IDH1-R132H enhances oncolytic HSV-1 therapy by facilitating viral entry and immune activation in glioma

Abstract

Oncolytic virotherapy represents a promising yet under-explored approach for precision cancer treatment, particularly when tailored to tumor-specific molecular profiles. Patients with high-grade isocitrate dehydrogenase (IDH) mutant astrocytomas have limited treatment options and poor prognoses. Here, we investigate the therapeutic efficacy of rQNestin34.5 v.2 (CAN-3110), an engineered oncolytic herpes simplex virus 1 (oHSV-1), in IDH1-R132H-mutant diffuse gliomas. We demonstrate that the IDH1-R132H mutation enhances glioma susceptibility to viral infection through upregulation of Nectin-1, the main HSV-1 entry receptor. Concurrently, IDH1-R132H-driven DNA hypermethylation suppresses interferon (IFN) signaling, creating a permissive microenvironment that facilitates viral replication and tumor cell apoptosis. In immunocompetent murine glioma models, intratumoral administration of rQNestin34.5 v.2 induces robust antitumor immune activation, including increased immune infiltration and systemic IFN-γ release. However, elevated expression of poliovirus receptor (PVR) and the immune checkpoint T-cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT) on tumor-infiltrating leukocytes suggests a potential resistance mechanism to virotherapy. Combining rQNestin34.5 v.2 with TIGIT blockade enhances therapeutic efficacy compared to monotherapy, identifying IDH1-R132H as a potential predictive biomarker for oncolytic virotherapy response.

Oncolytic virotherapy represents a promising approach for glioma therapy with challenges remain. Here, the authors show that IDH1-mutant gliomas exhibit increased susceptibility to oncolytic HSV-1 therapy due to impaired antiviral signaling, and that combining virotherapy with TIGIT blockade enhances antitumor immune responses and therapeutic efficacy.

Introduction

High-grade adult-type diffuse gliomas are defined by gain of function mutations in isocitrate dehydrogenase 1 (IDH1; codon 132) or 2 (IDH2; codon 172) genes, and together have an age-adjusted incidence of ~0.3 / 100,000 in the U.S¹. IDH-mutant gliomas are further classified into two histomolecular subgroups: a) oligodendrogliomas, characterized by codeletion of the 1p and 19q chromosome arms, and TERT-promoter mutations, and b) astrocytomas, characterized by absence of 1p/19q codeletion and frequented by loss-of-function mutations in the tumor suppressor protein 53 (TP53) gene and ATRX^2,3. Although lower grade IDH-mutant gliomas often display slower progression, the outcomes remain poor for high-grade IDH-mutant astrocytomas, with 1-year overall survival (OS) estimated to be 76% for patients with WHO CNS grade 4 IDH-mutant astrocytoma⁴. Current multimodal approaches include surgical resection, radiotherapy, chemotherapy (e.g., temozolomide, CCNU), novel IDH inhibitors (e.g., vorasidenib), anti-angiogenic drugs, and/or tumor-treating fields (TTF); however, overall durability of treatment responses remains suboptimal in patients^5–8. These challenges are compounded by the unique tumor microenvironment (TME) of gliomas, which displays immune evasion and therapy resistance through complex molecular and cellular interactions^9–12.

Ongoing research is uncovering how the distinct metabolic features of these tumors can be leveraged to improve efficacy of oncolytic virotherapy^13,14, and somatic mutations in key genes are increasingly recognized as predictors of therapeutic response^15,16. IDH1 p.R132H is the most common IDH1/2 mutation in gliomas and results from the substitution of arginine with histidine at residue 132, creating a neomorphic enzyme that converts α-ketoglutarate (α-KG) to the oncometabolite D-2-hydroxyglutarate (D-2-HG)^17,18. D-2-HG accumulation disrupts cellular metabolism, induces DNA and histone hypermethylation, and drives global epigenetic reprogramming, leading to altered differentiation and chromatin remodeling^10,19,20. These metabolic and epigenetic changes might significantly influence responses to oncolytic virotherapy, making IDH1-R132H a potential biomarker for stratifying patients in clinical settings.

The immunosuppressive TME of gliomas further hinders therapeutic outcomes²¹. Immune cells, including T cells and natural killer (NK) cells, are often inhibited from mounting effective anti-tumor responses^12,22. Immune suppression is partly mediated by the regulation of immune cell function through the CD155(PVR)/TIGIT/CD226 signaling axis²³. T-cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT), an inhibitory receptor expressed on T and NK cells, binds to poliovirus receptor (PVR or CD155) on glioma cells, suppressing immune activation and cytokine production^23,24. Because the co-stimulatory receptor CD226 competes with TIGIT for CD155 binding, targeting CD155/TIGIT has shown promise in restoring anti-tumor immunity, particularly when combined with immune checkpoint blockade, neoantigen vaccines, or oncolytic viruses^23,25–29.

Oncolytic virotherapy has emerged as a promising therapeutic strategy for gliomas, exemplified by the approval of teserpaturev/G47∆, an HSV-1-derived oncolytic virus (OV), for malignant gliomas in Japan, and T-VEC, another HSV-1-based OV, for metastatic melanoma in the U.S^30,31. CAN-3110 (former designation, rQNestin34.5 v.2³²) is a first-in-class, replication-competent HSV-1 oncolytic immunovirotherapy that elicits tumor cell oncolysis with immune activation. Unlike other clinical-stage oncolytic HSV-1 strains, CAN-3110 retains the HSV-1 neurovirulence gene ICP34.5 under the control of the cellular nestin promoter, which is selectively upregulated in malignant glioma cells^32,33. This design enables tumor-specific replication while minimizing neurotoxicity. Currently under evaluation in a Phase Ib clinical trial for recurrent IDH-wildtype glioma (NCT03152318), CAN-3110 has shown preliminary evidence for possible efficacy, with patients’ OS associated with increased T-cell infiltration of tumors³⁴. However, the impact of IDH mutations on this promising therapy remains unstudied.

In this study, we investigate how the IDH1-R132H mutation influences the efficacy of oHSV-1 rQNestin34.5 v.2 immunotherapy in high-grade diffuse glioma. We demonstrate that IDH1-R132H enhances glioma susceptibility to rQNestin34.5 v.2 infection. Intratumoral administration of rQNestin34.5 v.2 in IDH1-mutant gliomas elicits strong immunostimulatory effects, including increased infiltration of T and NK cells. We further identify CD155/TIGIT and PD-L1/PD-1 pathways as relevant immunoregulatory axes, with combined rQNestin34.5 v.2 and TIGIT blockade improving survival in mouse glioma models. These findings highlight the role of IDH1-R132H in shaping antiviral and antitumor immune responses and support its potential value as a biomarker for guiding oncolytic virotherapy in high-grade gliomas.

Results

IDH1-R132H Mutation Shapes the Cell Programs in Glioma

Mutations in TP53, and ATRX rank among the most frequent genetic alterations in IDH-mutant low-grade gliomas (LGGs) and in IDH-wild-type glioblastoma (GBM) (Fig. 1a). The IDH1 mutations are predominated by five distinct variants (R132C, R132G, R132H, R132S and R132L), with R132H and R132L being most and least common, respectively. For the common variants, IDH1 mutations are consistently associated with prolonged OS in glioma patients (Fig. 1b–e). Due to the rarity of R132S and R132L, limited data are available to evaluate their impact on patient outcomes. Given that IDH1-R132H is the most common of the IDH1 mutations, we evaluated how this mutation might change gene expression in MGG8 human glioma cells engineered to express IDH1-R132H, compared to wild-type IDH1/2 cells (Fig. 1f). The MGG8 cell line was generated from an IDH-wildtype glioblastoma with CDKN2A homozygous deletion, and the addition of IDH1 p.R132H mutation was hypothesized to confer similar characteristics as an WHO CNS grade 4 IDH-mutant astrocytoma. RNA-seq showed profound transcriptomic changes in MGG8 cells with the IDH1-R132H mutation, with 8,092 genes upregulated and 2,906 genes downregulated relative to IDH wildtype (Fig. 1g). To corroborate these findings, we leveraged the Cancer Genome Atlas (TCGA) gene expression dataset from treatment-naïve high-grade (grade ≥3) astrocytomas. This analysis revealed significant differential gene expression (28,588 coding and non-coding genes; FDR < 0.05) between IDH-wild-type astrocytoma (CNS WHO grade 4, GBM; n = 158) and IDH-mutant astrocytoma (CNS WHO grade 3; n = 58) (Fig. 1h). Analysis of the top 500 differentially expressed genes highlighted alterations in key biological processes, including DNA replication and nucleosome assembly, regulation of androgen receptor signaling, and interleukin-7-mediated signaling (Supplementary Data 1).

Fig. 1. IDH1-R132H shapes the molecular landscape of Astrocytoma.

a Distribution of the 20 most frequently mutated genes in LGGs and GBM based on data from the TCGA Research Network. b Kaplan-Meier survival curves comparing glioma patients with IDH1 mutations (n = 393; all variants) to those with wild-type IDH1 (n = 112). c Kaplan-Meier survival curves for glioma patients with IDH1-R132C missense mutations (n = 17) versus non-mutated cases (n = 485). d Kaplan-Meier survival curves for glioma patients with IDH1-R132G missense mutations (n = 11) versus non-mutated cases (n = 491). e Kaplan-Meier survival curves for glioma patients with IDH1-R132H missense mutations (n = 352) versus non-mutated cases (n = 150). Survival data in (b–e) were obtained from TCGA and analyzed using Kaplan-Meier methods. Log-rank test p-values are indicated within the figures. P < 0.05 was considered statistically significant. f Bulk RNA sequencing was conducted on human MGG8-IDH1-R132H and MGG8-IDH-wild-type glioma cells in culture (n = 3 replicates per group). The graphic was created in BioRender. Panagioti, E. (2026) https://BioRender.com/ttabs0q. g Gene expression enrichment analysis was performed on differentially expressed genes with significant alterations between IDH1-mutant and wild-type MGG8 cells. Upregulated genes (n = 8092) are highlighted in green, and downregulated genes (n = 2906) in light blue. h Analysis of mRNA sequencing data derived from TCGA human glioma biopsies was performed across cohorts comprising IDH-wild-type astrocytoma (CNS WHO grade 4, GBM; n = 158) and IDH-mutant astrocytoma (CNS WHO grade 3; n = 58). Gene expression enrichment analysis identified differentially expressed genes with significant alterations between IDH-mutant and wild-type astrocytoma (total=28,588 variables). The dotted line cutoff values for the log2 fold change were 1 and -1, and for the log10 (q-value) were 1.30. Upregulated (n = 2055) and downregulated (n = 9338) genes with significant q-values and log2 fold changes (FC) are highlighted in red. i Enrichment of the top 20 most significantly upregulated (green) and downregulated (light blue) pathways in MGG8 IDH1-mutated versus wild-type cells, with log differences displayed.

Gene ontology (GO) analysis of differentially expressed genes also shows significant alterations between IDH1-mutant and wild-type MGG8 cells, suggesting changes in numerous key biological processes and molecular pathways. Among the top 50 signaling pathways, several were related to innate host responses associated with viral infections such as HSV−1, Human immunodeficiency virus 1 (HIV−1), Epstein-Barr virus (EBV), Hepatitis B virus (HBV), Human papillomavirus (HPV), Human T-cell leukemia virus 1 (HTLV−1), Kaposi sarcoma-associated herpesvirus (KSHV), Human cytomegalovirus (HCMV), and SARS-CoV-2 coronavirus infection (Supplementary Fig. 1a). In addition to virus-related pathways, metabolic processes and oxidative phosphorylation were the most significantly upregulated in IDH1-R132H mutant cells. In contrast, pathways linked to ubiquitin-mediated proteolysis, mRNA surveillance, cell stemness (e.g., Wnt signaling), and cell cycle regulation (e.g., MAPK signaling) were among the most downregulated (Fig. 1i). To corroborate these findings, KEGG pathway analyses were performed on the top 500 upregulated and top 500 downregulated differentially expressed genes derived from TCGA glioma specimens, comparing IDH-wildtype astrocytoma (CNS WHO grade 4, GBM; n = 158) and IDH-mutant astrocytoma (CNS WHO grade 3; n = 58). This analysis revealed multiple pathway perturbations, including COVID-19-related pathways, Wnt signaling, neutrophil extracellular trap formation, viral carcinogenesis, and necroptosis among the top ten enriched pathways (Supplementary Fig. 1b, c). Virus susceptibility-focused gene set enrichment analysis (GSEA) further identified differential regulation of pathways associated with viral RNA transcription and replication, viral entry into host cells, and interferon-γ responses (Supplementary Fig. 1d). Collectively, these results suggest that IDH mutations may influence viral infection dynamics, with IDH-mutant and IDH-wild-type tumors potentially exhibiting differential susceptibility and resistance to oncolytic virotherapy.

IDH1-R132H mutation increases susceptibility to oncolytic HSV−1 rQNestin34.5 v.2 infection

Given that non-infected IDH1-mutant MGG8 tumor cells unexpectedly exhibited a transcriptional profile marked by upregulation of genes associated with HSV-1 infection, we hypothesized that the IDH1-R132H mutation might actually enhance susceptibility to HSV-1 infection. We first confirmed that these cells are PCR-negative for HSV-1, HSV-2, and a panel of sixteen other viruses (Human Basic Clear panel from Charles River) commonly screened in in-vitro culture systems. We next infected a panel of IDH1-R132H mutant and wild-type human high-grade glioma cell lines that included MGG8, SJGBM2 (which was derived from an IDH-wild-type glioblastoma that harbored inactivating mutations in ATRX and TP53), U87, and MGG119 with oHSV-1 rQNestin34.5 v.1, a green fluorescence protein (GFP)-expressing variant of rQNestin34.5 v.2³², and compared viral infectivity and cytotoxicity. Infections were monitored using green GFP expression and quantified at various multiplicities of infection (MOI). All IDH1-R132H mutant cell lines exhibited robust viral infection, characterized by enhanced GFP expression, cytotoxicity, and accelerated cell death compared to wild-type controls (Fig. 2a–d).

Fig. 2. The IDH1-R132H mutation enhances glioma cell susceptibility to oHSV-1 rQNestin34.5 v.2.

a Representative green fluorescence microscopy images showing viral infection in MGG8 IDH1-wild-type and IDH1-mutant cells 48 h post-infection with oHSV-1 rQNestin34.5 v.1 at MOIs of 0.1 and 1. rQNestin34.5 v.1 differs from rQNestin34.5 v.2 by retaining an eGFP transgene fused to the 3’end of the viral ICP6 gene. Scale bars, 100 μm (10× magnification). b Percentage of GFP-positive cells (rQNestin34.5 v.1 infected cells at MOIs 0.01, 0.1, and 1) in MGG8-IDH1-mutant vs. wild-type, SJGBM2-IDH1-mutant vs. wild-type, and U87-IDH1-mutant vs. wild-type cells. A two-way ANOVA followed by Šídák’s multiple comparisons test was used for statistical comparisons (n = 3 replicates per group). Data are presented as mean ± SD. Statistical significance is indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. c Time-kinetics of cell viability in MGG8 and SJGBM2 IDH1-mutant and wild-type cells infected with rQNestin34.5 v.2 at MOI: 0.1, and 1 measured by CellTiter-Glo assay (n = 6 replicates per group). Data are presented as mean ± SD. d Percentage of GFP-positive cells (rQNestin34.5 v.1-infected cells at MOIs 0.1, and 1) in the MGG119 patient-derived glioma cell line harboring endogenous IDH1 mutations at 48 h post-infection (n = 3 replicates per group). A two-way ANOVA followed by Šídák’s multiple comparisons test was used for statistical comparisons. Data are presented as mean ± SD. Time-course analysis of cell viability in MGG119 cells, as described in (c) (n = 4 replicates per group). e Schematic of bulk RNA sequencing: 5 × 10⁵ MGG8 IDH1-mutant and wild-type cells were infected with rQNestin34.5 v.2 at MOI: 1, and RNA was extracted 16 h post-infection for Bulk RNA sequencing (n = 3/group); Created in BioRender. Panagioti, E. (2026) https://BioRender.com/rbw9wni. f Enrichment of upregulated (green) and downregulated (light blue) genes in rQNestin34.5 v.2-infected MGG8 IDH-wild-type cells vs. non-infected cells. g Top upregulated pathways in MGG8-IDH-wild-type cells infected with rQNestin34.5 v.2. h Enrichment of upregulated (green) and downregulated (light blue) genes in rQNestin34.5 v.2-infected MGG8-IDH1-mutant cells vs. non-infected cells. i, Top 20 upregulated (green) and top 5 downregulated (light blue) pathways in MGG8-IDH1-mutant cells infected with rQNestin34.5 v.2. j Representative green fluorescence microscopy images of C3-IDH1-mutant vs. C54-IDH1-wild-type neurospheres 48 h post-infection with rQNestin34.5 v.1 at MOI: 0.1 and 1. Scale bars, 100 μm (10× magnification). k Percentage of GFP-positive cells in C3-IDH1-mutant vs. C54-IDH-wild-type cells 48 hours post-infection at MOI: 0.1 and 1 (n = 3 replicates per group). Data are reported as mean ± SD. A two-way ANOVA followed by Šídák’s multiple comparisons test was used for statistical comparisons. ****, P < 0.0001. l Time-kinetics of cell viability in C3-IDH1-mutant and C54-IDH1-wild-type cells infected with rQNestin34.5 v.2 at MOI: 0.1, 1, and 2, measured by CellTiter-Glo assay (n = 5 replicates per group). All experiments, except for bulk RNA-seq analyses, were independently repeated at least twice with consistent results.

Bulk RNA sequencing on rQNestin34.5 v.2 infected versus non-infected MGG8-IDH1-mutant and MGG8-IDH1/2 wild-type glioma cells (Fig. 2e) showed that infection of wild-type cells elicits only slight perturbation of cell function, with upregulation of only nine genes affecting two pathways: Signaling by nuclear receptors and Activation of anterior HOX genes in hindbrain development during early embryogenesis (Fig. 2f, g and Supplementary Table 1). This limited response likely reflects relative resistance of IDH1/2-wild-type cells to viral infection (Fig. 2b). In contrast, rQNestin34.5 v.2 infection of MGG8-IDH1-mutant cells caused differential expression of 380 genes, with 294 upregulated and 86 downregulated (Fig. 2h and Supplementary Data 2), associated with increased permissiveness of IDH1-mutant cells to viral infection. Key pathways significantly altered in infected MGG8-IDH1-mutant cells include the antiviral PID AP1 pathway, Negative regulation of cell differentiation, Kinase and transcription factor activation, Signaling by interleukins and Cellular senescence (Fig. 2i). Notably, RASD1, a small G-protein within the RAS superfamily³⁵, emerged as the most significantly upregulated gene (9.04-fold change, FDR-corrected p value = 6.3 × 10⁻⁴¹) in IDH1-mutant cells during viral infection (Supplementary Data 2).

We assessed infectivity of rQNestin34.5 v.2 in a syngeneic murine glioma cell line engineered to harbor the IDH1-R132H mutation (designated C3-IDH1-mutant)^36,37. These cells carry loss-of-function mutations in ATRX and TP53, mimicking genetic alterations frequently observed in human astrocytomas², but also harbor NRAS activating mutations, which are less frequently observed in human diffuse gliomas³⁸. As controls, we used syngeneic glioma cells with wild-type IDH1, NRAS mutations, and loss-of-function mutations in ATRX and TP53 (designated C54-IDH1-wild-type). Both neurosphere cultures are permissive to rQNestin34.5 v.2 infection; however, like human tumors, IDH1-mutant cells demonstrate enhanced viral infectivity and cytotoxicity compared to IDH1-wild-type cells (Fig. 2j–l). These results further validate the increased susceptibility of IDH1-R132H mutant gliomas to rQNestin34.5 v.2.

IDH1 mutation in glioma associated with elevated nectin-1 expression and altered interferon responsiveness

One potential explanation for the increased infectivity of IDH1-mutant cells with HSV-1 is enhanced expression of viral entry receptors. KEGG pathway analysis comparing IDH1-mutant and wild-type MGG8 cells reveals significant alterations at multiple HSV-1 entry points. This includes differential expression of herpesvirus entry mediator (HVEM), Nectin, paired immunoglobulin-like type 2 receptor alpha (PILRα), and α5 (Supplementary Fig. 2). RNA-Seq analysis of uninfected MGG8-IDH1 mutant cells showed 370 genes significantly altered in the HSV-1 infection pathway, with 343 upregulated and 27 downregulated, including upregulation of Nectin 1, TNFRSF14 (gene encoding HVEM), and PILRa (Supplementary Data 3). To confirm the upregulation of Nectin-1, the primary HSV-1 entry receptor, in IDH1-mutant cells, we performed flow cytometry analysis. This revealed increased Nectin-1 expression in engineered IDH1-mutant glioma cell lines (MGG8, SJGBM2, and U87) as well as in patient-derived glioma cell lines harboring endogenous IDH1 mutations (MGG119, LC1035, and SF10602) (Fig. 3a). To validate this observation clinically, we analyzed TCGA diffuse glioma specimens and compared Nectin-1 expression between IDH-mutant and IDH-wild-type tumors (Fig. 3a). Stratification by grade and IDH mutation status (IDH-mutant grade 4 astrocytoma; n = 12, IDH-wild-type grade 4 astrocytoma; n = 158, IDH-mutant grade 3 astrocytoma; n = 58, and IDH-wild-type grade 3 astrocytoma; n = 31) revealed the highest Nectin-1 expression in IDH-mutant grade 3 astrocytomas (Fig. 3a). These data indicate that Nectin-1 expression is associated with both tumor grade and IDH mutation status. Together, elevated expression of Nectin-1, the primary HSV-1 entry receptor, suggests that the IDH1-R132H mutation may facilitate viral entry and increase glioma cell susceptibility to infection.

Fig. 3. IDH1 mutant status drives upregulation of HSV-1 Nectin-1 receptor expression.

a The geometric mean fluorescence intensity (MFI) of human Nectin-1 expression was measured by flow cytometry in MGG8 (IDH1 wild-type vs. mutant), SJGBM2 (IDH1 wild-type vs. mutant), and U87 (IDH1 wild-type vs. mutant) human glioma cell lines, as well as in patient-derived glioma cell lines harboring endogenous IDH1 mutations (MGG119, LC1035, and SF10602). Statistical comparisons were performed using a one-way ANOVA followed by Tukey’s multiple comparisons test. Data are presented as mean ± SD. Statistical significance is indicated as follows: *, P < 0.05; ****, P < 0.0001; ns, not significant. Normalized mRNA counts of human Nectin-1 gene expression in IDH-wild-type versus IDH-mutant gliomas (q values between groups are shown). Gene expression profiles were obtained from TCGA datasets of human glioma biopsies, including IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). q values represent adjusted p-values calculated using an optimized FDR approach. A final q < 0.05 is considered significant. b MFI of IFNAR1, (c) IFNAR2, and (d) IRF3 expression were assessed by flow cytometry in glioma cell lines (MGG8, SJGBM2, U87, C3, and C54) and in patient-derived xenograft glioma lines harboring endogenous IDH1 mutations (MGG119, LC1035, and SF10602), comparing IDH1 wild-type and mutant variants. Statistical comparisons were performed using one-way ANOVA followed by Tukey’s multiple comparisons test. Data are presented as mean ± SD. Statistical significance is indicated as follows: *, P < 0.05; **, P < 0.01; ****, P < 0.0001; ns, not significant. Normalized IFNAR1, IFNAR2 and IRF3 gene expression in IDH-wild-type and IDH-mutant LGG and GBM cases. Data were derived from TCGA mRNA-sequencing datasets of human glioma biopsies, including IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). q values indicate FDR-adjusted P values, with q < 0.05 considered statistically significant. e RSAD2 gene expression in human glioma cases: IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). q values between groups are shown, with q < 0.05 considered significant. Data are presented as mean ± SD. f Single-sample gene set enrichment (ssGSEA) normalized scores for IFN-I response, (g) Inflammatory response, (h) IL6-JAK-STAT3 signaling, (i) IL2-STAT5 signaling and (j) Hypoxia-related HALLMARK signatures were analyzed across TCGA glioma cohorts comprising IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). Two-tailed Student’s t-test was used to compare groups, with adjusted P values shown; P < 0.05 was considered significant. Data are presented as mean ± SD. k ROS staining of MGG8, SJGBM2, C3 and C54 IDH1 mutant and wild-type cells. Cells were infected with rQNestin34.5 v.1 at an MOI of 1, and cellular ROS was assessed 24 h post-infection. Data from vehicle- and virus-treated groups are shown as MFI. A two-way ANOVA followed by Šídák’s multiple comparisons test was used for statistical comparisons (n = 3). Data are presented as mean ± SD. Statistical significance is indicated as follows: ***, P < 0.001; ****, P < 0.0001; ns, not significant. l Annexin V staining of SJGBM2 IDH1-mutant vs. wild-type glioma cells 48 h post-infection with rQNestin34.5 v.1. Cells were infected at an MOI of 1. Percentages of live, naked nuclei, early apoptotic, and late apoptotic cells (n = 3 replicates/group) are shown. A two-way ANOVA followed by Šídák’s multiple comparisons test was used for statistical comparisons. Statistical significance is *, P < 0.05; ****, P < 0.0001; ns, not significant. m GL261 murine glioma cells were lentivirally engineered to constitutively express human cell-surface Nectin-1, and the IDH1-R132H mutation was introduced. Expression of mutant IDH1 was confirmed in GL261-N1-mIDH1 tumors established in C57BL/6 mice. Representative immunohistochemistry demonstrated robust IDH1-R132H staining in mutant tumors, whereas no signal was detected in normal brain tissue or GL261-N1 tumors expressing wild-type IDH1 (Scale bars, 100 μm; 10× magnification). n MFI of Nectin-1 and o, IFNAR-1 cell-surface expression was measured in GL261-N1-wt-IDH1 and GL261-N1-mIDH1 cell lines (n = 3-4 replicates/group). Statistical comparisons were performed using Student’s t-test. Data are presented as mean ± SD. Statistical significance is indicated as **, P < 0.01 and ****, P < 0.0001. p Kaplan-Meier survival curves of mice bearing GL261-N1-wt-IDH1 or GL261-N1-mIDH1 tumors (n = 6 mice per group). Survival differences were analyzed using a log-rank (Mantel-Cox) test; P > 0.05, not significant. q Representative green fluorescence microscopy images showing viral infection in GL261-N1-wt-IDH1 and GL261-N1-mIDH1 glioma cells 72 h post-infection with rQNestin34.5 v.1 at an MOI of 3. Cells were pretreated with 10 µM ruxolitinib or vehicle (DMSO) and maintained in the drug for the duration of the experiment. Scale bars, 100 μm (10× magnification) r Percentage and MFI of GFP⁺ (HSV-1-infected) cells were quantified by flow cytometry 48 h post-infection (n = 3 biological replicates per group). Data are presented as mean ± SD. Statistical significance was assessed using two-tailed student’s t test (**, P < 0.01). s Concentration of mouse IFN-β (pg/ml) in the cell-free supernatant of GL261-N1-wt-IDH1 and GL261-N1-mIDH1 glioma cells 48 h post-infection with rQNestin34.5 v.2 at an MOI of 3. All samples were run in duplicate (n = 3 biological replicates per group). Data are presented as mean ± SD. Statistical comparisons were performed using two-way ANOVA followed by Šídák’s multiple comparisons test. Significance is indicated as *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. t ROS levels in GL261-N1-mIDH1 and wild-type cells were assessed 24 h post-infection with rQNestin34.5 v.1 (MOI 3) using fluorescence-based staining. Data from vehicle- and virus-treated groups are shown as MFI. Statistical comparisons were performed using two-way ANOVA followed by Šídák’s multiple comparisons test (n = 4 replicates per group). Data are presented as mean ± SD. Significance is indicated as *, P < 0.05; **, P < 0.01; ****, P < 0.0001; ns, not significant. All experiments were independently performed at least twice with reproducible results.

We next assessed the expression of interferon-αlpha/beta receptors 1 and 2 (IFNAR-1 and IFNAR-2) and the interferon-stimulated gene (ISG) interferon regulatory factor 3 (IRF3) in both IDH1-mutant and IDH1-wild-type glioma cell lines and TCGA clinical specimens. IFNAR-1 protein levels were consistently lower in IDH1-mutant engineered human lines (MGG8, SJGBM2, U87) and in patient-derived glioma cell lines harboring endogenous IDH1 mutations (MGG119, LC1035, and SF10602) (Fig. 3b). In mouse models, IFNAR-1 expression was significantly reduced in C3-mIDH1 cells compared with C54-wt-IDH1 cells, indicating that the IDH1-R132H mutation may suppress IFNAR-1 expression (Fig. 3b). Supporting this, in-vitro infection with rQNestin34.5 v.2 failed to induce interferon-beta (IFN-β) cytokine release in any of the tested IDH1-mutant human cell lines (data not shown), further indicating a dysfunctional IFNAR-1-mediated response, which is known to be critical for IFN-β induction³⁹. In contrast, IFNAR-2 and IRF3 expression levels were consistently higher across all IDH1-mutant glioma cell lines (Fig. 3c, d). However, analysis of glioma specimens from TCGA revealed reduced expression of IFNAR-1, IFNAR-2 and IRF3 in IDH-mutant groups (Fig. 3b–d). This discrepancy may reflect the influence of the TME and the complex crosstalk of type I interferon signaling between tumor and host immune cells, which is challenging to recapitulate in vitro. TCGA IDH-mutant grade 3 astrocytomas also exhibited reduced expression of innate antiviral signaling genes, including radical S-adenosyl methionine domain-containing 2 (RSAD2/viperin) (Fig. 3e).

We further assessed hallmark pathways regulating viral replication, inflammatory responses, cell survival and proliferation, and apoptosis, including type I interferon (IFN-I), IL6-JAK-STAT3, IL2-STAT5, and hypoxia signaling, using ssGSEA analyses, and found that all were consistently downregulated in IDH-mutant patient groups (Fig. 3f–j). We next measured reactive oxygen species (ROS) levels and apoptosis in IDH1-mutant and wild-type glioma cell lines at baseline and following in-vitro infection with rQNestin34.5 v.2. While baseline ROS levels were comparable, viral infection increased ROS and induced early apoptosis specifically in IDH1-mutant cells, indicating heightened sensitivity to oxidative stress-induced cell death (Fig. 3k, l). These differences in Nectin-1 expression, type I IFN responses, hypoxia signaling, and redox regulation between IDH1-mutant and wild-type gliomas prompted further investigation into how the IDH1-R132H mutation modulates immune responses to oHSV-1 rQNestin34.5 v.2 immunovirotherapy.

To determine whether enhanced viral infection in IDH1-mutant glioma is solely due to elevated Nectin-1 expression, we utilized the IDH1/2 wild-type GL261 orthotopic murine model. oHSV-1 rQNestin34.5 v.2 cannot infect GL261 cells, so we transduced them with a lentivirus constitutively expressing human Nectin-1 to enable viral entry. GL261-Nectin-1 cells were subsequently engineered to express the IDH1-R132H mutation (Fig. 3m), which further increased Nectin-1 protein levels and reduced IFNAR1 expression compared with GL261-Nectin-1 cells harboring wild-type IDH1 (Fig. 3n, o). While R132H expression remained stable in vivo, it did not extend survival in mice, suggesting that additional cellular mechanisms contribute to the phenotype of IDH1-mutant gliomas (Fig. 3p). Notably, viral infection was more robust in GL261-Nectin-1 cells expressing IDH1-R132H compared to wild-type IDH1 cells (Fig. 3q, r), and was accompanied by reduced IFN-β secretion and increased ROS production following infection (Fig. 3s, t). Pharmacological inhibition of the JAK/STAT pathway with the FDA-approved JAK1/2 inhibitor ruxolitinib⁴⁰ further enhanced viral infection, supporting a central role for this pathway in mediating oHSV-1 resistance in this model (Fig. 3q, r). Together, these findings indicate that the IDH1-R132H mutation promotes Nectin-1 expression and establishes a cellular program that enhances oHSV-1 rQNestin34.5 v.2 infection.

IDH1 mutation influences inflammatory mediators during oHSV-1 rQNestin34.5 v.2 infection

Given that IDH1-R132H cells are more susceptible to infection by rQNestin34.5 v.2, we next assessed how infection affects inflammation in these tumors. We first measured the extracellular release of two damage-associated molecular patterns (DAMPs), calreticulin (CALR) and high-mobility group box 1 (HMGB1), after infection with rQNestin34.5 v.2. C3-IDH1-mutant cells have significantly higher DAMP release compared to C54-IDH1-wild-type cells, which correlates with enhanced viral replication in IDH1-mutant cells (Supplementary Fig. 3a, b). C3-IDH1-mutant cells have less monocyte chemoattractant protein-1 (MCP-1/CCL2) and proinflammatory chemokine C-X-C motif ligand 1 (CXCL1) released after infection. IFN-γ treatment of IDH1-mutant cells stimulated release of chemoattractants CCL9 and CXCL10, whereas IDH1-wild-type cells exhibited minimal responses to IFN-γ (Supplementary Fig. 3c, d). Collectively, these findings suggest that IDH1-mutation influences rQNestin34.5 v.2 and IFN-γ mediated immune responses.

To confirm this phenotype in vivo, C57BL/6 mice bearing either C3-IDH1-mutant or C54-IDH1-wild-type intracranial tumors were treated with intratumoral rQNestin34.5 v.2 (1.5 × 10⁶ pfu) (Fig. 4a). Three days post-infection, mice with IDH1-mutant tumors exhibited elevated serum IFN-γ and interleukin-1 alpha (IL-1α), while mice bearing IDH1-wild-type tumors showed no changes in cytokine levels (Fig. 4b and Supplementary Fig. 4). These inflammatory responses were associated with increased frequencies of circulating NK cells, Ly6C⁺ CD8⁺ T cells, and myeloid cells. Additionally, there was reduced expression of CX3CR1⁺ and F4/80⁺ on CD11b⁺ cells in mice bearing IDH1-mutant tumors after infection, suggesting altered myeloid cell activation (Fig. 4c). In sharp contrast, no significant changes in peripheral immune responses were observed in IDH1-wild-type tumor-bearing mice after rQNestin34.5 v.2 infection. Interestingly, pre-infection blood levels of all inflammatory cytokines were higher in C3-IDH1-mutant tumor bearing groups, further emphasizing the immunologically distinct characteristics of IDH1-mutant gliomas (Fig. 4b and Supplementary Fig. 4). Some of the systemic differences also manifest within the TME, with IDH1-mutant tumors exhibiting higher numbers of CD45.2^high leukocytes, myeloid cells, and CD3⁺ T cells compared to IDH1-wild-type tumors (Fig. 4d). In IDH1-mutant tumors, rQNestin34.5 v.2 infection was associated with increased NK cell infiltration and, conversely, reduced frequencies of B220⁺ cells, CD4⁺ T cells, and TIGIT expression on NK cells. In contrast, IDH1-wild-type tumors exhibited increased NK cell and macrophage infiltration, along with decreased frequencies of CD4⁺ T cells and neutrophils after rQNestin34.5 v.2 treatment (Fig. 4d). Despite elevated CX3CR1 expression in peripheral CD11b⁺ cells of mice bearing IDH1-mutant tumors, CX3CR1 levels within the TME were comparable between IDH1-mutant and wild-type groups, both at baseline and following rQNestin34.5 v.2 infection (Fig. 4d). Together, these results indicate that rQNestin34.5 v.2 triggers much different innate immune responses in IDH1-mutant vs wild-type gliomas, resulting in significantly different TMEs.

Fig. 4. Intratumoral oHSV-1 rQNestin34.5 v.2 treatment triggers local and systemic immune activation in IDH1-mutant glioma.

a Treatment and sample collection: C57BL/6 mice were intracranially implanted with C3-IDH1-mutant or C54-IDH1-wild-type cells (40,000 cells) and treated with a single intratumoral dose of rQNestin34.5 v.2 (1.5 × 10⁶ pfu in 10 µL total volume) as outlined in the treatment schedule. On days 3 and 8 post-treatment, half of the mice from each group were perfused, and tumors were harvested under sterile conditions for immunophenotyping. Serum and peripheral blood samples were also collected for analysis. Graphic was created in BioRender. Panagioti, E. (2026) https://BioRender.com/bccox9x. b Levels of proinflammatory cytokines (IFN-γ and IL-1α) in the serum of C3-IDH1-mutant vs. C54-IDH1-wild-type tumor-bearing mice on day 3 post rQNestin34.5 v.2 treatment. Samples were run in duplicate for all mice (n = 5-6 mice per group), and the mean value was used for analysis. c Percentages of immune cell subsets in the peripheral blood of mice bearing C3-IDH1-mutant tumors on day 3 after rQNestin34.5 v.2 treatment. Data are shown as mean ± SD. d Percentages of CD45.2^high immune cell subsets in the TME of C3-IDH1-mutant vs. C54-IDH1-wild-type tumor-bearing mice on day 3 after rQNestin34.5 v.2 treatment. Data are presented as mean ± SD (n = 4–5 mice per group). A two-way ANOVA followed by Tukey’s multiple comparisons test was used for comparisons involving more than two groups (b, d), while a non-parametric Student’s t-test was applied for comparisons between two groups (c). Significance is indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; P > 0.05, ns, not significant. Experiments were independently performed at least twice with reproducible results.

TME differences persisted until at least day 8 post-infection, with IDH1-mutant tumors showing marked reductions in myeloid cells and increased frequencies of CD3⁺ T cells, CD4⁺ T cells, CD8⁺ T cells, and granzyme B⁺ CD8⁺ T cells. Additionally, TIGIT expression on CD4⁺ T cells and PD-1 expression on CD8⁺ T cells decreased (Supplementary Fig. 5a). In contrast, rQNestin34.5 v.2-treated IDH1-wild-type tumors showed much weaker immune responses, with only a reduction in myeloid cells and an increase in CD4⁺ T cells observed. Interestingly, a large population of regulatory T cells (Tregs; FOXP3⁺ CD4⁺ T cells) was observed in IDH1-mutant tumors compared to wild-type tumors, though this occurs independently of rQNestin34.5 v.2 treatment (Supplementary Fig. 5a). Serum cytokine analyses at day 8 revealed no differences between rQNestin34.5 v.2 -treated and vehicle-treated groups. However, IDH1-mutant tumor-bearing mice consistently displayed higher circulating cytokine levels compared to wild-type (Supplementary Fig. 5b). These findings align with the day 3 post-infection data, further supporting the role of the IDH1-R132H mutation in modulating chemokine secretion and immune responses.

To help corroborate these findings, we reanalyzed data from a recently published Phase I clinical trial of the oncolytic HSV-1 rQNestin34.5 v.2 (CAN-3110)³⁴ treated patients with recurrent IDH-mutant astrocytoma (n = 4) and oligodendroglioma (n = 5). We compared IDH-mutant and IDH-wild-type patients in this cohort to investigate the impact of IDH-mutation status on oncolytic HSV-1 therapy. Although statistical significance was not reached for many findings due to the limited sample size, several notable trends emerged that merit mention. IDH-mutant tumors from patients who were HSV-1 seronegative at baseline (confirmed by serological testing at enrollment) showed a trend toward higher PVRL1 (encoding Nectin-1) expression and lower levels of Growth Differentiation Factor 15 (GDF-15), SOD2, STAT1, and innate antiviral signaling genes, including IFNAR1, IRF3, and RSAD2/viperin, compared with HSV-1-seronegative IDH-wild-type tumors (Fig. 5a). Longitudinal transcriptional profiling of tumor biopsies before and after CAN-3110 infection demonstrated significant downregulation of STAT1 and RSAD2, along with a trend toward reduced PVRL1 expression in IDH-wild-type tumors, whereas no corresponding changes were observed in IDH-mutant tumors. While GDF-15 expression remained stable in IDH-mutant tumors, it was nominally upregulated in patients with IDH-wild-type tumors after treatment (Fig. 5b). Pathologic evaluation of matched pre- and post-treatment tumor samples from HSV-1 seronegative patients with IDH-mutant or wild-type gliomas revealed a trend toward an increase in CD4⁺ T cells post-treatment (Fig. 5c). Evaluation of the TME via ssGSEA analysis revealed trends towards increased ROS and hypoxia in IDH-wild-type patients (Fig. 5d), along with an increase in hypoxia signatures post-treatment (Fig. 5e).

Fig. 5. Immune cell responses in IDH-mutant and wild-type glioma biopsies pre and post-intratumoral CAN-3110 treatment.

a Bulk RNA sequencing analyses showing Transcripts Per Million (TPM) expression of PVRL1, SOD2, GDF15, JAK1, STAT1, IFNAR1, IRF3, and RSAD2 in IDH-mutant (n pre/post = 4) vs. IDH-wild-type (n pre = 15, n post = 17) patient biopsies collected pre- and post-CAN-3110 (clinical grade rQNestin34.5 v.2) treatment. These patients were HSV-1 seronegative at the start of treatment. Wilcoxon unpaired rank sum p-values are shown for comparisons of IDH-mutant vs. IDH-wild-type patients at each timepoint. b TPM expression of PVRL1, SOD2, GDF15, JAK1, STAT1, IFNAR1, IRF3 and RSAD2 in IDH-mutant (n = 3) vs. IDH-wild-type (n = 14) patient biopsies collected pre- and post-CAN-3110 treatment. Wilcoxon paired signed-rank p-values are shown above each comparison of pre- vs. post-CAN-3110 samples collected from IDH-mutant or IDH-wild-type patients. c Pathological analysis of biopsies comparing paired pre- and post-CAN-3110 treatment in IDH-mutant and IDH-wild-type cases. Graphs display CD8⁺ (IDHmut: n = 5; IDHwt: n = 22), CD4⁺ (IDHmut: n = 5; IDHwt: n = 20), and CD20⁺ (IDHmut: n = 5; IDHwt: n = 20) cells per mm³. Wilcoxon paired signed-rank p-values are shown above each comparison of pre- vs. post-CAN-3110 samples collected from IDH-mutant or IDH-wild-type patients. d HALLMARK gene ssGSEA scores for hypoxia and reactive oxygen species related signatures comparing IDH-mutant (pre/post: n = 4) vs. IDH-wild-type (pre: n = 16; post: n = 15) patients pre- and post-CAN-3110 treatment. Wilcoxon unpaired rank sum p-values are shown for comparisons of IDH-mutant vs. IDH-wild-type patients at each timepoint. e HALLMARK gene ssGSEA scores from (d) in paired pre- vs. post-CAN-3110 samples for IDH-mutant (n = 3) and IDH-wild-type (n = 14) patients. Wilcoxon paired signed-rank p-values are shown above each comparison of pre- vs. post-CAN-3110 samples collected from IDH-mutant or IDH-wild-type patients. Note that only these two signatures were assessed in the analyses for (d) and (e) to avoid a large multiple-testing penalty with a small sample size. f Kaplan-Meier survival curves for HSV-1 seronegative (n = 4, median survival = 5.4 months) vs. seropositive patients (n = 5, median survival = 39.9 months) with IDH-mutant gliomas treated with CAN-3110 (likelihood ratio test p = 0.05). Both astrocytoma and oligodendroglioma patients are included. Seropositive and seronegative patients were defined based on HSV-1 serology status assessed at enrollment, prior to CAN-3110 treatment. g Kaplan-Meier survival curves as in panel f with division of IDH-mutant oligodendrogliomas and astrocytomas. Oligodendroglioma HSV-1 seronegative (n = 1, survival = 32.1 months) vs. seropositive (n = 4, median survival = 49.5 months), and astrocytoma HSV-1 seronegative (n = 3, median survival = 3.5 months) vs. seropositive at (n = 1, survival = 22.5 months). Boxplots in (a–e) show the median (center line), 25 and 75th percentiles (box limits) and up to 1.5× the interquartile range or to the minimum/maximum values (if <1.5 × interquartile range distance from the box) (whiskers).

Strikingly, as has been previously shown with IDH-wild-type patients³⁴, the most robust clinical responses to CAN-3110 in IDH-mutant gliomas were observed in patients confirmed to be HSV-1 seropositive prior to treatment (Fig. 5f, g), suggesting the potential role of pre-existing HSV-1-specific memory B cells and T cells in enhancing anti-tumor immunity. Consequently, HSV-1 seropositive IDH-mutant patients may derive enhanced benefit from oncolytic HSV-1 therapy due to reduced antiviral signaling, virus-induced memory recall responses, and decreased susceptibility to hypoxia-mediated immune suppression compared to IDH-wild-type tumors.

IDH1-R132H gliomas and checkpoint immunotherapy

Given the significant differences in phenotypes and responses to virotherapy, we next investigated the efficacy of checkpoint-based immunotherapy in IDH1-mutant and wild-type gliomas. We first characterized two syngeneic murine models, harboring either mutant or wild-type IDH1, both in vitro and in vivo, to evaluate their suitability for modeling high-grade glioma. In-vitro, both C3-IDH1-mutant and C54-IDH1-wild-type cell lines exhibited robust neurosphere growth and expressed IDH1 at the cellular level (Fig. 6a). In vivo, IDH1-mutant tumors developed into high-grade lesions and retained stable expression of the IDH1-R132H mutation as the tumors grew to large volumes. In contrast, IDH1-wild-type tumors formed more aggressive, high-grade, and invasive lesions that remained negative for the R132H mutation (Fig. 6b). High expression of GDF-15 is associated with high-grade tumors and poor prognosis in glioblastoma patients^41,42. To further explore this, we compared GDF-15 expression in diffuse glioma biopsies stratified by grade and IDH mutation status (IDH-mutant grade 4 astrocytoma; n = 12, IDH-wild-type grade 4 astrocytoma; n = 158, IDH-mutant grade 3 astrocytoma; n = 58, and IDH-wild-type grade 3 astrocytoma; n = 31). Consistent with clinical observations from CAN-3110 trial patients described above, IDH-wild-type tumors exhibited significantly higher GDF-15 expression, which also increased with tumor grade (Fig. 6c). Consistently, in our mouse model, animals with IDH1-mutant tumor implants had lower serum levels of GDF-15 cytokine and showed prolonged survival compared to those with IDH1-wild-type tumors (Fig. 6d, e). These findings further support the association between elevated GDF-15 levels and more aggressive glioma phenotypes. Specifically, in C54-IDH1-wild-type tumors, median survival following injection of 30,000 cells was 25 days, whereas in C3-IDH1-mutant tumors, survival was extended to 46 days. Survival was also dependent on the number of cells injected; with 60,000 cells, median survival was reduced to 22 days for C54-IDH1-wild-type and 35 days for C3-IDH1 cell recipients (Fig. 6e). PD-L1 expression was assessed in-vitro, revealing no significant baseline differences between wild-type and IDH1-mutant cells, with both exhibiting low expression levels. Upon IFN-γ stimulation, PD-L1 expression was upregulated in both tumor groups; however, wild-type cells showed stronger induction of PD-L1 compared to mutant cells (Fig. 6f).

Fig. 6. IDH1-R132H glioma responds to anti-PD1 immunotherapy.

a IDH1 protein expression in C3-IDH1-mutant and C54-IDH1-wild-type cells in culture. Representative histograms compare IDH1 expression to isotype control. b Validation of IDH1-R132H expression in C3-IDH1-mutant glioma tissue from C57BL/6 mice with large tumors. Representative immunohistochemistry images show positive IDH1-R132H staining in C3-IDH1-mutant tumors, with no detectable staining in normal brain tissue or C54-IDH1-wild-type tumors. Scale bars, 100 μm (10× magnification). c Normalized human GDF-15 gene expression in IDH-wild-type and IDH-mutant diffuse glioma cases. Data were obtained from TCGA-deposited mRNA sequencing datasets of human glioma biopsies, comprising IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). Data are presented as mean ± SD. q values represent FDR-adjusted P values, with q < 0.05 considered significant. d C57BL/6 mice were intracranially implanted with either C3-IDH1-mutant or C54-IDH1-wild-type cells (40,000 cells). Serum was collected 10 days later, and GDF-15 levels (pg/ml) were measured (n = 4-5 mice per group). Data are shown as mean ± SD; comparisons were performed using a non-parametric Student’s t-test. *, P < 0.05. e Kaplan-Meier survival curves of C57BL/6 mice injected with 30,000 or 60,000 C3-IDH1-mutant and C54-IDH1-wild-type cells (n = 4-5 mice/ group). P values were determined using the log-rank Mantel-Cox test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P > 0.05, not significant. f MFI of PD-L1 expression in C3-IDH1-mutant and C54-IDH1-wild-type cells in-vitro, analyzed either untreated or after overnight stimulation with mouse IFN-γ (100 ng/well). Data represent three biological replicates per group (n = 3) and are shown as mean ± SD. Statistical comparisons were performed using two-way ANOVA with Šídák’s multiple comparisons test. g C57BL/6 mice bearing C3-IDH1-mutant or C54-IDH1-wild-type tumors were perfused, and brains were collected and fixed in 10% formalin for immunohistochemistry. Representative images show H&E, Ki67, CD8, and PD-L1 staining in C54-IDH1-wild-type tumors at day 18 post-implantation and in C3-IDH1-mutant tumors at days 18 and 35. Scale bars, 100 μm (20× magnification). h Quantification of tumor mitotic activity, percentages of Ki67⁺ neoplastic cells, CD8⁺ T cell infiltration (cells/mm²), and percentages of PD-L1⁺ neoplastic cells in C3-IDH1-mutant and C54-IDH1-wild-type tumors collected at days 18 and 35 post-implantation (n = 4 mice per group). Data are shown as mean ± SD. One-way ANOVA with Tukey’s multiple comparisons test was used for statistical analysis. Significance is indicated as **, P < 0.01; ****, P < 0.0001; ns, not significant (non-significant results are not marked in the graphs). i Anti-PD-1 treatment in C3-IDH1-mutant tumors: C57BL/6 mice implanted with 40,000 C3-IDH1-mutant cells were treated with anti-PD-1 antibodies (clones 29 F.1A12, 332.8H3, and RMP1-14) according to the indicated schedule. Antibodies were administered intraperitoneally at 200 µg per mouse every 2-3 days for a total of five doses. Blood and serum were collected on the day of the final dose. Created in BioRender. Panagioti, E. (2026) https://BioRender.com/flaack4. j Kaplan-Meier survival curves of mice treated with anti-PD-1 antibodies or IgG isotype control (n = 6 mice/ per group). Median survival and group comparisons were assessed using the log-rank test. Statistical significance is indicated as **, P < 0.01; P > 0.05, not significant. k Mice that remained tumor-free for 180 days (from panel j) were rechallenged with autologous C3-IDH1-mutant cells in the hemisphere contralateral to the primary tumor injection. Naïve mice injected with tumor cells served as controls (n-2-5 mice per group). Kaplan-Meier survival curves are shown. l Anti-PD-1 treatment in C54-IDH1-wild-type tumors: C57BL/6 mice implanted with 40,000 C54-IDH1-wild-type cells were treated intraperitoneally with anti-PD-1 antibodies (clones 29 F.1A12, 332.8H3, and RMP1-14) or IgG isotype control at 200 µg per mouse according to the indicated schedule (n = 6 mice per group). Created in BioRender. Panagioti, E. (2026) https://BioRender.com/flaack4. m Body weight (grams) kinetics of mice bearing C54-IDH1-wild-type tumors and treated with anti-PD1 clones (n = 6 mice per group). Data are shown as mean ± SD. n Kaplan-Meier survival curves of mice bearing C54-IDH1-wild-type tumors treated with anti-PD-1 antibodies (n = 6 mice per group). Survival comparisons were performed using the log-rank test. P > 0.05, not significant. All experiments, except for the survival analyses in panels e and i-n and the immunohistochemistry results, were independently repeated twice with similar outcomes.

Tumor-infiltrating immune cells were assessed in both C3-IDH1-mutant and C54-IDH1-wild-type tumors using hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC). Due to the significantly faster growth rate of IDH1-wild-type tumors, two time points were selected for analysis of the slower-growing IDH1-mutant tumors: day 18, corresponding to the peak growth of wild-type tumors (when mice became moribund), and day 35, when IDH1-mutant tumors reached their comparable endpoint. When tumors are enlarged and symptomatic, IDH1-mutant tumors (day 35) show comparable CD8⁺ T cell infiltration, mitotic activity, and Ki67 expression to those of enlarged and symptomatic (day 18) IDH1-wild-type tumors. Notably, PD-L1 expression was significantly higher and more uniformly distributed in IDH1-mutant-type tumors than in IDH1-wild-type tumors (Fig. 6g, h). These findings suggest that the IDH1-R132H mutation influences immune checkpoint PD-L1 expression, resulting in distinct immune landscapes despite comparable tumor sizes. When comparing early (day 18) to late (day 35) IDH1-mutant tumors, early tumors demonstrated significantly higher CD8⁺ T cell infiltration, with no changes in PD-L1 expression, Ki67 levels, or mitotic activity (Fig. 6g, h). These results suggest that harnessing infiltrating CD8⁺ T cells early during tumor development may improve therapeutic outcomes in IDH1-mutant tumors.

Given the differences in PD-L1 expression between IDH1-wild-type and IDH1-mutant models, therapeutic inhibition of PD-1/PD-L1 binding was explored as a potential strategy. We systemically administered one of three in vivo PD-1 blocking antibody clones targeting mouse PD-1: clone 29 F.1A12, clone RMP1-14 (both rat IgG2a), or clone 332.8H3 (mouse IgG1) to mice with IDH-1 mutant or wild-type tumors. For IDH1-mutant tumors, treatment was initiated four days later than in the IDH1-wild-type model due to their slower growth kinetics (Fig. 6i). Anti-PD-1 inhibition was highly effective in the C3-IDH1-mutant glioma model. The 29 F.1A12 and RMP1-14 clones exhibited the greatest efficacy, preventing tumor death in 80% and 50% of the mice, respectively. In contrast, the 332.8H3 clone was less effective, curing only 33% of the mice (Fig. 6j). Long-term C3-IDH1-mutant tumor survivors were rechallenged by implanting the same glioma cells into the opposite hemisphere of the primary tumor, and no new tumors developed, suggesting the establishment of immunological memory (Fig. 6k). In contrast, none of the clones prevented tumor death in the C54-IDH1-wild-type model. However, a trend toward improved outcomes was observed in mice treated with the 29 F.1A12 and RMP1-14 anti-PD1 clones, as evidenced by slower body weight loss and prolonged survival (median survival: 17 days in isotype control-treated mice versus 20 and 19.5 days in RMP1-14- and 29 F.1A12-treated mice, respectively) (Fig. 6l–n).

We next performed immunophenotyping on peripheral blood and serum collected on the day of the final anti-PD-1 treatment to assess immune cell responses in mice with C3-IDH1-mutant gliomas. Monoclonal 29 F.1A12-treatment was associated with increased systemic IFN-γ release and upregulation of PD-1 on circulating B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells, suggesting immune cell activation (Supplementary Fig. 6a–e). In contrast, 332.8H3-treatment led to increased frequencies of myeloid cells (CD11b⁺), which may have contributed to the lower therapeutic efficacy observed with this clone, as well as increased TIGIT expression on NK cells. (Supplementary Fig. 6b, e). Interestingly, RMP1-14 treatment induced changes solely in TIGIT expression on CD8⁺ T cells and NK cells, as well as an increase in KLRG1 expression on CD8⁺ T cells, suggesting a multimodal mechanism of action (Supplementary Fig. 6d, e). PD-1 inhibition may also influence TIGIT expression, highlighting their functional interplay²⁵. However, because only a single time point was assessed, the possibility of broader immune activation during the initial doses of PD-1 blockade cannot be concluded. Collectively, while no existing models fully recapitulate the complexity of human gliomas, the immunocompetent C3 (IDH1-mutant) and C54 (IDH1-wild-type) models reproduce many hallmark features of the disease, providing a valuable platform for studying tumor biology, immune interactions, therapeutic responses, and mechanisms of resistance.

Screening immune checkpoint targets for combination with oHSV-1 rQNestin34.5 v.2

Although in situ administration of rQNestin34.5 v.2 activates the immune system and modulates the TME in both IDH1-mutant and wild-type syngeneic murine gliomas; we expected it to be unlikely as monotherapy to be highly effective in gliomas^30,34. Therefore, we explored a combination strategy to enhance immune responses and improve therapeutic outcomes. To identify potential combination targets, we performed in-vitro screening of key inhibitory checkpoint pathways currently under investigation in clinical trials (Fig. 7a), evaluating checkpoint ligands on tumor cells with and without rQNestin34.5 v.2 infection.

Fig. 7. The PVR/TIGIT/CD226 pathway is overexpressed in IDH1-R132H mutated glioma.

a During the effector phase of immune responses, T-cell receptors (TCRs) on immune cells recognize peptide-MHC complexes on tumor cells, initiating an antitumor immune response. Simultaneously, immune checkpoint receptors on immune cells interact with ligands expressed on cancer cells, modulating the immune response. Several inhibitory checkpoint pathways are implicated in glioma, including PD-1/PD-L1, TIM-3/Galectin-9, TIGIT/CD155(PVR)/CD112, LAG-3/MHC class II, HVEM/BTLA, and CTLA-4/CD80/CD86 (Created in BioRender. Panagioti, E. (2026) https://BioRender.com/lqdexxp). b C3-mIDH1 and c C54-wt-IDH1 cells were infected in-vitro with rQNestin34.5 v.1 at various MOIs (0.01, 0.05, 0.1, and 0.5), and checkpoint ligand expression was assessed 24 h later by flow cytometry. As controls, cells were stimulated with mouse IFN-β and IFN-γ (100 ng/well). MFI of checkpoint ligand expression is shown as heat maps (n = 3 replicates per group). ND=no detectable expression. d Comparison of CD155 and e PD-L1 expression across TCGA-derived human glioma biopsies, including IDH-mutant grade 4 astrocytoma (n = 12), IDH-wild-type grade 4 astrocytoma (n = 158), IDH-mutant grade 3 astrocytoma (n = 58), and IDH-wild-type grade 3 astrocytoma (n = 31). RNA counts are normalized for each group. Data are presented as mean ± SD. q values represent FDR-adjusted P values, with q < 0.05 considered significant. C57BL/6 mice bearing intracranial C3-mIDH1 and C54-wt-IDH1 tumors were treated intratumorally with rQNestin34.5 v.2, and tumors were collected on day 3 post-infection after perfusion. MFI of PD-L1 (f), CD155 (g), TIGIT (h), and CD226 (i) expression in CD45^low and CD45.2^high populations (myeloid cells, macrophages, neutrophils, CD4⁺ T cells, CD8⁺ T cells, NK cells and B cells) is shown. Comparisons were made between vehicle- and rQNestin34.5 v.2-treated groups for each cell subset (n = 3 mice per group). Statistical significance was determined by two-way ANOVA with Tukey’s multiple comparisons test. Data are presented as mean ± SD. Significance is indicated as *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. Non-significant results are not indicated with a symbol in the graphs. j Pie charts show the proportions of single CD226⁺, single TIGIT⁺, and double CD226⁺/TIGIT⁺ CD4⁺ and CD8⁺ tumor-infiltrating T cells in tumors collected as described above (n = 3 mice per group). Statistical significance was determined by two-way ANOVA with Šídák’s multiple comparisons test. Data are presented as mean ± SD. Non-significant results are not indicated with a symbol in the graphs. All experiments were independently repeated twice with similar outcomes.

In C3-IDH1-mutant cells, infection increased CD112 expression, while in C54-IDH1-wild-type cells, upregulation of CD155 and BTLA was observed (Fig. 7b, c). Notably, baseline expression of CD155 was dramatically higher in the IDH1-mutant cells. Additionally, IDH1-mutant cells displayed higher baseline levels of CD80 and BTLA, whereas IDH1-wild-type cells showed higher expression of MHC class I and PD-L1. Stimulation with exogenous mouse IFN-γ and IFN-β further increased MHC class II and checkpoint ligand expression in both models, with IFN-γ inducing stronger upregulation of MHC class I, CD155, CD80, and PD-L1 compared to IFN-β (Fig. 7b, c). In-vitro screening was also extended to the CT-2A glioma cell line, which is wild-type for IDH1/2 and was lentivirally transduced to express human Nectin-1, enabling HSV-1 entry, replication, and cytotoxic activity (Supplementary Fig. 7a–c). Infection of CT-2A-Nectin-1 cells with rQNestin34.5 v.2 induced an increase in HMGB1 levels, whereas calreticulin remained undetectable (Supplementary Fig. 7d). Additionally, infection did not stimulate cytokine release; only MCP-1 and IL-6 were detected in-vitro, regardless of the MOI used (Supplementary Fig. 7e). These results suggest immune resistance of CT-2A and help to explain the limited immunostimulatory effect of rQNestin34.5 v.2 in this model. CT-2A-Nectin-1 cells exhibited high CD155 expression, and treatment with exogenous IFN-γ led to upregulation of MHC class I and PD-L1 (Supplementary Fig. 7f). TIGIT expression was detected on CD4⁺ and CD8⁺ T cells, as well as NK cells within CD45.2^high tumor-infiltrating leukocytes (TILs). Intratumoral administration of rQNestin34.5 v.2 increased TIGIT expression on NK cells by day 3 post-infection (Supplementary Fig. 7g).

Additionally, reanalysis of TCGA gene expression data from grade 3 and 4 IDH-mutant and IDH-wild-type diffuse glioma biopsies confirmed increased expression of CD155 and PD-L1 in both groups (Fig. 7d, e), underscoring the potential relevance of the TIGIT and PD-1 pathways in glioma immune evasion. Notably, elevated CD155 and PD-L1 expression was associated with both IDH status and tumor grade, with IDH-wild-type grade 4 astrocytoma (GBM) exhibiting the highest levels.

Following this in-vitro analysis, attention shifted to in vivo evaluation of PD-L1 and CD155 expression in the glioma TME of C3-IDH1-mutant and C54-IDH1-wild-type gliomas, with a focus on immune cell subsets expressing these ligands and the impact of rQNestin34.5 v.2 treatments on their expression. We concomitantly evaluated the distribution of TIGIT and CD226, both of which bind to CD155, in the TME after rQNestin34.5 v.2 infection. Infection upregulated PD-L1 and CD155 expression on myeloid subsets in both IDH1-mutant and wild-type tumors (Fig. 7f, g). TIGIT was highly expressed on CD8⁺ tumor-infiltrating T cells in both models, and CD226 was prominently expressed on CD4⁺ and CD8⁺ T cells, NK cells, and B cells, with upregulation on neutrophils following infection (Fig. 7h, i). Interestingly, while the majority of tumor-infiltrating CD4⁺ T cells expressed only CD226, a significant proportion of tumor-infiltrating CD8⁺ T cells co-expressed TIGIT and CD226, suggesting a possible dual signaling mechanism in which these cells receive co-stimulatory signals from CD226 and inhibitory signals from TIGIT. Notably, TIGIT expression predominated over CD226 on CD8⁺ T cells from IDH1-wild-type tumors, whereas CD226 expression exceeded TIGIT in IDH1-mutant tumors (Fig. 7j). Viral infection induced modest shifts in cell composition, with a nominal increase in CD226⁺ single-positive cells and a decrease in TIGIT⁺ single-positive CD8⁺ T cells in both models, particularly in IDH1-wild-type (Fig. 7j). These findings highlight a potential role for the CD155/TIGIT/CD226 axis in modulating glioma immunity and suggest that TIGIT blockade may represent a promising therapeutic strategy to enhance antitumor efficacy.

Intratumoral oHSV-1 rQNestin34.5 v.2 and anti-TIGIT blockade increase survival and establish immunological memory

Given the changes that we observed in CD226 and TIGIT expression, we tested combinations of rQNestin34.5 v.2 with anti-PD1 and anti-TIGIT antibodies in mice bearing orthotopic IDH1-mutant or IDH1-wild-type gliomas. Mice received single intratumoral doses of rQNestin34.5 v.2 (1.5 × 10⁶ pfu) and/or checkpoint inhibitors as outlined in Fig. 8a, e. For anti-PD-1, clone 332.8H3 was selected due to its moderate efficacy as monotherapy, providing a window for further improvement in survival and the potential to enhance survival when combined with TIGIT monoclonal antibodies, as previously reported⁴³. To block TIGIT, antibody clone 1B4 was chosen for its selective inhibition²³. Both checkpoint inhibitors were delivered systemically, with doses given 2 to 3 days apart for a total of five treatments. The combination therapies were well tolerated, with no acute systemic adverse effects or neurological toxicity observed.

Fig. 8. oHSV-1 rQNestin34.5 v.2 in combination with anti-TIGIT checkpoint blockade enhances therapeutic efficacy in IDH1-R132H mutated glioma.

a C57BL/6 mice were intracranially implanted with 40,000 C3-mIDH1 cells and treated with rQNestin34.5 v.2 (1.5 × 10⁶ pfu in 10 µL total volume intratumorally) or HBSS (vehicle) on day 6, followed by five doses of anti-TIGIT (200 µg/mouse intraperitoneally) administered every 2-3 days, as outlined in the treatment scheme. Created in BioRender. Panagioti, E. (2026) https://BioRender.com/jvf11tp. b Kaplan-Meier survival curve showing the survival response of C3-mIDH1 tumor-bearing mice for each group according to treatments (n = 7-8 mice/ group). c Long-term survivors (160 days tumor-free) from (b) were rechallenged with autologous C3-mIDH1 cells contralateral to the primary injection site, and survival was monitored. d Long-term survivors from the rQNestin34.5 v.2 + anti-TIGIT group (b) were rechallenged with heterologous B16F10 melanoma cells, and survival was monitored. Naïve mice injected with B16F10 cells served as controls. e C57BL/6 mice were intracranially implanted with 40,000 C54-wt-IDH1 tumor cells and treated as described in the treatment scheme. Created in BioRender. Panagioti, E. (2026) https://BioRender.com/5to2b6u. f Kaplan-Meier survival curve of C54-wt-IDH1 tumor-bearing mice (n = 6 mice/ group) for each group according to treatments. g C57BL/6 mice were intracranially implanted with 50,000 CT-2A-Nectin-1 cells (wt-IDH1) and treated as described in the treatment scheme. Created in BioRender. Panagioti, E. (2026) https://BioRender.com/sldh7k1. h Kaplan-Meier survival curves of CT-2A-Nectin-1 glioma-bearing mice treated as described above (n = 8 mice/ group) for each group according to treatments. i Long-term survivors (day 100 tumor-free) from the rQNestin34.5 v.2 + anti-TIGIT group from (h) were rechallenged with autologous CT-2A-Nectin-1 tumor cells contralateral to the primary injection site, and survival was monitored. Naïve mice injected with CT-2A-Nectin-1 tumor cells served as controls. Kaplan-Meier survival curves and log-rank tests were used to determine median survival times and compare survival among groups. Statistical significance is indicated as follows: *, P < 0.05; **, P < 0.01; P > 0.05, not significant. Survival analyses shown in (a–d) were independently reproduced twice with similar outcomes.

Mice treated with rQNestin34.5 v.2 and anti-TIGIT demonstrated a significant survival benefit compared to vehicle control mice (p value = 0.0391). Notably, 50% of mice bearing C3-IDH1-mutant tumors were long-term survivors for at least 160 days post-tumor implantation (Fig. 8b). These results suggest a synergistic effect of the combination therapy, given the more modest efficacy of rQNestin34.5 v.2 or anti-TIGIT monotherapies. To assess the durability of the observed protection, long-term survivors (day >150) were rechallenged with autologous glioma cells injected contralaterally to the primary tumor site and monitored for survival. Mice that survived following rQNestin34.5 v.2 monotherapy or the rQNestin34.5 v.2 + anti-TIGIT combination was protected against rechallenge, indicating that these treatments induced durable, long-term immunity (Fig. 8c). Interestingly, among survivors of anti-TIGIT monotherapy, only half were protected against tumor rechallenge, suggesting that this treatment induces incomplete protective memory. To assess the specificity of the immune response, long-term survivors from the rQNestin34.5 v.2 + anti-TIGIT group were challenged with B16F10 melanoma cells, and all developed tumors, confirming that the protective response is tumor-specific (Fig. 8d).

For mice implanted with IDH1-wild-type tumors, the combination of rQNestin34.5 v.2 and anti-TIGIT demonstrated lower efficacy compared to the IDH1-mutant model, with only 2 of 18 mice surviving long term across combination and monotherapy arms. This suggests that differences in intrinsic cellular functions and/or responses to rQNestin34.5 v.2 in IDH1-mutant cells that we previously observed likely contribute to the difference in survival outcomes in vivo (Fig. 8f). The combination treatment also showed efficacy in CT-2A-Nectin-1 gliomas, with 25% of mice achieving long-term survival and developing protective immune memory (Fig. 8g–i). These results further support the therapeutic potential of combining rQNestin34.5 v.2 with immune checkpoint blockade to elicit durable antitumor immunity.

Anti-PD-1 monotherapy outperformed anti-TIGIT monotherapy (50% vs 25% survival) but did not significantly enhance the efficacy of rQNestin34.5 v.2 (five of eight survivors with combination therapy group versus 4 of 8 with PD-1 monotherapy group) (Supplementary Fig. 8a, b). Treated mice that became long-term survivors (day >200) were protected upon rechallenge with the same tumor model, regardless of the treatment received, indicating establishment of lasting tumor-specific immunity (Supplementary Fig. 8c). We next evaluated the combination of anti-TIGIT and anti-PD-1, with or without rQNestin34.5 v.2, following the treatment schedule shown in Supplementary Fig. 8d. No synergistic or additive survival benefit was observed with the triple combination. Furthermore, co-administration of anti-TIGIT with anti-PD-1 did not enhanced survival beyond that achieved with anti-PD-1 monotherapy (Supplementary Fig. 8d), indicating limited cooperation between these agents under the treatment schedule assessed.

Discussion

High-grade astrocytomas pose formidable clinical challenges due to their aggressive progression and resistance to standard treatments. Recent advances in oncolytic virotherapy, particularly with engineered oHSV-1 strains, have shown promise in selectively targeting tumor cells and stimulating anti-tumor immune responses³⁴. However, like all therapeutic modalities, the identification of biomarkers to refine patient selection and optimize treatment application is crucial. In this study, we investigated the impact of the IDH1-R132H mutation in response to oHSV-1 rQNestin34.5 v.2 virotherapy. Our results demonstrate that the IDH1-R132H mutation in glioma cells influences HSV-1 virus entry, innate and adaptive immune activation, and therapeutic outcomes, identifying it as a potential biomarker for the stratification and personalization of rQNestin34.5 v.2 therapy in high-grade diffuse gliomas. Additionally, we demonstrate a synergistic therapeutic effect of combining rQNestin34.5 v.2 with TIGIT immune checkpoint blockade in IDH1- mutant gliomas.

A key finding of this study is the increased susceptibility of IDH1-R132H-mutant gliomas to oHSV-1 virotherapy. In contrast, infection of IDH1-wild-type cells was limited, potentially dampening downstream antiviral and immune gene induction and confounding RNA-seq analyses. These results align with prior reports that IDH1-R132H-driven epigenetic reprogramming, including DNA hypermethylation, suppresses type I interferon responses and promotes a more permissive state for viral replication, as demonstrated with vesicular stomatitis virus (VSV-Δ51)^11,13. Here, we further delineated disruption of type I IFN signaling and interferon-stimulated genes (e.g., RSAD2) and observed a notable discrepancy in IFNAR-1, IFNAR-2, and IRF3 expression between in vitro glioma cells and patient tumors. IDH1-mutant glioma cells cultured in-vitro exhibited uniformly low IFNAR-1 protein levels with minimal variability across lines, but increased IFNAR2 and IRF3 expression. In contrast, bulk RNA-seq data from tumor biopsies showed the opposite pattern. This difference likely reflects tissue composition, as biopsy-derived RNA represents the TME, whereas in-vitro analyses assess tumor cells in isolation. IFNAR1/2, IRF3, and STAT1 signaling may be upregulated within the TME, contributing to immune evasion. Consistent with this, STAT1 expression is elevated in IDH-wild-type compared to IDH-mutant gliomas and associates with poor prognosis^44–46, while STAT1 inactivation promotes glioma stem cell survival by enabling escape from type I IFN-mediated suppression⁴⁷. IFNAR-1/2, IRF3 and STAT1 are broadly expressed in the brain-resident cells and regulate inflammatory responses^45,48–50.

Our study further identifies molecular and metabolic alterations that promote a virus-permissive TME in IDH1-mutant gliomas. Notably, we observed upregulation of Nectin-1, the primary HSV-1 entry receptor, with the highest expression detected in IDH-mutant grade 3 astrocytoma, suggesting enhanced viral entry in this subgroup. Both grade 3 and grade 4 IDH-mutant diffuse gliomas exhibited reduced IL-6-JAK-STAT3 and IL-2-STAT5 signaling, as well as diminished inflammatory response and hypoxia signatures. Hypoxia has been shown to downregulate viral entry receptors, suppress viral replication, and induce interferon-stimulated genes that restrict viral spread^51–54. Likewise, IFN-I^55,56, IL-6-JAK-STAT3⁵⁷, and IL-2-STAT5^58,59 pathways regulate cell proliferation, survival, and induction of effective antiviral responses. Consistent with a more permissive state, IDH1-mutant tumors displayed increased apoptosis and an enhanced oxidative stress response following oHSV-1 rQNestin34.5 v.2 virotherapy.

The IDH1-R132H mutation altered immune activity both within the TME and systemically. IDH1-mutant tumors exhibited increased secretion of pro-inflammatory cytokines, and treatment with the rQNestin34.5 v.2 instigated an immunogenic cell death cascade that enhanced tumor immunogenicity. When combined with anti-TIGIT immune checkpoint therapy, rQNestin34.5 v.2 further amplified immune activation, leading to improved therapeutic efficacy. However, while this combination therapy was effective in IDH1-mutant gliomas, its efficacy was diminished in IDH1-wild-type tumors, emphasizing the necessity for personalized treatment strategies. Genetic alterations in cancer cells can sculpt the TME and modulate immune responses. For instance, ATRX-deficient gliomas are associated with enhanced T-cell infiltration, increased cytokine production, and improved survival^14,60. Co-occurring mutations in oncogenes and tumor suppressors, including PTEN, NF1, and NOTCH1, can further influence both anti-tumor and anti-viral immune responses^14,61. Integrating tumor transcriptomic and mutational profiles, alongside functional validation of their impact on the TME, may identify patients most likely to benefit from oncolytic virotherapy and immune checkpoint blockade⁶².

In addition to cancer cell genetic mutations, the timing and selection of immune checkpoint inhibitors are critical determinants of successful oncolytic virotherapy. While our study demonstrated synergy with anti-TIGIT, combining rQNestin34.5 v.2 with anti-PD-1 did not yield comparable benefits. This lack of synergy may stem from factors such as antibody dosing, treatment timing, or the specific anti-PD-1 clone employed. Notably, earlier administration of anti-PD-1 (clone 332.8H3) proved more effective than delayed treatment, indicating that optimizing treatment schedules could enhance the efficacy of PD-1 blockade when combined with oncolytic virotherapy. The therapeutic potential of TIGIT blockade in IDH1-mutant gliomas was particularly compelling in our study. As an inhibitory receptor expressed on immune cells, including CD8⁺ T cells, TIGIT plays a central role in immune evasion within the TME. Blocking TIGIT, in combination with rQNestin34.5 v.2 virotherapy, enhanced effector T cell activation and proliferation, leading to a stronger antitumor immune response. In contrast, dual PD-1/TIGIT blockade did not provide synergistic or additive benefits in our IDH1-mutant models, either alone or with rQNestin34.5 v.2. Although preclinical studies suggest promise for dual PD-1/TIGIT inhibition across solid tumors^{25–27,63,64}, clinical results remain mixed, emphasizing the need for predictive biomarkers and patient stratification^65–70. Our model may serve to investigate mechanisms of resistance to combination anti-PD-1/anti-TIGIT therapy.

The widespread prevalence of HSV-1 accentuates the importance of accounting for pre-existing antiviral immunity when developing oncolytic virotherapy strategies, as it may significantly influence therapeutic efficacy. While virus-specific immunity has the potential to neutralize oncolytic virus vectors, it may also enhance antitumor immunity by engaging pre-existing immune memory^71,72. Indeed, HSV-1 seropositivity has been associated with prolonged OS in IDH-wild-type glioma patients treated with clinical-grade rQNestin34.5 v.2 (CAN-3110)³⁴. Perhaps more interesting, the most durable clinical responses to CAN-3110 in IDH-mutant gliomas were observed in HSV-1 seropositive patients, implicating a role for HSV-1-specific memory T cells in potentiating antitumor immune responses⁷³. These findings suggest that antiviral immunity may promote immune cell infiltration, foster a pro-inflammatory TME, and support repeat-dosing of HSV-1-based oncolytic viruses. Although not directly assessed here, establishing the impact of baseline HSV-1 serostatus will be critical for patient selection and optimizing the efficacy of immunovirotherapy.

In conclusion, this study identifies the IDH1-R132H mutation as a key biomarker for oHSV-1 therapy in high-grade astrocytoma. Enhanced viral susceptibility and immune infiltration in IDH-mutant tumors support the potential of combining rQNestin34.5 v.2 with TIGIT blockade. These findings support clinical evaluation of oncolytic virotherapy strategies guided by tumor-specific genetic alterations.

Methods

Cell lines

IDH1 wild-type U-87 MG (HTB-14IG) and IDH1 mutant U-87 isogenic (HTB-14IG) adult human glioblastoma cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Following thaw, these cell lines were cultured as monolayers in Eagle’s Minimum Essential Medium (ATCC, catalog 30-2003) supplemented with 10% heat-inactivated FBS (Thermo Fisher Scientific, catalog A5670801) and 1% Penicillin-Streptomycin (Gibco, Fisher Scientific, catalog 15-140-122). Cells were maintained at 37 °C in a humidified incubator with 5% CO2 to ensure optimal growth conditions. For passaging, Accutase cell dissociation reagent (BioLegend, catalog 423201) was used to detach adhered cells from the flask surface.

IDH1 wild-type MGG8 and SJGBM2 human glioma cell lines, along with their IDH1-R132H mutant counterparts and LC1035, SF10602, and MGG119 patient-derived IDH1-mutant glioma cells, were generously provided by Dr. Maria Castro’s laboratory at the University of Michigan Medical School (Ann Arbor, MI). MGG8 was derived from an IDH-wildtype glioblastoma with CDKN2A/B homozygous deletion⁷⁴. SJGBM2 was derived from an IDH-wildtype glioblastoma from a pediatric patient⁷⁵. IDH1 mutant MGG8 and SJGBM2 cell lines were generated as previously described³⁶. Briefly, 1.5 × 10⁵ cells per well were seeded in 6-well plates and transfected 24 h later with the pCMV-IDH1-R132H-Entry plasmid (Origene, catalog RC400096) using jetPRIME (Polyplus, catalog 114-07). Transfected cells were selected with G418 (Geneticin; Gibco, catalog 10131-035) at 1000 µg/mL for MGG8 and 800 µg/mL for SJGBM2. After 15 days, resistant colonies were isolated, expanded, and validated for IDH1-R132H expression by Western blotting. These cell lines were cultured as monolayers in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, Thermo Fisher Scientific, catalog 12430062), supplemented with 20% heat-inactivated FBS, 1% antibiotic-antimycotic streptomycin amphotericin B penicillin (Gibco, Thermo Fisher Scientific, catalog 15240062), 1% L-glutamine (Gibco, catalog 25030081), 1% non-essential amino acids (Gibco, catalog 11140050), 1% sodium pyruvate (Sigma-Aldrich, catalog S8636), and 0.2% Normocin (Invivogen, catalog ant-nr-2) under the same humidified conditions described above. G418 Geneticin (Thermo Fisher Scientific, catalog 10131035) was added to the IDH1 mutant MGG8 cell line at a concentration of 1000 µg/mL for selection. The SJGBM2-IDH1 wild-type and -mutant cell lines were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM, Gibco, Thermo Fisher Scientific, catalog 12440061), supplemented with 20% heat-inactivated FBS, 1% antibiotic-antimycotic streptomycin, amphotericin B, penicillin, and 0.2% Normocin. For selection, SJGBM2-IDH1-mutant cells were treated with Puromycin Dihydrochloride (InvivoGen, ThermoFisher Scientific, catalog CAS 58-58-2) at a concentration of 0.2 µg/mL. Cells were passaged twice weekly using Accutase cell dissociation reagent.

LC1035, SF10602, and MGG119 patient derived IDH1-mutant glioma cells were were cultured as neurospheres in Neurobasal Medium (Gibco, Thermo Fisher Scientific, catalog 21103-049), supplemented supplemented with 1% Antibiotic Antimycotic (100X) Streptomycin Amphotericin B Penicillin (Gibco, Thermo Fisher Scientific, catalog 15240062), 1x B-27 with Vitamin A (Gibco, Thermo Fisher Scientific, catalog 17504044), 1x N-2 (Gibco, Thermo Fisher Scientific, catalog 17502048), 0.2% Normocin (Invivogen, catalog ant-nr-2), 1% L-glutamine (Gibco, catalog 25030081), 1% sodium pyruvate (Sigma-Aldrich, catalog S8636), 20 ng/mL hFGF (Peprotech, catalog AF-100-18B), 20 ng/mL hEGF (Peprotech, catalog AF-100-15) and 20 ng/mL PDGFα (Peprotech, catalog AF-100-13A) under the same humidified conditions described above.

CT-2A-Nectin-1 orthotopic murine glioma cell line was obtained from in-house frozen stocks, and GL261 orthotopic murine glioma cell line was obtained from the NCI-Frederick Cancer Research Tumor Repository (Frederick, MD). Both cell lines, which we previously showed to be IDH-wildtype⁷⁶, were cultured as monolayers in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated FBS and 1% Penicillin-Streptomycin, under the humidity and temperature conditions described above. Mouse C54-IDH1-wild-type neurospheres and mouse C3-IDH1-mutant neurospheres³⁶ were generated and generously provided by Dr. Maria Castro’s laboratory. The Sleeping Beauty (SB) Transposase System was used to generate genetically engineered, immunocompetent IDH1-wild-type and IDH1-mutant mouse glioma models, which harbor ATRX and TP53 loss⁷⁷. The C54-IDH1-wild-type glioma model includes genetic alterations in NRAS^G12V, Atrx, and TP53 knockdown, while the C3-IDH1-mutant glioma model includes NRAS^G12V, Atrx, and TP53 knockdown, along with the addition of IDH1-R132H. C54-IDH1-wild-type neurospheres, derived from SB glioma-bearing female mice, and C3-IDH1-mutant neurospheres, derived from SB glioma-bearing male mice, were used for the preclinical in vivo experiments in this study. C54-IDH1-wild-type and C3-IDH1-mutant neurospheres were cultured in serum-free DMEM-F12 media (Gibco, Thermo Fisher Scientific, catalog 11320033) supplemented with 1% Penicillin-Streptomycin, 1x B-27 with Vitamin A (Gibco, Thermo Fisher Scientific, catalog 17504044), 1x N-2 (Gibco, Thermo Fisher Scientific, catalog 17502048), 100 µg/mL Normocin, 20 ng/mL hFGF (Peprotech, AF-100-18B), and 20 ng/mL hEGF (Peprotech, catalog AF-100-15). Neurospheres were passaged by dissociation with Accutase detachment reagent. Cells were cultured in suspension in ultra-low attachment cell culture flasks (Corning, catalog 4616) and maintained in a humidified incubator at 37 °C with 95% air and 5% CO2, passaged every 2-4 days. All cell lines were tested for human and mouse pathogen contamination upon acquisition using the Mouse and Human Basic Clear PCR panels from Charles River Laboratories. Early-passage cells were cryopreserved using serum-free cell freezing medium (Bambanker, catalog NC2960954) and stored at −80 °C. Additionally, they were routinely tested for mycoplasma contamination to ensure their validity.

Engineered cell lines: transduction and expression of human nectin-1 and IDH1-R132H

oHSV-1 rQNestin34.5 v.2 cannot enter GL261 murine glioma cells. To overcome this limitation, GL261 cells were transduced to express human Nectin-1, a cellular protein that mediates HSV-1 entry into human cells. Briefly, low-passage GL261 cells (8 ×10⁴ cells) were plated in a 6-well tissue culture-treated plate (Corning, catalog 353224) 24 h before infection with Nectin-1 Lentifect Purified Lentiviral Particles (GeneCopoeia, catalog LPP-Q0038-Lv115-050). Cells were infected at a multiplicity of infection (MOI) of 5 in the presence of 0.5 µg/mL Polybrene (MilliporeSigma, catalog TR-1003) and incubated at 37 °C with 5% CO2. A mock non-infected well using PBS- (CORNING, catalog MT21031CM) treated cells served as a control. Cell culture media was replaced with 2 mL of fresh media 24 h post-infection, and cells were incubated for an additional 3 days under the same conditions. For positive selection of transduced cells, 50 µg/mL Hygromycin B (Invitrogen, Thermo Fisher Scientific, catalog 10687010) was used. Stably transduced cells were selected and confirmed by flow cytometry analysis (LSRFortessa, BD Biosciences) using an anti-human Nectin-1 antibody (1:200 dilution, BioLegend, catalog 340404). GL261-Nectin 1 and parental GL261 cells were transfected to express IDH1-R132H. Cells were seeded in a 6-well plate (2 ×10⁵ each per well), and after 24 h they were transfected with p-CMV-IDH1- R132H-Entry plasmid (Origene, catalog RC400096) using jetPRIME transfection system (Polyplus, catalog 114-07) according to the manufacturer’s instructions. One day after transfection, the media was replaced with selection media containing G418 Geneticin at a concentration of 800 µg/mL. On day 15 of selection, individual cell colonies were taken from the well using autoclaved filter paper (Whatman, catalog 1001-185) squares (2×2 mm approx.) previously embedded in HyQTase Cell Detachment Solution (GE Heathcare Life Science, HyClone, catalog SV30030.01) and placed in a well of 24-well plate with the appropriate cell culture media. After 24 h the medium was replaced with selection medium. Each well corresponded to an isolated colony that was expanded for in-vitro experiments. IDH1-R132H protein expression was confirmed by immunofluorescence (IF) using the IDH1-R132H mutation-specific antibody, clone H09 (Dianova, catalog DIA-H09, 1:300 dilution).

Viruses

Preclinical lots of the oncolytic herpes simplex viruses (oHSVs), CAN-3110 (rQNestin34.5 v.2)⁷⁸ and the Green Fluorescence Protein (GFP)-expressing variant rQNestin34.5 v.1-GFP (rQNestin34.5 v.1)⁷⁸, were generated and provided by the laboratory of Dr. William Goins at the University of Pittsburgh, stored at −80 °C, and kept on wet ice during experiments. The stock titers were 1.2 × 10⁷ PFU/ml for both rQNestin34.5 v.2 and rQNestin34.5 v.1. For in-vivo experiments, a dose of 1.5 × 10⁶ PFU of virus was diluted in Hanks’ Balanced Salt Solution (HBSS from Gibco, Life Technologies, catalog 1300251) to a total volume of 10 μL, and administered intratumorally. Preclinical toxicology studies in glioma xenograft and naïve murine brain models showed that this dose did not result in lethality and is safe³². Virus injection was performed according to the same coordinates used for intracranial tumor implantation. HBSS was used as the vehicle in the control experimental groups. For in-vitro experiments, treatment doses were calculated based on the multiplicity of infection (MOI) appropriate for the experimental conditions and cell type.

Cytotoxicity assay

Cell cytotoxicity kinetics following oHSV-1 rQNestin34.5 v.2 infection in MGG8-IDH1-mutant, MGG8-IDH1-wild-type, SJGBM-IDH1-mutant, SJGBM-IDH1-wild-type, C3-IDH1-mutant, C54-IDH1-wild-type, and CT-2A-Nectin-1 murine glioma cells were measured using the CellTiter-Glo 3D Assay (Promega, catalog G9681). Briefly, C3-IDH1-mutant, C54-IDH1-wild-type cells were seeded onto 96-well flat-bottom ultra-low attachment plates (Corning, catalog 3447) at a density of 3000 cells/well and immediately infected with rQNestin34.5 v.2 at MOIs of 2, 1, and 0.1. Non-infected wells served as controls. The remaining glioma cell lines were seeded onto 96-well flat-bottom TC-treated plates (Corning, catalog 3598) at a density of 3000 cells/well and allowed to attach for 24 h before infection. Cell viability was assessed at 0, 12, 24-, 48-, 72-, and 96-hours post-infection using the CellTiter-Glo 3D Assay Kit, following the manufacturer’s instructions. Luminescent signals were measured using a plate reader (Polar Star Omega). Values proportional to the luminescent signal produced by each glioma cell line were plotted in Excel. The average luminescence of the non-infected group was used as the vehicle control. The percentage of cell viability was determined by expressing each MOI replicate value as a percentage of the vehicle control. Quintuplicate or sextuplicate samples for each MOI were plotted as a percentage of the vehicle control in Prism 9 to generate graphs.

Cellular ROS and apoptosis detection assays

To assess cellular ROS levels in IDH1 mutant and wild-type glioma cells in-vitro, 1 × 10⁵ cells were incubated with CellROX Green Reagent (Invitrogen, catalog C10444) at a concentration of 1000 nM per well for 45 min at room temperature in the dark. ROS production was then measured by flow cytometry using the GFP channel. ROS levels were assessed in cells infected with rQNestin34.5 v.2 or vehicle (HBSS) at the indicated multiplicities of infection (MOI: 3 for GL261-N1 and GL261-N1-R132H; and MOI: 1 for all other cell lines). Data were collected 24 hours post-infection. Apoptosis was assessed using Annexin V-Brilliant Violet APC / 7-amino-actinomycin D (7AAD) staining (BioLegend, catalog 640930) according to the manufacturer’s instructions. Briefly, SJGBM2 IDH1 wild-type and mutant cells were treated with either vehicle (HBSS) or rQNestin34.5 v.1 (MOI: 1) were harvested 48 h post-infection, washed with cold PBS, and resuspended in binding buffer. Cells were then incubated with Annexin V-APC and 7AAD for 20 min at room temperature in the dark. Samples were analyzed by flow cytometry within one hour.

Assessment of immunogenic cell death

The ability of oHSV-1 rQNestin34.5 v.2 to induce immunogenic cell death was examined in C3-IDH1-mutant, C54-IDH1-wild-type, and CT-2A-Nectin-1 murine glioma cells. Extracellular release of HMGB1 and cell surface export of calreticulin (CALR) by infected glioma cells were assessed. To measure Calreticulin expression, C3-IDH1-mutant, C54-IDH1-wild-type, and CT-2A-Nectin-1 glioma cells were infected with rQNestin34.5 v.1-GFP at MOIs: 0.01, 0.1, and 1. At 24- and 48- hours post-infection, cells were collected and stained with a Calreticulin antibody (Abcam, catalog GR3373966-1) and 7-AAD viability dye (Thermo Fisher Scientific, catalog A1310). Samples were analyzed using a BD FACSymphony A5 Cell Analyzer (BD Biosciences), and data were processed with FlowJo software. For HMGB1 measurement, cell-free supernatant was collected from triplicate samples of C3-mIDH1, C54-wt-IDH1, and CT-2A-Nectin-1 cells at 72 h post-rQNestin34.5 v.2 infection, following quick centrifugation at 2500xg for 5 min and storage at −20 °C. HMGB1 release was quantified using the HMGB1 ELISA kit (IBL International GmbH, catalog 30164033) according to the manufacturer’s instructions. Statistical analysis was performed using Prism 9.

Chemokine and cytokine detection

The release of proinflammatory cytokines/chemokines was examined in fresh, cell-free culture supernatant samples from C3-IDH1-mutant, C54-IDH1-wild-type, and CT-2A-Nectin-1 glioma cells, at 48 h following in-vitro infection with rQNestin34.5 v.2 at MOIs of 0.01, 0.1, and 1. Glioma cells stimulated with 250 IU/mL mouse IFN-γ (PeproTech, catalog 315-05) served as controls. Cytokine quantification was performed using the BioLegend Mouse LEGENDplex bead-based immunoassay: the anti-virus response panel (BioLegend, catalog 740621) to measure 13 soluble analytes (CXCL1, TNF, MCP-1, IL-12p70, RANTES, IL-1β, CXCL10, GM-CSF, IL-10, IFN-β, IFN-α, IFN-γ, and IL-6), and the proinflammatory chemokine panel (BioLegend, catalog 740007) to quantify 13 soluble analytes (MCP-1, CCL5, CXCL10, CCL11, CCL17, CCL3, CCL4, CXCL9, CCL20, CXCL5, CXCL1, CXCL13, and CCL22). Results were analyzed with LEGENDplex Data Analysis Software (BioLegend). Quantitative determination of mouse GDF-15 levels in serum was performed using the R&D Systems ELISA kit (catalog MGD150) according to the manufacturer’s instructions. Additionally, the LegendMax Mouse IFN-γ ELISA Kit (BioLegend, catalog 430807), LegendMax Mouse IFN-β ELISA Kit (BioLegend, catalog 437), and the Human IFN-β ELISA Kit (BioLegend, catalog 449507) were used following the manufacturers’ protocols. Results were acquired using the BD FACSymphony A5 Cell Analyzer (BD Biosciences). For the measurement of cytokine/chemokine panels in the mice serum at the indicated time points mentioned in the figure legends, fresh blood was collected from individual mice using Microvette® CB 300 Serum capillary collection tubes (SARSTEDT, catalog 16.440.100) and allowed to clot for 2 hours at 4 °C. After centrifugation at 2000 rpm for 15 min, the serum was collected and stored at −20 °C for subsequent analysis.

Bulk/RNA-seq

5 × 10⁵ MGGB wild-type and mutant IDH1 cells were plated in a 6-well plate and, 24 h later, infected with rQNestin34.5 v.2 at an MOI of 1. Four hours after infection, the medium containing the virus was removed and replaced with fresh culture medium. Sixteen hours post-infection, the monolayer cells were washed, lysed, and homogenized. RNA was then extracted using the PureLink RNA Mini Kit (Invitrogen, catalog 12183018 A), followed by one-column PureLink DNase treatment. RNA yield and quality were assessed using a Nanodrop spectrophotometer (Fisher Scientific, AccuSkan FC) and stored at −80 °C. RNA library preparation, sequencing, and analysis were conducted at Azenta Life Sciences (South Plainfield, NJ, USA) as follows: RNA samples were quantified using a Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, USA), and RNA integrity was checked using an Agilent TapeStation 4200 (Agilent Technologies, Palo Alto, CA, USA). ERCC RNA Spike-In Mix (ThermoFisher Scientific, catalog 4456740) was added to normalized total RNA before library preparation following the manufacturer’s protocol. RNA sequencing libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina using the manufacturer’s instructions (NEB, Ipswich, MA, USA). Briefly, mRNAs were initially enriched with Oligod(T) beads. Enriched mRNAs were fragmented for 15 minutes at 94 °C. First-strand and second-strand cDNA were subsequently synthesized. cDNA fragments were end-repaired and adenylated at 3’ends, and universal adapters were ligated to cDNA fragments, followed by index addition and library enrichment by PCR with limited cycles. The sequencing library was validated on the Agilent TapeStation (Agilent Technologies, Palo Alto, CA, USA), and quantified by using Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA) as well as by quantitative PCR (KAPA Biosystems, Wilmington, MA, USA). The sequencing libraries were clustered on a flow cell. After clustering, the flow cell was loaded on the Illumina NovaSeq instrument according to the manufacturer’s instructions. The samples were sequenced using a 2x150bp Paired-End (PE) configuration, targeting 30 M reads/sample. Raw sequence data (.bcl files) generated by the sequencer were converted into fastq files and de-multiplexed using Illumina’s bcl2fastq 2.20 software. After investigating the quality of the raw data, sequence reads were trimmed to remove possible adapter sequences and nucleotides of poor quality. The trimmed reads were mapped to the reference genome available on ENSEMBL using the STAR aligner v.2.5.2b. Unique gene hit counts were calculated by using feature Counts from the Subread package v.1.5.2. Only unique reads that fell within exon regions were counted. Using DESeq2, a comparison of gene expression between the groups of samples was performed. The Wald test is used to generate p-values and Log2 fold changes. Genes with adjusted p-values < 0.05 and absolute log2 fold changes > 1 were called as differentially expressed genes for each comparison. A gene ontology analysis was performed on the statistically significant set of genes by implementing the software GeneSCF. The human Gene Ontology (GO) was used to annotate genes regarding biological processes. Principal Component Analysis (PCA) was performed using the “plotPCA” function within the DESeq2 R package using the 500 genes with the highest variance, which were used to generate the plot.

Mice

Female C57BL/6 J mice, 4–5 weeks old, were obtained from Envigo (NY, USA, catalog 044) and Jackson Laboratories (ME, USA, catalog 000664). Experimental and control mice were 5–6 weeks old at the start of the experiments and were housed in a BL2 barrier facility. All experimental procedures involving animals were conducted in accordance with an approved animal protocol reviewed by the Institutional Animal Care and Use Committees (IACUC) at Brigham and Women’s Hospital and Beth Israel Deaconess Medical Center. The procedures adhered to established guidelines and regulations. All mice were quarantined according to the standard operating procedures (SOP) of the testing facility. Housing and animal care practices complied with current Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) standards, as well as the requirements outlined in the Guide for the Care and Use of Laboratory Animals⁷⁹. Animals were maintained under standard housing conditions, including a 12:12 h light/dark cycle, controlled ambient temperature (20 °C–22 °C), and maintained humidity (30–70%), with ad libitum access to irradiated NTP-2000 wafer feed (Zeigler Brothers, Gardners, PA, USA) in accordance with the testing facility’s SOP.

Surgical procedure

C57BL/6 mice were prepared for intracranial injection of murine glioma cells by moderate sedation with isoflurane inhalation. Analgesics, including Carprofen (1 mg/kg) and Buprenorphine SR (0.6 mg/kg), were administered before the procedure. Throughout the surgery, the animal's vital signs and the depth of anesthesia were monitored by assessing respiratory rate, skin color, and the toe pinch reflex. Once adequate anesthesia was achieved, the mice were secured in a stereotaxic frame (Stoelting, catalog 51615U). Eye lubricant cream (Optixcare) was applied to prevent ocular dehydration. The skin over the skull was shaved (Manscaped, catalog MAN-TR3-01) and cleaned with a cotton swab (Puritan, catalog 867-WC) soaked in surgical scrub solution (Henry Schein Medical, catalog 87818-150-01), followed by an alcohol-soaked cotton swab in a circular motion. A midline sagittal incision, approximately 8–10 mm in length, was made using a sterile disposable scalpel (no. 10, Fisher Scientific, catalog 12-460-451). After identifying the bregma, a small burr hole was drilled (Dremel, catalog 8050-35) into the skull at the following stereotactic coordinates: 1 mm anterior and 2 mm lateral to the right of the bregma, for intraparenchymal injection.

A sterile Hamilton syringe with a 27-gauge needle (Hamilton, catalog CAL80000) was loaded with one of the following: 5 μL of tumor cell suspension in HBSS (day 0), 10 μL of the appropriate concentration of oHSV-1 rQNestin34.5 v.2 solution in HBSS (on days 5 or 7, according to the treatment schedule), or 5 μL of HBSS (on days 5 or 7, according to the treatment schedule). The syringe was securely fixed to the stereotactic frame, and the needle was inserted intracranially at coordinates 2 mm right lateral, 1 mm anterior, and 3 mm deep to the bregma. The needle was slowly lowered into the burr hole to puncture the dura, advancing to a final depth of 3 mm into the brain. The injection was administered at a rate of 0.5 μL every 60 s, followed by slow needle retraction at a rate of 0.5 mm every 60 s to minimize the risk of cell leakage. After the injection, the needle was carefully withdrawn, and the burr hole and skin incision were closed using 4.0 nylon sutures (Ethycon, catalog J845G). Mice were placed on small animal heating pads (K&H Pet Products) until they regained consciousness and mobility, after which they were returned to their cages. The stereotactic apparatus was cleaned with 70% alcohol between surgeries, and Hamilton syringes were cleaned with three cycles of PBS. Postoperative analgesia included Buprenorphine SR (0.6 mg/kg, subcutaneously) immediately after the animal regained consciousness, and Meloxicam (0.2 mg/kg, once daily for 3 days following surgery). Mice were monitored daily, and euthanasia was performed in a CO₂ chamber once humane endpoint criteria were met, including significant >30% weight loss, grimacing, hunched posture, or lethargy.

Antibody treatments

The following anti-PD-1 antibodies were administered: RPM1-14 (BioLegend, catalog 114116), 29 F.1A12⁸⁰ (BioXCell, catalog BE0273), and 332.8H3 (produced in-house from hybridoma at the laboratory of Dr. Gordon Freeman, DFCI, Boston, MA)⁸¹, or isotype control (polyclonal mouse IgG, BioXCell, catalog BE0093). These antibodies were administered intraperitoneally (i.p.) at 200 μg per mouse in sterile PBS every 2-3 days for a total of 5 doses. Murine anti-TIGIT antibody (clone 1B4, produced in-house at the laboratory of Dr. Vijay Kuchroo, BWH, Boston, MA) or isotype control (polyclonal mouse IgG, BioXCell, catalog BE0093) was given i.p. at 200 μg per mouse in sterile PBS every 2-3 days for 5 doses.

Immunohistochemistry (IHC) of paraffin samples

Following perfusion, mouse brains were fixed in 10% neutral buffered formalin (Thermo Scientific, catalog 5701) for 48 h at 4 °C, after which they were transferred to 70% ethanol (Fisherbrand, catalog HC13001GL). The tissues were then processed and embedded in paraffin at the BIDMC Histology Core Facility using a Leica ASP 300 paraffin tissue processor and Tissue-Tek embedding station (Leica). Sections were cut at 5 µm using a Leica rotary microtome. Hematoxylin and Eosin (H&E) staining was performed on tissue sections of each sample. Antigen retrieval and immunohistochemistry (IHC) were performed on paraffin-embedded sections using heat-induced epitope retrieval (HIER). The slides treated with IDH1-R132H and CD8a were processed for antigen retrieval using Tris-EDTA pH9 retrieval buffer (BioLegend, catalog 422704). Slides treated with Ki-67 and PD-L1 were processed for antigen retrieval using Sodium Citrate pH 6.0 retrieval buffer (Millipore Sigma, catalog C9999). Immunostaining was performed manually at the BIDMC-Immunostaining and Microscopy Core Facility (RRID: SCR_012312). Tissue sections were blocked for 60 minutes in PBS containing 2% BSA, then incubated overnight at 4 °C for about 16 h with primary antibodies: mouse monoclonal IDH1-R132H (1:300, Dianova, catalog DIA-H09), rabbit monoclonal Ki67 (1:500, Cell Signaling, catalog 9129), PD-L1 (1:50, Cell Signaling, catalog 13684), and CD8a (1:500, Cell Signaling, catalog 98941). Each antibody was optimized through repeated rounds of testing, which included evaluating different antigen retrieval conditions, diluents, and a wide range of antibody concentrations to ensure the fidelity of the staining. After primary antibody incubation, sections were incubated for 90 min with goat anti-rabbit IgG HRP polymer (Abcam, catalog ab214880) at a 1:2 dilution in 2% BSA. Detection was performed using a DAB detection kit (Vector Laboratories, catalog SK-4105). IDH1-R132H antibody slides were developed for 2 min, Ki-67 for 9 min, PD-L1 for 7 min, and CD8a for 10 min, following the manufacturer’s instructions. Stained slides were scanned using a PhenoImager HT (Akoya Biosciences) and analyzed with QuPath 0.5.1 software. Immunostaining quantification was performed by a blinded pathologist using an automated analysis program within MATLAB’s image processing toolbox. The tumor mitotic index and areas of necrosis were assessed through visual inspection by the pathologist. Representative regions of interest were selected, and multiple fields of view were captured at 20x magnification as multispectral images. Graphs were generated in GraphPad Prism 9, and statistical analysis was performed using the Wilcoxon signed-rank test.

TILs isolation

Orthotopic gliomas were established as described above. On days 3 and 8 post intratumoral oHSV rQNestin34.5 v.2 administration mice were euthanized, transcardially perfused with chilled PBS, and brains were harvested for isolation of tumor-infiltrating leukocytes (TILs) by Ficoll Paque density gradient centrifugation. To isolate TILs, the required media were first prepared. The base medium, R10, was composed of 500 mL RPMI-1640 (Thermo Fisher Scientific, catalog 11875093), 10 mL FBS (2%), 5 mL PenStrep (1%), and 5 mL HEPES (Thermo Scientific, catalog J61275.AK). For a complete medium, R10 was supplemented with 50 mL FBS (10%), 5 mL PenStrep (1%), and 5 mL HEPES. A Percoll Plus solution (Cytiva, catalog 17-5445-01) was also prepared by mixing 45 mL Percoll Plus (90%), 2.5 mL 20x PBS solution (5%), and 2.5 mL molecular grade water (5%). Additionally, Percoll fractions were prepared by mixing stock Percoll solutions with PBS or HBSS to create 30, 37, and 70% Percoll fractions. Brain/tumor tissue samples were cut into small pieces using a sterile scalpel. The tissue pieces were transferred to a GentleMACS C tube (Miltenyi Biotec, catalog 130-093-237), which contained an enzyme digestion mix composed of 2.35 mL R10 medium (10% FBS), 100 µL Enzyme D, 50 µL Enzyme R, and 12.5 µL Enzyme A (Brain tumor dissociation kit, Miltenyi Biotec, catalog 130-095-942). The tube was tightly closed and attached upside down to the sleeve of the GentleMACS dissociator (Miltenyi Biotec, catalog 130-093-235), where the dissociation program m_impTumor_02 was selected. Following the dissociation, the C tube was incubated on continuous rotation at 37 °C for 40 min. After incubation, the C tube was placed back in the dissociator, and program m_impTumor_03 was run for further tissue dissociation. The homogenized cell suspension was then transferred through a 100 µm strainer (Falcon, catalog 352360) into a 15 mL tube containing 10 mL of R10 medium. The cells were centrifuged at 400 g for 5 min at 21 °C. The pellet was visualized, and the supernatant was discarded. The pellet was resuspended in R10 medium up to 10 mL, followed by another centrifugation step. The cells were then washed with T cell isolation buffer (2% FBS) at the same centrifugation settings. For the density gradient centrifugation, the cell pellet was resuspended in a Percoll gradient. A new 15 mL tube was prepared with 4 mL of 70% Percoll (bottom layer), 4 mL of 37% Percoll (middle layer), and 4 mL of 30% Percoll (top layer). The gradient was spun at 500xg for 40 min at 21 °C with 5 acceleration and 0 brake settings. After centrifugation, the fat and debris were carefully removed from the top of the gradient using a manual pipette, and the TILs were collected from the interface between the 37 and 70% Percoll layers. The TILs were resuspended in 5x volume of R2 media and centrifuged at 400xg for 5 min at 4 °C. The pellet was washed with R2 media, followed by another centrifugation step. The supernatant was discarded, and the cells were resuspended in R10 medium. Finally, the isolated cells were counted using a hemocytometer or an automated cell counter, and then they were used for subsequent immunostaining procedures.

Single-cell suspensions were prepared from mouse spleens by first collecting the spleens in 15 mL tubes containing 3 mL of cold RPMI with 10% heat-inactivated FBS and immediately placing the tubes on ice. Under sterile conditions in a flow cabinet, the spleens were transferred onto a 70 mm strainer (Falcon, catalog 08-771-2) and gently crushed using the plunger flange of a sterile 1 mL syringe, with 6 mL of RPMI media added to facilitate cell release, resulting in a final volume of 10 mL. The suspension was centrifuged for 5 min at 1600 RPM, 4 °C, and the supernatant was discarded. The pellet was incubated with 3 mL of 1X red blood cell (RBC) lysis buffer (eBioscience, catalog 00-4333-57) for 3 min, followed by the addition of 7 mL of RPMI to stop the lysis. Cells were washed twice with RPMI, and a small aliquot was taken for counting. The final cell concentration was adjusted to 15 × 10⁶ cells/mL, with 1.5 × 10⁶ cells/100 µL used for flow cytometry staining. Peripheral blood (25 µL per mouse) was collected (Sarstedt Microvette, catalog 16.443.100) from the tail vein of mice at the indicated time points. Red blood cells (RBCs) were incubated for 40 min at 4 °C with 150 µL of lysis buffer, then washed three times with staining buffer. The cells were subsequently stained with panels of immune markers.

Multiparameter flow cytometry

For immunophenotyping of TILs, spleenocytes (1.5 × 10⁶ cells/well), peripheral blood mononuclear cells (PBMCs, 30 µL blood/well), and cultured glioma cells (3 × 10⁵ cells/well) were plated in a non-tissue culture-treated 96-well U-bottom plate (Corning, catalog 351177) and stained for a panel of immune cell markers. For in-vitro experiments with mouse glioma cell lines, the following antibodies for cell surface markers were used: Glioma cells were labeled with Viability Ghost Red APC/Cy7 (TONBO Biosciences, catalog 13-0865, 1:800) or 7-Aminoactinomycin D 7AAD (Thermo Fisher Scientific, catalog A1310, 1:800) in staining buffer (2% FBS, 0.05% sodium azide in PBS) for 15 min at room temperature. Cells were then blocked with Fc block (BD, clone 2.4G2, catalog 553142, 1:100) for 10 min at 4 °C. Following blocking, cells were stained with a mixture of fluorochrome-conjugated antibodies specific for: PD-L1 PE/Cyanine7 (BioLegend, clone 10 F.9G2, catalog 124314, 1:200), H-2Kb APC (eBioscience, clone AF6-88.5.5.3, catalog 17-5958-82, 1:200), IFNAR−1 Pe (BioLegend, clone MAR1-5A3, catalog 127311, 1:200) and TLR2 Pe (BioLegend, clone CB225, catalog 148604, 1:100), CD62L Brilliant Violet 605 (BD Biosciences, clone MEL-14, catalog 564108, 1:200), CD112 BV786 (BD Biosciences, clone 829038, catalog 748050, 1:200), CD80 BV786 (BD Biosciences, clone 16-10A1, catalog 740888, 1:200), I-A/I-E (MHC class II) FITC (BioLegend, clone M5/114.15.2, catalog 107605, 1:100), Galectin 9 PerCP/Cyanine5.5 (BioLegend, clone RG9-35, catalog 136112, 1:100), CD272 (BTLA) Pe (BioLegend, clone 8F4, catalog 134803, 1:200), CD86 PE/Cyanine7 (BioLegend, clone GL-1, catalog 105014, 1:200), CD155 BUV496 (BD Biosciences, clone TX56, catalog 749992, 1:200), and CD155 BV510 (BD Bioscience, clone TX56, catalog 748237, 1:200).

For in-vitro experiments with human glioma cell lines, the following antibodies for cell surface markers were used: PD-L1 Pe (BioLegend, clone 29E.2A3, catalog 329706, 1:150), CD155 (PVR) APC (BioLegend, clone SKIL4, catalog 337617, 1:100), CD111 (Nectin 1) Pe (BioLegend, clone R1.302, catalog 567019, 1:200), CD112 (Nectin 2) APC (BioLegend, clone TX31, catalog 337411, 1:100), CD270 (HVEM, BioLegend, clone 122, catalog 318809, 1:100), IDH1 Pe (BD Biosciences, clone RMab-03, catalog 567019, 1:200), CD44 Pe (BD Biosciences, clone G44-26, catalog 559942, 1:100), CD86 Brilliant Violet 421 (BioLegend, clone IT2.2, catalog 305425, 1:100), HLA-DR, DP, DQ (MHC Class II) FITC (BioLegend, clone Tu39, catalog 361705, 1:100), HLA-A,B,C (MHC class I) APC/Cyanine7 (BioLegend, clone W6/32, catalog 311425, 1:100), IFN-α/β R1 Pe (R&D Systems, clone 85228, catalog FAB245P, 1:100), and IFNAR2 APC (Invitrogen, clone 122, catalog MA5-4052, 1:100). For intracellular staining cells were fixed and permeabilized FOXP3 Fix/Perm buffer (BioLegend, catalog 421403) and stained for IRF3 Alexa Fluor 647 (BD Biosciences, clone SL-12.1, catalog 566347, 1:100) according to manufacturer’s instructions.

Mouse TILs, splenocytes, and PBMCs from in vivo experiments were stained with surface marker panels in staining buffer for 30 min at 4 °C. The following antibodies were used for surface staining: Zombie UV Fixable viability DAPI (BioLegend, catalog 423107, 1:1000), CD45.2 APC-Cy7 (BioLegend, clone 104, catalog 109823, 1:200), CD45 PerCP (BioLegend, clone 30-F11, catalog 103130, 1:200), TIGIT Pe (BioLegend, clone GIGD7, catalog 12-9501-80, 1:200), TIGIT PeCy7 (BioLegend, clone GIGD7, catalog 25-9501-82, 1:200), TIGIT Alexa 700 (Invitrogen, clone GIGD7, catalog 46-9501-82, 1:200), CD8a Violet Fluor 450 (BioLegend; clone 2.43; catalog 75-1886-U100; 1:100), CD8a FITC (BioLegend, clone 53-6.7, catalog 100706, 1:150), CD8a BUV805 (BD Biosciences, clone 53-6.7, catalog 612898, 1:200), B220 FITC (BioLegend; clone RA3-6B2; catalog 553087; 1:100), B220 BUV496 (BD Biosciences; clone RA3-6B2; catalog 612950; 1:200), PD1 BV605 (BioLegend, clone RMP1-30, catalog 748267, 1:200), PD1 FITC (eBioscience, clone RPMI-30, catalog 4347279, 1:200), CD3 BV711 (BioLegend, clone 17A2, catalog 100241, 1:200), CD3 BV421 (BioLegend, clone 145-2C11, catalog 562600, 1:100), CD3 APC (BioLegend, clone 145-2C11, catalog 553066, 1:200), CD3 Pe/Cy7 (BioLegend, clone 145-2C11, catalog 100320, 1:800), CD44 BV786 (BioLegend, clone IM7, catalog 563736, 1:200), CD44 APC-Cy7 (BioLegend, clone IM7, catalog 560568, 1:200), CD4 Spark Blue 515 (BioLegend, clone GK1.5, catalog 100493, 1:100), CD4 BUV563 (BD Biosciences, clone GK1.5, catalog 612923, 1:200), CD4 BV786 (BD Biosciences, clone RM4-5, catalog 563727, 1:200), NK1.1 Pe/Cy7 (eBioscience, clone PK136, catalog 25-5941-81, 1:200), CD64 FITC (BioLegend, clone X54-5/7.1, catalog 139315, 1:100), F4/80 APC-Cy7 (BioLegend, clone BM8, catalog 123118; 1:200), F4/80 Pe (TONDO biosciences, clone BM8.1, catalog 50-4801, 1:200), Tim3 PerCP/Cy5.5 (BioLegend, clone B8.2C12, catalog 134011, 1:100), Tim3 BUV395 (BD Biosciences, clone 5D12/TIM3, catalog 747620, 1: 200), KLRG1 FITC (BioLegend, clone 2F1/GKRG1, catalog 138410, 1:100), KLRG1 Brilliant Violet 711 (BioLegend, clone MAFA, 2F1-Ag, catalog 138427, 1:200), CD19 Pe-CF594 (BioLegend, clone 6D5, catalog 115554, 1:200), Gr1 A700 (BioLegend, clone RB6-8C5, catalog 108422, 1:800), CD62L BV650 (BioLegend, clone MEL-14, catalog 564108, 1:200), CX3CR1 BV785 (BioLegend, clone SA011F11, catalog 149029, 1:200), Ly6G Pe-CF594 (BioLegend, clone 1A8, catalog 562700, 1:500), Ly6G A700 (BioLegend, clone 1A8, catalog 127622, 1:1000), CD226 (DNAM-1) Pe (BioLegend, clone 10E5, catalog 128817, 1:200), Ly-6C APC (BioLegend, clone HK1.4, catalog 128015, 1:200), CD103 BUV395 (BD Biosciences, clone M290, catalog 740238, 1:100), Ly6C APC (BioLegend, clone HK1.4, catalog 128015, 1:200), CD11b BUV737 (BD Biosciences, clone M1/70, catalog 612801, 1:600), CD11b Brillian Violet 650 (BioLegend, clone M1/70, catalog 101239, 1:150), CD11b BV510 (BioLegend, clone M1/70, catalog 101245, 1:200), CD25 APC (BioLegend, clone PC61.5, catalog 108412, 1:200), CD24 BV711 (BioLegend, clone M1/69, catalog 563450, 1:200), PD-L1 BV421 (BioLegend, clone B7-H1, catalog 124315, 1:200), CD155 PerCP/Cy5.5 (BioLegend, clone TX56, catalog 131513, 1:100), CD155 (PVR) BUV496 (BD Biosciences, clone TX56, catalog 749992, 1:200), MHC class II BV605 (BioLegend, clone M5/114.15.2, catalog 107639, 1:500). After antibody staining, cells were washed twice with staining buffer and then resuspended in 100 µL of fixation buffer (BD Biosciences) for 10 min at 4 °C. Following fixation, cells were washed twice with staining buffer before acquisition at the flow cytometer. For intracellular staining cells were fixed and permeabilized with Cytofix/Cytoperm buffer (BD Biosciences, catalog 554714) and stained for Granzyme B Pe (BD Biosciences, clone GB11, catalog 561142, 1:100) or Granzyme B FITC (BioLegend, clone GB11, catalog 515403, 1:100) and Ki67 Alexa 700 (BD Biosciences, clone 16A8, catalog 562419, 1: 800), or resuspended in FOXP3 Fix/Perm buffer (BioLegend, catalog 421403) and stained for FOXP3 PE (eBioscience, clone FJK-16a, catalog 12-5773-80, 1:200) or FOXP3 APC (Invitrogen, clone FJK-16s, catalog 17-5773-82, 1:200) according to manufacturer’s instructions. Samples were resuspended in 150 μl staining buffer and acquired in a LSRFortessa (BD Biosciences), BD FACSymphony A5 Cell Analyzer (BD Biosciences), and Cytec Aurora (Cytec Biosciences). FlowJo software (Treestar Inc. ver 10.) and FCS Express (De Novo Software) were used for data analysis. The gating strategy for immunophenotyping of TILs is shown in Supplementary Fig. 9.

TCGA data analysis

List of the distribution of the most frequently mutated genes in LGG and GBM obtained from The Cancer Genome Atlas (TCGA) database⁸². Survival analyses comparing IDH1-mutated and IDH1-not mutated glioma patients were performed using clinical and transcriptomic data accessed from TCGA⁸². Differential gene expression analysis of TCGA data was performed on treatment-naïve high-grade (grade ≥3) diffuse glioma from male and female patients of diverse ethnic backgrounds, predominantly White. Tumor samples were annotated in TCGA under historical classification criteria as low-grade gliomas (LGG) or glioblastoma (GBM) and further stratified by IDH mutation status. The TCGA cohorts comprised LGG-IDH-mutant (n = 58), LGG-IDH-wild-type (n = 31), GBM-IDH-mutant (n = 12), and GBM-IDH-wild-type (n = 158). All LGG tumors (IDH-mutant and IDH-wild-type) corresponded to grade 3 astrocytoma. GBM samples were all primary tumors. Although not formally classified at the time of TCGA annotation, histopathological features that would previously have supported a diagnosis of GBM in IDH-mutant tumors (i.e., necrosis and/or microvascular proliferation) are now sufficient for classification as IDH-mutant, CNS WHO grade 4 astrocytoma under current WHO criteria. Detailed information on TCGA GBM and LGG patient cohorts, including ethnicity, age, tumor location, and histological subtypes, is provided in Supplementary Data 4. To align with current WHO classification criteria, we defined these cohorts based on tumor grade and IDH status (IDH-mutant grade 4 astrocytoma; n = 12, IDH-wild-type grade 4 astrocytoma; n = 158, IDH-mutant grade 3 astrocytoma; n = 58, and IDH-wild-type grade 3 astrocytoma; n = 31) throughout the manuscript. RNA-Seq and clinical data from TCGA for these samples were downloaded from the Genomic Data Commons portal⁸³, using R’s TCGAbiolinks (v2.30.4) package⁸⁴. Briefly, the analysis was performed using DESeq2⁸⁵, after which ClusterProfiler (v4.10.0)⁸⁶ was utilized for downstream functional investigations. Plots were generated in R using ggplot2 (v3.4.4), EnhancedVolcano version (v1.20.0), and ComplexHeatmap (v2.18.0)⁸⁷. A single-sample gene set enrichment analysis was performed on normalized expression data using R’s GSVA package (v1.50.5)⁸⁸.

Analysis of CAN-3110 trial data

Bulk RNA sequencing, patient survival, and pathology count data from Ling et al. ³⁴ were analyzed in R version 4.2.1 using the following libraries: openxlsx v4.2.5, survival v3.3-1, survminer v0.4.9, RColorBrewer v1.1-3, ggplot2 v3.5.2, ggpubr v0.4.0, rstatix v0.7.0, GSVA v1.46.0, and msigdbr v24.1.0. ssGSEA analyses were performed using the GSVA package with hallmark gene sets obtained using msigdbr. ssGSEA scores were calculated using TPM values of a filtered set of genes in which genes with TPM < 0.5 were removed, followed by removal of the lowest 10% of genes by variance. ssGSEA scores were normalized using GSVA’s ssgsea.norm default functionality.

Statistics

Statistical comparisons were performed using Prism (GraphPad Software; version 9) and R (version 3.5.0 or later). Comparisons between groups were made according to their assignments. Significance of differences between two groups was evaluated by an unpaired, 2-tailed Student's t-test. Multiple comparisons were analyzed using analysis of variance (ANOVA) models, with ad hoc pairwise comparisons adjusted using post hoc multiple comparison tests (Tukey’s or Šidák’s). Kaplan-Meier survival curves were used to assess animal survival, and comparisons were performed using the log-rank Mantel-Cox test for correction for multiple comparisons. For bulk RNA-seq data, statistical significance was set at false discovery rate (FDR) and P values less than 0.05 and not significant (NS) greater than 0.05. Final graphs and tables were formatted using Adobe Illustrator (version CC).

Graphical illustrations

The experimental design illustrations were created using BioRender. (https://biorender.com/).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization, E.P. and E.A.C.; Methodology: E.P., H.J.K., A.L.L., L.S., S.S., W.F.G., D.R., S.S., K.O.D., J.B.I., I.S.V., M.G.C., S.E.L., G.J.F., M.B.Y., V.K.K., C.H.C., and E.A.C.; acquisition of data (performed experiments, statistical analysis, biostatistics, computational analysis, generation of figures), E.P., H.J.K., A.L.L., L.S., S.S., D.R., and J.B.I.; analysis and/or interpretation of data, E.P., H.J.K., A.L.L., L.S., S.S., W.F.G., S.S., J.B.I., M.G.C., S.E.L., G.J.F., M.B.Y., V.K.K., C.H.C., and E.A.C.; writing of the original manuscript, E.P.; writing-review and/or editing of the manuscript: All authors; administrative, technical, or material support, E.P., H.J.K., A.L.L., L.S., S.S., W.F.G., K.O.D., M.G.C., G.J.F., V.K.K., C.H.C., and E.A.C.; study supervision, E.P., C.H.C., and E.A.C.

Peer review

Peer review information

Nature Communications thanks Antonio Dono, Shaun Xiaoliu Zhang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

E.A.C. discloses support for this research from NIH grants (U01NS061811, P01CA163205, R01NS110942), as well as a grant from the Alliance for Cancer Gene Therapy (Greenwich, CT). E.A.C., G.J.F., and V.K.K. also disclose support for this research from NIH grant P01CA236749. S.E.L. and C.H.C. disclose support for this research from NIH grant R01CA263324. E.P. discloses support for the publication of this work from the BIDMC 2023 Early-Stage Career Investigator Award (GRT65864).

Data availability

Processed bulk RNA-seq data from MGG8 glioma cells have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE301461. Gene expression profiles from human glioma specimens were obtained from The Cancer Genome Atlas (TCGA). The data are publicly available with no restrictions. All other data are available from the corresponding authors upon reasonable request, as they contain additional information that will be used in separate publications. Source data are provided with this paper.

Code availability

TCGA processed data and analysis scripts are available at https://github.com/lasejour/IDH1-TCGA-Analysis. RNA-seq and clinical data from the Phase I clinical trial of CAN-3110 were accessed through the Database of Genotypes and Phenotypes (dbGaP; http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs003378.v1.p1).

Competing interests

E.A.C. is named inventor on patents related to rQNestin34.5 v.2. E.A.C. is an advisor to Bionaut Labs, Seneca Therapeutics, ReIgnite and Calidi. He has equity options in Bionaut Laboratories, Seneca Therapeutics, ReIgnite, Ternalys Therapeutics. He is co-founder and on Board of Directors of Ternalys Therapeutics. He has received research support from NIH, US Department of Defense, American Brain Tumor Association, National Brain Tumor Society, Alliance for Cancer Gene Therapy, Neurosurgical Research Education Foundation, Advantagene, NewLink Genetics, and Amgen. The Patents related to oHSV and CAN-3110 are under the possession of Brigham and Women’s Hospital (BWH) with E.A.C. named as inventor. These patents have been licensed to Candel Therapeutics, Inc. Present and future milestone license fees and future royalty fees are distributed to BWH from Candel. I.S.V. reports grants from NCI, National Heart, Lung, and Blood Institute, National Institute of Diabetes and Digestive and Kidney Diseases, Harvard Stem Cell Institute and consulting for Mosaic, AlphaSights, Chronicle Medical Software Inc., and Guidepoint Global outside of the submitted work. G.J.F. has patents/pending royalties on the PD-L1/PD-1 pathway from Roche, Merck MSD, AstraZeneca, Bristol-Myers-Squibb, Merck KGA, Boehringer-Ingelheim, Dako, Leica, Mayo Clinic, Eli Lilly, Coherus BioSciences, and Novartis. G.J.F. has served on advisory boards for iTeos, NextPoint, IgM, GV20, IOME, Bioentre, Santa Ana Bio, Simcere of America, and Geode. G.J.F. has equity in Nextpoint, iTeos, IgM, Invaria, GV20, Bioentre, and Geode. V.K.K. has an ownership interest and is a member of the SAB for Tizona and Trishula Therapeutics. V.K.K. is an inventor on patents related to Th17 cells and immunometabolism. V.K.K. has an ownership interest in Tizona Therapeutics, Trishula, Celsius Therapeutics, Bicara Therapeutics, Larkspur Therapeutics and Werewolf Therapeutics. V.K.K. has financial interests in Biocon Biologic, Compass, Elpiscience Biopharmaceutical, Equilium, PerkinElmer and Syngene. V.K.K. is a member of SABs for Cell Signaling Technology, Elpiscience Biopharmaceutical, Larkspur, Tizona Therapeutics, Tr1X and Werewolf. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73974-5.