of the synapsin family acting as a tumor suppressor in GBM. This discovery reveals a novel role for SYN3, diverging from its traditional neurological functions. Specifically, SYN3 uniquely induces neuronal-like transdifferentiation in GSCs and mitigates tumorigenicity through the modulation of the Notch signaling pathway. Moreover, our research uncovers a novel interaction between NRG3 and JAG1 that effectively inhibits the transmission of Notch signaling. Therefore, these findings highlight the distinct therapeutic potential of SYN3 in GBM, setting it apart from other synapsin family members and offering a novel avenue for targeted GBM treatment strategies.

Glioblastoma (GBM) is an aggressive brain cancer with limited treatment options and a poor prognosis.¹ Despite significant advancements in therapies such as surgery, radiation, and chemotherapy, GBM remains highly resistant to treatment.^2,3 The core of this challenge lies in GBM stem cells (GSCs), known for their capabilities of self-renewal, unlimited proliferation, and aberrant differentiation.⁴ GSCs are pivotal in driving the progression of tumors, contributing to resistance against conventional therapies, and leading to frequent relapses and high mortality rates.⁵ Therefore, the advancement of GBM treatment hinges on the targeted elimination of GSCs, underscoring the urgent need for understanding the biological mechanisms that maintain the stemness of GSCs and contribute to their resistance.

Functionally, GSCs resemble neural stem cells (NSCs) in their capacities for self-renewal and differentiation into multiple lineages.⁶ Importantly, NSCs undergo differentiation into terminally differentiated neurons, a process that inherently leads to the loss of their self-renewal capability.⁷ It has been suggested that promoting the differentiation of GSCs into neuronal cells could serve as an effective strategy to reduce their tumorigenic potential.⁸ This approach, known as differentiation therapy, involves various methods, including the modulation of specific transcription factors and the application of differentiation-inducing agents, to encourage the terminal differentiation of GSCs.^9–11 Although initial studies have established a foundation for differentiation therapy, further understanding of the molecular mechanisms that drive GSC differentiation is crucial for developing more targeted and effective therapies. Consequently, differentiation therapy is at the forefront of innovative treatments, offering new prospects in the fight against GBM by directly targeting the adaptability and resistance of GSCs.

The Notch signaling pathway is fundamentally critical in regulating the neuronal differentiation of NSCs, effectively inhibiting their maturation into neuronal lineages. This pathway is equally crucial in maintaining the self-renewal capacity of GSCs, thereby preventing their differentiation into mature cell types.^12,13 Given its central role in sustaining the undifferentiated state of GSCs, the modulation of Notch signaling represents a promising therapeutic strategy to reduce the tumorigenic potential and enhance the treatment efficacy in GBM. Specifically, the strategic application of Notch inhibitors or other molecular interventions designed to disrupt this pathway can drive GSCs from their stem-like state towards a more differentiated and less malignant neural phenotype.

The role of tumor suppressor genes in modulating stem cell differentiation has emerged as a critical area of study. Recent findings indicate that the inactivation of p53 fosters a state of undifferentiation with a high self-renewal potential in both GSCs and NSCs,¹⁴ suggesting a promising strategy to promote differentiation, thereby mitigating the aggressive nature of GBM and improving therapeutic outcomes. An intriguing candidate is synapsin III (SYN3), a member of the synapsin protein family predominantly expressed in neurons, playing a pivotal role in neuronal development and synaptic function.¹⁵ Studies have shown that SYN3 tends to accumulate in the caudate and putamen regions in individuals diagnosed with Parkinson’s disease (PD).¹⁶ Interestingly, emerging evidence suggests a reverse association between PD and cancer,¹⁷ particularly highlighting the shared genes involved in both PD and GBM that exhibit contrasting patterns of dysregulation.¹⁸ For example, the tumor suppressor genes p53 and PTEN are found to be upregulated in PD,^19,20 whereas oncogenes EGFR and HIF-α are upregulated in GBM but are downregulated in PD.^21,22 This contrast in gene expression and pathway involvement between PD and GBM suggests a potential tumor-suppressive function for SYN3 in GBM, as its associated pathways and regulatory mechanisms in neurons might also play a significant role in influencing tumor development and progression.

In this study, we demonstrate that SYN3 exhibits inherent anti-tumor activity and facilitates the neuronal-like transdifferentiation of GSCs, leading to significant suppression of their tumorigenic potential. Mechanically, SYN3 increases the expression of neuregulin 3 (NRG3), a protein enriched in neural tissue containing the EGF-like domain.²³ NRG3 competitively binds to Jagged 1 (JAG1), inhibiting the canonical JAG1-Notch1 signaling pathway. Collectively, our findings suggest that SYN3 initiates a program promoting advanced neuronal-like differentiation in GSCs and merits further investigation as a potential therapeutic target for GBM.

Materials and Methods

Details of this section can be found in the Supplementary Materials.

Ethics Statements

All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee of Laboratory Animal Center, Henan University, China (HUSOM2022-065).

Co-immunoprecipitation (Co-IP)

IP lysis buffer (Beyotime) was used to lyse cells. Then cell lysates were incubated with 40 μL of protein A/G magnetic beads (Bimake) and 5 μg of primary target antibody overnight at 4 °C. After washing the samples with IP lysis buffer 5×, co-immunoprecipitated proteins were detected by western blot analysis.

Datasets

GBM datasets were downloaded from GlioVis data portals (http://gliovis.bioinfo.cnio.es/) and GEPIA (http://gepia.cancer-pku.cn/detail.php).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 8.0 software. All data are presented as mean ± SEM. Two-tailed student’s t-test was performed to determine the statistical significance of the difference. While multivariate analyses were conducted using ANOVA. P value < .05 was considered statistically significant (^*P < .05, ^**P < .01, ^***P < .001, ^****P < .0001, NS = nonsignificant).

Results

SYN3 as the Sole Synapsin Family Member Exhibiting Anti-Tumor Activity in GBM

Synapsins, including SYN1, SYN2, and SYN3, are primarily recognized for their essential roles in regulating synaptic vesicle dynamics at presynaptic terminals, sharing conserved domains.²⁴ However, recent studies reveal that these synapsin proteins have distinct neuronal functions. Notably, SYN3 is distinguished from SYN1 and SYN2 by its unique J domain in the C-terminal region, which imparts specialized functionality.²⁵ This prompted us to investigate whether the anti-tumor potential is shared across the synapsin family or a unique feature of SYN3, utilizing available public datasets. We initially investigated the relationship between synapsin family protein expression and patient prognosis using data from the TCGA and REMBRANDT databases. The result showed that high SYN3 mRNA expression was significantly associated with improved patient survival, while high SYN1 and SYN2 expressions did not correlate with better outcomes (Supplementary Figure 1A-F). Further analysis showed that SYN3 expression was markedly lower in GBM compared to other glioma subtypes and normal tissue, suggesting its potential involvement in GBM pathology (Supplementary Figure 1G and H). SYN3 mRNA expression was generally consistent across GBM subtypes, with a modest increase in the proneural subtype (Supplementary Figure 1I-K). Moreover, SYN3 expression negatively correlated with the stemness marker Nestin and positively correlated with the neuronal marker MAP2, suggesting a role in promoting neuronal differentiation (Supplementary Figure 1L and M).

To further explore the anti-tumor potential of SYN3 in comparison to SYN1 and SYN2, we overexpressed each synapsin member via lentiviral transduction in multiple GSCs and assessed their effects on GSC proliferation and self-renewal. RT-qPCR and western blot confirmed successful overexpression (Supplementary Figure 2A-I). Cell proliferation assays revealed that overexpression of SYN1 or SYN2 did not limit GSC proliferation (Figure 1A, Supplementary Figure 2J-M). In contrast, SYN3 overexpression significantly inhibited GSC growth (Figure 1A, Supplementary Figure 2N and O). Further immunocytochemical staining showed that SYN3 localized in the cytoplasm and membrane after overexpression and significantly reduced the proportion of Ki67-positive cells in GSCs. Specifically, SYN3 overexpression resulted in a 2-, 3-, and 4-fold reduction in the proportion of Ki67-positive cells in the 448, 131, and 83 GSCs, respectively (Figure 1B and C, Supplementary Figure 2P-S), indicating effective inhibition of GSC proliferation by SYN3.

Figure 1.

SYN3 acts as a tumor suppressor and orchestrates the neuronal-like transdifferentiation of GBM stem cells (GSCs). (A) Cell proliferation assays were performed using 448 GSC infected with SYN1 (left), SYN2 (middle), and SYN3 (right) or vector lentiviral constructs. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^***P < .001, NS, not significant. (B) 448 GSC after overexpressing SYN3 or GFP alone, were examined for Ki67 and SYN3 expression by immunocytochemical staining. DAPI was used to counterstain cell nuclei. (C) The percentage of Ki67 + cells and SYN3 + cells were quantified shown in (B). Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^*P < .05, ^***P < .001. (D) Schematic of GSCs infected with lentivirus. (E) Western blot analysis of Flag, Nestin expression in 448 GSC infected with SYN1-Flag (left), SYN2-Flag (middle), and SYN3-Flag (right) or vector lentiviral constructs. GAPDH or β-actin was used as a loading control. (F) LDAs were performed using 448 GSC infected SYN1 (left), SYN2 (middle), and SYN3 (right) expression or vector lentiviral constructs. (n = 24, ^****P < .0001; NS, not significant, t test). (G) Western blot analysis of NRG3, MAP2, NeuN, SYN3, and SYN2 expression in 448 GSC infected with SYN3 expressing or vector lentiviral constructs. β-actin was used as a loading control. (H) RT-qPCR analysis of SYN3, ASCL1, MAP2, NRG3, and NeuN in 448 GSC infected with SYN3 expressing vector lentiviral constructs. β-actin was used as a loading control. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^*P < .05, ^**P < .01, ^****P < .0001. (I) Immunocytochemical staining of SYN3, MAP2, and Nestin of 448 GSC induced with SYN3 or GFP alone. Nuclei were stained by DAPI. (J) The percentage of MAP2 + cells and Nestin + cells were quantified shown in (I). Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^*P < .05, ^***P < .001.

Next, we examined whether the synapsin members are involved in regulating GSC stemness (Figure 1D). Overexpression of SYN1 or SYN2 did not reduce Nestin expression or sphere-forming ability in the 448, 131, and 83 GSCs, whereas SYN3 overexpression led to a significant reduction in stemness across all three GSCs (Figure 1E and F, Supplementary Figure 3A-O). Additionally, SYN3 overexpression resulted in significant upregulation of MAP2, NeuN, and NRG3 expression (Figure 1G and H, Supplementary Figure 4A-D). Immunofluorescence analysis revealed morphological changes in GSCs overexpressing SYN3, including the formation of neuroblast-like structures with elongated axons and neurites, further supporting neuronal-like transdifferentiation (Figure 1I and J, Supplementary Figure 4E-H). These cells showed increased MAP2 expression and decreased Nestin expression, confirming successful differentiation.

Overall, our findings suggest that SYN3, unlike SYN1 or SYN2, significantly inhibits GSC proliferation and self-renewal while promoting neuronal differentiation. These results underscore SYN3’s potential as a unique therapeutic target for GBM treatment, offering a strategy to reduce tumorigenic properties by inducing the transdifferentiation of GSCs into less malignant, neuron-like cells.

SYN3 Induces GSC Transdifferentiation into Neuron-Like Cells by Inhibiting the Notch Signaling Pathway

To further explore the relationship between SYN3 and lineage differentiation, we analyzed the correlation between SYN3 and lineage differentiation signatures including astrocytic, oligodendrocytic, and neuronal lineages in the TCGA, CGGA, and REMBRANDT databases.²⁶ SYN3 showed the strongest positive correlation with neuronal differentiation compared to other lineages (Figure 2A, Supplementary Figure 5A-F).

Figure 2.

SYN3 induces GBM stem cells (GSCs) to transdifferentiate into neuron-like cells by impairing the Notch signaling pathway. (A) Correlation between the SYN3 expression and an astrocytic differentiation signature (left), neuron differentiation signature (middle), and oligodendrocyte differentiation signature (right) in the TCGA GBM dataset. Data are presented as mean ± SEM (n = 166). Two-tailed student’s t test. (B) Gene set enrichment analysis (GSEA) plots for the Notch pathway signature, P = .006 (left), the Wnt pathway signature, P = 0.126 (middle), and the Shh pathway signature, P = .192 (right), comparing high and low SYN3 expression levels. NES, normalized enrichment score. FDR-q, false discovery rate q-value. (C) Western blot analysis of neuron markers and Notch pathway proteins expression in 448 GSC infected with SYN3 expressing or vector lentiviral constructs. Vinculin was used as a loading control. (D) Western blot analysis of NICD, Nestin, Hes1, MAP2, NRG3, and NeuN in 448 GSC treated with DAPT (left, 2 μM) and Nirogacestat (right, 10 μM) for 48 hours. α-tubulin was used as a loading control. (E) Cell proliferation assays were performed using 448 GSC treated with DAPT (left, 2 μM, once 2 days) and Nirogacestat (right, 10 μM, once 2 days) or vehicle. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^***P < .001, ^****P < .0001. (F and G) LDAs were performed using 448 GSC treated with DAPT (F, 2 μM) and Nirogacestat (G, 10 μM) or vehicle. (n = 24, ^****P < .0001, t test). (H) Western blot analysis of SYN3, NICD, MAP2, Nestin, and Hes1 expression in 448 GSC infected with a vector or SYN3-expressing lentiviral construct, both SYN3- and NICD-expressing lentiviral constructs, or a control construct. α-tubulin was used as a loading control. (I) Cell proliferation assays were performed using 448 GSC infected with a vector or SYN3-expressing lentiviral construct, both SYN3- and NICD-expressing lentiviral constructs, or a control construct. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^***P < .001, ^****P < .0001. (J) LDAs were performed in 448 GSC infected with a vector or SYN3-expressing lentiviral construct, both SYN3- and NICD-expressing lentiviral constructs, or a control construct. (n = 24, ^****P < .0001, t test).

To investigate the molecular mechanisms underlying SYN3-mediated transdifferentiation of GSCs into neuron-like cells, we focused on key signaling pathways involved in maintaining GSC self-renewal, including the Notch, Wnt, and Sonic Hedgehog (Shh) pathways.^27–29 Gene enrichment analysis (GSEA) was conducted to examine the association between SYN3 and key signaling pathways using gene sets from the TCGA databases. The results revealed a significant negative correlation between SYN3 and the Notch signaling pathway, while no significant associations were found with the Wnt or Shh pathways (Figure 2B). To determine whether SYN3 regulates downstream signaling through the Notch pathway in GSC cultures, we overexpressed SYN3 in GSCs and examined its effect on downstream gene expression and neuronal-like transdifferentiation. Western blot analysis confirmed that SYN3 overexpression reduced NICD and Hes1 expression, while significantly increasing MAP2, NeuN, and NRG3 expression (Figure 2C, Supplementary Figure 5G and H). These findings suggest that SYN3 is sufficient to induce neuronal-like transdifferentiation of GSCs by inhibiting the Notch signaling pathway.

To further validate the involvement of the Notch signaling pathway in GSC neuronal-like transdifferentiation and the inhibitory effect of SYN3 on GSC stemness and proliferation through this pathway, we disrupted Notch signaling using 2 different γ-secretase inhibitors, DAPT and Nirogacestat, which block the final cleavage of Notch1. GSCs were treated with DAPT (2 μM) or Nirogacestat (10 μM) for 48 hours, followed by evaluation of neuronal-like transdifferentiation using western blot and RT-qPCR assays. Treatment with γ-secretase inhibitors led to a reduction in NICD and Hes1 expression, and significantly increased MAP2, NeuN, and NRG3 expression at the protein level in GSCs (Figure 2D, Supplementary Figure 5I-L). Consistent results were observed at the mRNA level (Supplementary Figure 6A-F). Furthermore, Notch signaling inhibition led to decreased stemness and proliferation in GSCs (Figure 2E-G, Supplementary Figure 6G-N).

To further confirm the role of SYN3 in regulating neuronal-like transdifferentiation via the Notch signaling pathway, we performed a rescue experiment by simultaneously overexpressing NICD and SYN3 in GSCs. Notably, NICD overexpression counteracted the effect of SYN3, as evidenced by reduced expression of MAP2 and increased levels of Nestin and Hes1 (Figure 2H, Supplementary Figure 7A and B). Additionally, restoring NICD expression in SYN3-overexpressing GSCs reinstated their self-renewal and proliferation abilities (Figure 2I and J, Supplementary Figure 7C-F). To investigate the combined effects of SYN3 overexpression and Notch inhibition, GSCs were treated with SYN3-expressing lentiviral constructs and DAPT. The results showed that the dual treatment further reduced NICD, Hes1, and Nestin levels, while significantly increasing MAP2, NeuN, and NRG3 expression (Supplementary Figure 7G-I). Collectively, these findings provide strong evidence that SYN3 regulates GSC neuronal-like transdifferentiation and self-renewal by modulating the Notch signaling pathway.

SYN3 Inhibits the Notch Signaling by Promoting NRG3 Expression

To clarify the molecular mechanism by which SYN3 promotes the transdifferentiation of GSCs into neuron-like cells via Notch signaling, we explored how SYN3 triggers the inhibition of this pathway. The Notch signaling cascade is initiated when transmembrane ligands on one cell bind to Notch receptors on an adjacent cell, leading to a series of events that regulate gene expression in the receiving cell.³⁰ Given that SYN3 regulates the release of neurotransmitters such as serotonin, dopamine, and glutamate through synaptic vesicle fusion,³¹ we hypothesized that SYN3 may also influence the release of secreted proteins in GSCs through similar vesicular mechanisms and that the secreted proteins interact with cell surface receptors, thereby modulating Notch signaling.

To test this hypothesis, we collected conditioned medium (CM) from SYN3-overexpressing GSCs (448, 131, and 83) and control GSCs, and then treated the GSCs with the respective CM samples (Figure 3A). Western blot analysis revealed that GSCs treated with CM from SYN3-overexpressing GSCs showed decreased expression of NICD, Hes1, and Nestin, along with increased levels of MAP2, NeuN, and NRG3, compared to those treated with control CM (Figure 3B). These results suggest that secreted factors present in the CM inhibit the Notch signaling pathway, with their release being mediated by SYN3. However, the specific factors controlled by SYN3 remain unidentified. To address this, we conducted an extensive analysis to identify potential correlations between SYN3 and various rate-limiting enzymes involved in neurotransmitter synthesis, as well as secreted proteins in the brain. Using the CGGA database, we identified the top 15 proteins most strongly correlated with SYN3 (Supplementary Figure 8A). Notably, our data revealed that SYN3 showed the highest correlation with NRG3 (r-value = 0.6, P < .0001; Figure 3C). Interestingly, the aforementioned experimental results demonstrated that NRG3 expression was upregulated upon SYN3 overexpression in GSCs (Figure 2C, Supplementary Figure 4A and B). Based on these findings, we hypothesized that SYN3 promotes GSC neuronal-like transdifferentiation by increasing the expression and release of NRG3, thereby inhibiting the Notch signaling pathway.

Figure 3.

SYN3 inhibits the Notch pathway by promoting the expression of NRG3. (A) Schematic of conditioned medium (CM) treatment in the GBM stem cells (GSCs) model. (B) Western blot analysis of NICD, Hes1, Nestin, MAP2, NRG3, and NeuN expression in GSCs treated with CM from GSCs infected with SYN3 or vector lentivirus. α-tubulin was used as a loading control. (C) Correlation dot-plot of NRG3 and SYN3 from the CGGA database. Data are presented as mean ± SEM (n = 651). Two-tailed student’s t test. (D) NRG3 expression in each type of glioma from the TCGA database. Data are presented as mean ± SEM. Two-tailed student’s t test. (E) Kaplan–Meier survival curves for GBM patients with high (top 50% contribution) and low (down 50% contribution) NRG3 expression based on the median expression in the CGGA dataset. n = 220, log-rank test. (F) Western blot analysis of MAP2, NeuN, Nestin, NICD, and Hes1 expression in 448 GSC (top), 131 GSC (middle), and 83 GSC (bottom) treated with rNRG3 (20 ng/ml) or vehicle for 48 hours. α-tubulin was used as a loading control. (G) Cell proliferation assays were performed using 448 GSC (top), 131 GSC (middle), and 83 GSC (bottom) treated with rNRG3 (20 ng/ml) or vehicle, treated cells with rNRG3 once 2 days. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^*P < .05, ^***P < .001. (H) LDAs performed using 448 GSC (top), 131 GSC (middle), and 83 GSC (bottom) treated with rNRG3 (20 ng/ml) or vehicle. (n = 24, ^****P < .0001, t test).

To explore the function of NRG3 in GBM, we analyzed its mRNA expression in 4 grades of glioma and found that NRG3 expression was downregulated in the highest-grade GBM (Figure 3D). Additionally, high expression of NRG3 was significantly associated with a better prognosis compared to low expression in GBM patients (Figure 3E, Supplementary Figure 8B). These results prompted us to investigate whether NRG3 induces GSC neuronal-like transdifferentiation through the Notch signaling pathway. We modulated NRG3 expression in GSCs by treating them with recombinant NRG3 (rNRG3) protein. A preliminary dose–response experiment (0, 5, 10, 20, and 50 ng/ml) identified 20 ng/ml as the optimal concentration, as it significantly reduced NICD expression (Supplementary Figure 8C and D). Based on this, GSCs were treated with 20 ng/ml rNRG3, and western blot analysis showed a reduction in the expression of NICD, Hes1, and Nestin, alongside an increase in the expression of MAP2 and NeuN after treatment. (Figure 3F). Moreover, the rNRG3 protein effectively inhibited the proliferation of GSCs (Figure 3G), and the number of sphere formations in GSCs was significantly reduced in response to rNRG3, indicating that NRG3 impairs the self-renewal ability of GSCs (Figure 3H). Collectively, our findings strongly suggest that SYN3 inhibits the Notch signaling pathway by regulating the expression of NRG3, thereby promoting the transdifferentiation of GSCs into neuron-like cells and suppressing the stemness of GSCs.

NRG3 Disrupts the Interaction Between JAG1 and Notch1

Next, we explored how NRG3 directly regulates the Notch signaling pathway. Previous studies have shown that Notch1’s extracellular domain contains up to 36 epidermal growth factor (EGF)-like domains and a negative regulatory region, while JAG1 comprises 16 EGF-like domains, an N-terminal C2 domain, a Delta-Serrate-Lag-2 (DSL) domain, and a cysteine-rich domain. The high-affinity binding region of the JAG1-Notch1 interaction involves EGF domains 8 and 12 of Notch1 and the EGF3 and C2 domains of JAG1.³² Interestingly, NRG3 has been reported to contain an extracellular domain with EGF-like domains that bind to the extracellular domain of ErbB4.^33,34 Based on these findings, we hypothesized that NRG3 competes with the EGF-like domains of JAG1 or Notch1, blocking the JAG1-Notch1 binding site, reducing the release of the Notch intracellular domain, and ultimately inhibiting the Notch signaling pathway.

To validate this hypothesis, we constructed overexpression plasmids for the extracellular domains of Notch1 and JAG1 and transfected 293T cells (Figure 4A). Successful overexpression of the specific Notch1 and JAG1 regions was confirmed by RT-qPCR (Supplementary Figure 9A and B). We further validated the interaction between Notch1 and JAG1 through co-immunoprecipitation (Co-IP) assay in 293T cells overexpressing Flag-tagged Notch1 and HA-tagged JAG1. Co-IP using Flag or HA as the precipitating antibody demonstrated the interaction between Notch1 and JAG1 (Supplementary Figure 9C and D). Additionally, we constructed a His-tagged NRG3 overexpression plasmid, confirming its expression by RT-qPCR (Supplementary Figure 9E). Co-IP with His as the precipitating antibody showed that NRG3 interacts with JAG1 but not with Notch1 (Figure 4B). Furthermore, Co-IP with HA antibody confirmed the interaction between JAG1 and NRG3 (Figure 4C), while Flag-tagged Notch1 does not interact with NRG3 (Figure 4D).

Figure 4.

NRG3 disrupts the interaction between JAG1 and Notch1. (A) Individual domains of Notch1 (EGF8-EGF12) and JAG1 (C2-EGF3) from the Notch1-JAG1 binding site. (B) Co-IP analysis for the interaction of NRG3 and JAG1, or NRG3 and Notch1 in 293T cells transfected with His-tagged NRG3 and HA-tagged JAG1 or His-tagged NRG3 and Flag-tagged Notch1. Cell lysates were precipitated with anti-His antibody. (C) Co-IP analysis for the interaction of JAG1 and NRG3 in 293T cells transfected with HA-tagged JAG1 and His-tagged NRG3. Cell lysates were precipitated with anti-HA antibody. (D) Co-IP analysis for the interaction of Notch1 and NRG3 in 293T cells transfected with Flag-tagged Notch1 and His-tagged NRG3. Cell lysates were precipitated with an anti-Flag antibody. (E) Binding affinity between NRG3 and JAG1 tested by MST assay in vitro. The concentration of NRG3 proteins is kept constant at 50 nM, while the JAG1 concentration varies from 0.8 µM to 0.012 nM. The binding curve yields a Kd of 687 nM. Inset, thermophoretic movement of fluorescently labeled proteins. Fnorm = F1/F0 (Fnorm: normalized fluorescence; F1: fluorescence after thermodiffusion; F0: initial fluorescence or fluorescence after T-jump). Kd, dissociation constant. (F) The binding affinity between NRG3 and Notch1 was tested by MST assay in vitro. The concentration of NRG3 proteins is kept constant at 50 nM, while the Notch1 concentration varies from 2.5 µM to 0.076 nM. (G and H) Co-IP of 448 GSC transfected with HA-tagged JAG1 (G) or His-tagged NRG3 (H) using antibodies targeting HA, His, or normal IgG. (I and J) Co-IP of 131 GSC transfected with HA-tagged JAG1 (I) or His-tagged NRG3 (J) using antibodies targeting HA, His, or normal IgG. (K) Co-IP analysis for the interaction quantify of Notch1 and JAG1 or NRG3, Notch1, and JAG1 in 293T cells transfected with Flag-tagged Notch1 and HA-tagged JAG1 or His-tagged NRG3, Flag-tagged Notch1, and HA-tagged JAG1. Cell lysates were precipitated with anti-Flag antibody. (L) Schematic of NRG3 competitively binds to JAG1 to disrupt the interaction between JAG1 and Notch1.

To further examine the direct interaction between NRG3 and JAG1, we conducted the microscale thermophoresis (MST) assay to determine the binding affinity and dissociation constant (Kd) in vitro. The results showed a clear binding curve for NRG3-JAG1 with a Kd of 681 nM (Figure 4E), indicating a high-affinity interaction. However, no binding affinity between NRG3 and Notch1 was observed (Figure 4F). Furthermore, we performed Co-IP using HA or His antibody in 448 and 131 GSCs transfected with HA-tagged JAG1 or His-tagged NRG3, interactions between JAG1 and NRG3 were observed (Figure 4G-J). These findings suggest that NRG3 serves as an interaction partner for JAG1, effectively obstructing the conventional interaction between Notch1 and JAG1, thereby impeding downstream signal transduction. To investigate whether NRG3 competes with the binding site where Notch1 and JAG1 interact, we also performed a Co-IP assay using Flag antibody in 293T cells. The results showed a significant reduction in the complex formation between Notch1 and JAG1 when the NRG3 overexpression plasmid was introduced (Figure 4K), indicating that NRG3 can competitively occupy the active binding sites of Notch1 and JAG1 (Figure 4L), consequently blocking the inherent ligand-receptor binding signaling and inhibiting the Notch signaling pathway.

SYN3 Suppresses GSC Tumorigenicity

After confirming that SYN3-driven neuronal-like transdifferentiation in GSCs is mediated by the repression of Notch signaling in vitro, we investigated whether SYN3 could inhibit GSC tumorigenicity in vivo. To test this, 83-luc GSCs were transduced with lentivirus encoding SYN3 or a control GFP-encoding vector prior to orthotopic xenotransplantation (Figure 5A). Tumor bioluminescence, body weight, and survival were monitored throughout the study.

Figure 5.

SYN3 overexpression in GBM stem cells (GSCs) represses GBM growth. (A) Schematic of GBM orthotopic model constructed with SYN3 overexpression. (B) Kaplan–Meier survival curves of mice implanted with 5 × 10³ 83-luc GSCs infected with a SYN3 expressing or vector lentivirus (n = 6 in each group, from independent biological replicates). Vector vs SYN3, log-rank test. (C) Representative H&E staining of the whole brain showing tumor mass (left), necrotic areas (middle), and tumor cellularity (right) in mice bearing orthotopic xenografts of 83-luc GSC infected with SYN3-expressing or vector control lentiviral constructs. The samples were collected 24 days after cell injection. Necrotic areas were identified through morphological analysis, and tumor cellularity was measured by counting the number of cell nuclei in images at × 40 magnification of the H&E-stained sections. (D) Relative cellularity of total tumor from mice implanted with 5 × 10³ 83-luc GSCs infected with SYN3 expressing or vector lentivirus. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^*P < .05. (E) Representative images and quantification of IHC staining for SYN3, MAP2, NeuN, Nestin, and Ki67 of the mice bearing orthotopic xenografts of 83-luc GSC infected with SYN3 expressing or vector lentiviral constructs. The sample was collected 24 days after cell injection. IOD, Integrated Optical Density. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed Student’s t test, ^*P < .05, ^**P < .01, ^***P < .001. (F) Kaplan–Meier survival curves of mice implanted with 1 × 10⁴ X01 GSCs infected with a SYN3 expressing or vector lentivirus (n = 6 in each group, from independent biological replicates). Vector versus SYN3, log-rank test. (G) H&E staining of X01 GSC infected with SYN3 expressing or vector lentiviral constructs. The sample was extracted 30 days after cell injection. (H) Representative images and quantification of IHC staining for SYN3, MAP2, NeuN, Nestin, and Ki67 of the mice bearing orthotopic xenografts of X01 GSC infected with SYN3 expressing or vector lentiviral constructs. The sample was extracted 30 days after cell injection. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed Student’s t test, ^**P < .01, ^***P < .001.

The results demonstrated that SYN3 overexpression effectively reduced the volume of intracranial tumors, as evidenced by bioluminescence measurements (Supplementary Figure 10A and B), and slowed the rate of weight loss in mice compared to the control group (Supplementary Figure 10C). Notably, enhancing SYN3 expression extended tumor latency, increasing median survival from 25 to 30 days (Figure 5B). H&E staining revealed a substantial reduction in tumor mass in the orthotopic mouse model with SYN3 overexpression, along with less extensive necrotic areas and reduced tumor cellularity (Figure 5C and D). Furthermore, we assessed the activation status of downstream proteins regulated by SYN3 in orthotopic mouse tissues derived from 83 GSC using immunohistochemistry (IHC) and immunofluorescence. SYN3-overexpressing tumors showed increased expression of NeuN and MAP2, along with decreased expression of Nestin and Ki67 (Figure 5E, Supplementary Figure 10D). These findings underscore the inhibitory effects of SYN3 on GSC tumorigenicity through the induction of neuronal-like transdifferentiation in the 83-GSC orthotopic xenograft models.

To further validate these results, we modulated SYN3 overexpression in X01 GSC, which had demonstrated high tumorigenicity in our previous studies.³⁵ Prior to conducting the orthotopic injections, we first confirmed the in vitro function of SYN3 in X01 GSC. RT-qPCR and western blot analysis confirmed the successful overexpression of SYN3 (Supplementary Figure 10E). We investigated whether SYN3 functions as a tumor suppressor by promoting neuronal-like transdifferentiation and inhibiting self-renewal in X01 GSC. SYN3 overexpression significantly increased MAP2 expression at both mRNA and protein levels, while downregulating NICD and Nestin expression and upregulating NeuN and NRG3 expression (Supplementary Figure 10F). LDAs demonstrated a reduced sphere-forming ability in SYN3-overexpressing X01 GSC (Supplementary Figure 10G), and microscopy revealed elongated axon-like structures, indicative of neuronal-like transdifferentiation (Supplementary Figure 10H). Subsequently, we performed an orthotopic mouse model using X01-derived GSC (Figure 5A). The results demonstrated that mice bearing SYN3-overexpressing X01 GSC had significantly longer median survival than those in the control group (Figure 5F). H&E staining revealed that SYN3 overexpression substantially reduced tumor volume (Figure 5G). Additionally, SYN3-overexpressing tumors exhibited elevated expression levels of MAP2 and NeuN, while Nestin and Ki67 expression was lower compared to controls (Figure 5H).

To address potential sex-based differences in neuronal differentiation,³⁶ we conducted in vivo experiments using male and female GBM mouse models. Male mice exhibited shorter survival times and faster tumor progression compared to females, consistent with previous findings (Supplementary Figure 11A-E).³⁷ However, SYN3 overexpression prolonged survival similarly in both sexes (Supplementary Figure 11A). Additionally, IHC staining for SYN3 and MAP2 showed no sex-specific variations in neuron markers (Supplementary Figure 12A-C), confirming that SYN3’s therapeutic effects are independent of sex.

Collectively, these findings from orthotopic mouse models using 83-luc and X01 GSCs strongly suggest that SYN3 functions as a tumor suppressor by promoting GSC transdifferentiation into neuron-like cells and reducing GSC tumorigenicity.

SYN3 as a Therapeutic Target for GSC Differentiation Therapy

Given the observed inhibition of GBM progression due to SYN3 overexpression in GSCs, further research was conducted to explore the viability of SYN3 as a therapeutic target for GBM. To this end, we utilized adeno-associated virus serotype 9 (AAV9), which enables targeted gene delivery directly to the GBM regions, maximizing therapeutic efficacy while minimizing potential off-target effects.³⁸ Additionally, AAV9 can sustain transgene expression for over six months,³⁹ providing prolonged therapeutic effects crucial for addressing the aggressive nature of GBM. These properties make AAV9 an ideal vector for the intracranial delivery of SYN3.

To assess the efficiency of the AAV in delivering SYN3 (AAV-SYN3) in vitro, X01 and 83 GSCs were exposed to either the control virus or AAV-SYN3 for 48 hours to induce SYN3 overexpression. Western blot and RT-qPCR analysis confirmed the successful overexpression of SYN3 (Figure 6A-D). Subsequently, cell proliferation assays demonstrated that AAV-mediated SYN3 overexpression significantly inhibited the growth of GSCs (Figure 6E and F) and markedly reduced their sphere-forming capacity (Figure 6G and H). These results clearly illustrate that AAV vectors can effectively mediate SYN3 overexpression, highlighting their potential utility in gene therapy applications.

Figure 6.

SYN3 as a therapeutic target for GBM stem cells (GSCs) differentiation therapy. (A and B) Western blot analysis of SYN3 expression in X01 GSC (A) and 83 GSC (B) treated with the control vector or AAV-SYN3. GAPDH was used as a loading control. (C and D) RT-qPCR analysis of SYN3 expression in X01 GSC (C) and 83 GSC (D) treated with the control vector or AAV-SYN3. GAPDH was used as a loading control. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed Student’s t test, ^**P < .01, ^***P < .001. (E and F) Cell proliferation assays were performed using X01 GSC (E) and 83 GSC (F) treated with the control vector or AAV-SYN3. Data are presented as mean ± SEM (n = 3 independent experiments). Two-tailed student’s t test, ^**P < .01, ^***P < .001. (G and H) LDAs performed using X01 GSC (G) and 83 GSC (H) treated with the control vector or AAV-SYN3. (n = 24, ^****P < .0001, t test). (I) Illustration of AAV treatment in GBM model mice. (J) Kaplan–Meier survival curves of mice implanted with 5 × 10³ 83-luc GSCs treated with the control vector or AAV-SYN3 (n = 6 in each group, from independent biological replicates). Vector vs SYN3, log-rank test. (K) H&E staining and IHC analysis of SYN3, MAP2, NeuN, CD44, and Ki67 of the mice bearing orthotopic xenografts of 83-luc GSC treated with the control vector or AAV-SYN3. The sample was extracted 19 days after cell injection.

To evaluate the therapeutic potential of elevating SYN3, AAV-SYN3 monotherapy was administered via intracranial injection on days 3 and 10 post-tumor inoculations (Figure 6I).³⁹ AAV-mediated SYN3 gene therapy inhibited GBM growth and reduced the rate of weight loss in mice compared to the control group (Supplementary Figure 13A and B). Notably, AAV-SYN3 therapy significantly prolonged overall survival in the 83 GBM models (Figure 6J). H&E staining revealed a substantial reduction in tumor mass in the orthotopic mouse model following AAV-mediated SYN3 overexpression (Figure 6K). Additionally, elevated levels of MAP2 and NeuN were observed in mice treated with AAV-SYN3, while the expression of CD44 and Ki67 was lower in SYN3-overexpressing tumors compared to the control virus (Figure 6K, Supplementary Figure 13C). Overall, these findings indicate that SYN3-mediated differentiation therapy effectively represses GBM progression.

Discussion

GBM, an aggressively invasive and predominantly incurable brain malignancy, presents formidable challenges in both treatment efficacy and prognostic outcomes, primarily due to the existence of GSCs.⁴⁰ This compels a deeper exploration into the molecular underpinnings of GSCs to unearth potential therapeutic targets. Our results demonstrate that SYN3 is a potentially critical tumor suppressor in GBM development by impairing the self-renewal ability and tumorigenicity of GSCs, challenging the traditional view of SYN3 being primarily associated with neurological functions. We elucidate a new role of SYN3 in orchestrating the transdifferentiation of GSCs into neuron-like cells, shedding light on the underlying molecular mechanisms. Notably, SYN3 overexpression in GSCs upregulates the expression of NRG3, which competes with JAG1 to impede the activation of the Notch signaling pathway, thus inhibiting the stemness and proliferation abilities of GSCs. Taken together, this groundbreaking revelation positions SYN3 at the forefront of inducing GSCs transdifferentiation into neuron-like cells, unveiling a novel molecular paradigm.

GSCs present a significant challenge for conventional therapies, which predominantly target differentiated tumor cells. GSCs exhibit properties akin to NSCs, including robust self-renewal capacity and the ability to differentiate into diverse cell lineages.⁴¹ Consequently, steering GSCs toward terminal differentiation stands out as a promising strategy in the battle against GBM.⁴² Our research validates that upon differentiation induced by SYN3, GSCs forfeit their tumorigenic traits, concurrently exhibiting a gene expression profile indicative of neuronal differentiation and maturation. This observation aligns with prior studies underscoring the notion that inducing differentiation in cancer stem cells can inhibit their proliferation.^11,43 In the context of human GBM, previous studies showed that GSCs can induce differentiation toward a neuronal lineage through the synergistic effect of combining ASCL1 high expression and γ-secretase inhibitor,⁹ or by enhancing the expression of NGN2 coupled with SOX11.⁴¹ Furthermore, the underlying mechanisms driving this process have remained elusive. Our study makes a pivotal advancement in this realm by demonstrating that SYN3 alone suffices to reprogram GSCs into mature neuron-like cells, characterized by elongated, extensive neurites. Moreover, we delve into the intricacies of the SYN3 signaling pathway, elucidating its role in fostering neuronal-like transdifferentiation. Compared to the well-known tumor suppressor p53 gene therapy, which has shown limited effectiveness in extending survival in GBM models,⁴⁴ SYN3 demonstrates a more robust tumor-suppressive effect. In addition to its direct impact on GSC differentiation, we also found SYN3 negatively correlates with both anti-tumor (activated CD8⁺ T cells, Th1 cells, and NK cells) and immune-suppressive cells (Tregs and MDSCs), while IHC analysis showed no changes in microglia/macrophages (Supplementary Figure 14). Thus, SYN3 impedes GSC proliferation by promoting neuronal identity, presenting a promising GBM therapy, though its role in the immune microenvironment requires further investigation.

There are 3 synapsin proteins in the synapsin family - SYN1, SYN2, and SYN3 - that play roles in synaptogenesis and the modulation of neurotransmitter release.⁴⁵ SYN1 and SYN2 are linked to epilepsy susceptibility.^46,47 while SYN2 and SYN3 are implicated in schizophrenia due to reduced expression in patients.^48,49 Structurally, each synapsin isoform has 5 main structural domains, with conserved N-terminal domains (A-C) and variable C-terminal regions defining isoform-specific functions. Recent studies highlight the distinct roles of these isoforms at presynaptic terminals.⁵⁰ Notably, our study identifies SYN3, unlike SYN1 and SYN2, as significantly associated with better prognosis in GBM patients, suggesting that the unique J domain in SYN3 may contribute to its anti-tumor properties.

Mechanistically, our findings illustrate that SYN3 induces GSCs to transdifferentiate into neuron-like cells by inhibiting the Notch signaling pathway, which maintains NSC homeostasis and inhibits neuronal differentiation by suppressing proneural transcription factors expression.⁵¹ Furthermore, Hes1, a downstream factor of the Notch pathway, can be inhibited to increase MAP2 expression,⁵² supporting the idea that the SYN3 signaling axis promotes GSC neuronal-like transdifferentiation by inhibiting Notch signaling. Our findings revealed that a putative signaling pathway involving SYN3 can enhance NRG3 which is a neural-enriched protein found in vesicles be released into extracellular space, which is consistent with the specific biological activities and method of action of SYN3.³¹ Interestingly, NRG3 shows strong binding affinity for the EGF-like domains of JAG1, as shown by co-immunoprecipitation and MST experiments. These data align with previous reports on the EGF-like domains involved in the binding of NRG3 to ErbB4.⁵³ Importantly, we present the first evidence of JAG1 binding to a molecule other than Notch receptors. Our findings confirm that NRG3 modulates the Notch pathway by competitively binding to JAG1, preventing its interaction with Notch receptors and inhibiting Notch pathway activation. However, it is important to note that NRG3 may also interact with other signaling pathways beyond JAG1 in GBM, particularly through its binding to ErbB4, potentially modulating NRG3-mediated Notch inhibition, suggesting that the limitation of this study lies in not comparing the effect of NRG3 on its canonical receptor, ErbB4. Additionally, NRG3 may influence neuronal and glial biology, including cell migration, synaptogenesis, and glial cell differentiation,^54,55 processes that were not fully explored here. Future research should investigate these interactions and broader effects to better understand the role of NRG3 in GBM progression. In addition, while our study focused on GBM, it is important to note that SYN3/NRG3 and Notch signaling might be the more complex relationship in normal neurodevelopment. We observed a negligible correlation (r < 0.2) between SYN3/NRG3 expression and Notch signaling across human Carnegie stages and mouse embryonic days (Supplementary Figure 15), suggesting that SYN3 and NRG3 may not suppress Notch signaling during development. This implies that the SYN3-NRG3-Notch axis might not be involved in conserved neural differentiation mechanisms. Instead, our findings indicate that SYN3-mediated inhibition of Notch signaling promotes GSC differentiation into neuron-like cells in GBM. However, one limitation of this study is the lack of investigation into this mechanism in developmental models. Future studies in non-cancerous neurodevelopmental models are needed to determine whether this pathway is cancer-specific or involved in normal neurogenesis.

Particularly given the neuronal-like transdifferentiation of GSCs by SYN3, our data suggest that SYN3 holds potential as a single-agent differentiation therapy for GBM. However, it is important to acknowledge that this approach is still in the preclinical stages and faces significant challenges, particularly the lack of clinical validation. In addition, while AAV9 shows promise as a vector for intracranial delivery of SYN3, its clinical application in tumor therapy remains an area requiring further investigation to overcome challenges such as delivery efficiency and tumor-specific targeting. Our data show the effect of combining SYN3 overexpression with DAPT and Nirogacestat, which suppress GSC development in vitro. Notably, Nirogacestat has recently been approved for the treatment of desmoid tumors,⁵⁶ nonetheless, its side effects remain a concern. Therefore, in an attempt to reduce the toxic and adverse effects of Nirogacestat and present a viable therapeutic option for GBM treatment, we are interested in combining Nirogacestat with SYN3 differentiation therapy in future research. Additionally, given that temozolomide (TMZ) is currently employed to treat GBM patients with a major effect on outcome, nevertheless, due to the existence of GSCs, GBM patients will develop resistance to TMZ, while GBM may become more drug-sensitive upon differentiation.⁵⁷ Therefore, therapeutic approaches that combine TMZ medications, Notch pathway inhibitors, and SYN3 nanomedicines may also provide useful additional treatment options for GBM.

In conclusion, our results elucidate that the mechanism underlying the efficacy of SYN3 as a tumor suppressor involves driving GSCs to transdifferentiate into neuron-like cells by blocking the signal transduction of the Notch pathway. It is crucial to highlight that NRG3, under the regulation of SYN3, plays a pivotal role in this process by binding to JAG1 and effectively curtailing Notch signaling. The sustained suppression of self-renewal, proliferation, and tumorigenic potential induced by SYN3 provides a foundation for future research to explore more clinically applicable methods of targeting SYN3 in GBM treatment.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

Funding

This work was supported by grants from the National Natural Science Foundation of China (82173228) and Henan Provincial Technology R&D Plan Joint Fund (232301420017).

Conflict of interest statement

The authors declare no potential conflicts of interest.

Authorship statement

Y.D., B.S., and J.Y. designed the experiments. Y.D. performed the experiments. Y.D., X.J., Z.Y., and R.G. performed the bioinformatics analysis. Z.W. and S.R. conducted the adeno-associated virus. X.J., J.B.P., and B.S. provided intellectual support to this study. Y.D. and Z.Y. conducted mouse models and performed tissue staining of mouse tumor samples. Y.D. and J.Y. wrote the manuscript. J.Y. supervised the study.

Data availability

All data generated or analyzed during this study are included in this article, and all data supporting the findings of this study are available from the lead contact, Jinlong Yin on reasonable request.