Delta-24-RGD, an Oncolytic Adenovirus, Increases Survival and Promotes Proinflammatory Immune Landscape Remodeling in Models of AT/RT and CNS-PNET

Brain tumors are the solid tumors most frequently affecting children under 15 years old and have a devastating impact on childhood mortality, as they contribute to nearly 30% of all cancer-related deaths in children (1). By histologic grouping, embryonal tumors comprise up to 24.9% of central nervous system (CNS) malignancies in infants under 1 year old, as well as 20.2% of those in children 1–4 years old (2). Embryonal CNS tumors include a heterogeneous group of highly cellular and mitotically active immature-appearing neoplasms that are able to invade surrounding tissues and disseminate through the cerebrospinal fluid (3, 4). Among these embryonal CNS tumors, in this study, atypical teratoid/rhabdoid tumors (AT/RT) and CNS primitive neuroectodermal tumors (CNS-PNET), two aggressive brain tumors that comprise nearly 30% of embryonal tumors in children, were the focus (1).

AT/RTs are highly malignant brain tumors that are characterized by biallelic loss of function of SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily b-member 1 (SMARCB1) or, more rarely, SMARCA4 (5–7). AT/RT peak incidence occurs during the first 2 years of life and represents the most common type of malignant CNS tumor in children under 1 year old (2, 8, 9). Despite intensive multimodal therapies, most patients with AT/RT inevitably experience recurrence, with a median survival time of 6–11 months (10, 11). Unlike AT/RTs, CNS-PNETs represent a group of heterogeneous diseases, as revealed by molecular profiling. While researchers are still exploring molecular markers to precisely classify these tumors, the World Health Organization has reclassified tumors that were defined previously as CNS-PNETs into a new group known as CNS embryonal tumors, not otherwise specified (12). Currently, most PNETs are treated with multimodal therapeutic protocols designed for high-risk medulloblastomas (13, 14), which induce serious side effects in these patients. The prognosis of patients affected by these embryonal tumors remains dismal; therefore, it is essential to further explore novel therapeutic approaches to improve the life expectancy and quality of life of these children.

Oncolytic virotherapy is a promising therapeutic strategy based on the administration of cancer-selective viruses (15). Delta-24-RGD is a conditionally replicative adenovirus that contains two modifications that endow it with tumor specificity (16, 17). Delta-24-RGD has shown promising therapeutic results in preclinical studies (17–20), leading to clinical trials for the treatment of ovarian cancer (21) and recurrent high-grade gliomas (NCT00805376 and NCT01956734; ref. 22). Previous results have revealed that intratumoral injection of Delta-24-RGD induces an initial phase of oncolysis, followed by a delayed inflammatory response with a reduction in the tumor burden (21, 23). Importantly, an ongoing clinical trial employing Delta-24-RGD as a therapeutic agent in children affected by diffuse intrinsic pontine gliomas (NCT03178032; refs. 24, 25) has shown no signs of toxicity, thus underscoring the feasibility of Delta-24-RGD for the treatment of pediatric brain tumors.

In this study, we explored the use of the oncolytic adenovirus, Delta-24-RGD, as a therapeutic agent in preclinical models of AT/RTs and CNS-PNETs in vitro and in vivo. We found that Delta-24-RGD replicated in and induced immunogenic cell death marker expression in AT/RT and PNET cultures in vitro. Furthermore, in vivo, Delta-24-RGD treatment resulted in extended overall survival in several murine models of these diseases. Moreover, Delta-24-RGD treatment exerted significant antitumor effects on a metastatic model of AT/RTs. Delta-24-RGD treatment of immunocompetent humanized mouse models bearing AT/RTs or PNETs also led to an increase in overall survival and triggered antitumor immune responses that involved tumor immune microenvironment remodeling. Taken together, these results underscore the therapeutic potential of Delta-24-RGD for treatment of local and metastatic AT/RTs and PNETs and provide a foundation for the use of Delta-24-RGD in future clinical trials targeting these embryonal tumors.

The public datasets used for this analysis were downloaded from the Gene Expression Omnibus data repository (http://www.ncbi.nlm.nih.gov/geo; RRID:SCR_005012). For PNET and normal brain samples, the normalized data matrix for GSE14295 experiments was downloaded. Raw data for AT/RT samples (GSE70678) were also downloaded and normalized using RMA (26). First, a filtering process was performed in R/Bioconductor (27). Genes with expression levels lower than the noise signal in more than 50% of the samples of all the studied conditions (fetal brains, adult brains, PNETs, and AT/RTs) were considered to be not expressed. Then, the relative expression abundance was calculated for each gene by gene expression standardization with a mean of 0 and a SD of 1.

A total of 2 × 10⁵ cells (6-well plates) were infected with Delta-24-RGD/GFP at a multiplicity of infections (MOI) of 0, 0.1, 1, 10, or 100 plaque-forming units (PFU)/cell. At 48-hour postinfection, transduction was verified by fluorescence microscopy, after which samples were harvested and washed with PBS. Then, the percentage of GFP⁺ cells was determined by flow cytometry (FACSCantoII) and analyzed with FlowJo V10 (BD Biosciences, RRID:SCR_008520).

A total of 3 × 10⁵ cells (24-well plates) were transfected (Fugene 6; E2691; Promega) with 250 ng of plasmid, E2F-1-Luc, expressing a firefly luciferase reporter under the control of an E2F-1–responsive promoter (28). In addition, 250 ng of pRL-cytomegalovirus (CMV), which constitutively expresses Renilla luciferase (E2261; Promega), was cotransfected as a transfection control. Twenty-four hours later, firefly and Renilla luciferase activities were measured using a Dual-Luciferase Reporter Assay System according to the manufacturer's instructions (E1910; Promega).

A total of 2 × 10⁵ PNET or AT/RT cells (6-well plates) were infected with Delta-24-RGD at an MOI of 0, 5, 10, 25, or 50 PFUs/cell. At 16 hours postinfection, the cultures were washed with PBS, and fresh growth medium was added. At 48 hours postinfection, the cells were harvested and washed with PBS. Proteins were extracted with PBS 0.1% SDS, and protein concentrations were measured with Protein Assay Dye Reagent (Bio-Rad Laboratories, RRID:SCR_008426). Then, the presence of viral proteins (E1A and fiber) was detected by Western blotting (see the Supplementary Materials and Methods for the detailed protocol).

Cell cultures (10⁵ cells/well; 6-well plates) were infected with 10 PFUs/cell of Delta-24-RGD. After 72 hours, complete cell cultures were collected and freeze–thawed three times, and total infectious titers were determined by serial dilutions in HEK293 cells by hexon staining (29).

Adherent cell lines were established as follows. We performed 3-day and 5-day kinetic studies to establish the most appropriate number of cells per 96-well plate, which were as follows: BT-12, HB, and PFSK-1: 2,000 cells; CHLA-266: 6,000 cells; and CHLA-06: 15,000 cells. Then, cultures were infected with Delta-24-RGD at different MOIs ranging from 0 to 50. Cell viability was measured 3 and 5 days later using the CellTiter 96 Aqueous One Solution Cell Proliferation Assay (G3581, Promega), as described previously (30).

Suspension cell lines were established as follows. Both JHU-CTX-GBMP1 (CTX) and BT-183 cells grown in suspension were plated at a density of 2 × 10⁵ cells per well in 6-well plates. At 3 and 5 days after infection with the same doses as above, cells were stained with trypan blue to enable cell counting. Dose–response curves were analyzed using GraphPad Prism 8 (Statistical Software for Sciences, RRID:SCR_002798) to determine the IC₅₀ value of Delta-24-RGD in these cells.

Each cell line was plated at a density of 2 × 10⁵ cells per well (6-well plates) and infected with Delta-24-RGD at the corresponding 3-day IC₅₀ value. Seventy-two hours later, the concentrations of the damage-associated molecular pattern (DAMP) markers Hsp90α (ADI-EKS-895, Enzo Life Sciences Inc.), HMGB1 (ST51011, IBL International), and ATP (ENLITEN ATP Assay System, FF2000, Promega) were measured in supernatants of infected and mock-infected cultures.

Calreticulin (CRT) translocation to the plasma membrane was determined by immunofluorescence staining of 2 × 10⁴ cells infected with Delta-24-RGD at a dose corresponding to the 3-day IC₅₀ value (or PBS as a negative control). At 4 hours postinfection, cells were fixed in 4% methanol-free formaldehyde (28906, Thermo Fisher Scientific) for 15 minutes at 37°C and then stained with appropriate antibodies. The extended immunofluorescence protocol is included in the Supplementary Materials and Methods.

In vivo experiments with PFSK-1 and JHU-HB-GBMP1 tumor xenograft models were carried out in 4- to 6-week-old female athymic nude (nu/nu) mice (Envigo). Studies of BT-183, BT-12, CHLA-06, and CHLA-266 xenograft models were performed in 4- to 6-week-old Balb/c-Rag2tFwa-Il2rg mice. In addition, hCD34-humanized mice (hu-CD34-NSG-SGM3, The Jackson Laboratory) were used to develop supratentorial PFSK-1 (10⁵ cells) and CHLA-06 (2 × 10⁶ cells) tumors in an immunocompetent environment.

Tumor cells and treatments were injected following the guide-screw system described by Lal and colleagues (31). For the orthotopic supratentorial model, CHLA-266 (2 × 10⁶), PFSK-1 (10⁵), or JHU-HB-GBMP1 (10⁶) cells were injected into the striatum at the following coordinates with respect to the bregma: 2.5 mm lateral, 1 mm cranial, and 2 mm deep (3.5 mm including screw height). The infratentorial xenograft model was established by injecting 5 × 10⁵ tumor cells (BT-12, CHLA-06, or CHLA-266) into the cerebellum: 1 mm lateral, 0.8 mm posterior, and 2 mm deep (3.5 mm including screw height), with respect to lambda. The disseminated model that we used was based on the intraventricular administration procedure described by Studebaker and colleagues (32). For this model, BT-12-GFP/luc (10⁶) cells were injected into the right lateral ventricle at 1 mm lateral, 0.5 mm posterior, and 3.2 mm deep with respect to the bregma. Mice were screened via bioimaging prior to being treated and were withdrawn in cases in which no reporter signal was detected at the ventricle.

Intratumoral treatments were carried out by injecting 3 μL of PBS or Delta-24-RGD following the same guide-screw system at the coordinates of the corresponding tumor model. In survival experiments, mice were euthanized when symptoms of disease (e.g., loss of weight and hunched position) were evident. A summary of the in vivo experiments performed is described in Supplementary Table S1.

Ethical approval for all animal studies was granted by the Animal Ethical Committee of the University of Navarra (CEEA, Pamplona, Navarra, Spain) under the protocols CEEA/091-18, CEEA/094-15, and CEEA/066-18. All animal studies were performed at the veterinary facilities of the Center for Applied Medical Research (Pamplona, Navarra, Spain) in accordance with institutional, regional, and national laws and ethical guidelines for experimental animal care.

Formalin-fixed, paraffin-embedded (FFPE) brain sections (4 μm thick) were stained by the hematoxylin and eosin (H&E) method or were immunostained as follows: anti-hexon (1:2,000, AB1056, Merck Millipore), anti-E1A (1:1,000, sc-430, Santa Cruz Biotechnology), anti-CD3 (1:300, RM9107, Thermo Fisher Scientific), anti-Iba-1 (1:4,000, 019-19741, Wako), and anti-GFP (1:1,000, ab6556, Abcam). Signals were developed with Vectastain ABC Kits (Vector Laboratories Inc.) according to the manufacturer's instructions.

A multiplex immunolabeling protocol based on tyramide signal amplification and Opal fluorophores was developed and validated essentially as described previously (33). Single-plex chromogenic IHC assays were used as the gold standard for cell antigen visualization and compared against the corresponding fluorescence channel for multiplexed immunofluorescence on sequential FFPE sections. An extended explanation is included in the Supplementary Materials and Methods.

Multiplexed immunofluorescence slides were scanned on a Vectra-Polaris Automated Quantitative Pathology Imaging System (Akoya Biosciences), as described previously (33, 34). An extended explanation is included in the Supplementary Materials and Methods.

Statistical analyses were performed using GraphPad Prism 8 (Statistical Software for Sciences). Dose–response curves for infectivity and viability were obtained by nonlinear regression. For quantitative variables, data with normal distributions were assessed by Shapiro–Wilk tests, and then comparisons among groups were performed using two-tailed (one-tailed when indicated) nonparametric tests with 95% confidence intervals (CI) for datasets that were nonnormally distributed (Mann–Whitney or Kruskal–Wallis tests) or parametric tests when normality was confirmed (Student t test or one/two-way ANOVA). Dunn (nonparametric) or Dunnett/Tukey (one- and two-way ANOVA, respectively) corrections were applied for post hoc pairwise comparisons. Kaplan–Meier plots were analyzed by log-rank (Mantel–Cox) tests.

Because infection by Delta-24-RGD is mediated by integrins and coxsackie and adenovirus receptor (CAR; refs. 16, 35, 36), we first assessed the relative expression levels of α_v integrin (ITGAV), β₃ integrin (ITGB3), β₅ integrin (ITGB5), and CAR (CXADR) in silico in AT/RT and PNET samples, as well as in normal fetal and adult brain samples (Fig. 1A). The expression levels of α_v integrins and CAR were observed in both AT/RT and PNET samples, and their relative abundances were above the mean expression of their respective transcriptomes. In contrast, normal fetal brain samples showed lower relative mRNA expression of α_v integrin. In terms of the β₅ integrin and β₃ integrin genes, their relative abundances in PNET samples were similar to those in normal brain samples and were lower than those in AT/RT samples. We obtained similar results when we performed analyses that accounted for the different molecular subgroups of AT/RTs (Supplementary Fig. S1A). Collectively, these in silico findings revealed that AT/RTs and PNETs express receptors for Delta-24-RGD. Confirming the potential susceptibilities of AT/RTs and PNETs to Delta-24-RGD infection, AT/RT and PNET cell lines displayed robust CAR and integrin β₅ expression levels (Fig. 1B). In accordance with the gene expression of β₃ integrin in tumor samples, this receptor was nearly absent in the four cell lines evaluated. Finally, to determine whether the expression of these viral receptors enables endocytosis of the oncolytic virus, Delta-24-RGD, AT/RT and PNET cell lines were infected with Delta-24-RGD expressing the reporter GFP at increasing MOIs ranging from 0.1 to 100 PFUs/cell. We observed dose-dependent expression of GFP in AT/RT and PNET cultures (Fig. 1C). More than 80% of AT/RT and PNET cells were infected at an MOI of 1, and virtually 100% infection was achieved with an MOI of 10 (Fig. 1D). In fact, cells infected at an MOI of 100 were already detached and dying at 48 hours (Fig. 1C). The MOI needed to yield a 50% infection rate was below 0.1 PFUs/cell for BT-12 and CHLA-06 cells, 0.18 PFUs/cell for CHLA-266 cells, and 0.23 PFUs/cells for PFSK-1 cells (Fig. 1D).

Because the heterogeneity and scarcity of CNS-PNET models are major drawbacks for conducting preclinical studies, we next studied the oncolytic effects of Delta-24-RGD on other “PNET-like” tumors to cover a broader range of embryonal tumors, excluding medulloblastomas. Therefore, we assessed the expression levels of viral receptors in the BT-183 cell line, which was derived from an embryonal tumor with multilayered rosettes (ETMR; ref. 37), and the CTX and JHU-HB-GBMP1 (HB) cell lines, which correspond to models of glioblastomas (GBM) with PNET-like components (i.e., currently known as GBMs with primitive neuronal components). CAR expression was detected in the ETMR cell line, BT-183 (Supplementary Fig. S1B), as well as in GBM-PNET–like cultures (Supplementary Fig. S1B). Similar to our results for AT/RT and PNET cultures, the expression levels of α_vβ₃ integrin in ETMR cells and GBM-PNET–like cells were negligible, while α_vβ₅ integrin was detected in the three cell lines tested. Overall, the expression of adenoviral receptors was lower in these models, which is in agreement with their higher resistance to adenoviral infection (Supplementary Fig. S1C). According to these data, we concluded that AT/RT and PNET cell lines could be readily infected by Delta-24-RGD.

Delta-24-RGD replication in tumor cells requires a constitutively active E2F pathway. Compared with those in normal human adult and fetal brain samples, the relative E2F-1 levels in PNET and AT/RT samples were higher, indicating that these tumor cells may be permissive to Delta-24-RGD replication (Fig. 2A). The expression levels of other positive cell-cycle regulators, such as CCND1 and CDK4/6, were enriched in different AT/RT molecular subgroups (Supplementary Fig. S2A). In addition, AT/RT cell lines displayed deletion of the tumor suppressor gene, SMARCB1 (Supplementary Table S2), and the PFSK-1 cell line harbored p53 mutations; both aberrations resulted in cell-cycle deregulation. Regarding direct alterations in the pRb-E2F pathway, only BT-12 cells showed loss of the CDKN2A and CDKN2B genes encoding the CDK4/6 inhibitors, p16^INK4a and p15^INK4b, respectively. Corroborating this finding, we found a significant increase in E2F-1 promoter activity in PNET and AT/RT cell lines compared with normal astrocytes (Fig. 2B). This result demonstrates that although the pRb-E2F pathway is not directly affected at the genomic level in some tumors, other epigenetic mechanisms may upregulate E2F-1 activation, thus suggesting that these tumor cells are potential targets for Delta-24-RGD replication.

We next assessed viral cycle progression in AT/RT and PNET cultures. The E1A protein, which is an adenoviral protein that is transcribed immediately after infection, was detected in all infected AT/RT and PNET cultures at 16 and 48 hours after infection (Supplementary Fig. S2B). Similarly, fiber, a protein that is expressed after viral genome replication, was detected at 48 hours, but not at 16 hours, after infection (Supplementary Fig. S2B). The viral titers in all infected AT/RT and PNET cultures were significantly increased at 72 hours after infection (Fig. 2C; Supplementary Fig. S2C). These results confirm that Delta-24-RGD undergoes a replicative cycle in AT/RT, PNET, and PNET-like cells, although the viral burst size is dependent on the specific cellular model.

Next, we assessed the antitumor effects of Delta-24-RGD on these in vitro models (Fig. 2D; Supplementary Fig. S2D). Dose-dependent cytolysis was observed at 3 days after infection in all cultures, which was in agreement with the generation of infectious viral particles observed at 72 hours after infection. At 5 days postinfection, the initial dose required to reach cell death in 50% of a culture dropped dramatically to 1 PFU/cell or less due to the increase in infection/replication/cytolysis rounds.

In addition to its cytolytic activity, Delta-24-RGD also promotes a proinflammatory environment at tumor sites by boosting antitumor immune responses (20, 22). One of the mechanisms that participates in the enhancement of immune responses is the secretion of DAMPs into the extracellular medium by infected tumor cells. We observed significant increases in levels of secreted DAMPs, Hsp90α and HMGB1, following Delta-24-RGD infection in comparison with mock infection in all tested cell lines (Fig. 2E). In contrast, Delta-24-RGD infection did not significantly alter ATP concentrations in any cell line. Translocation of the endoplasmic reticulum–resident protein, CRT, to the cell surface of tumor cells is another well-known mechanism that contributes to the immunogenicity of tumor cells by acting as an “eat me” signal to promote tumor cell phagocytosis by macrophages (38). Early Delta-24-RGD infection triggered translocation of CRT to the plasma membrane (Fig. 2F). Thus, we concluded that Delta-24-RGD induces antitumor effects in vitro.

Next, we investigated the therapeutic potential of Delta-24-RGD in mouse models of PNETs and AT/RTs. To this end, mice bearing orthotopically injected PFSK-1 cells were treated with a single intratumoral injection of Delta-24-RGD (10⁷ or 10⁸ PFUs/mouse) or PBS (mock-treated group). The mice treated with the lowest dose of Delta-24-RGD did not show a significant increase in median survival; however, this treatment yielded 22% long-term survivors (2/9 mice) free of disease. The mice treated with the highest dose of Delta-24-RGD showed a significant increase in median survival, and 56% of the mice were long-term survivors (Fig. 3A). The long-term survivors were euthanized on day 97 after infection, and no visible tumor lesions were detected via H&E staining (Supplementary Fig. S3A). Histologic analyses performed at 14 days after injection revealed the presence of highly vascularized tumor masses spanning most of the right brain hemisphere in mock-treated mice. E1A and hexon staining in Delta-24-RGD–treated mice demonstrated in vivo infection and replication, respectively (Fig. 3B). Furthermore, activated microglia were found at a high density at the tumor edge, labeled with the marker Iba-1, and indicated tumor infiltration by reactive amoeboid microglia, which was in contrast with the resting microglia (stellate shape) present in most of the normal parenchyma (Fig. 3B). In addition, in Delta-24-RGD–treated mice, the highest degree of tumor infiltration by microglia was detected in areas that matched the infected areas, thus indicating that the virus induced substantial changes in the immune landscape surrounding the tumor microenvironment.

To determine whether the efficacy of this virus may be relevant to a broad range of different types of tumors in vivo, we next assessed the effects of Delta-24-RGD on PNET-like mouse models with primitive features, such as ETMRs and GBMs with PNET-like components. E1A and hexon staining confirmed that Delta-24-RGD infected and replicated in ETMR-induced tumors (BT-183 cell lines; Supplementary Fig. S3B). However, unlike in mice with PFSK-1 cell–induced PNETs, ETMRs showed no effects of treatment with Delta-24-RGD on tumor growth or infiltration of Iba-1–positive cells (microglia/macrophages). Unexpectedly, although replication of Delta-24-RGD was observed both in vitro and in vivo, survival curve comparisons showed no therapeutic benefit of Delta-24-RGD at the doses used to treat ETMR-induced tumors in mice (Supplementary Fig. S3C). Similar experiments were performed with the GBM PNET-like cell line, HB. Histologic analyses of HB tumors at 14 days after Delta-24-RGD treatment showed the presence of E1A and hexon proteins at the tumor site in the Delta-24-RGD–treated mice (Supplementary Fig. S3D). In this model, mice treated with the highest Delta-24-RGD dose displayed a significant increase in overall survival compared with mock-treated mice (Supplementary Fig. S3E).

Next, we assessed the antitumor effects of Delta-24-RGD on several AT/RT infratentorial models (CHLA-06, BT-12, and CHL-266 cells) following the protocol described above. At the highest dose, Delta-24-RGD treatment led to significant increases in median survival and long-term survival in the three models. The increase in median survival ranged from 10 days for CHL-06 tumors to 45 days for BT-12 and CHLA-266 tumors (Fig. 3C). The lowest dose tested also showed a significant improvement in the median survival of mice bearing CHL-06 tumors (13.5-day increase). Delta-24-RGD infection and replication were detected in tumor masses by E1A and hexon staining, respectively. Similar to PFSK-1 tumors, CHLA-06 tumors treated with Delta-24-RGD showed accumulation of microglia at the tumor rim and infiltration of reactive microglia, which were exacerbated in the Delta-24-RGD–containing regions of the tumors (Fig. 3D). Furthermore, the presence of the adenoviral proteins, E1A and hexon, was detected in tumor masses nearly 100 days after injection of Delta-24-RGD (Supplementary Fig. S4A and S4B), indicating that replication of the virus lasted as long as the time course of tumor cell propagation.

Because AT/RTs can also arise at supratentorial locations, CHLA-266 cells were engrafted orthotopically in the striatum (Fig. 3E). In this model, we observed a significant increase in the median survival of Delta-24-RGD–treated mice (39 days), and 29% of these mice (2/7) were long-term survivors with no symptoms of disease. These data underscore the therapeutic benefit of Delta-24-RGD in preclinical mouse models of localized AT/RTs in vivo.

Next, we addressed whether treatment with Delta-24-RGD could be effective in established tumors, which better recapitulate clinical conditions in human patients. Thus, we first established the time frames in which mice bearing PFSK-1 or CHLA-266 cells (supratentorial PNET and AT/RT models, respectively), as well as those bearing CHLA-06 cells (infratentorial AT/RT model), displayed visible tumors (Supplementary Fig. S4C). PFSK-1 and CHLA-06 tumor–bearing mice were treated 7 and 8 days after cell implantation, respectively, while treatment of CHLA-266 tumor–bearing mice was delayed to 29 days. Of importance, Delta-24-RGD treatment at the highest dose resulted in a significant increase in survival and led to long-term survival in the three models assessed. Moreover, the lowest dose also showed a significant benefit in the PFSK-1 and CHL266 models (Fig. 3F). Anatomopathologic analysis showed the capacity of the virus to spread through the tumors (Supplementary Fig. S4D). In addition, these analyses revealed ongoing infection and replication in the tumor mass (E1A and hexon staining, respectively), as well as a high degree of reactive microglial infiltration (Supplementary Fig. S4E). Furthermore, we observed an enrichment of reactive microglia at viral replication foci. These data indicate that virus treatment results in microglial activation.

Because the presence of disseminated disease is a major hurdle in the efficacy of antitumor therapies in patients with AT/RT (39), we next evaluated the efficacy of Delta-24-RGD in a preclinical mouse model of this condition. Luminometry and histology in pilot studies confirmed that the BT-12 cell line was able to generate disseminated lesions upon intraventricular injection, as we observed luciferase signals at distal regions in the spinal cord and the presence of extracranial tumor lesions (Fig. 4A). In vivo experiments showed that by day 26, 58% of mock-treated mice developed secondary tumors (Fig. 4B). Bioluminescence images captured at early stages (days 11 and 22) revealed that Delta-24-RGD–treated mice displayed a significantly lower tumor burden than mock-treated mice (Fig. 4C–E). Regarding the efficacy of Delta-24-RGD in this disseminated model, administration of the lower dose (10⁷ PFUs) increased median survival by 21.5 days and produced two long-term survivor animals; however, the difference was not significant. In contrast, treatment with the highest Delta-24-RGD dose (10⁸ PFUs) significantly increased median survival compared with PBS-treated mice and led to 70% of the mice becoming long-term survivors (Fig. 4F). Anatomopathologic analysis of the spinal cord of long-term survivors revealed that they were free of disease (Supplementary Fig. S4F). These data underscore the antitumor effect of Delta-24-RGD even in the context of disseminated disease.

Previous preclinical and clinical data have provided evidence that Delta-24-RGD enhances the antitumor immune response in brain tumors (20, 22, 40). Therefore, we evaluated the effects of Delta-24-RGD in an immunocompetent background using NSG-SMG3 mice humanized with hCD34⁺ progenitor cells that recapitulate both the innate and adaptive branches of the immune system (41). HLA typing was performed in PNET and AT/RT cell lines with the aim of choosing a donor with at least HLA-A matching (Supplementary Materials and Methods).

Survival studies demonstrated that treatment of mice bearing PFSK-1 or CHLA-06 cells with a single injection of Delta-24-RGD (10⁸ PFUs) resulted in a significant increase in overall survival compared with mock treatment (28 vs. 55 days for PFSK-1 cells and 23 vs. 34 days for CHLA-06 cells; Fig. 5A and B). Moreover, one animal survived the PFSK-1 challenge at the end of the study (Fig. 5A). Assessment of mouse body weight throughout the experiment revealed no signs of toxicity associated with virus-mediated inflammation induced by Delta-24-RGD (Supplementary Fig. S5A).

Histologic analyses of brain samples from CHLA-06 tumor–bearing mice indicated ongoing viral infection and replication in Delta-24-RGD–treated mice (Fig. 5C). Multispectral immunofluorescence panels of different immune populations revealed that most immune cells accumulated in the tumor-invasive margin (Fig. 5C; Supplementary Fig. S5B) and in Delta-24-RGD–containing regions (Fig. 5C). In both Delta-24-RGD–treated mice and mock-treated mice, the presence of CD20⁺ B cells was nearly negligible (Fig. 5D). Furthermore, Delta-24-RGD also induced reductions in the frequency of CD68⁺ cells (macrophages) in the tumor and at the invasive margin, although the reduction was less pronounced at the latter site. No differences were observed in terms of the total CD11b⁺ myeloid population, although the antibody that was used also detects the murine orthologue; thus, this particular assessment may have been hampered by the high abundance of mouse microglia. Interestingly, even though no differences were observed between the groups in regard to total T cells (CD3⁺), there was a remarkable increase in the CD8⁺ subpopulation in the Delta-24-RGD–treated group, which was nearly absent in the mock-treated group. To validate our results, the multispectral immunofluorescence results were confirmed by conventional IHC (Supplementary Fig. S5C).

Surprisingly, PFSK-1 tumors showed no evidence of viral infection or T-cell infiltration despite observation of clear therapeutic effects (Fig. 5E). Nevertheless, analysis of the macrophage/microglial marker, Iba-1, revealed pronounced activation and recruitment of these cells at the tumor margin in Delta-24-RGD–treated mice.

Finally, survival experiments were performed for an ETMR model (BT-183) established with humanized mice. However, although no signs of toxicity were observed (Supplementary Fig. S5D), Delta-24-RGD showed no therapeutic benefit (Supplementary Fig. S5E), and there was no T-cell infiltration despite evidence of replication of Delta-24-RGD (Supplementary Fig. S5F).

Although the outcomes of the different models were variable, these results are encouraging and may reflect the heterogeneity of the type of tumor, donor, and immune response (T-cell responses vs. predominant macrophage activation).

Potential tumor targets of Delta-24-RGD must be permissive to viral infection and replication. We confirmed previously the expression of adenoviral receptors and the transcription factor, E2F-1, in gene expression datasets from tumor biopsies (42, 43), providing direct translational relevance. Unlike the common features of other brain tumors, canonical oncogenic mutations in components of the pRb-E2F pathway, such as CDK4/CCND1 amplification and CDKN2A/B deletion, are uncommon in CNS-PNETs (44, 45) and are nearly absent in AT/RTs (46, 47). Nonetheless, epigenetic deregulation mediated by loss of SMARCB1 in AT/RTs, as well as loss of p53 and amplification of different tyrosine kinases, results in aberrant cell-cycle progression that allows viral replication.

Although virotherapy has been evaluated in embryonal tumors, mainly medulloblastomas (48–51), there is a paucity of preclinical studies investigating these types of tumors. Nonetheless, an oncolytic measles virus was found to have therapeutic effects on localized and disseminated models of AT/RTs (32). Another oncolytic adenovirus, VCN-01 (armed with hyaluronidases), was demonstrated to be efficacious in a model of supratentorial PNETs (52). These previous findings together with our current results underscore the potential of virotherapies and the need to translate them to clinical trials.

Despite our use of immunodeficient mice in this study, our in vivo models still developed immune responses in the myeloid compartments, and we observed that Delta-24-RGD triggered tumor infiltration of amoeboid reactive microglia, thus underscoring the ability of this virus to “warm” the tumor microenvironment to a proinflammatory status, which might contribute to enhancing antitumor immune responses (53). In our humanized mouse AT/RT model, we observed a decrease in the frequency of CD68⁺ cells after Delta-24-RGD treatment, which has been proposed to negatively impact the survival of patients with MYC-AT/RTs (54). Therefore, we speculate that Delta-24-RGD may have a direct effect on this population, inducing remodeling of the tumor microenvironment and stimulating the expansion and activity of CD8⁺ CTLs to exert their antitumor effects. Interestingly, in the PFSK-1–humanized model, although we could not detect viral replication or T-cell infiltration, we observed the existence of reactive glia. These data are in agreement with a potential antitumor immune response. However, it might be that at 15 days, the immune response was already fading. In addition, we cannot rule out the possibility that other mechanisms were taking place (i.e., early innate response and cytotoxicity), and further studies are needed to test this hypothesis.

One of the limitations of this study was the paucity of preclinical models. First, the availability of well-characterized human-derived CNS-PNET models is scarce; therefore, tumor models do not recapitulate the variability found in patients. Second, human adenoviruses do not exhibit high levels of replication in murine cells (55), so the oncolytic and immunostimulatory properties of Delta-24-RGD must be evaluated independently using human xenografts and syngeneic models, respectively (20). Transgenic mouse models have already been developed for spontaneous CNS-PNETs, AT/RTs, and ETMRs (56–58); however, these mice suffer from severe health conditions, and spontaneous tumor models are difficult to treat via intratumoral injections. The development of stable cell cultures from these tumors may be a desirable approach to overcome these difficulties. In this study, we used humanized mice as an approach to evaluate the combined effects of viral replication and immunomodulation because such mice allow the study of xenografted human tumors in an immunocompetent environment. However, this model is per se complex and presents with intrinsic limitations, such as the age of the mice used in the analyses. Because of the engraftment process of human CD34⁺ cells, these mice are already adults by the time they can be used in experiments (∼15 weeks) and are, therefore, not an ideal model for trying to assess the immune system of children. Another important issue is the HLA matching of the immune system of donor and tumor. Therefore, caution should be exerted when making definitive conclusions, and further studies are required to understand this process in depth.

In summary, we demonstrated that treatment with the oncolytic adenovirus, Delta-24-RGD, is a potential therapeutic approach for AT/RTs and PNETs. Our data showed that virus administration was safe and effective in controlling local, both supratentorial and infratentorial, advanced, and disseminated tumors in models of AT/RTs and CNS-PNETs. These encouraging data open the door for oncolytic viruses to be considered serious candidates in the treatment of pediatric brain tumors and strongly support initiating clinical studies.

S. Labiano reports personal fees from Roche Pharma Research & Early Development outside the submitted work. J.A. Chan reports grants from Kids Cancer Care Foundation of Alberta/Alexander's Quest fund and Alberta Children's Hospital Foundation during the conduct of the study. C. Gomez-Manzano reports grants from Department of Defense during the conduct of the study, other from DNAtrix, Inc. outside the submitted work, as well as has a patent covering Delta-24-RGD to DNAtrix. J. Fueyo reports grants from Department of Defense during the conduct of the study, other from DNAtrix, Inc. outside the submitted work, as well as has a patent for IP for Delta-24-RGD licensed to DNAtrix, Inc. M.M. Alonso reports grants from DNAtrix outside the submitted work. No disclosures were reported by the other authors.

M. Garcia-Moure: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. M. Gonzalez-Huarriz: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing-review and editing. S. Labiano: Data curation, formal analysis, investigation, visualization, methodology, writing-review and editing. E. Guruceaga: Data curation, formal analysis, investigation, visualization, methodology, writing-review and editing. E. Bandres: Data curation, investigation, methodology, writing-review and editing. M. Zalacain: Data curation, investigation, methodology, writing-review and editing. L. Marrodan: Data curation, investigation, methodology, writing-review and editing. C. de Andrea: Data curation, formal analysis, investigation, visualization, methodology, writing-review and editing. M. Villalba: Data curation, formal analysis, investigation, visualization, methodology, writing-review and editing. N. Martinez-Velez: Data curation, investigation, methodology, writing-review and editing. V. Laspidea: Data curation, investigation, methodology, writing-review and editing. M. Puigdelloses: Data curation, investigation, methodology, writing-review and editing. J. Gallego Perez-Larraya: Data curation, investigation, methodology, writing-review and editing. I. Iñigo-Marco: Data curation, investigation, methodology, writing-review and editing. R. Stripecke: Data curation, investigation, methodology, writing-review and editing. J.A. Chan: Data curation, investigation, methodology, writing-review and editing. E.H. Raabe: Data curation, investigation, methodology, writing-review and editing. M. Kool: Data curation, formal analysis, investigation, methodology, writing-review and editing. C. Gomez-Manzano: Data curation, formal analysis, investigation, methodology, writing-review and editing. J. Fueyo: Data curation, formal analysis, investigation, methodology, writing-review and editing. A. Patiño-García: Data curation, formal analysis, investigation, methodology, writing-review and editing. M.M. Alonso: Conceptualization, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

We thank Nature Authors Services for its linguistic assistance during the preparation of this article. The performed work was supported through the Departamento de Salud del Gobierno de Navarra (54/2018-APG), Predoctoral Fellowship from Gobierno de Navarra (to V. Laspidea), Instituto de Salud Carlos III y Fondos Feder (PI19/01896 MMA, PI18/00164 to A. Patiño-García), Amigos de la Universidad de Navarra (to M. Puigdelloses), Fundación La Caixa/Caja Navarra (to A. Patiño-García and M.M. Alonso), Fundación El sueño de Vicky, Asociación Pablo Ugarte-FuerzaJulen (to A. Patiño-García and M.M. Alonso), and Department of Defense (DOD) Team Science Award under grant (CA 160525 MMA to C. Gomez-Manzano and J. Fueyo). This project also received funding from the European Research Council under the European Union's Horizon 2020 Research and Innovation Programme (817884 ViroPedTher to M.M. Alonso).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.