B3GALT4 remodels the tumor microenvironment through GD2-mediated lipid raft formation and the c-met/AKT/mTOR/IRF-1 axis in neuroblastoma

A total of 81 NB and 25 ganglioneuroma (GN) tumor tissue samples were collected from February 2014 to April 2019 at Tianjin Medical University Cancer Institute and Hospital. The clinical, prognostic, and pathological information of patients was included. The guarantees were fully informed and consented to the research. The study’s procedures were authorized by the Research Ethics Committee of Tianjin Medical University Cancer Institute and Hospital (E20210027) and conducted according to the Code of Ethics of the World Medical Association (Declaration of Helsinki). The clinical characteristics of the patients included in this study are shown in Table S1.

The gene expression data and clinical information for GSE49710 [20], GSE85047 [21], and TARGET-NB were downloaded from the GEO database and the TARGET database. There were 498, 283, and 154 patients in each dataset respectively. Additionally, the GSE112447 and GSE90689 datasets were retrieved to analyze the GD2 molecular phenotype. The relationship between B3GALT4 expression, CD8A expression, and INSS stages was determined. Kaplan-Meier and log-rank tests were used to determine the relationship between B3GALT4 expression levels and NB prognosis. Pearson correlation coefficients were used to assess associations between interferon regulatory factor-1 (IRF-1), CXCL9, and CXCL10 expression. A comprehensive analysis of the mRNA expression levels of genes from the GSE49710 dataset was conducted and categorized. To identify the B3GALT4-mediated biological function in NB progression, gene set enrichment analysis (GSEA) of the GSE49710 dataset was performed. A false discovery rate < 0.25 and P < 0.05 were considered statistically significant for the enriched gene sets. The R software (version 3.3.2) was used to conduct all bioinformatics and statistical analyses.

Briefly, histopathological sections (4 μm thickness) were baked at 60 °C overnight and then deparaffinized and dehydrated. Antigen repair and blocking of endogenous peroxidase activity were performed. After that, these slices were incubated with various primary antibodies for 24 h at 4 °C in a damp container and then incubated with matching secondary antibodies for 1 h at ambient temperature. Finally, Diaminobenzidine (DAB) was utilized for 5 min until a brown reaction result was observed. Digital images were acquired using a light microscope. Anti-GD2, anti-B3GALT4, anti-CD8, anti-caveolin-1, anti-CXCL9, and anti-CXCL10 were used as primary antibodies (Table S2).

The results of the IHC assays were analyzed by two professional pathologists who were blinded to the experimental procedures. The frequency of positive cells was classified into five categories: 0 (negative), 1 (1–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%). The staining degree was categorized as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The IHC score was estimated by multiplying the two outcomes mentioned above. The CD8A-positive cells in 200× photographs of tumor tissues were identified to establish the frequency of CD8⁺ T cells.

The NB cell lines 9464D and 975A2 (gifts from Dr. Rimas Orentas at Seattle Children’s Research Institute) were cultured in high-glucose dulbecco’s modified eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS, BI) and 1% P/S (Gibco) at 37 °C, in a 5% CO₂ humidified incubator. PCR was used to ensure that the cells were free of mycoplasma infection every month. After freezing, the cells were employed in 20 passages and cultivated for less than 6 months. The cell lines were described in previous publications [22].

ShRNA targeting B3GALT4 was synthesized and inserted into the pLKO.1-puromycin lentiviral vector to generate the B3GALT4 knockdown vector (sh-B3GALT4). The scrambled shRNA was inserted into the pLKO.1-puromycin lentiviral vector and defined as a negative control (Scramble). The shRNA sequences are shown in Table S3. The pCDHO-P-FLAG-Puro lentiviral vector encoding mouse B3GALT4 was constructed by RiboBio (Guangzhou, China) and defined as B3GALT4. An empty vector was defined as a negative control (Vector). The cloned plasmids and packaging plasmids (pMDLg/pRRE, pRSV-Rev, and pMD2.G) were transfected into 293 T cells to synthesize the lentiviral particles used to infect NB cells. Stably transfected 9464D and 975A2 cells were selected with puromycin (4 μg/ml and 3 μg/ml) for 48 h, and then the expression levels of B3GALT4 were assessed by qRT-PCR and western blot.

To explore the effect of lipid rafts on B3GALT4-mediated biological functions, 9464D cells transfected with sh-B3GALT4 were treated with the lipid raft inhibitor methyl-β-cyclodextrin (MβCD, 5 mM, Sigma-Aldrich, USA). After treatment, the cells were harvested in preparation for further experiments.

Cells were seeded in 96-well plates. After incubating for 24 h, 48 h, and 72 h, the medium was replaced with CCK-8 solution (100 μl, Solarbio, China), and the culture was incubated for 2 h. The sample absorbance (450 nm) was measured by a microplate reader (Biotek Instruments Inc., USA), and each sample was measured three times.

Cells were plated in 6-well plates. After 2 weeks, the colonies were photographed after being fixed in 4% paraformaldehyde and stained with 0.1% crystal violet solution. The colonies were counted using ImageJ software.

Cells suspended in serum-free medium were seeded into the top compartment of 24-well transwell plates with an 8 μm pore size (Corning, USA). The top compartments were plated with Matrigel (BD Biosciences, USA) for the invasion assay. Subsequently, 600 μl full medium was added to the bottom compartment. After 24 h, the nonmigrating cells were removed. The cells in the lower chamber were fixed with 4% paraformaldehyde, dyed with 0.1% crystal violet solution, and photographed.

To measure GD2 expression in tumor cells, first, the cells were digested with 0.25% trypsin. Subsequently, a single cell suspension was prepared and incubated with 5 μl of anti-GD2 antibody (Clone 14.G2a, BD Biosciences) for 30 min at 4 °C. After being washed twice, the cells were treated with a fluorescence-labeled secondary antibody for 30 min at 4 °C. To assess the frequency of CD8⁺ T cells in tumor tissues, single-cell suspensions of tumor tissues from 9464D tumor-bearing mice were obtained using the Tumor Dissociation Kit (Miltenyi Biotech, Germany). Cells were treated with FITC-CD3, APC-Cy7-conjugated CD4, and PE-Cy7-conjugated CD8 antibodies (Table S2) for 30 min at 4 °C. Negative control experiments (treatment with the appropriate isotype control Ab) were conducted. The data were collected using a CytoFLEX LX flow cytometer (Beckman, USA) and analyzed by Flow Jo software.

Cell climbing sheets were prepared, fixed with 4% paraformaldehyde for 10 min, and then blocked with 10% goat serum for 30 min. Next, the cells were treated with anti-GD2 and anti-caveolin-1 overnight at 4 °C (Table S2). Then, the cells were incubated with FITC-conjugated and Cy3-conjugated antibodies for 1 h at room temperature. Negative control experiments were conducted. Following nucleus staining with 4′,6-diamidino2-phenylindole (DAPI) (Solarbio, China), images were obtained using a Zeiss LSM880 confocal microscope (Zeiss, Germany) at 630× magnification. Fluorescence colocalization analysis was performed by ImageJ software.

First, the transfected cells were treated with the anti-GD2 antibody ch14.18/CHO (dinutuximab beta, 1000 ng/ml, Rentschler Biopharma SE, Germany), disialoganglioside GD2 (NH4 + salt) (1000 ng/ml, Matreya LLC, USA), and recombinant mouse hepatocyte growth factor (HGF, 40 ng/ml, Solarbio, China). Then, lipid rafts were extracted using the Minute Plasma Membrane-Derived Lipid Raft Isolation Kit (Invent Biotechnologies, USA) according to the manufacturer’s instructions.

The secretion levels of the chemokines CXCL9 (MIG) and CXCL10 (IP-10) in the cell culture medium were determined using the mouse CXCL9 and CXCL10 ELISA Kit (Biolab, China). The chemokine concentrations were determined by plotting a curve constructed using the recombinant molecule.

CD8⁺ T cells were isolated from the spleens of wild type (WT) C57BL/6 mice using the mouse CD8a⁺ T-Cell Isolation Kit (Miltenyi Biotec, Germany). Cells were cultured with RPMI 1640 supplemented with 10% FBS, 1% P/S, and 100 U/ml recombinant murine IL-2. To perform a chemotaxis assay, 24-well Transwell chambers with 5-μm polycarbonate membranes were used. Recombinant murine CXCL9 (100 ng/ml, PeproTech, USA), CXCL10 (50 ng/ml, PeproTech, USA), and culture supernatants from transfected cells after 48 hours of culture were added to the bottom wells. Then, 1 × 10⁵ CD8⁺ T cells were seeded in the top chamber. After incubation for 2 h, the number of migrating cells was assessed using a hemocytometer.

TRIzol reagent (Invitrogen, USA) was used to obtain total RNA, and PrimScript RT Master Mix was used to perform reverse transcription (Takara, Japan). Then, qRT-PCR was performed using the SYBR Green PCR Kit (Takara, Japan) in an ABI QuantStudio 5 (Q5) system (Applied Biosystems, USA). The cycling characteristics were as described below: 95 °C/5 s and 60 °C/34 s for 40 cycles. The mRNA level was normalized to GAPDH. The relative expression level was determined using 2^−ΔΔCt. All reactions were performed in triplicate. The primers for the target genes are listed in Table S4.

RIPA lysis buffer (Solarbio, China) was used to harvest total protein. The BCA approach (Thermo, USA) was used to determine the protein concentration. Protein samples were separated using SDS-PAGE gels and transferred to PVDF membranes. After blocking with 5% skimmed milk powder for 1 h, the membranes were treated with the primary antibodies at 4 °C overnight. The primary antibodies are listed in Table S2. The membranes were treated with the secondary antibodies for 1 h at room temperature. The band images were created by a GelView 6000Plus system (Biolight Biotechnology, China) and quantified by ImageJ software. β-actin was used as an internal control.

All animal experiments were carried out following the Guide for the Care and Use of Laboratory Animals and were authorized by the Institutional Animal Care and Use Committee of the Tianjin Medical University Cancer Institute and Hospital (AE-2021112). Female C57BL/6 mice (6–8 weeks, weighing 18–20 g) were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd. (Beijing, China) and housed under specific pathogen-free conditions. In each treatment group, 5 mice were used to investigate changes in the tumor phenotype.

To examine the inhibitory effects of B3GALT4 on tumor growth in vivo, the vector and B3GALT4-overexpressing 9464D cells (2 × 10⁶/100 μl PBS per mouse) were injected into the backs of the mice. To further confirm whether a lipid raft inhibitor can reverse the knockdown of B3GALT4-mediated immunosuppressive effects in vivo, sh-B3GALT4 9464D cells (2 × 10⁶/100 μl PBS per mouse) were injected into the backs of the mice. On Day 6 after tumor implantation, mice were treated with the lipid raft inhibitor MβCD (300 mg/kg) or PBS every 3days for 18 days. To assess the antitumor effect of MβCD combined with anti-GD2 mAb in NB xenograft models, 9464D cells (2 × 10⁶/100 μl PBS per mouse) were injected into the backs of the mice. On Day 6 after tumor implantation, MβCD was injected intraperitoneally (i.p.) (300 mg/kg) every 3 days, and anti-GD2 mAb (200 μg/per mouse, i.p.) was injected twice weekly for 18 days. According to the algorithm, the tumor volume was analyzed every 3 days: (length × width²)/2. After the mice were euthanized, the tumors were separated, weighed, and harvested for subsequent analysis.

Data analysis and graphing were conducted with Prism 8.0 software (Graph Pad Software Inc.). Quantitative data are presented as the mean ± standard deviation. An unpaired Student’s t test was used to analyze differences between the two groups. Statistically significant differences were analyzed using one-way ANOVA (Bonferroni test) among multiple groups. A P value of less than 0.05 was regarded as statistically significant, and “ns” indicated “not significant”. The P values in the figures are depicted with the following symbols: ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001, ^**** P < 0.0001.

Disialoganglioside GD2 is an important target for immunotherapy of NB. Here, we confirmed that GD2 was overexpressed in NB tissues compared with GN tissues by IHC (Fig. 1A). In addition, high GD2 expression was related to advanced INSS stage, bone marrow metastasis, and COG high-risk classification (Fig. 1B-D). Kaplan-Meier analysis results showed that a higher GD2 level was correlated with a shorter 3-year overall survival (OS) of NB patients (Fig. 1E).

As shown in Fig. 1F, B3GALT4 is a crucial glycosyltransferase involved in synthesizing ganglioside GD2. The GSEA results also demonstrated that B3GALT4 was mainly enriched in hydrolyzing N-glycosyl compounds involved in ganglioside GD2 metabolism (Fig. 1G). According to the analysis of the GEO databases, B3GALT4 expression was significantly reduced in GD2 (+) NB tissues compared with GD2 (−) NB tissues (P < 0.001) (Fig. 1H). We subsequently evaluated GD2 and B3GALT4 protein expression levels in human NB clinical samples by IHC. We also observed a strong negative connection between the levels of GD2 and B3GALT4 (r = − 0.931, P < 0.0001) (Fig. 1I). To further evaluate whether the GD2 expression level was mediated by B3GALT4, we established stable NB cell lines with B3GALT4 knockdown (sh-B3GALT4) and B3GALT4 overexpression (B3GALT4) NB cell lines (Fig. 1J and K) and analyzed the expression level of GD2 in these cells using a flow cytometry assay. We found that B3GALT4 knockdown significantly upregulated GD2 expression levels in the NB cell lines 9464D and 975A2, while B3GALT4 overexpression significantly downregulated GD2 levels (Fig. 1L). The above data indicated that GD2 was overexpressed in NB tissues and was mediated by B3GALT4.

Given the role of B3GALT4 in regulating ganglioside GD2 synthesis, we explored the association between B3GALT4 expression levels and the progression of NB. NB data from two GEO datasets (GSE49710 and GSE85047) showed that low expression of B3GALT4 was associated with advanced INSS stage (Fig. 2A and B). NB patients with low B3GALT4 expression had shorter OS (Fig. 2C and D), event-free survival (EFS) (Fig. 2E), and relapse-free survival (RFS) (Fig. 2F) than patients with elevated expression. We then analyzed the expression of B3GALT4 in NB samples from patients with different clinical stages by IHC. The results showed that B3GALT4 IHC scores in advanced stages of NB tissues were significantly lower than those in early-stage tissues (P < 0.0001) (Fig. 2G). The above findings suggested that B3GALT4 is a potential prognostic biomarker in patients with NB.

We then investigated the biological function of B3GALT4 in NB cells. The results showed that overexpression of B3GALT4 decreased the proliferation, migration, and invasion of 9464D and 975A2 cells. In contrast, knockdown of B3GALT4 in the 9464D and 975A2 cell lines enhanced proliferation, migration, and invasion (Fig. 3A-E).

To better evaluate the effect of B3GALT4 on NB growth in vivo, we established a subcutaneous model with B3GALT4-overexpressing 9464D cells in C57BL/6 mice (Fig. 3F). The results showed that upregulation of B3GALT4 dramatically inhibited tumorigenesis (Fig. 3G and H). In addition, IHC staining of Ki-67 revealed that the number of the Ki-67-positive cells in tumors of the B3GALT4 group were lower than those in the vector group (P < 0.01) (Fig. 3I), suggesting that B3GALT4 overexpression inhibited the proliferation of 9464D tumors. These data demonstrated that B3GALT4 serves as a tumor suppressor gene in NB.

To assess whether the antitumor activity of B3GALT4 overexpression was related to modulation of the TME, we first conducted GSEA based on data from the GSE49710 dataset. In comparison to the B3GALT4 low expression group, the high expression group was primarily enriched in the regulation of T-cell chemotaxis, T-cell receptor complex, C-C chemokine binding, and the chemokine signaling pathway (Fig. 4A). Increasing evidence indicates that aberrant glycosylation of cancer cells plays a key role in the modulation of the TME [23‐25]. Previously, we found that the glycosyltransferase signature was related to CD8⁺ T-cells infiltration of NB [26]. However, the correlation between B3GALT4 expression and T-cell infiltration is still unknown in NB. Therefore, we investigated the correlations between B3GALT4 expression levels and known human chemokines, T-cell infiltration and function, and immunosuppression based on the GSE49710 dataset. The results indicated that higher B3GALT4 expression was related to elevated expression levels of T-cell-related chemokines (CXCL9 and CXCL10) and T-cell markers (CD3, CD4, and CD8A) (Fig. 4A). Several studies verified the relationship between enhanced T lymphocyte infiltration and better outcomes in patients with NB [27‐29]. Given the roles of CD8⁺ T cells in NB outcome, we analyzed the relationship between CD8A level and tumor stages and the prognosis of NB patients from the GSE49710 dataset. As shown in Fig. 4C, decreased expression of CD8A was observed in advanced stages of NB. High CD8A expression was associated with longer OS (P < 0.0001) and EFS (P < 0.0001) (Fig. 4D). Moreover, the expression level of B3GALT4 was significantly positively correlated with CD8A, CXCL9, and CXCL10 levels in clinical NB samples from the GSE49710 dataset (Fig. 4E). Recently, CXCL9 and CXCL10 were reported to be necessary for the trafficking of activated CD8⁺ T cells into tumor sites [30‐32]. To explore whether B3GALT4 mediated CD8⁺ T-cell infiltration via CXCL9 and CXCL10, we conducted a migration assay in vitro. First, CD8⁺ T cells were isolated from the spleens of WT C57BL/6 mice. Then, a chemotaxis assay was carried out (Fig. 4F). The results demonstrated that the supernatant of B3GALT4-overexpressing tumor cells markedly promoted the migration of CD8⁺ T cells. However, the supernatant of sh-B3GALT4-transfected NB cells strongly inhibited the migration of CD8⁺ T cells (Fig. 4G). Furthermore, recombinant murine CXCL9 or CXCL10 alone significantly induced CD8⁺ T-cell migration (Fig. 4G). A rescue study was performed to observe the effect of the chemokines CXCL9 and CXCL10 on B3GALT4 inhibition-induced CD8⁺ T-cell recruitment. Sh-B3GALT4-transfected 9464D and 975A2 cellular supernatants were treated with the chemokines CXCL9 and CXCL10, respectively. The results showed that concurrent treatment with chemokines induced sh-B3GALT4-inhibited CD8⁺ T-cell migration in both cell lines (Fig. 4G). To confirm the impact of upregulation of B3GALT4 on CD8⁺ T-cell migration in vivo, flow cytometry was used to determine the number of CD8⁺ T cells in tumors from subcutaneous xenograft mouse models. The results demonstrated that the CD8⁺ T cells were accumulated more in the stroma of B3GALT4-overexpressing tumors than in the stroma of vector-treated tumors (Fig. 4H). We further confirmed that the expression levels of CXCL9, CXCL10, and CD8A in the B3GALT4 overexpression group were higher than those in the vector group by an IHC assay (Fig. 4I). In addition, B3GALT4 expression was positively correlated with CXCL9/CXCL10 expression and CD8⁺ T-cell infiltration in human NB clinical samples (Fig. 4J). Accordingly, these data provide strong evidence that B3GALT4 overexpression may promote CD8⁺ T lymphocyte recruitment into tumor sites via CXCL9 and CXCL10 in NB.

Studies have shown that lipid rafts are essential in regulating tumor progression [33‐35]. Gangliosides are mainly localized in lipid rafts, and gangliosides GM1 and GD3 modulate lipid raft activity [36, 37]. We investigated whether ganglioside GD2 could similarly regulate lipid raft formation. As caveolin-1 is a lipid raft marker, we first explored whether GD2 was colocalized with caveolin-1 by double immunofluorescence staining. The results demonstrated that GD2 was colocalized with caveolin-1 on the plasma membrane of NB cells (Fig. 5A). Then, we further analyzed the expression levels of GD2 and caveolin-1 by IHC in tumors from subcutaneous xenograft mouse models. The results confirmed that the expression level of caveolin-1 was decreased in the B3GALT4 overexpression group (low GD2 expression) and upregulated in the B3GALT4 low expression group (high GD2 expression) compared with the vehicle group (Fig. 5B). To further investigate the role of ganglioside GD2 mediated by B3GALT4 in lipid raft formation, we extracted lipid rafts and evaluated the expression of caveolin-1 in NB cells with different treatments. First, purified ganglioside GD2 alone significantly enhanced caveolin-1 protein levels in lipid raft fractions (Fig. 5C). Caveolin-1 expression levels were significantly decreased in B3GALT4-overpressing NB cells compared with vector-treated cells. However, the protein expression levels of caveolin-1 were significantly increased in B3GALT4-overpressing cells after purified ganglioside GD2 treatment. Meanwhile, the protein expression levels of caveolin-1 were increased by B3GALT4 silencing, whereas the addition of anti-GD2 mAb reversed this effect (Fig. 5C). These results showed that GD2 mediated by B3GALT4 played an essential role in lipid raft formation.

We next explored how upregulation of B3GALT4 induced CXCL9/CXCL10 expression. The HGF/c-Met pathway in lipid rafts plays a significant role in the tumor microenvironment [38, 39]. It has recently been reported that the clustering of GD2 activates c-Met [40]. Based on previous studies, we sought to determine whether B3GALT4 regulates CXCL9 and CXCL10 production through c-Met signaling in lipid rafts. We first extracted lipid rafts and observed the phosphorylated c-Met (p-c-Met) levels by western blot. The results showed that the expression levels of p-c-Met in lipid rafts were significantly increased in B3GALT4-knockdown cells and decreased in B3GALT4-overexpressing cells compared with the vehicle cells (Fig. 6A), indicating that B3GALT4 downregulation activated the c-Met signaling pathway in lipid rafts (Fig. 6A). Since the activation of c-Met depends on HGF, we next investigated whether HGF could affect the activation of c-Met mediated by B3GALT4 downregulation. Western blot analysis showed that HGF treatment increased p-c-Met levels in the vehicle cells, but no changes in the levels of p-c-Met were observed in B3GALT4-knockdown or B3GALT4-overexpressing cells after HGF treatment (Fig. 6A). These results indicate that B3GALT4 downregulation can activate c-Met independent of HGF. GSEA was conducted to explore the downstream signaling pathway based on data from the GSE49710 dataset. The results showed that the PI3K/AKT/mTOR and IFN-γ signaling pathways were significantly enriched in samples with high B3GALT4 expression (Fig. 6B). It is well established that the IFN-γ signaling pathway regulates CXCL9 and CXCL10 expression in cancer cells [41]. IRF-1 is an IFN-γ/p-STAT1 responsive transcription factor that governs immune-related genes, such as genes of the CXCL and CCL families [42]. We found that the expression of IRF-1 was significantly positively correlated with the expression of CXCL9 and CXCL10 in NB tissues from the NB dataset (GSE49710) (Fig. 6C). Accordingly, western blot analysis showed that the level of IRF-1 was dramatically decreased in B3GALT4-knockdown cells and increased in B3GALT4-upregulated cells (Fig. 6D). A previous study showed that the inhibition of the PI3K/AKT/mTOR signaling pathways elevates the expression of the chemokines CXCL10 and CXCL11 expression via IRF-1 [19]. In our study, western blot analysis revealed that the expression levels of p-AKT and p-mTOR were significantly increased in B3GALT4-knockdown cells and significantly decreased in B3GALT4-overexpression cells (Fig. 6D). Moreover, the results showed that overexpression of B3GALT4 enhanced CXCL9 and CXCL10 synthesis, but downregulation of B3GALT4 reduced the levels of these two chemokines, as shown by western blotting (Fig. 6E) and the ELISA test (Fig. 6F). These findings confirmed the conclusion that the c-Met/AKT/mTOR/IRF-1 pathway mediated by B3GALT4 is involved in CXCL9 and CXCL10 production.

To further verify whether B3GALT4 downregulation promoted NB progression through lipid rafts, rescue experiments were carried out. MβCD can remove cellular cholesterol and is commonly used as a lipid raft inhibitor [43]. As observed previously, B3GALT4-knockdown 9464D cells exhibited higher proliferation, invasion, and migration abilities than untreated 9464D cells. However, this difference was abolished by MβCD treatment (Fig. 7A and B). The expression levels of caveolin-1, CXCL9, and CXCL10 were significantly increased after MβCD treatment compared with the B3GALT4 knockdown group (Fig. 7C). We next demonstrated that MβCD reversed the effect of B3GALT4 knockdown on the c-Met/AKT/mTOR pathway (Fig. 7D). Meanwhile, MβCD markedly reversed the inhibitory effect of B3GALT4 knockdown on the migration of CD8⁺ T cells (Fig. 7E). Subcutaneous tumor models with B3GALT4-knockdown 9464D cells confirmed that MβCD relieved the knockdown of B3GALT4-mediated tumor growth (Fig. 7F and G). In accordance with the chemotaxis assay data, MβCD treatment enhanced the number of CD8⁺ T cells in tumor tissues of C57BL/6 mice treated with B3GALT4-knockdown 9464D cells (Fig. 7H). IHC staining of tumor tissues from mice treated with B3GALT4 knockdown 9464D cells confirmed the expression levels of CXCL9 and CXCL10, and the percentage of CD8A-positive cells was increased after MβCD treatment, while caveolin-1 was decreased (Fig. 7I). Taken together, these data demonstrated that MβCD attenuated the knockdown of B3GALT4-induced tumor progression and immunosuppression.

Anti-GD2 mAbs have shown certain clinical efficacy for patients with high-risk NB [44]. Unfortunately, some NB patients do not respond to anti-GD2 mAb treatment due to various immunosuppressive factors, such as a deficiency of CD8⁺ T-cell infiltration [45]. Based on the previous findings, we investigated whether the inhibition of lipid rafts could improve the efficacy of anti-GD2 therapy by examining the antitumor effect of the anti-GD2 mAb in combination with MβCD using a 9464D subcutaneous xenograft tumor mouse model. After xenografts were established, PBS, anti-GD2 mAb, MβCD, or the combination of anti-GD2 mAb and MβCD was administered to the mice (Fig. S1A). The results showed that anti-GD2 mAb combined with MβCD therapy significantly suppressed tumor growth compared with PBS treatment or treatment with MβCD (Fig. S1B-D). The combination of anti-GD2 mAb and MβCD therapy showed only a nonsignificant trend for tumor growth delay compared with anti-GD2 mAb treatment alone. Meanwhile, no statistically significant variations in the average body weight of mice were identified (Fig. S2).

To further explore the effect of anti-GD2 mAb plus MβCD on tumor immunity, we detected the protein expression of CD8A, caveolin-1, CXCL9, and CXCL10 by IHC staining. We found that the protein levels of CXCL9 and CXCL10 were obviously elevated, and the protein level of caveolin-1 was downregulated in the tumor cells of mice treated with anti-GD2 mAb plus MβCD compared with the other three groups (Fig. 8A and B). Next, we found a higher percentage of CD8A-positive cells in the tumor tissues of mice given anti-GD2 mAb plus MβCD than that in the other three groups (Fig. 8A and B). Furthermore, flow cytometry analysis also showed that the percentage of CD8⁺ T cells was higher in tumor tissues of mice treated with anti-GD2 mAb plus MβCD than in the other three groups (Fig. 8C).