Volum 18, Issue 5 (2026)

Introduction

Glioblastoma is the most common type of neuroepithelial brain tumor in adults, accounting for 48.6% of all malignant neoplasms of the central nervous system [1, 2]. Currently, glioblastoma therapy, in accordance with standards of care, is based on maximally radical surgical resection of the tumor, followed by fractionated radiotherapy concurrently with adjuvant chemotherapy using temozolomide [3]. As a second-line treatment, bevacizumab, carmustine, and lomustine are used [4, 5]. The median survival with standard glioblastoma therapy does not exceed 15 months [2].

According to the main hypothesis of glioblastoma pathogenesis, the tumor originates fr om glioma stem cells (GSCs) derived from resident neural stem cells through oncogenic mutations and other factors [6]. GSCs are particularly resistant to radiotherapy and chemotherapy, which is why it is hypothesized that quiescent GSCs are the main factor in glioblastoma recurrence [7–9]. There exists a bidirectional interaction between GSCs and immune cells within the tumor microenvironment (TME). This interaction can define and modify the stem cell niche for GSCs, as well as influence the immune composition of the TME [10]. Therefore, combined therapeutic approaches that target both highly proliferative tumor cells and quiescent GSCs, as well as immune cells within the TME, are particularly promising [11, 12].

Various immunotherapy approaches are being considered as alternative therapies for solid tumors, including molecularly targeted therapy, chimeric antigen receptor T cell (CAR-T) therapy, natural killer cell (NK cell) therapy, PD-1/PD-L1 immune checkpoint blockade, antitumor neoantigen peptide and mRNA vaccines, and oncolytic virotherapy [13–16]. The application of these approaches in glioblastoma has not yet yielded any significant success due to the high genetic heterogeneity of glioblastoma, which arises from clonal evolution of the tumor [17], as well as its status as an immunologically “cold” tumor — i.e., a tumor with low immunogenicity and a high level of immunosuppression within the TME [18]. Therefore, approaches aimed at immunological “priming” of glioblastoma and overcoming the immunosuppressive influence of the TME become particularly relevant.

The principles of oncolytic therapy are based on the unique ability of certain recombinant viruses — such as modified herpes simplex virus, parvovirus, vesicular stomatitis virus, reovirus, Newcastle disease virus, Rift Valley fever virus, Sindbis virus, poliovirus, Seneca virus, myxovirus, adenovirus, and vaccinia virus — to selectively replicate within tumor cells, induce their cytolysis, and trigger the release of tumor-specific antigens. This, in turn, elicits an immune response within the TME and leads to immunogenic cell death of tumor cells [19]. Furthermore, genetically modified tumor-targeting viruses can be used for the selective “gene delivery” to tumors. Such transgenes may deliver immunomodulatory agents — for instance, chemokines (CCL3, CCL5) or interleukins (IL-15, IL-21, etc.) — to enhance the antitumor immune response within the TME [20–22].

Oncolytic virotherapy using recombinant vaccinia viruses (rVV) can be applied both as monotherapy and in combination with cellular immunotherapy or immune checkpoint inhibitors [23–25]. In particular, treatment of experimental glioma with a rVV expressing IL-21 cDNA, combined with immune checkpoint inhibition, yielded promising results [21]. A number of other pre-clinical studies on recombinant oncolytic viruses also support recommending this approach for clinical trials in glioblastoma [16, 22, 26, 27].

Previously, through collaborative efforts by several research groups, novel rVV vectors were developed based on the Lister strain from the Institute of Viral Preparations (LIVP), deficient in thymidine kinase and carrying IL-15 cDNA. The oncolytic efficacy of the resulting recombinant strain, LIVP-mIL15, was demonstrated in preclinical studies using mouse models of colorectal cancer and breast cancer [28]. Given that glioblastoma is known to be susceptible to rVV infection, the current study is a logical continuation of our previous work.

The aim of the study was to evaluate the impact of the recombinant vaccinia virus strain carrying the interleukin-15 cDNA (LIVP-mIL15) on the tumor microenvironment in the GL261 experimental glioma model.

Materials and Methods

Viruses. The rVV strain (LIVP-mIL15) with a thymidine kinase deletion was developed based on the LIVP strain by a research team led by Academician P.M. Chumakov at the Engelhardt Institute of Molecular Biology, Russian Academy of Sciences [27, 28]. To monitor targeted delivery of the protein of interest (mouse IL-15) to tumors, LIVP strains were engineered to additionally express the reporter gene of the tagRFP (red fluorescent protein) under the control of the 7.5 K promoter (designated LIVP-mIL15-RFP) or the firefly luciferase gene (FLuc; Photinus pyralis luciferase; designated LIVP-mIL15-FLuc). These constructs were successfully tested in previous studies [28, 29]. All recombinant viruses were purified by sucrose gradient centrifugation for both in vitro and in vivo experiments. Viral titres were determined as previously described [28].

Cell line cultivation and glioblastoma spheroid modelling. GL261 cells and the GL261-GFP subclone (stably transduced with a lentivirus carrying GFP cDNA [green fluorescent protein]) were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37𐩑C, 5% CO₂ in a humidified atmosphere, and passaged every 2–3 days.

Spheroids of murine glioblastoma GL261 cells were generated in 96-well ultra-low attachment round-bottom plates (Ultra Low Cluster, 96 wells with lid, CoStar Group, USA). A total of 3000 cells were seeded into each well. On day 3 after spheroid formation, the wells were supplemented with the rVV (LIVP-mIL15-RFP) at several serial dilutions in culture medium.

From each well, 100 µl of conditioned medium was collected, and 100 µl of fresh culture medium was immediately added, containing serially diluted viral particles ranging from 10 to 0.3 µl, with a viral titre of 2·10⁸ plaque-forming units (PFU)/ml. Thus, the maximum dose per spheroid was 2·10⁶ PFU, achieved by adding 10 µl of the viral preparation to 90 µl of medium.

Spheroid assembly and proliferation were monitored using the Invitrogen EVOS M7000 imaging system (Thermo Fisher Scientific, USA) over a period of 6 days (approximately 144 h).

Animal study. Animal experiments were conducted in accordance with ethical principles established by European Convention for the Protection of Vertebrata used for Experimental and Other Scientific Purposes (the Convention was passed in Strasbourg, March 18, 1986, adopted in Strasbourg, June 15, 2006). The experimental protocols were approved by the local ethics committees of the following institutions: the Federal Research and Clinical Centre of the Federal Medical-Biological Agency of Russia (protocol No.7 dated September 6, 2022), the Engelhardt Institute of Molecular Biology, Russian Academy of Sciences (protocol No.1 dated April 3, 2024), and the Pirogov Russian National Research Medical University (protocol No.27/2025 dated July 24, 2025). All animals had free access to food and water and were housed under controlled temperature and humidity conditions.

Orthotopic glioma modelling. Orthotopic glioma was modelled in 20 female C57BL/6 mice. For intracranial implantation, aliquots of GL261 cells were prepared at a concentration of 150·10³ cells in 5 µl of sterile Dulbecco’s phosphate-buffered saline (DPBS). Under experimental operating conditions, the mice were first sedated with isoflurane, then anesthetized with Zoletil (40 mg/kg, 50 µl; Virbac SA, France) and Rompun (10 mg/kg, 50 µl; Bioveta, Czech Republic). Once deep anesthesia was achieved, the animals were placed in a stereotactic system (RWD Life Science, China).

Assessment of immune status and intravital imaging. To assess the immune status by flow cytometry, a stereotactic injection of 150·10³ unlabelled GL261 cells in 5 µl of sterile DPBS was performed into the striatum of the right cerebral hemisphere (n=10). For intravital visualization using confocal microscopy, mice were implanted with 150·10³ GL261-GFP cells in 5 µl of sterile DPBS into the cortex of the right hemisphere (n=10). The modelling protocol and stereotactic implantation coordinates are described in detail in our previous work [30].

Magnetic resonance imaging (MRI). Weekly MRI scans were performed on mice with orthotopic glioma at the Shared Research Facilities Centre “Medical and Biotechnological Nanotechnologies” of the Pirogov Russian National Research Medical University, using a ClinScan 7T scanner (Bruker Biospin, Germany). The MRI protocol comprised several sequences. For T2-weighted imaging in the axial plane, a Turbo Spin Echo pulse sequence with magnetization recovery pulse and respiratory triggering was applied. The turbo factor was set to 14, with 2 averages, TR/TE parameters of 3500/46 ms, spectral fat suppression, a field of view (FOV) of 20×45 mm, a slice thickness of 0.5 mm, and a matrix size of 130×192. For T2-weighted imaging in the coronal plane, the same Turbo Spin Echo sequence with magnetization recovery pulse and respiratory triggering was used, but with a turbo factor of 18, 2 averages, TR/TE of 2500/46 ms, spectral fat suppression, an FOV of 15×20 mm, a slice thickness of 0.5 mm, and a matrix size of 130×192. Susceptibility-weighted imaging in the axial plane was performed using a 3D Gradient Echo pulse sequence with radiofrequency signal suppression and flow compensation. This sequence had TR/TE parameters of 33/8 ms, a flip angle of 14°, 1 average, an FOV of 20×20 mm, a slice thickness of 0.5 mm, and a matrix size of 192×159.

Administration of the recombinant vaccinia virus strain. A single administration of the recombinant strain was performed after confirming tumor growth by MRI, on day 5 (for flow cytometry) or day 10 (for intravital visualization). The animals were randomly assigned to three groups, which received either LIVP-mIL15-RFP, LIVP-mIL15-FLuc, or 100 µl of DPBS. For each injection, 2.5·10⁵ PFU of vaccinia virus were used in 100 µl of DPBS. The administration of the rVV strain was carried out at the National Research Center for Epidemiology and Microbiology named after Honorary Academician N.F. Gamaleya.

In vivo bioluminescent imaging. Intravital bioluminescence imaging was performed using an IVIS Spectrum CT system (Perkin Elmer, USA) in accordance with the manufacturer’s instructions. During the procedure, mice were anesthetized with 2.5% isoflurane in oxygen. To investigate the targeted delivery kinetics, LIVP-mIL15-FLuc was administered via two routes: intravenously and stereotactically. Bioluminescence signals from the rVV strain carrying the bioluminescent FLuc marker were recorded 15 min after intraperitoneal injection of D-luciferin sodium salt (BioVision, USA) at a dose of 150 mg/kg. Images were analyzed using LivingImage 4.2 software (Caliper Life Sciences, USA). The bioluminescence signal was quantified using total flux values within the region of interest, which was manually delineated to encompass the entire signal area of the brain tumor.

Intravital confocal microscopy. Intravital microscopy was performed in mice with orthotopic GL261-GFP glioma 48 h after stereotactic intratumoral administration of LIVP-mIL15-RFP. The examination was carried out using an inverted Nikon A1 confocal microscope (Nikon, Japan) equipped with a custom-designed stage that allowed secure fixation of the animal. One hour prior to imaging, the mice received an intravenous injection of primary fluorescently labelled antibodies targeting lymphocyte surface receptors (anti-mouse CD45 Brilliant Violet 421; 103134; BioLegend, USA).

Under deep anesthesia, the mouse head was fixed in a stereotactic frame (RWD), and an intracranial window was created above the glioma focus under sterile conditions, without breaching the pia mater. The mouse was then secured on a heated custom-designed stage so that the tumor was positioned directly above the objective lens. The imaging procedure lasted approximately 15–20 min. After the experiment was completed, the mice were decapitated, and the brains were extracted for subsequent ex vivo studies.

Colocalization analysis was performed as follows. Using the NIS Elements AR software (Nikon, Japan), representative tumor image regions with fluorescent signals from both LIVP-mIL15-RFP-infected cells and GL261-GFP cells were selected, applying a binary mask. The selection threshold was set at the minimum level required to remove background noise.

Next, colocalization of signals from CD45, LIVP-mIL15-RFP, and GL261-GFP was analyzed in the selected zones. To assess colocalization, the Manders overlap coefficient was used, as it is less sensitive to signal intensity variations and the influence of background noise. To ensure statistical reliability of the data, at least 15 FOV were analyzed. Confocal microscopy images were processed using the NIS Elements AR software (Nikon, Japan).

Flow cytometry. The freshly isolated mouse brain was minced in RPMI culture medium using a DSC-400 tissue dissociator (RWD Life Science, China) according to the manufacturer’s protocol. To obtain a single-cell suspension, the tissue was passed through a 100 µm mesh filter, followed by a 70 µm mesh filter. The cells were washed in DPBS and diluted to a concentration of 1^–5·10⁶/ml in cold DPBS buffer supplemented with 10% FBS. The cells were incubated at 4°C for 20 min with selected antibodies against CD45 (cat. No.103134, 103116), CD3 (cat. No.100209), CD4 (cat. No.100407), CD8 (cat. No.100753), CD19 (cat. No.115549), NK1.1 (cat. No.108741), CD11b (cat. No.101208), CD11c (cat. No.117312), Ly6G (cat. No.127626), and F4/80 (cat. No.123110) from BioLegend (USA) in DPBS supplemented with 2% FBS, 1 mM ethylenediaminetetraacetic acid, and 0.1% sodium azide (washing buffer). Next, the cells were washed twice with the washing buffer, resuspended in the same buffer, centrifuged at 300 g for 10 min, and transferred to a 96-well plate for further analysis. Flow cytometry analysis was performed using a MACSQuant Analyzer 16 (Miltenyi Biotec, Germany) and FlowJo software. The study was performed 48 h after a single stereotactic administration of LIVP-mIL15-RFP.

Statistical data analysis. Statistical analysis was performed using GraphPad Prism software, v. 8.0.2 (GraphPad Software, Inc., USA). The data were presented as follows: median (Me); first and third quartiles [Q1; Q3]; minimum and maximum values (min, max). To assess the significance of differences, non-parametric tests for small samples were applied. For data with a normal distribution (photon count), the results were presented as mean ± standard deviation (SD). Statistically significant differences were defined as those with p<0.05.

Results

In vitro study using a GL261 tumor spheroid model. In vitro, LIVP-mIL15-RFP effectively infected GL261 cells in 3D spheroids, as detected by RFP ex * pression (Figure 1). Red fluorescence in target cells exposed to LIVP-mIL15-RFP was observed from 48–72 up to 144 h. The data were evaluated qualitatively. The maximum brightness of RFP signal in tumor spheroids was observed at doses ranging from 1.2·10⁵ to 2.5·10⁵ PFU per spheroid. When 5·10⁵ PFU were added, the expression level began to decrease significantly due to a cytopathic effect. At the highest dose (10 µl, 2·10⁶ PFU of the virus), a pronounced cytopathic effect was observed from the very beginning, leading to complete spheroid disintegration after 96 h of observation (see Figure 1).

Figure 1. Infection of tumor spheroids derived from GL261 cells with the recombinant vaccinia virus strain LIVP-mIL15-RFP. Fluorescence images showing RFP signal in infected cells. The maximum level of RFP expression was observed on days 4–5 (indicated by arrows). The viral stock titre was 2·10⁸ PFU/ml

Based on the in vitro experimental results, a dose of 2.5·10⁵ PFU of LIVP-mIL15-RFP was selected. This dose, calculated relative to the initial number of 150·10³ implanted glioma cells, does not induce a rapid cytotoxic response but is optimal in terms of the ability to infect host cells and achieve maximum expression of the target protein.

Intravenous and stereotactic single administration of recombinant vaccinia virus in mice with orthotopic glioma. After implantation of GL261 cells into mice, MRI scans were performed every 5–7 days for a period of 21–22 days. By the third week, tumor volume reached 15 mm³ according to MRI data (Figure 2).

Figure 2. MRI visualization of the tumor: (a) GL261 cells were injected into the right striatum of female C57BL/6 mice; MRI scans were performed at 1, 2, and 3 weeks post-tumor implantation; representative images are presented: the top row shows frontal projections, the bottom row shows sagittal projections; the arrows indicate the localization of the tumor within the tissue; (b) dynamics of implanted tumor volume (presented as Me [Q1; Q3], min, max)

After confirming tumor growth of glioma on MRI data on the 5^th day after implantation, the mice were administered the rVV strain LIVP-mIL15-FLuc, and infection was assessed by luciferase expression using the IVIS Spectrum CT optical imaging system (Perkin Elmer, USA) (Figure 3).

Figure 3. Intravenous and stereotactic delivery of the recombinant vaccinia virus strain. GL261 cells were inoculated into the right striatum of female C57BL/6 mice. Detection of LIVP-mIL15-FLuc bioluminescence using the IVIS Spectrum CT system 15 min after luciferin injection. The signal was monitored using an in vivo imaging system. Representative images are shown: (a) mouse that received intravenous administration of LIVP-mIL15-FLuc (left) or DPBS (right) 24 h after injection; (b) the same mice 144 h after injection; (c) mouse that received stereotactic administration of LIVP-mIL15-FLuc 48 h after injection. The graph (d) shows the dynamics of intracranial luminescence of GL261 cells infected with LIVP-mIL15-FLuc after intravenous injection (Mean±SD)

After stereotactic administration of LIVP-mIL15-FLuc, bioluminescence in the area of the orthotopic glioma was detected 48 h after injection, and its intensity increased up to day 6. This indicated active expression of the target protein by tumor cells following infection with LIVP-mIL15-FLuc (Figure 3 (c)).

In the case of intravenous administration of LIVP-mIL15-FLuc, bioluminescence in the area of the orthotopic glioblastoma was detected 72 h after injection. A strong signal of approximately the same intensity was observed for 6 days (144 h) after administration of LIVP-mIL15-FLuc (Figure 3 (d)). Thus, LIVP-mIL15-FLuc — whether administered intravenously or stereotactically — infected tumor cells and induced expression of the target protein.

In vivo confocal microscopy of LIVP-mIL15-RFP. The delivery of LIVP-mIL15-RFP to the tumor and the periglioma zone was investigated using in vivo confocal microscopy. Infected glioma cells and cells of the periglioma zone were detected by RFP fluorescence, and the nature of the infected cells was determined by colocalization of the RFP signal with other reporter signals.

Based on preliminary results obtained using optical imaging with the IVIS Spectrum CT optical imaging system (Perkin Elmer, USA), the study of targeted delivery of LIVP-mIL15-RFP to glioblastoma cells and the periglioma zone was performed 48 h after administration of LIVP-mIL15-RFP. Through a transcranial window, fluorescent signals from the blue, green, and red channels were simultaneously detected: BV421 (CD45⁺ of the TME); GFP (GL261 cells); RFP (rVV strain LIVP-mIL15-RFP). To assess which cells — GL261 or CD45⁺ cells — are predominantly targeted by LIVP-mIL15-RFP after stereotactic administration, a colocalization analysis of the signals was performed (Figure 4).

Figure 4. Results of intravital microscopy following stereotactic administration of LIVP-mIL15-RFP into the site of an orthotopic GL261-GFP glioma: (a, b) expression of RFP is visualized both in infected tumor cells and in cells of the tumor microenvironment (TME); the arrow indicates the orthotopic GL261-GFP glioma; a number of infected RFP⁺ cells are observed along the track from the stereotactic virus injection; fluorescence in the blue channel (anti-CD45-BV421) indicates the presence of immune cells in the TME; (c, d) colocalization analysis of the RFP signal with signals from GL261-GFP and anti-CD45-BV421; the highest degree of colocalization is shown for RFP and GFP; however, there is also a subpopulation of CD45⁺RFP⁺ cells (i.e., a subset of TME leukocytes are also infected by LIVP-mIL15-RFP)

According to the processed results, the majority of LIVP-mIL15-RFP infection signals colocalized with glioma cells (GL261). In addition, a subpopulation of CD45-positive immune cells within the TME was observed, in which the signal from LIVP-mIL15-RFP also colocalized (Figure 4 (d)). These data indicate a limited selectivity of LIVP-mIL15-RFP towards tumor cells.

Study of the tumor microenvironment in experimental glioma after a single administration of LIVP-mIL15-RFP. The subpopulation composition of immune cells in the experimental glioma site was investigated by flow cytometry 48 h after a single stereotactic administration of LIVP-mIL15-RFP. Two groups of 10 mice each were formed, with 5-day gliomas confirmed by MRI: the first group received LIVP-mIL15-RFP, and the second group received DPBS (control). For each animal, the tissue from the left (intact) hemisphere served as an internal control.

Flow cytometry analysis showed that the total proportion of CD45⁺ leukocytes 48 h after LIVP-mIL15-RFP injection not only fails to increase, but tends to decrease in both the intact and tumor-affected hemispheres (Figure 5 (а)). Analysis of the subpopulation composition of lymphocytes within TME revealed that after stereotactic injection of LIVP-mIL15-RFP, the proportion of CD45⁺NK1.1⁺ cells (natural killer cells) and CD45⁺CD8⁺ lymphocytes (cytotoxic cells) significantly increased in the ipsilateral (affected) hemisphere compared to the control (Figure 5 (b, c)). Notably, the proportions of the CD45⁺CD4⁺ subpopulations (helper T cells, T regulatory cells) and CD45⁺CD49b⁺ cells (NK, NKT, and cytotoxic lymphocytes) increased symmetrically in both the intact and tumor-bearing brain hemispheres (Figure 5 (d, e)). Collectively, these data indicate that a single stereotactic administration of LIVP-mIL15-RFP induces an immune cells activation effect 48 h later, particularly affecting natural killer cells and cytotoxic lymphocytes within the TME.

Figure 5. Subpopulation composition of CD45⁺ immune cells in the intact and tumor-affected brain hemispheres 48 h after stereotactic injection of LIVP-mIL15-RFP. Analysis by flow cytometry: (a) total population of CD45⁺ cells; (b) subpopulation of CD45⁺NK1.1⁺ cells (natural killer cells); (c) subpopulation of CD45⁺CD8⁺ cells (cytotoxic lymphocytes): (d) subpopulation of CD45⁺CD4⁺ cells (helper T cells); (e) subpopulation of CD45⁺CD49b⁺ cells (NK, NKT, and cytotoxic lymphocytes)

Discussion

Recombinant oncolytic viruses capable of delivering cDNA of immunomodulatory proteins into glioma cells and cells of the TME are considered promising agents for immunological “priming” of solid tumors [27, 29]. In this study, we investigated the prospects of expanding the application spectrum of LIVP-mIL15-RFP — a virus that induces IL-15 expression in infected cells and has demonstrated efficacy in models of colon and breast cancer [28]. Given the immunologically “cold” nature of experimental glioblastoma, it represents a promising target for IL-15, which can activate the recruitment of NK cells and cytotoxic lymphocytes into the TME [30].

The GL261 cell line-based model was chosen as the most relevant experimental murine glioblastoma. GL261 cells harbor point mutations in the TP53 and KRAS genes (encoding the p53 and K-RAS proteins, respectively) [31]. These cells are radiotherapy-resistant, relatively immunogenic, express the major histocompatibility complex class I (MHC I), and have a relatively high neoantigen load [32]. Notably, the pathways most frequently affected in human glioblastoma are the p53, RB, and RTK/RAS/PI(3)K pathways [33]. Thus, GL261 cell models to some extent mimic the pathogenesis of human glioblastoma [34].

It was demonstrated that LIVP-mIL15-RFP is capable of infecting GL261 cells cultured in a 3D spheroid format, and also successfully infects and induces reporter gene expression in cells of orthotopic experimental glioblastoma in vivo. Interestingly, it has previously been shown that rVV stimulate the migration of infected cells, thereby enhancing their own spread [35].

Our data, showing that intravenously administered LIVP can cross the blood-brain barrier in the tumor area and reach cells of intracranial glioma, demonstrate its potential for systemic administration. This is important because exogenous IL-15, when administered systemically or subcutaneously, may exhibit toxic effects [36, 37], while its targeted delivery to the tumor using rVV could be a safer option.

It is known that vaccinia viruses, via the VCP protein (vaccinia virus complement-control protein), can suppress the complement system, thereby partially evading the host’s immune surveillance in the bloodstream [38]. Moreover, tumor-associated blood vessels are typically characterized by increased permeability [39], which facilitates the penetration of relatively large rVV into tumor tissue following intravenous administration.

It has been previously reported that glial cells of the brain can also be susceptible to rVV infection without supporting active viral replication. Therefore, non-tumor microglial cells and astrocytes may become “rVV traps” — cells that accumulate the virus but lim it its spread [40]. Moreover, the susceptibility of glial cells to rVV appears to be modulated by their phenotype/activation state. For example, the M2 phenotype of monocytic macrophages (the so-called “pro-tumor” phenotype, typical for macrophages in the TME) is more favorable for viral replication compared to the “anti-tumor” M1 phenotype [40].

We tested in vivo whether the LIVP-mIL15-RFP strain primarily targets glioblastoma cells or is also distributed to glial cells. Although a marginal subpopulation of glial cells near the tumor tissue was also infected, LIVP-mIL15-RFP was mainly detected within the tumor mass, as evidenced by colocalization of RFP and GFP signals in intravital microscopy.

It was previously shown that CD45⁺ neutrophils are capable of accumulating viruses [41, 42]. We can hypothesize that the observed colocalization of reporter signals with CD45 is also due to active uptake of rVV by neutrophils.

The vaccinia virus has a double-stranded DNA genome (dsDNA), and its replication occurs in the cytoplasm. Therefore, when a rVV enters a cell, it activates cytosolic pathways for recognition of the host cell’s dsDNA [43–45]. This, in turn, leads to the release of type I interferons (IFN-I) and other signalling molecules, triggering an innate antiviral response. The first line of defense in the TME is the recruitment of NK cells — a subset of effector lymphocytes that play a key role in innate immunity. These cells recognize and eliminate malignant and virus-infected cells. Notably, NK cells appear to predominantly destroy glioblastoma stem cells (or glioblastoma stem cells are more susceptible to NK cell-mediated lysis) [46, 47]. The leading role of NK cells in the innate immune response to brain tumors is discussed in detail in [48], but it has its limitations. Within the TME, NK cells are capable of secreting immunomodulatory molecules (IFN-γ, CCL4, TNF-α) that regulate the recruitment and phenotype of many other types of immune cells.

In turn, as summarized in [49], the TME also regulates the phenotype and functions of NK cells, which may explain the limited success of NK cell-based therapies against solid tumors. In addition to responding to viral infection and performing immunomodulatory functions, NK cells are also a promising type of immune cell for the development of chimeric antigen receptor (CAR)-based antitumor therapy [50]. Data on the timing of NK cell recruitment to the glioblastoma microenvironment induced by a single dose of recombinant virus are largely lacking.

In our experiments, a notable increase in the NK cell subpopulation within the TME was observed 48 h after a single injection of LIVP-mIL15-RFP, which was likely driven by IL-15 expression in cells infected with LIVP-mIL15. It has previously been shown that NK cells can form and maintain “memory cell” subpopulations [51]. Therefore, it can be hypothesized that a second round of rVV injection may evoke an even more pronounced NK cell response. IL-15 is a cytokine that regulates the differentiation, activation, and survival of NK cells [52, 53]. Therefore, enhancing LIVP with IL-15 cDNA appears to us a promising approach for modulating the antitumor NK cell response.

Conclusion

The recombinant vaccinia virus LIVP-mIL15-RFP is capable of infecting GL261 cells in tumor spheroids, as well as intracranial glioma cells in vivo, both via stereotactic and intravenous administration routes. Infection with LIVP-mIL15-RFP — without causing massive detectable cell death of tumor cells — affects the immune composition of the tumor microenvironment by increasing infiltration with CD4⁺, CD8⁺, and NK cells. This may have a positive impact on antitumor cellular immunity.

Acknowledgements. The authors express their gratitude to Alexey Stepanenko for providing mouse glioblastoma GL261 cells (Neurobiology Department, V. P. Serbsky National Medical Research Centre for Psychiatry and Narcology, Moscow, Russia).

Study funding. This study was funded by a grant from the Russian Science Foundation (No.22–64-00057) and by the State Assignment of the Federal Medical-Biological Agency (FMBA) of Russia.

Conflict of interest. The authors declare no conflict of interest.