Diffuse intrinsic pontine glioma (DIPG) bears a dismal prognosis. A genetically engineered brainstem glioma model harboring the recurrent DIPG mutation, Activin A receptor type I (ACVR1)-G328V (mACVR1), was developed for testing an immune-stimulatory gene therapy.
We utilized the Sleeping Beauty transposase system to generate an endogenous mouse model of mACVR1 brainstem glioma. Histology was used to characterize and validate the model. We performed RNA-sequencing analysis on neurospheres harboring mACVR1. mACVR1 neurospheres were implanted into the pons of immune-competent mice to test the therapeutic efficacy and toxicity of immune-stimulatory gene therapy using adenoviruses expressing thymidine kinase (TK) and fms-like tyrosine kinase 3 ligand (Flt3L). mACVR1 neurospheres expressing the surrogate tumor antigen ovalbumin were generated to investigate whether TK/Flt3L treatment induces the recruitment of tumor antigen–specific T cells.
Histologic analysis of mACVR1 tumors indicates that they are localized in the brainstem and have increased downstream signaling of bone morphogenetic pathway as demonstrated by increased phospho-smad1/5 and Id1 levels. Transcriptome analysis of mACVR1 neurosphere identified an increase in the TGFβ signaling pathway and the regulation of cell differentiation. Adenoviral delivery of TK/Flt3L in mice bearing brainstem gliomas resulted in antitumor immunity, recruitment of antitumor-specific T cells, and increased median survival (MS).
This study provides insights into the phenotype and function of the tumor immune microenvironment in a mouse model of brainstem glioma harboring mACVR1. Immune-stimulatory gene therapy targeting the hosts' antitumor immune response inhibits tumor progression and increases MS of mice bearing mACVR1 tumors.
The therapeutic efficacy of anti-diffuse intrinsic pontine glioma (DIPG) immune responses is limited because of a low number of immune cells in the tumor microenvironment. We have uncovered a novel treatment strategy that utilizes adenoviral delivery of therapeutic genes, thymidine kinase (TK) and fms tyrosine kinase 3 ligand (Flt3L), into the tumor eliciting a reprograming of the host's own immune system to recognize and kill tumor cells. We demonstrate that TK/Flt3L therapy generates an effective antitumor response and can be safely delivered into the brainstem. This treatment approach could provide a novel translational approach toward potentiating an anti-DIPG immune response to overcome the current limitations in the treatment of patients with DIPG.
Diffuse intrinsic pontine glioma (DIPG) is an aggressive, diffusive brain tumor that originates in the pons and is diagnosed on the basis of clinical and radiological criteria (1). DIPG occurs mostly in children (median age, 6–7 years). It is an inoperable brain tumor and despite multiple trials testing various chemotherapeutic strategies, none has demonstrated a survival benefit. Thus, focal radiation remains the standard of care (1, 2). The median overall survival for DIPG is 10.8 months, and the 2-year survival is approximately 5.2% (3). In 2014, four independent studies that comprised data from 195 DIPGs, identified Activin A receptor type I (ACVR1) as the most recurrently mutated gene following the histone H3 K27M mutation (4–7). Six recurrent, somatic mutations in ACVR1 were found in 24% of DIPG cases (8). The characteristics of patients with DIPG presenting with the ACVR1 mutation include younger age of onset (∼5 years of age) and longer overall survival time (4, 9). Sequencing revealed that DIPG tumors have a lower mutation frequency compared with adult glioblastoma (GBM), and that classical tumor pathways such as the PI3K and p53 pathways are also altered in these tumors (10).
ACVR1 encodes a type 1 serine/threonine kinase that is part of the TGFβ family. It is noteworthy to mention that ACVR1 mutations are only found in DIPG tumors; ACVR1 is not mutated in any other cancers. Thus, uncovering the role of this gene mutation in DIPG tumorigenesis and disease progression could provide novel mechanistic insights useful to develop novel therapeutic targets.
Because DIPGs are not resectable and they are highly invasive, it would be advantageous to harness the power of the immune system to elicit effective antitumor immunity in patients with DIPG. Immune-mediated treatment modalities have yielded promising clinical benefits in melanoma, non–small cell lung cancer, renal cell cancer, and prostate cancer (11–14). We have also previously shown the efficacy of an immune-stimulatory gene therapy approach in several rat and mouse models of adult GBM (15, 16). This approach has recently completed its phase I clinical trial accrual for the treatment of adult patients with newly diagnosed glioblastoma multiforme [World Health Organization grade 4 (NCT01811992)]. This immune-mediated gene therapy approach is based on adenoviral delivery of herpes simplex virus type 1-thymidine kinase (TK) and Fms-like tyrosine kinase 3 ligand (Flt3L). Upon administration of the prodrug, ganciclovir, proliferating tumor cells expressing TK undergo immunogenic cell death (15, 16). Dying tumor cells release damage-associated molecular patterns (DAMPs), such as high-mobility group B1 protein (HMGB1), calreticulin, and ATP (15, 17). Meanwhile, Flt3L elicits the recruitment of dendritic cells to the tumor microenvironment (15). HMGB1 released by dying tumor cells activates dendritic cells through Toll-like receptor 2 (TLR2)-mediated signaling (15). Activated dendritic cells pick up the tumor antigens and traffic to the draining lymph nodes, where they generate a specific antitumor cytotoxic T-cell response (15, 18). Herein, we aimed to test the efficacy of this immune-stimulatory approach in an immunocompetent mouse model of mutant ACVR1 (mACVR1) brainstem glioma.
To test the efficacy of this immune-mediated gene therapy approach in syngeneic mouse brainstem glioma models, we developed a genetically engineered mouse model encoding mACVR1 using the Sleeping Beauty (SB) transposase system (19–22). SB is a transposase that is able to recognize inverted repeats/direct repeats sites on DNA transposons and carry out a cut-and-paste reaction to integrate transposon DNA into a host genome (23). The SB system can be utilized to generate endogenous tumors that resemble gliomas through delivery of DNA transposons that encode oncogenes and tumor suppressors (19–22, 24). Transcriptome analysis of ACVR1-mutant neurospheres identified an increase in the TGFβ signaling pathway and signaling pathways regulating the pluripotency of stem cells, and a decrease in the focal adhesion pathway. Treatment with TK/Flt3L immune-stimulatory gene therapy significantly improved median survival (MS) compared with standard of care. The TK/Flt3L gene therapy induced a strong antitumor cytotoxic immune response demonstrated by an increase in the frequency of tumor antigen–specific CD8 T cells in mice treated with TK/Flt3L therapy when compared with saline controls. Our results suggest that immune-mediated gene therapy could be a promising therapeutic approach for DIPG.
Materials and Methods
All animal studies were conducted according to guidelines approved by the Institutional Animal Care and Use Committee at the University of Michigan (Ann Arbor, MI). Animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited animal facility and had constant access to food and water; they were monitored daily for tumor burden. Males and females were used. The strain of mice utilized in the study was C57BL/6 (Jackson Laboratory, 000664).
A murine model of brainstem glioma was generated by employing the SB transposon system to integrate plasmid DNA into the genome of postnatal day 1 (P1) mouse pups. The plasmids utilized were as follows: (i) SB transposase and luciferase (pT2C-LucPGK-SB100X, henceforth referred to as SB/Luc), (ii) a short hairpin against p53 (pT2-shp53-GFP4, henceforth referred to as shp53) or shp53-NO-GFP, and (iii) a constitutively active mutant of NRAS (pT2CAG-NRASV12, henceforth referred to as NRAS) with or without (4) mACVR1 G328V (pkt-ACVR1-G328V-IRES-Katushka; henceforth referred to as mACVR1; refs. 19, 20, 22). To create the mACVR1 plasmid, we cloned pCMV5-ALK2-WT into pKT2-IRES-Katushka by blunt cloning. Then, we used the QuikChange II Site-Directed Mutagenesis Kit (Agilent, 200523) to introduce (c.983G>T, p.Gly328Val) mutation into pKT2-ALK2-WT-IRES-Katushka, to generate pKT2-ACVR1-G328V-IRES-Katushka (Addgene plasmid #77437). The resultant mACVR1 plasmid was confirmed by Sanger sequencing. SB/Luc, shp53, and NRAS plasmids were the generous gift of Dr. John Ohlfest (University of Minnesota, Minneapolis, MN; now deceased). The pCMV5-ALK2-WT plasmid was a generous gift from Jeff Wrana (University of Toronto in the Department of Molecular Genetics) (Addgene plasmid #11741). All experiments were performed using P1 or P2 wild-type C57BL/6 mice. The plasmid combinations injected were as follows: (ref. 1; i) shp53 and NRAS (henceforth referred to as wt-ACVR1) and (ii) shp53, NRAS, and ACVR1m. Mice were injected according to a previously described protocol (20) and described in detail in the Supplementary Materials and Methods.
IHC of paraffin-embedded brains
IHC staining was performed as described previously (20, 22) and detailed in the Supplementary Materials and Methods. Antibody information is available in Supplementary Table S1.
Mouse neurospheres were generated from tumors that were developed using the SB system by injection of the following plasmid combinations (ref. 1; i) shp53 and NRAS, or (ii) shp53, NRAS, and ACVR1m into the lateral ventricle (1.5-mm AP, 0.7 mm lateral, and 1.5 mm deep from the λ-suture) following previously described protocols and detailed in the Supplementary Materials and Methods (20–22, 25). Western blot analysis and inhibitor treatment with LDN-214117 is also detailed in the Supplementary Materials and Methods and antibody information is available in Supplementary Table S1. SU-DIPG-VI and SU-DIPG-XX1 were obtained from Dr. Michelle Monje (Stanford University, Stanford, CA) in accordance with an institutionally approved protocol at each institution. Experimental details of immunocytochemistry on DIPG cultures are available in the Supplementary Materials and Methods.
RNA sequencing was performed in collaboration with the University of Michigan Sequencing Core. Total RNA was isolated from tumor neurosphere using the RNeasy Plus Mini Kit (Qiagen, 74134) and 100 ng of purified RNA was sent for analysis. The libraries were prepared using RiboGone (Takara, 634836) and TruSeq Stranded Total RNA Human/Mouse/Rat (Illumina, 20020596) with 100 ng of input and 13 PCR cycles. Sequencing was performed by the UM DNA Sequencing Core, using the Illumina Hi-Seq platform. RNA-sequencing analysis is available in the Supplementary Materials and Methods.
Implantable syngeneic murine brainstem glioma models
Female C57BL/6 mice between 6–8 weeks were used for all implantation experiments. Intracranial tumors were generated by stereotaxic injection of 1,000 ACVR1m tumor neurospheres into the pons using a 5-μL Hamilton syringe with a removable 33-gauge needle with the following coordinates: (0.8 mm posterior; 1.00 mm lateral to the λ-suture; and 5 mm deep). Animals were anesthetized, then the skin over the incision site was cut and retracted, and a burr hole was drilled into one side of the skull using a 0.45-mm drill bit corresponding to the pons coordinates. Tumor neurospheres were delivered in a 2-μL volume after holding the needle in place for 2 minutes. Each injection was performed over the course of 7 minutes; the needle was left in place for an additional minute before being slowly withdrawn from the brain. Intratumoral injection of adenoviral vectors and radiation treatment, flow cytometry, T-cell proliferation analysis, and the cytotoxic T-cell assay are detailed in the Supplementary Materials and Methods. Details of complete blood cell count and serum chemistry analysis, neuropathologic analysis, and hematoxylin and eosin (H&E) staining of liver sections are elaborated in the Supplementary Materials and Methods.
All data were analyzed using GraphPad Prism version 8, or R (version 3.1.3). All animal studies were carried out with at least three animals per group (specified in each experiment). The statistical test used is indicated in each figure. A P ≤ 0.05 was considered significant.
Brainstem gliomas induced by transforming neural progenitor cells using the SB transposase system
To assess the impact of activating ACVR1 mutations in brainstem glioma pathogenesis and response to therapeutics, we generated a genetically engineered mouse model of brainstem glioma using the SB transposase system (19, 20). We induced brainstem tumors by activation of the receptor tyrosine kinase (RTK)–RAS–PI3K pathway, which is upregulated in a large percentage of DIPGs, and through inactivation of TP53, also commonly mutated in DIPG (3, 26, 27). This was achieved through the delivery of the following plasmids: NRASV12, a short hairpin targeting tumor protein p53 (TP53) (shP53), and SB transposase/firefly luciferase, with or without ACVR1G328V (Fig. 1A; Supplementary Fig. 1S) into the fourth ventricle of neonatal mice. We used bioluminescence imaging to monitor transfection efficiency and tumor development (Fig. 1B). The two experimental groups were: (i) wild-type ACVR1 (wt-ACVR1; NRASV12 and shp53) and (ii) mACVR1 (NRASV12, shp53, and ACVR1G328V). The MS of mice in the mACVR1 group was 127 days postinjection (dpi), while the MS for the wt-ACVR1 group was MS = 85 dpi; P = 0.0014, Mantel–Cox test (Fig. 1C). All tumors, regardless of ACVR1 mutation status, displayed high cellularity, nuclear atypia, invasive features, and grew in the brainstem (Fig. 1D). All tumors expressed oligodendrocyte transcription factor 2 (Olig2; Fig. 1E), glial fibrillary acidic protein (GFAP; Fig. 1F), nestin (Fig. 1G), and transcription factor Sox2 (Fig. 1H). Both wt-ACVR1 and mACVR1 tumors were positive for the proliferative marker, Ki67 (Fig. 1I), phosphorylated extracellular signal-regulated kinase (pERK) 1/2 (Fig. 1J), and phospho-MEK1/2 (Supplementary Fig. S2A). Expression of pERK 1/2 (Supplementary Fig. S2B) and pMEK1/2 (Supplementary Fig. S2C) was also confirmed on human DIPG cultures: SU-DIPG VI (wt-ACVR1) and SU-DIPG XX1 (mACVR1).
mACVR1 brainstem gliomas exhibit elevated levels of phosphorylated Smad1/5
We next investigated whether tumors encoding mACVR1 exhibited activated Smad1/5/8 transcription factors, downstream mediators of ACVR1 signaling (Fig. 2A). We observed that, indeed, mACVR1 brainstem gliomas displayed elevated levels of phosphorylated (phospho)-smad1/5 (Fig. 2B). This correlated with increased levels of the downstream canonical target gene, inhibitor of DNA binding 1 (Id1; Fig. 2C). Tumor neurospheres expressing mACVR1 (mACVR1 NS) also exhibited elevated levels of phospho-smad1/5 (Fig. 2D) and elevated levels of Id1 (Fig. 2D). To test the effect of a specific inhibitor of ACVR1, LDN-214117 on phospho-Smad1/5 signaling (Fig. 2E), we treated mACVR1 NS with increasing concentrations of LDN-214117. We observed decreased levels of phospho-Smad1/5 and Id2, while the levels of total Smad1 remain unchanged. Expression of ID1 was also confirmed on human DIPG cultures: SU-DIPG VI (wt-ACVR1) and SU-DIPG XX1 (mACVR1; Supplementary Fig. S2D). Overexpression of wt-ACVR1 did not result in activation of the phospho-Smad1/5 pathway, suggesting that phopho-Smad1/5 activation is only activated in the mACVR1 modal (Supplementary Fig. S3).
mACVR1 differentially regulates TGFβ pathway and pathways related to stem cell maintenance and focal adhesion
RNA-sequence analysis of mACVR1 versus wt-ACVR1 NS identified genes that were differentially regulated (1.5-fold; FDR corrected, P < 0.05; Fig. 3A). Gene ontology (GO) terms that were overrepresented in the set of differentially expressed genes (DE) included: developmental processes, cell migration, neuron differentiation, neurogenesis, nervous system development, cellular developmental process, cell development, cell proliferation, neuron development, cell surface receptor signaling pathway, neuron projection development, cell adhesion, positive regulation of developmental process, and positive regulation of cell differentiation (FDR correction; minimum 30 DE genes per term; Fig. 3B). The top three pathways that were impacted by the mutation in ACVR1 were focal adhesion (FDR corrected, P = 0.004), the TGFβ signaling pathway (FDR corrected, P = 0.004), and signaling pathways regulating pluripotency of stem cells (FDR corrected, P = 0.009; Fig. 3C). Overexpressed genes within the TGFβ pathway included ACVR1 and inhibitor of DNA binding genes Id1, Id2, and Id3 (Fig. 3C). Gene set enrichment analysis (GSEA) suggests an enrichment in the response to bone morphogenetic protein (BMP) and the regulation of cell differentiation (Fig. 3D–F). Because one of the top signaling pathways was involved the regulation of pluripotency in stem cells, we evaluated whether mACVR1 tumors express CD133, CD44, and aldehyde dehydrogenase 1 family member A1 (Aldh1), stem cell markers. Intracranial wt-ACVR1 or mACVR1 tumors were established in the pons of adult mice using SB-derived neurosphere. The results demonstrate that mACVR1 tumors have increased expression of the cancer stem cell markers CD133 (P = 0.0002) and CD44 (P = 0.0112; Fig. 4A). We did not observe differences in the expression of another cancer stem cell marker, that is, Aldh1 (Fig. 4A). We next investigated the tumor-initiating potential of wt-ACVR1 and mACVR1 NS in vivo. With wt-ACVR1 NS, the minimum number of cells required to generate brainstem gliomas with 100% penetrance was 1,000 cells, whereas, with mACVR1 NS it was possible to generate brainstem gliomas with 100% penetrance using 500 cells (Fig. 4B and C). These results indicate that mACVR1 NS have a greater tumor-initiating potential.
Preclinical testing of immune-stimulatory gene therapy using SB-derived neurosphere
Because of the invasive nature of brainstem gliomas and their refractive response to current therapies, we wanted to assess the efficacy of immune-stimulatory TK/Flt3L gene therapy using an immune-competent, intracranial mouse model of brainstem glioma. To generate this model, we utilized neurosphere expressing mACVR1. The release of DAMPs is crucial for the success of TK/Flt3L-mediated therapy, therefore, we first assessed whether mACVR1 NS released DAMPs in vitro after treatment with ganciclovir alone, TK alone, or ganciclovir + TK, with, or without 3 Gray (Gy) of irradiation (IR). We found that mACVR1 NS treated with ganciclovir + TK released increased levels of calreticulin (P < 0.0001; Fig. 5A), high mobility group box 1 protein (HMGB1; P < 0.0001; Fig. 5A), and ATP (P = 0.0071; Fig. 5A). We also observed that the combination of ganciclovir + TK with IR further increased the release of calreticulin (P < 0.0001; Fig. 5A), HMGB1 (P = 0.0375; Fig. 5A), and ATP (P = 0.0158; Fig 5A) release by the mACVR1 NS.
Having established that mACVR1 NS release calreticulin, HMGB1, and ATP in vitro following treatment with either IR or TK + ganciclovir, we next implanted mACVR1 NS into the pons of immune-competent adult C57BL/6 mice to generate a transplantable brainstem model amenable to preclinical therapeutic implementation. Brainstem tumors derived using the transplantable model were positive for Ki67 (proliferating cells), GFAP (astrocyte marker), Vimentin (astrocytes and ependymal cells), and Iba1 (microglia), but negative for MBP (a marker of mature oligodendrocytes; Supplementary Fig. S4). The tumors were also positive for pERK1/2 and pMEK1/2 demonstrating activation of the RTK–RAS–PI3K signaling pathway (Supplementary Fig. S4). We also observed expression of Id1, a downstream target gene of BMP–Smad1/5 signaling (Supplementary Fig. S4). Five days' postintracranial tumor implantation into the pons (Fig. 5B), mice were assigned to four treatment groups as indicated in Fig. 5C. Our results demonstrate that Ad-TK/Ad-Flt3L therapy is more effective in prolonging the MS of mACVR1 brainstem glioma compared with standard-of-care alone [MS = 36 days postimplantation (dpi) for TK/Flt3L group vs 23 dpi for IR group; P = 0.0014, Mantel–Cox test] or versus saline control (MS = 18 dpi; P = 0.0015, Mantel–Cox test; Fig. 5C). Our data also indicate that radiation does not have a detrimental effect on the efficacy of TK/Flt3L therapy (Fig. 5C). This experiment is validated in Supplementary Fig. S5.
We next aimed to investigate whether Ad-TK/Ad-Flt3L treatment recruits antibrainstem glioma–specific T-cell infiltration into the tumor immune microenvironment (TME). To do this, we used the surrogate tumor antigen ovalbumin (OVA) (18, 28). Using OVA-expressing mACVR1 cells (mACVR1-OVA NS), we were able to quantify tumor-specific CD8 T cells in the TME through the use of the SIINFEKL-H2Kb tetramer (Fig. 6A and B). The full gating strategy is presented in Supplementary Fig. S6. We observed a 3.8-fold increase in the frequency of tumor-specific CD8 T cells in the TME after treatment with Ad-TK/Ad-Flt3L gene therapy (***, P = 0.0007; Fig. 6B). To test the impact of Ad-TK/Ad-Flt3L gene therapy on the activation status of CD8 T cells in the TME, we stained effector T cells for the expression of IFNγ after restimulating CD8 T cells isolated from the TME with mACVR1-OVA NS-lysate for 24 hours. Our data show that IFNγ is increased 3.77-fold (**, P < 0.0038) in CD8 T cells from Ad-TK/Ad-Flt3L gene therapy–treated mice compared with saline controls (Fig. 6C). To test whether Ad-TK/Ad-Flt3L gene therapy affected antigen-specific T-cell proliferation, we labeled splenocytes with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE), and stimulated them with the OVA cognate SIINFEKL peptide. Our results show that the percentage of T cells that proliferated in response to the SIINKEKL peptide was greater (3.2-fold; ****, P < 0.0001) in mice treated with Ad-TK/Ad-Flt3L gene therapy compared with the saline-treated control group (Fig. 6D). We also measured IFNγ levels in supernatants of samples from the T-cell proliferation assay and our results show IFNγ levels were 2-fold (****, P < 0.0001) higher in the supernatant of splenocytes isolated from mice treated with Ad-Tk/Ad-Flt3L gene therapy compared with the saline-treated control group, after stimulation with SIINFEKL peptide (Fig. 6D). In addition, the cytolytic activity of T cells isolated from the spleen of animals treated with Ad-TK/Ad-Flt3L gene therapy was observed to be significantly higher (1.95-fold; P < 0.0001 at 20:1 ratio; Fig. 6E) when compared with the saline-treated group indicating that gene therapy significantly enhanced the cytolytic activity of splenic T cells. To determine whether the cytolytic activity is mACVR1-OVA specific, we used B16F10 murine melanoma cells as a negative control. When splenic T cells from mACVR1-OVA–bearing mice were cocultured with mACVR1-OVA cells, we observed an increase (4.25-fold; P < 0.0001 at 20:1 ratio) in tumor cell death in comparison with when they were cocultured with B16-F12 cells.
To evaluate any potential adverse effects of delivering gene therapy into the brainstem, we performed a detailed histopathologic analysis of brains from nontumor-bearing animals. After the animals were treated intracranial with Ad-TK/Ad-Flt3L, ganciclovir was administered intraperitoneally 1 day after adenoviral injection for 7 days. Brains were harvested for neuropathology analysis 24 hours after the last dose of ganciclovir was administered. Architectural integrity was assessed by H&E staining and by IHC using GFAP (astrocytes), MBP (myelin sheaths and oligodendrocytes), and Iba1 (microglia). No gross tissue abnormalities were observed in response to TK/Flt3L therapy compared with the saline controls (Supplementary Fig. S7). There was also no change in GFAP or MBP expression in response to TK/Flt3L therapy indicating that the brain architecture was unaffected (Supplementary Fig. S7). There was no increase in Iba1 expression in the animals treated with TK/Flt3L therapy indicating that gene therapy does not induce inflammation in normal brain tissue 8 days posttreatment (Supplementary Fig. S7).
To assess potential systemic toxicity due to TK/Flt3L or IR treatment, we performed H&E staining on liver sections and complete hematologic and serum biochemical analysis in tumor-bearing animals from the saline, TK/Flt3L, and TK/Flt3L + IR treatment groups. The liver sections from all treatment groups had normal hepatocyte architecture and did not show signs of inflammation or necrosis (Supplementary Fig. S8A). The white blood cell counts were within normal range for the saline, TK/Flt3L, and TK/Flt3L + IR groups, but significantly decreased in the IR group in comparison with the saline-treated group (P < 0.0001; Supplementary Fig. S8A). This is consistent with a study that found that whole-brain radiation significantly decreased total white blood cell counts (29). Retrospective studies have found that total lymphocyte counts are associated with decreased response to immune checkpoint inhibitors in multiple cancers (30, 31). Nevertheless, radiation has been shown to trigger immunogenic cell death, which can sometimes have a favorable effect (17). We saw that radiation when combined with TK/Flt3L therapy did not have a detrimental effect on total white blood cell counts. Red blood cell, hemoglobin, hematocrit, platelet, lymphocyte, neutrophil, and monocyte counts were not significantly affected by TK/Flt3L or IR therapy (Supplementary Fig. S8B–S8I). We did not find any significant changes in important enzymes involved in liver (alanine aminotransferase and aspartate aminotransferase) and kidney (blood, urea, and nitrogen) function as a result of TK/Flt3L or IR therapies (Supplementary Fig. S8J–S8L).
DIPG remains an incurable tumor with a poor prognosis (32). The tumors originate in the pons and infiltrate into sensitive regions of the brainstem precluding surgical resection (33). DIPGs are also resistant to radiation and chemotherapy (34). DIPG studies involving the development and implementation of novel therapies have utilized patient-derived xenograft models, where human DIPG cells are implanted into the brain of immune-deficient mice or rats to establish intracranial tumors (35, 36). One limitation with those models is that immune-suppressed animals cannot be used to test immunotherapies or perform immune-related mechanistic studies. In this study, we utilized the SB transposase system to generate an endogenous mouse model of mACVR1 brainstem glioma. ACVR1 is frequently mutated to ACVR1-G328V in DIPG (4–7). The SB system efficiently and reproducibly integrates plasmid DNA into the neural progenitor cells' host chromosomal DNA of neonatal mice, allowing for the functional assessment of the role of candidate DIPG genes in promoting tumor progression (19–22). Mutations in ACVR1 confer ligand-independent activation of ACVR1, which is a mediator of the BMP signaling pathway (4, 5, 37, 38). We show that SB tumors are localized in the brainstem and have increased downstream signaling of BMP as demonstrated by increased phospho-smad1/5 levels and Id1 (Fig. 2B and C). Transcriptional profiling indicated an enrichment in the BMP signaling pathway in mACVR1 NS compared with wt-WT ACVR1 tumor neurosphere (Fig. 3D). In addition, our tumors express Olig2, a biomarker for DIPG tumors that is highly expressed in 70%–80% of human DIPGs (39).
Our data also suggests that mACVR1 regulates stemness and tumor initiation potential. Interestingly, clonal evolution analysis of a human DIPG tumor found that the ACVR1 mutation was present along with the H3.1 K27M mutation in all tumor clones, in contrast to secondary mutations found only in some subclones, implicating mACVR1in DIPG tumor initiation (40). In addition, Hoeman and colleagues observed that ACVR1 R206H contributed to increased tumor incidence using the RCAS model driven by p53 loss and PDDGF-A, supporting our results that ACVR1 plays a role in tumor initiation (41). It was also recently reported that only amino acid substitutions of ACVR1 at the G328 residue confer a significant increase in survival in patients with DIPG (9). Therefore, our data suggests independent roles for ACVR1 on survival and tumor progression.
Neurospheres harboring mACVR1 were implanted into the pons of adult immune-competent mice, enabling the generation of transplantable DIPG models and amenable for preclinical therapeutic studies. This implantable model is more stringent for assessment of therapeutic efficacy than genetically engineered models because tumor growth rates and time of death vary less between animals. Currently, the treatment option for children with DIPG is limited to radiation to provide palliative care (1). Although radiation can temporarily provide symptomatic relief and extend survival by a few months, it can cause detrimental effects on the developing brain (42, 43). Therefore, there is a dire need for new therapeutic interventions. This research focuses on the response of mACVR1 brainstem glioma tumors to immunotherapies.
There has been significant progress in the field of cancer immunotherapy leading to improvements in overall survival in many types of solid tumors (11–13, 44, 45). Advances in immunotherapies include the development of immune therapies targeting immune checkpoints, vaccine approaches against tumor antigens or dendritic cell vaccines designed to stimulate the adaptive immune response, adoptive cell therapy, oncolytic viral therapy, and immune-stimulatory gene therapy (45–47). This has led to a growing number of clinical trials testing immunotherapies in DIPG (47).
In recent years, it has been established that activated immune cells are able to migrate and enter into the brain parenchyma, and that the brain has a functional lymphatic system that allows for transport of central nervous system antigens to the draining lymph nodes (48, 49). Thus, the brain is capable of mounting T-cell–mediated adaptive immune responses, but, because there are very low numbers of local professional antigen-presenting cells within the normal brain, it is not possible to prime a potent immune response against antigens localized in the brain parenchyma (50, 51).
In adult GBM there is an evidence of immune cell infiltration, but an immunosuppressive environment precludes effective antitumor immunity (46, 52, 53). GBM tumors establish an immunosuppressive environment by the release of immunosuppressive cytokines, such as TGFβ and IL10, by the recruitment or induction of immunosuppressive cells, such as regulatory T cells, myeloid-derived suppressor cells, or tumor-associated macrophages, and by the expression of immune checkpoint receptor ligands (52–58). In comparison with adult GBM, initial studies of the tumor microenvironment in DIPG have found that there is a low number of immune infiltrates in human DIPG tumors and that they do not express inflammatory cytokines and chemokines (59, 60). These data provide support for the use of an immune-modulatory therapeutic strategy to enhance the recruitment of immune cells into the tumor, with the aim of mounting an effective anti-DIPG immune response.
We have previously demonstrated that combined immune-stimulatory gene therapy mediated through the delivery of adenoviruses encoding herpes simplex virus type 1 TK and Flt3L leads to tumor regression and long-term survival in several rodent models of GBM (15, 16, 61–63). This therapy is based on inducing tumor cell death through expression of suicide gene TK (64, 65). Tumor antigens and DAMP molecules, such as calreticulin, HMGB1, and ATP, are released by dying tumor cells (17, 66). The effectiveness of this combination therapy also relies on Flt3L to recruit dendritic cells into the tumor microenvironment, while the release of HMGB1 stimulates TLR2-dependent activation of dendritic cells (15, 65, 67). Activated dendritic cells can then transport antigens to the draining lymph nodes and induce tumor-specific T-cell responses (15, 65). Initial results from the first-in-human phase I clinical trial of combined adenoviral delivery of TK and Flt3L for the treatment of adult GBM are promising and report that the therapy was well tolerated (68). However, it has been established that the biology of DIPG and GBM is different (69). Therefore, it is necessary to assess the safety and efficacy of TK/Flt3L therapy in preclinical models of brainstem glioma. Herein, we demonstrate that treatment with TK/Flt3L gene therapy in mice bearing mACVR1 brainstem gliomas stimulates a strong antitumor cytotoxic immune response leading to a significant increase in survival (Fig. 5C). Treatment with TK/Flt3L increased the frequency of tumor-specific CD8 T cells in the tumor microenvironment and increased toxicity as demonstrated by enhanced IFNγ production (Fig. 6B and C).
In addition, delivery of TK/Flt3L into the normal brainstem did not induce any local or systemic cytotoxicity. H&E staining and immunostaining for GFAP (astoryctes), MBP (myelin sheaths and oligodendrocytes), and Iba1 (activated macrophages and microglia) was used to assess local toxicity, and we observed no architectural abnormalities or overt inflammation as a result of TK/Flt3L therapy. Histologic examination of H&E-stained liver tissue did not show signs of inflammation, necrosis, or alterations in normal hepatocyte structure. Hematologic toxicity, assessed by complete blood count and serum chemistry analysis, indicated that TK/Flt3L therapy did not induce any toxicity as values from a TK/Flt3L-treated group were not significantly altered when compared with saline-treated animals. Our results are consistent with other preclinical studies that report the brainstem can tolerate adenoviral-mediated immunotherapy (70, 71). Results from a clinical trial (NCT03178032) utilizing adenoviral vector delivery of an oncolytic virus into the pons of patients with DIPG will also shed light on the feasibility and toxicity of intratumoral adenoviral delivery into the brainstem (72).
In conclusion, we demonstrate the amenability of the SB transposase to be used to develop genetically and histologically accurate models of DIPG expressing mACVR1 and provide compelling evidence that warrants further development of conditionally cytotoxic immune-stimulatory gene therapy for the treatment of DIPG. We anticipate that in the clinic, this approach could also be used in combination with immune checkpoint blockade to further enhance the therapeutic efficacy of the TK/Flt3L-mediated antibrainstem glioma immune response.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Conception and design: F. Mendez, P. Kadiyala, F.J. Nunez, P.R. Lowenstein, M.G. Castro
Development of methodology: F. Mendez, P. Kadiyala, F.J. Nunez, M. Edwards, P.R. Lowenstein, M.G. Castro
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Mendez, P. Kadiyala, F.J. Nunez, F.M. Nunez, J.C. Gauss, R. Ravindran, S. Haase, P.R. Lowenstein, M.G. Castro
Analysis and interpretation of data (eg, statistical analysis, biostatistics, computational analysis): F. Mendez, P. Kadiyala, F.J. Nunez, S. Carney, F.M. Nunez, J.C. Gauss, P.R. Lowenstein, M.G. Castro
Writing, review, and/or revision of the manuscript: F. Mendez, P. Kadiyala, F.J. Nunez, S. Carney, F.M. Nunez, J.C. Gauss, P.R. Lowenstein, M.G. Castro
Administrative, technical, or material support (ie, reporting or organizing data, constructing databases): P. Kadiyala, S. Carney, M.B. Garcia-Fabiani, P.R. Lowenstein, M.G. Castro
Study supervision: P.R. Lowenstein, M.G. Castro
Other (data collection): S. Pawar
This study was supported by NIH/National Institute of Neurological Disorders & Stroke (NIH/NINDS) grants R37-NS094804, R01-NS105556, and R21-NS107894 (to M.G. Castro); NIH/NINDS grants R01-NS076991, R01-NS082311, and R01-NS096756 to (P.R. Lowenstein); NIH/NIBIB R01-EB022563 (to M.G. Castro and P.R. Lowenstein); and the Department of Neurosurgery; Leah's Happy Hearts Foundation, ChadThough Foundation, Pediatric Brain Tumor Foundation, and Smiles for Sophie Forever Foundation (to M.G. Castro and P.R. Lowenstein). RNA Biomedicine grant F046166 (to M.G. Castro). NIH/NINDS-F31NS103500 and Rackham Pre-doctoral Fellowship (to F. Mendez); S. Carney was supported by NIH/NCI-T32-CA009676. M.B. Garcia-Fabiani was supported by the American Brain Tumor Association Basic Research Fellowship “in Memory of Bruce and Brian Jackson.” We thank J. Ohlfest (University of Minnesota, Minneapolis, MN, deceased) for providing the SB model plasmids and Jeff Wrana for providing the wt-ACVR1 plasmid. We thank the support and academic leadership of Dr. Karin Muraszko, and the administrative support of Angela Collada.
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.