MicroRNA-138 regulation promotes the efficacy of oncolytic herpes virus G47Δ for malignant brain tumor therapy

Abstract

G47Δ, a third-generation oncolytic herpes simplex virus type 1, is a promising therapeutic agent for malignant glioma. We investigated whether inhibition of host microRNA-138 (miR-138), a regulator of the viral immediate-early protein ICP0, could enhance its efficacy. In vitro, miR-138 suppression increased viral protein expression, progeny virus production, and cytotoxicity in human glioblastoma and murine Neuro2a cells, whereas overexpression suppressed these effects. Neuro2a cells stably silenced for miR-138 showed tumor regression in 9 of 13 syngeneic mice, but not in athymic mice or following CD8⁺ T cell depletion, indicating dependence on adaptive immunity. Mechanistically, miR-138 directly targeted signal transducer and activator of transcription 1 (STAT1) mRNA: Its inhibition increased STAT1 and enhanced MHC class I expression. In bilateral subcutaneous tumors, miR-138 inhibition suppressed growth on the treated side, but contralateral tumor suppression required combination with G47Δ, suggesting systemic antitumor immunity involvement. In vivo electroporation-mediated miR-138 inhibition in subcutaneous tumors elevated STAT1 and suppressed growth. G47Δ combined with in vivo miR-138 inhibition significantly augmented efficacy compared with G47Δ alone. These findings identify miR-138 as both an antiviral regulator and immune modulator, and demonstrate that its inhibition potentiates G47Δ efficacy, providing rationale for a novel therapeutic strategy for malignant brain tumors.

Graphical abstract

MiR-138 inhibition potentiates oncolytic activities of G47Δ by enhancing viral replication and tumor cell killing via regulating the viral protein ICP0, and by augmenting antitumor immunity via STAT1-mediated pathways. The microRNA-targeted combination is a rational therapeutic strategy for improving the efficacy of oncolytic virus therapy for malignant brain tumors.

Introduction

Oncolytic virus therapy has emerged as a promising strategy for cancer treatment.¹ G47Δ is a third-generation oncolytic herpes simplex virus type 1 (HSV-1) with triple mutations that enhance tumor-selective virus replication and antitumor immunity elicitation.² Phase 1/2 and phase 2 clinical trials of G47Δ in patients with glioblastoma confirmed its safety and significant survival benefit, leading to the approval as the first oncolytic virus drug for malignant glioma in 2021.^3,4 Nevertheless, variability in efficacy among patients highlights the need for strategies that further potentiate G47Δ activity, overcoming potential resistance, both intrinsic and acquired, while reinforcing adaptive immune responses.

MicroRNAs (miRNAs) are small non-coding RNAs that orchestrate gene regulatory networks and regulate diverse biological processes, including oncogenesis.⁵ They also play crucial roles in host-virus interactions, serving either as antiviral defense factors or as viral cofactors.⁶ Several host miRNAs restrict HSV-1 replication,^7,8,9 including microRNA-138 (miR-138) that suppresses the viral immediate-early protein ICP0 and promotes viral latency.^10,11 ICP0 is a key regulator of HSV-1 lytic reactivation, making its regulation by miR-138 particularly relevant in the brain, where miR-138 is neuron-specific and enriched.^12,13,14 These findings suggest that targeting miR-138 may provide a means to enhance the oncolytic activity of G47Δ in malignant brain tumors. In this study, we investigate the role of miR-138 in G47Δ therapy. We show that miR-138 inhibition enhances viral replication and tumor cell killing, while also augmenting antitumor immunity through signal transducer and activator of transcription 1 (STAT1)-mediated pathways. STAT1 functions as a central mediator of interferon (IFN) signaling and tumor immunosurveillance.^15,16 This dual effect provides a compelling rationale for combining G47Δ with miR-138 inhibition as a potential therapeutic strategy.

Results

MiR-138 expression and function in glioblastoma cells

We first quantified miR-138 expression in glioblastoma cell lines, clinical specimens, and patient-derived glioma stem-like cells (GSCs) using quantitative RT-PCR (RT-qPCR). In contrast to non-central nervous system (CNS) tumor cell lines such as HeLa, HEK293T, FaDu, SCC4, and Vero, where miR-138 has been reported to be downregulated,^10,17,18 several glioblastoma cell lines, several GSCs, and other CNS tumor cell lines DAOY (human medulloblastoma) and Neuro2a (HSV-1 susceptible murine neuroblastoma), exhibited relatively elevated expressions (Figure 1A). Some GSCs showed low expressions of miR-138. In paired patient-derived samples, miR-138 expression was significantly higher in glioblastoma tumors compared with their matched GSCs (Figures 1B and 1C). Recurrent glioblastoma specimens showed no statistically significant difference in miR-138 expression compared with primary tumors (Figure 1B). To assess endogenous miR-138 activity, we performed dual-luciferase reporter assays. The Renilla luciferase reporter was designed to contain or lack a 23-bp miR-138 target sequence in the 3′UTR, whereas the firefly luciferase reporter served as a transfection control. An increase in the levels of functional endogenous miR-138 within cells corresponds to a decrease in the expression of the miR-138 reporter Renilla luciferase activity normalized by the firefly luciferase activity (RLuc/FLuc ratio). Several CNS tumor cells, such as U87, Neuro2a and DAOY, and some GSCs exhibited reduced reporter activities proportional to their elevated miR-138 expression levels (Figure 1D). Normal human astrocytes also showed a decreased reporter activity. These results demonstrate that a subset of glioblastoma cells and GSCs expresses functional miR-138.

Figure 1.

MiR-138 expression and functional characteristics in glioblastomas

(A) Relative expression of microRNA (miR)-138 in human astrocytes, CNS tumor cells, glioma stem-like cells (GSCs) and non-CNS tumor cells by quantitative RT-PCR (RT-qPCR). Data are shown as mean of triplicates. Bars, SEM. (B) Expressions in 12 GSCs and 18 glioblastoma specimens. Statistical analysis was performed using Student’s t test. ∗p < 0.05; n.s., not significant. (C) Expressions in paired GSCs and glioblastoma tissues from eight patients. Expression levels were normalized to normal human astrocytes with U6 snRNA as reference. Data are shown as mean of triplicates. Bars, SEM. (D) Dual-luciferase assay of miR-138 activity. Relative activity was calculated as the ratio of Renilla to firefly luciferase and normalized to non-target controls. An increase in the levels of functional endogenous miR-138 within cells corresponds to a decrease in the expression of the miR-138 reporter Renilla luciferase activity (miR-138 target) normalized by the firefly luciferase activity (RLuc/FLuc ratio). Data are shown as mean of triplicates. Bars, SEM.

MiR-138 regulates G47Δ-ICP0 expression and modulates host interferon signaling

To investigate whether host endogenous miR-138 affects lytic gene expression during G47Δ infection, we performed western blot analyses. Stable suppression of miR-138 was achieved by lentivirus TuD transduction, and overexpression by transient transfection with a miR-138 mimic. In U251 human glioblastoma cells with minimal endogenous miR-138 activity, transfection of a miR-138 mimic resulted in reduced ICP0 expression (Figure 2A). Conversely, inhibition of miR-138 in Neuro2a cells led to increased ICP0 expression, accompanied by enhanced expression of the immediate-early protein ICP4, the early protein ICP8, and the late protein gC at later time points (Figure 2B). We next examined the impact of miR-138 inhibition on interferon signaling, considering the role of ICP0 in antagonizing this pathway. In Neuro2a cells harboring TuD-miR-138 (Neuro2a-TuD-miR-138), G47Δ infection partially reduced STAT1/STAT2 phosphorylation and blunted the induction of the IFN-inducible gene ISG15 (Figures 2C and 2D). Elevated STAT1 protein levels in Neuro2a-TuD-miR-138 cells further increased upon IFN-α treatment (Figure 2D), suggesting that miR-138 inhibition enhances G47Δ activity not only through ICP0 upregulation but also via modulation of host interferon signaling.

Figure 2.

MiR-138 directly regulates G47Δ-ICP0 and contributes to the inhibition of type I interferon (IFN) signaling

(A) U251 cells transfected with miR-138 mimic (Mimic138) or negative control (NC) were infected with G47Δ (MOI = 0.1) and harvested at indicated times. (B and C) Neuro2a cells stably expressing miR-138 inhibitor (TuD138) or negative control (TuD NC) were infected with G47Δ (MOI = 1) and harvested at indicated times. (D) TuD138 or TuD NC cells pretreated with IFN-α (500 U/mL) were infected with G47Δ (MOI = 1, 12 h). Protein levels of ICP0, ICP4, ICP8, glycoprotein C (gC), STAT1, phosphorylated STAT1 (pSTAT1), pSTAT2, and interferon-stimulated gene 15 (ISG15) were analyzed by western blotting with β-actin as a loading control.

In vitro effects of miR-138 inhibition on G47Δ replication and cytotoxicity

To evaluate the effects of miR-138 on G47Δ infection, we examined viral replication and cytolytic activity in vitro. Suppression or overexpression of miR-138 was confirmed by functional assays, including dual-luciferase reporter validation (Figure S2). In Neuro2a cells, which express functional miR-138, inhibition of miR-138 increased viral progeny and enhanced tumor cell killing at MOI = 1 (Figure 3A). Conversely, in U251 cells lacking functional miR-138, enforced miR-138 expression reduced viral replication and attenuated cytolytic activity at MOI = 0.1 (Figure 3B).

Figure 3.

MiR-138 inhibition increases replication and cytopathic effects of G47Δ in vitro

(A) Neuro2a and (C) U87 cells were transduced with TuD-138 or TuD-NC, and (B) U251 and (C) U87 cells were transfected with miR-138 mimic (Mimic138) or NC mimic (Mimic NC). Virus titers were measured by plaque assays at 48 h after G47Δ infection (MOI = 0.1 for Neuro2a, 0.01 for U251 and U87). Cell viability was assessed by cell counting (Neuro2a, U87) or MTS assay (U251). Data are shown as mean of triplicates. Bars, SEM. Statistical analysis was performed using Student’s t test or two-way ANOVA. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To validate these observations, we conducted both gain- and loss-of-function experiments in U87 human glioblastoma cells. MiR-138 inhibition increased virus production and cytotoxicity at MOI = 0.1 and 0.01, whereas miR-138 overexpression reduced virus yield, despite relatively low endogenous miR-138 activity in these cells (Figure 3C). Additional analyses showed that miR-138 overexpression decreased cell viability in U87 cells but not in U251 cells, and that miR-138 inhibition in Neuro2a, DAOY, and U87 cells did not alter basal cell viability (Figure S3). The results are in line with prior reports that miR-138 suppresses tumorigenicity in glioblastoma.¹⁹

Effects of miR-138 inhibition on antitumor immune responses and STAT1/MHC class I expression in vivo

We examined the impact of miR-138 inhibition in vivo using subcutaneous tumor models. Tumors of Neuro2a cells and Neuro2a cells transduced with a control TuD (Neuro2a-TuD-NC) grew progressively when implanted in syngeneic A/J mice, whereas Neuro2a-TuD-miR-138 tumors grew initially but subsequently regressed in the majority of mice (9/13 mice) (Figure 4A). These mice also rejected subsequent rechallenge with parental Neuro2a cells, indicating the development of protective immunity. In contrast, Neuro2a-TuD-miR-138 tumors in immunodeficient mice failed to regress and grew progressively, similarly to controls (Figure 4B). Increased infiltration of CD4⁺ and CD8⁺ T cells was detected in Neuro2a-TuD-miR-138 tumors (Figure 4C), and depletion of CD8⁺ T cells, but not CD4⁺ T cells, abrogated the tumor regression (Figure 4D).

Figure 4.

MiR-138 directly targets STAT1 and indirectly regulates antitumor immunosurveillance

(A) Tumor growth of syngeneic mice injected with TuD-138 or TuD-NC Neuro2a cells (n = 10–13 per group). (B) Tumor growth in nude mice (n = 10 per group). (C) Flow cytometric quantification of tumor-infiltrating CD4⁺ and CD8⁺ T cells (n = 3). (D) Tumor growth in mice depleted of lymphocytes before implantation (n = 5 per group). (E) STAT1 mRNA expression in tumors measured by RT-qPCR (n = 3). (F) STAT1 protein expression in tumor lysates by western blotting. (G) Immunohistochemistry of STAT1 in tumor sections (scale bar, 50 μm). (H) (i) Predicted miR-138 binding sites within STAT1 3′UTR and corresponding mutants. (ii) Dual-luciferase assay in Neuro2a-TuD138 or Neuro2a-TuD NC cells transfected with wild-type or mutant STAT1 3′UTR vectors. (I) Flow cytometric analysis of MHC class I expression in cells and tumors. (i) Representative histograms. (ii) Quantified mean fluorescence intensity (MFI). Data are shown as mean. Bars, SEM. Statistical analysis: Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To explore the underlying mechanism, we focused on STAT1, a critical mediator of type I IFN-induced gene expression. STAT1 mRNA and protein levels were elevated in Neuro2a-TuD-miR-138 tumors from immunocompetent mice (Figures 4E and 4F). Immunohistochemistry confirmed increased cytoplasmic and nuclear STAT1 staining, with higher nuclear localization in Neuro2a-TuD-miR-138 tumors (Figure 4G). These findings parallel the STAT1 induction observed after IFN-α treatment in Neuro2a cells (Figure 2D).

Luciferase reporter assays using wild-type and mutant 3′UTR sequences of STAT1 demonstrated that miR-138 directly targets STAT1 transcripts to a certain extent (Figure 4H), although indirect mechanisms may also be involved. Finally, flow cytometric analysis revealed upregulation of MHC class I molecules (H2K^k and H2D^d) on Neuro2a-TuD-miR-138 tumor cells, whereas no difference was observed in cultured cells (Figure 4I). These findings indicate that miR-138 inhibition enhances antitumor immune responses in vivo, at least in part, through STAT1 upregulation and consequent induction of MHC class I expression.

In vivo efficacy of miR-138 inhibition combined with G47Δ therapy

We investigated whether the efficacy of G47Δ could be enhanced by combining with miR-138 inhibition in mouse models. First, we assessed direct oncolytic effects under an immunodeficient condition. Intratumoral administration of G47Δ significantly suppressed Neuro2a-TuD-miR-138 tumor growth compared with non-treated Neuro2a-TuD-NC tumors (Figure 5A). However, viral genome replication and viral yields showed no significant differences between the groups (Figures 5B and 5C), likely reflecting the relatively low susceptibility of murine cells to G47Δ compared with human cells.

Figure 5.

G47Δ increases direct oncolytic and antitumor immune effects in miR-138-suppressed tumors

(A–C) Direct oncolytic effects in athymic mice bearing TuD-NC or TuD-138 tumors. Tumor-bearing mice were intratumorally injected with G47Δ (1 × 10⁷ pfu). (A) Tumor growth of TuD NC or TuD138 tumors injected with or without G47Δ on days 0 and 3 (n = 5 per group). (B) Viral DNA copy number in tumors at 1, 24, and 48 h after infection measured by droplet digital PCR (ddPCR; n = 3–4 per group). (C) Viral titers in tumors at 1, 24, and 48 h (n = 4 per group). (D) Antitumor immune effects in syngeneic mice. Bilateral subcutaneous tumors were established, and left tumors were treated with intratumoral injections of G47Δ (1 × 10⁶ pfu on days 0 and 3). Growth of treated and contralateral tumors is shown (n = 8–9 per group). Data are shown as mean. Bars, SEM. Statistical analysis: two-way ANOVA with Bonferroni post-hoc test. ∗p < 0.05.

Second, to evaluate potential antitumor immune responses, we employed a syngeneic bilateral subcutaneous tumor model. On the injected side, G47Δ significantly inhibited the growth of Neuro2a-TuD-NC tumors, while Neuro2a-TuD-miR-138 tumors exhibited marked suppression of growth with or without G47Δ treatment (Figure 5D). On the contralateral side, Neuro2a-TuD-miR-138 tumors showed partial growth suppression compared with Neuro2a-TuD-NC tumors. Combining G47Δ further caused a significant growth inhibition of Neuro2a-TuD-miR-138 tumors, causing complete regression of untreated contralateral tumors in 4 of 9 mice. These findings suggest that miR-138 inhibition potentiates both the direct and immune-mediated antitumor effects of G47Δ, with host immunity playing a pivotal role in mediating this therapeutic augmentation.

In vivo regulation of miR-138 by synthetic TuD electroporation

To evaluate the therapeutic applicability of miR-138 inhibition, we employed in vivo electroporation of S-TuD instead of lentivirus-based stable transduction. Electroporation of S-TuD-miR-138 successfully suppressed miR-138 expression in both Neuro2a cells and subcutaneous tumors, although variability was observed depending on transfection efficiency (Figure 6A). We then examined whether S-TuD electroporation could regulate STAT1 expression. In cultured cells, western blotting at 48 h after electroporation showed no change in STAT1 protein, consistent with the mRNA levels, while cells with stably suppressed miR-138 showed a small increase in STAT1 protein (Figures 6B, S4A, and S4B). In contrast, STAT1 mRNA was significantly upregulated in S-TuD-miR-138 electroporated tumors (Figure 6B). This difference between in vitro and in vivo could be attributed to STAT1 enhancement in immunocompetent environment. We also visualized miRNA regulation using in vivo S-TuD electroporation. When a dual luciferase reporter plasmid containing the perfectly matched miR-138 sequence was co-transfected with S-TuD by in vivo electroporation, luciferase activity was reduced in miR-138-abundant Neuro2a tumors. Remarkably, this reduced luminescence was recovered by S-TuD-miR-138 (Figures 6C and 6D).

Figure 6.

Delivery and antitumor efficiency of in vivo electroporation-mediated miR-138 inhibition

Mice bearing subcutaneous Neuro2a tumors were treated with S-TuD electroporation (500 pmol/tumor) with or without intratumoral G47Δ injection (2 × 10⁵ pfu). (A and B) miR-138 and STAT1 mRNA expressions after electroporation in cells (50 nM) or tumors (n = 3), measured by RT-qPCR and normalized to controls. (C) Representative bioluminescence images of tumors co-transfected with dual-luciferase miR-138 reporter (Luc-T138) and S-TuD constructs. (D) Inhibitory effect of S-TuD-miR-138 on endogenous miR-138 activity quantified as Renilla/firefly luciferase (RLuc/FLuc) ratio (n = 7 per group). (E) Tumor growth after electroporation therapy alone (n = 5 per group). (F) Tumor growth following the combination therapy. Electroporation was performed on days 0 and 3, and G47Δ was injected intratumorally on days 1 and 4 (n = 8–9 per group). Data are shown as mean. Bars, SEM. Statistical analysis: two-way ANOVA with Bonferroni post-hoc test. ∗p < 0.05.

Next, we investigated the efficacy of combining G47Δ with in vivo S-TuD-miR-138 electroporation. Single electroporation of S-TuD-miR138 to Neuro2a tumors in syngeneic A/J mice significantly suppressed the tumor growth (Figure 6E). The combination of intratumoral injection of G47Δ (2 × 10⁵ pfu) and S-TuD-miR-138 electroporation caused a significant growth suppression of Neuro2a tumors compared with G47Δ combined with control electroporation (Figure 6F), suggesting that miR-138 inhibition can potentiate the oncolytic activities of G47Δ.

Discussion

This study identifies miR-138 as a dual regulator of viral replication and host immunity, providing a novel strategy to potentiate the efficacy of a third-generation oncolytic HSV-1, G47Δ. Previous studies have exploited miRNA regulation to enhance the specificity of oncolytic viruses, including HSV-1, by incorporating miRNA-responsive elements into viral genomes.^{20,21,22,23,24} In contrast, our approach leverages the regulatory function of a host miRNA to augment viral activity without further genetic modification of the virus. Our findings suggest that miR-138 inhibition can overcome intrinsic barriers to viral replication while simultaneously reinforcing adaptive immune responses.

MiR-138 has been reported to act as a tumor suppressor in various cancers,²⁵ yet its role in glioblastoma appears variable—being reduced in some reports,¹⁹ enriched in GSCs,²⁶ and increased at recurrence.²⁷ We confirmed that a subset of glioblastoma and GSCs express miR-138 at functional levels and demonstrated that it directly suppresses ICP0, an immediate-early protein essential for HSV-1 lytic reactivation. It has been demonstrated that a disruption of miR-138 target sites in ICP0 enhanced viral gene expression and host mortality, underscoring its antiviral function.¹⁰ Subsequent studies revealed that miR-138 also targets OCT-1 and FOXC1, further suppressing HSV-1 lytic cycle genes.^12,28 Our results extend these findings by showing that inhibition of miR-138 enhances G47Δ replication and tumor cell killing both in vitro and in vivo. Importantly, expression levels alone did not necessarily predict activity, highlighting the need for functional assays such as luciferase reporters.

Mechanistically, we found that G47Δ infection under miR-138 inhibition attenuated type I IFN signaling, thereby facilitating viral replication. The IFN system is a central antiviral defense,²⁹ and HSV-1 has evolved countermeasures, including suppression of JAK–STAT phosphorylation.^30,31 ICP0 plays a key role in this process, as ICP0-null mutants are highly sensitive to IFN pretreatment,^32,33,34 and ICP0 directly interferes with STAT1.³² We further demonstrated that miR-138 directly targets STAT1 mRNA to a certain extent, and its inhibition upregulates STAT1 protein and MHC class I expression in tumor cells. STAT1 has dual functions: it promotes antitumor immunity by enhancing antigen presentation and CD8⁺ T cell cytotoxicity,^16,35,36 and its intratumoral expression correlates with improved prognosis across multiple cancers,^37,38 but it also mediates antiviral IFN responses.²⁹ Beyond immune regulation, STAT1 can exert direct anti-proliferative and pro-apoptotic effects on tumor cells,³⁹ although in our in vitro system, cell proliferation appeared independent of STAT1 following miR-138 inhibition. Unphosphorylated STAT1, which accumulates in response to IFNs,⁴⁰ may also contribute to sustaining immune activation. This duality of STAT1 function may vary across tumor contexts, and further evaluation will be needed to determine which effects predominate in glioblastoma. Our data suggest that miR-138 inhibition offsets the latter by blunting IFN signaling, thereby preserving viral oncolysis while still reinforcing STAT1-driven immune surveillance.

The results in immunocompetent A/J mice using Neuro2a cells engineered to stably suppress miR-138 (Neuro2a-TuD-miR-138) and using in vivo inhibition of miR-138 by electroporation of S-TuD-miR-138 to Neuro2a tumors indicate that miR-138 inhibition potentiates the efficacy of G47Δ. Our data highlight the functional duality of miR-138: While it has tumor-suppressive properties, it simultaneously acts as an antiviral factor that restricts oncolytic activity. By targeting miR-138, it is possible to enhance both viral replication and antitumor immunity, thereby converting this duality into therapeutic advantage. For clinical translation, the development of safe and effective delivery systems for nucleic acids remains essential. In vivo electroporation has already been shown to deliver siRNA and plasmids effectively in tumors,^41,42 and our study provides proof-of-concept for applying this method to a miRNA inhibitor, thereby offering a feasible strategy for therapeutic application.

In conclusion, our study demonstrates that miR-138 functions as both an antiviral regulator and an immune modulator. MiR-138 inhibition potentiates oncolytic activities of G47Δ at two complementary levels: enhancing viral replication and augmenting adaptive antitumor immunity. Our findings provide a rationale for miR-138-targeted combination strategies in malignant brain tumors for improving the efficacy of oncolytic virus therapy.

Materials and methods

Cell lines and virus

G47Δ is a triple-mutated HSV-1 that has deletions or inactivation of the γ34.5, ICP6, and α47 genes.² G47Δ was propagated and titered on Vero cell cultures, as described previously.² U87MG, DAOY, HeLa, HEK293, FaDu, and Vero cells were purchased from the American type culture collection, and U251MG, T98G, A172, Neuro2a, and SCC4 cells were obtained from the Japanese collection of research bioresources. SCC4 cells were cultured in DMEM/nutrient mixture F-12 (F12), FaDu cells in minimum essential medium (MEM), and all other cell lines in DMEM. Each medium was supplemented with 10% FBS (Invitrogen) without antibiotics. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂. Normal human astrocytes were obtained from ScienCell Research Laboratories (Carlsbad, CA, USA) and cultured according to the manufacturer’s instructions.

Glioma stem-like cells

Patient-derived glioma stem-like cells (GSCs) were isolated from surgical specimens obtained at IMSUT Hospital (Tokyo, Japan). Neurospheres were maintained in serum-free DMEM/F-12 supplemented with B27 (Invitrogen), 20 ng/mL epidermal growth factor (EGF; PeproTech, Rocky Hill, NJ, USA), and 20 ng/mL basic fibroblast growth factor (bFGF; PeproTech). Expression of CD133 was evaluated by flow cytometry, and expression of Nestin and SOX2 was confirmed by western blotting (Figure S1). Patient clinical characteristics are summarized in Table S1. Patient tumor samples were obtained with written informed consent under a protocol reviewed and approved by the Institutional Review Board at the University of Tokyo Hospital and the Institute of Medical Science, the University of Tokyo.

Reagents

Mouse IFN-α was purchased from R&D Systems (Minneapolis, MN, USA). Anti-CD8 (clone 53–6.72) and anti-CD4 (clone GK1.5) monoclonal antibodies were obtained from BioX Cell (Lebanon, NH, USA).

Vector construction

Tough decoy (TuD)-miR-138 and TuD-negative control (NC) expression plasmids were constructed as previously described.^43,44 Briefly, the oligonucleotides listed in Table S2 were synthesized and inserted into the BsmBI-digested ph7SK-shuttle. For lentivirus production, each h7SK-TuD cassette, the 0.4 kb BamHI-EcoRI fragment, was cloned into BamHI-EcoRI site of the lentiviral vector pLSP. To construct luciferase reporter plasmids with the miR-138 target site, STAT1 binding site, or STAT1 binding site with point mutations, oligonucleotides listed in Table S2 were synthesized and cloned into the XhoI-NotI site of psiCHECK-2 vector (Promega).

Lentiviral transduction and transfections

Stably-transduced cell lines were generated by lentiviral transduction with TuD constructs followed by puromycin selection. Transient overexpression of miR-138 was achieved using mirVana miRNA mimics (Ambion) transfected with lipofectamine RNAiMAX (Invitrogen). Synthetic TuD (S-TuD) oligonucleotides (sequences listed in Table S2) were introduced into Neuro2a cells by electroporation (Neon Transfection System, Life Technologies, Carlsbad, CA, USA) at a working concentration of 50 nM.

In vivo S-TuD electroporation

S-TuD delivery into Neuro2a tumors was performed by in vivo electroporation using a plate-and-fork-type electrode CUY663-5X8 (NEPA GENE) and a Type II electroporator (NEPA GENE) with constant electrical parameters, as previously reported.^41,42 A total of 500 pmol S-TuD in sterile PBS (50 μL) was directly injected into the tumor, and electrical pulses were immediately delivered three times at 50 ms pulse length and 50 V field strength.

Cell-based assays

Cell survival after G47Δ infection was monitored by daily cell counts using a Coulter counter (Beckman Coulter, Brea, CA, USA). Viability following transfection was measured using the CellTiter 96 AQueous MTS assay (Promega, Madison, WI, USA).

Virus yield

Plaque assays were performed on Vero cells after infection of tumor cells at MOI = 0.01–0.1. For in vivo titration, tumors were harvested, homogenized, and viral DNA copies quantified by droplet digital PCR (ddPCR) targeting ICP6.

Luciferase reporter assays

Cells were co-transfected with psiCHECK-2 reporter plasmids and miRNA mimics or S-TuD, and luciferase activity measured using Dual-Glo (Promega). Reporter constructs containing wild-type or mutant STAT1 3′UTR sequences (Table S2) were also used to validate direct targeting by miR-138. For in vivo imaging, tumors were electroporated with psiCHECK-2 (Promega) and S-TuD, followed by substrate injection and bioluminescence imaging using IVIS Lumina III.

Animal experiments

Five-week-old BALB/c nude mice and A/J mice (Japan SLC, Shizuoka, Japan) were used. Subcutaneous tumors were established by injecting 5 × 10⁶ cells into the flanks. G47Δ was administered intratumorally when tumors reached ≥5 mm. For in vivo electroporation of S-TuD, tumors ≥7 mm were injected with S-TuD followed by electroporation and subsequent G47Δ infection. For T cell depletion, mice were treated intraperitoneally with 250 μg anti-CD4 or anti-CD8 antibodies every 5 days, achieving >95% depletion. Tumor volume was calculated as length × width × height (mm³). All animal experiments were performed in accordance with institutional guidelines and were approved by the Ethics Committee for Animal Experimentation of the University of Tokyo.

RNA isolation and RT-qPCR

Total RNA was extracted using the miRNeasy Mini Kit (Qiagen, Hilden, Germany). For miR-138 quantification, TaqMan MicroRNA assays (Applied Biosystems, Foster City, CA, USA) were used with U6 snRNA as control. For STAT1 mRNA quantification, cDNA was generated with ReverTra Ace (TOYOBO, Osaka, Japan) and amplified with TaqMan probes (Applied Biosystems), normalized to β-actin.

Western blotting

Protein lysates were prepared with RIPA buffer (cells) or T-PER reagent (tissues), separated by SDS-PAGE, transferred to PVDF membranes, and probed with antibodies against viral proteins (ICP0, ICP4, ICP8, and gC), STAT1, phosphorylated STAT1 (pSTAT1), p-STAT2, ISG15, SOX2, Nestin, and β-actin. Bands were visualized using the ECL system (Pierce, Thermo Fisher Scientific, Rockford, IL, USA) and quantified using ImageJ.

Flow cytometry

Tumors were dissociated with a Miltenyi Biotec (Bergisch Gladbach, Germany) kit and stained with antibodies against CD3, CD4, CD8, CD45, H2D^d, H2K^k, and CD133, along with viability dye. Samples were acquired on a CytoFLEX (Beckman Coulter, Brea, CA, USA) and analyzed with FlowJo.

Immunohistochemistry

Excised tumors were formalin-fixed, paraffin-embedded, sectioned, and stained with anti-STAT1 antibody (Cell Signaling Technology, Danvers, MA, USA).

Statistics

Data were analyzed using GraphPad Prism v8.0. Comparisons were made using Student’s t test (two groups) or two-way ANOVA with Bonferroni post-hoc test for growth curves. p < 0.05 was considered statistically significant.

Data and code availability

The data that support the findings of this study are available from the corresponding author on reasonable request.

Acknowledgments

This research was supported in part by AMED under grant no JP18nk0101378 to H.I. and T.T., and by JSPS KAKENHI under grant no JP22H00483 to T.T. and grant number JP19K09474 to S.K.

We thank Dr. Miwako Iwai for her valuable advice and support during this study.

Author contributions

S.K. contributed to conceptualization, data curation, investigation, methodology, and writing (original draft preparation). T.H. and H.I. performed data analysis and interpretation. M.T. performed data analysis and writing (manuscript preparation). T.T. contributed to conceptualization, supervision, writing (manuscript preparation) and funding acquisition.

Declaration of interests

T.T. owns the patent right for G47Δ in Japan, and is a member of the Editorial Board of Molecular Therapy Oncology. M.T. is a member of a division endowed by Denka Company.