Poliovirus Receptor (CD155) Expression in Pediatric Brain Tumors Mediates Oncolysis of Medulloblastoma and Pleomorphic Xanthoastrocytoma

Abstract

Poliovirus oncolytic immunotherapy is a putatively novel approach to treat pediatric brain tumors. This work sought to determine expression of the poliovirus receptor (PVR), CD155, in low-grade and malignant pediatric brain tumors and its ability to infect, propagate, and inhibit cell proliferation. CD155 expression in pleomorphic xanthoastrocytoma (PXA), medulloblastoma, atypical teratoid rhabdoid tumor, primitive neuroectodermal tumor, and anaplastic ependymoma specimens was assessed. The ability of the polio: rhinovirus recombinant, PVSRIPO, to infect PXA (645 [BRAF V600E mutation], 2363) and medulloblastoma (D283, D341) cells were determined by viral propagation measurement and cell proliferation. PVR mRNA expression was evaluated in 763 medulloblastoma and 1231 normal brain samples. CD155 was expressed in all 12 patient specimens and in PXA and medulloblastoma cell lines. One-step growth curves at a multiplicity of infection of 10 demonstrated productive infection and peak plaque formation units at 5–10 hours. PVSRIPO infection significantly decreased cellular proliferation in 2363, 645, and D341 cell lines at 48 hours (p < 0.05) and resulted in cell death. PVR expression was highest in medulloblastoma subtypes Group 3γ, WNTα, and WNTβ (p < 0.001). This proof-of-concept in vitro study demonstrates that PVSRIPO is capable of infecting, propagating, prohibiting cell proliferation, and killing PXA and Group 3 medulloblastoma.

INTRODUCTION

Brain cancer is now the leading cause of mortality for patients under age 19, contributing to nearly 30% of all pediatric cancer deaths (1). Medulloblastoma is the most common solid malignant cancer (WHO Grade IV) in children (2). Medulloblastoma is not a single entity but consists of at least 12 molecular subgroups with different genetics, demographics, clinical characteristics, and prognoses (3, 4). The mainstay of medulloblastoma therapy consists of maximum safe surgical resection followed by adjuvant chemotherapy and radiation. This regimen affords patients a 5-year overall survival of 60%–85% (5), but nearly all experience toxicity from chemotherapy and craniospinal irradiation.

In contrast, pleomorphic xanthoastrocytoma (PXA) is a rare condition comprising <1% of all primary brain tumors. It is a Grade II tumor; however, anaplastic transformation occurs in approximately 20% of patients (6). The 5-year overall survival is 80%–100% for Grade II and 75%–85% for Grade III PXA (6–8). Therapy consists of surgical resection, typically followed by observation following a gross total resection or chemotherapy and radiation for recurrence or subtotal resection of a Grade III tumor. There is a need for novel therapies for both low-grade and malignant pediatric brain tumors that improve survival and minimize the need for toxic cytotoxic chemotherapy and radiation.

The highly attenuated polio: rhinovirus chimera, PVSRIPO, is the poliovirus type 1 (Sabin) vaccine containing a heterologous internal ribosomal entry site of human rhinovirus type 2 (9). This genetic engineering abolishes neurovirulence, which was demonstrated conclusively in investigational new drug (IND)-directed toxicology evaluations in nonhuman primates (10) and in Phase I clinical investigations with intracerebral delivery in adult patients with recurrent glioblastoma (11). The internal ribosomal entry site of PVSRIPO renders it neuro-incompetent thereby leaving normal neural tissue unscathed (9, 12–15). PVSRIPO efficacy rests on virtually universal ectopic expression of the poliovirus receptor (PVR) (CD155) in solid neoplasia including malignant glioma (16). However, the ability of PVSRIPO to target, infect, and kill pediatric low-grade and malignant brain tumors has not previously been described. Herein, we discovered robust CD155 expression in a variety of pediatric brain tumors. As a proof-of-concept, we focus on 2 relatively dissimilar pediatric brain tumors, medulloblastoma and PXA, to demonstrate the ability of PVSRIPO to infect, replicate, and prohibit proliferation in cells derived from these tumors.

MATERIALS AND METHODS

PVR Protein Analysis

De-identified cases of pediatric brain tumor archival tissue were obtained from the Duke Brain Tumor Biorepository (Duke IRB Pro00007434). Twelve brain tumor specimens were analyzed for PVR expression using immunohistochemistry: 2 medulloblastomas, 2 PXAs, 3 atypical teratoid rhabdoid tumors (ATRT), 2 primitive neuroectodermal tumors ([PNET], diagnoses made prior to WHO brain tumor classification update), and 3 anaplastic ependymomas (Table). Specimens were prepared and analyzed as previously described in (17). Briefly, 5-µm formalin-fixed paraffin-embedded sections were processed on a Bond-maX Processing Module (Leica Microsystems, Buffalo Grove, IL) using Bond Polymer Refine Detection kit. After deparaffinization, eptitope retrieval, and blocking, the following primary antibodies were applied: Rabbit anti-human PVR (Cell Signaling Technology no. 13544, Danvers, MA) at 5 µg/mL for 60 minutes or Rabbit IgG (Cell Signaling Technology no. 3900S); Rabbit IgG served as a control for each specimen.

TABLE.

Patient Demographics

Tumor Diagnosis and Specimen Number	Age (Years)	Gender	Tumor Location	Metastatic Status at Diagnosis
Medulloblastoma 1	2.6	Female	Right Cerebellar	M0
Medulloblastoma 2	5.6	Female	Fourth ventricle	M0
PXA 1	21	Male	Left Temporal	M0
PXA 2*	7.5	Male	Left Frontoparietal	M0
ATRT 1	2.6	Female	Right Temporal	M0
ATRT 2	44	Female	Right Parieto-occipital	M0
ATRT 3	42	Female	Right Frontal	M0
PNET 1	7.2	Female	Left Frontal	M0
PNET 2	5.7	Female	Right Parietal	M0
AE 1	5.1	Male	Left Frontoparietal	M+ (Thoracic)
AE 2	17	Male	Right posterior central sulcus	M0
AE 3	14	Male	Right Frontal	M0

^*BRAF mutation.

PXA, pleomorphic xanthoastrocytoma; PNET, primitive neuroectodermal tumor (diagnosis made prior to 2016 WHO update); ATRT, atypical teratoid rhaboid tumor; AE, anaplastic ependymoma.

Two Group 3 medulloblastoma cell lines (18), D283 and D341, and 2 PXA cell lines, 645 and 2363, were used for in vitro studies. All cell lines used were generated and validated at the Duke University Medical Center. Western blot analysis was completed as previously described in (19). Briefly, cell lysates were resolved by electrophoresis in 4%–12% SDS-Page gels (Invitrogen, Carlsbad, CA), transferred to nitrocellulose membrane, blocked by Starting Block (Thermo Scientific, Waltham, MA no. 37539) for 1 hour, incubated with anti- PVR/CD155 antibody (1:1000, D3G7H- Cell Signaling Technologies) overnight, washed with TBST 3 times, incubated with secondary anti-rabbit antibody (1:5000, Cell Signaling Technologies no. 7074S) for 1 hour, washed 3 times with TBST, then imaged using ECL Western blotting detection reagents (Thermo Scientific no. 34078). Immunoblots were reprobed with ribosomal protein S6 (rpS6) as a loading control (Cell Signaling Technologies).

Gene Expression Profiling

PVR mRNA expression in 10 normal brain regions (cerebellar cortex, frontal cortex, hippocampus, medulla, occipital cortex, putamen, substantia nigra, temporal cortex, thalamus, and white matter) was analyzed across 1231 samples (20) and profiled on the Affymetrix Exon 1.0 T array (Probe 3835645). PVR mRNA expression was analyzed across 763 primary medulloblastoma samples, profiled on the Affymetrix Gene 1.1 ST array as previously described and normalized using the RMA method, and subgrouped using similarity network fusion (GSE85217) (3). Medulloblastoma subtype was based on work from Cavalli et al (3). The number of patients of each medulloblastoma subtype was as follows: 49 WNTα, 21 WNTβ, 65 SHHα, 35 SHHβ, 47 SHHγ, 76 SHHδ, 67 Group 3α, 37 Group 3β, 40 Group 3γ, 98 Group 4α, 109 Group 4β, 119 Group 4γ. Differences across subgroups and subtypes were evaluated using ANOVA in the R statistical environment using a t-test with multiple comparisons (v3.4.2).

BRAF Mutation Analysis

PXA samples 2363 and 654 underwent DNA extraction using High Pure PCR Template Purification Kit (Roche, Indianapolis, IN) according to the manufacturer’s specifications. Both samples were tested at the Duke University Medical Center Molecular Diagnostics Lab for BRAF V600E and V600K mutations using TaqMan real-time PCR (Thermo Fisher). Two real-time PCR reactions were performed using oligonucleotide primer pairs that amplify both the mutant and wild-type BRAF alleles and allele specific TaqMan probes. A threshold cycle value (Ct) was measured for each TaqMan probe. 5% BRAF V600E and V600K sensitivity controls, an approximately 50% BRAF mutant control, and a wild-type (normal) BRAF control were included in each assay. Results were interpreted based on the patient delta Ct value relative to the controls. This test was performed using an ABI Prism 7500 Sequence Detection System (Thermo Fisher).

Virus Infection of Tumor Cells

PVSRIPO was derived and propagated as previously described in (21). One step infection of tumor cells was completed as previously described in (19). Briefly, growth media was removed from the cells, and PVSRIPO was added at a multiplicity of infection (MOI) of 10 in serum-free media. Media without PVSRIPO served as mock controls. After 30 minutes incubation, unbound virus was washed. Cells were overlayed with a new portion of media containing 2% fetal bovine serum and incubated for different time intervals. Cells were frozen at specified time point and processed for plaque assay to assess viral propagation. Plaque assays were performed as described elsewhere (19). Briefly, infected cells from each time point were freeze/thawed twice to release the virus; serial 10-fold dilutions were prepared and applied to cells in 6-well plates. After 30 minutes rocking at room temperature cells were overlayed with 1:1 mix of 1.2% Tragacanth Gum (Sigma-Aldrich, St Louis, MO): 2× minimum essential medium (MEM) (Thermo Fisher) and incubated for 2 days followed by crystal violet staining. PVSRIPO’s effect on cell proliferation by was determined using an MTT assay (Thermo Fisher) per the manufacturer’s protocol in triplicate. Briefly, 5000 cells were plated, incubated for 8 hours, then infected with PVSRIPO or UV-inactivated PVSRIPO (controls). PVSRIPO’s effect on cell death was determined using poly (adenosine diphosphate–ribose) polymerase (PARP) cleavage (1:1000, Cell Signaling Technology) at 0, 4, 8, 24, and 48 hours.

RESULTS

Robust PVR Expression in Patient Specimens

In all patient samples analyzed, robust expression of PVR was detected (Fig. 1). This finding is comparable to previous work in which PVR expression was detected in 62 of 63 glioblastoma specimens (17). In these 12 samples, we found PVR staining to demonstrate both intratumor and intertumoral heterogeneity. However, our group has previously shown that PVR expression does not correlate with susceptibility to or cancer cell killing by PVSRIPO (17). Notably, PVR expression is found in normal brain (Supplementary DataFig. S1) as well as in anterior horn cells (17).

FIGURE 1.

CD155/PVR is robustly expressed in a variety of pediatric brain tumor specimens. Left panel demonstrates controls, right panel demonstrates PVR staining (brown). Endothelial cells within each specimen all stain for PVR. All photos taken at 40×, white bar = 200 µm.

PVR RNA expression across 10 normal brain sites is shown in Figure 2. In comparing PVR RNA expression across the 12 medulloblastoma subtypes as recently described by Cavalli et al (3) in 763 patient samples, PVR expression was significantly higher in Group 3γ, WNTα, and WNTβ compared with the other subtypes (p < 0.001) (Fig. 3; Supplementary DataTable S1).

FIGURE 2.

PVR expression across 10 brain sites in 1231 samples. Parentheses denote patient numbers for each site.

FIGURE 3.

PVR expression across the 12 medulloblastoma subgroups in 763 patients. Box and whisker plot comparing RNA expression of PVR between the 12 medulloblastoma subgroups in 763 medulloblastoma samples. Parentheses denote patient numbers for each subtype. PVR is preferentially expressed in the WNTα, WNTβ, and Group 3γ (p < 0.001, Supplementary DataTable S1). **p < 0.01.

Viral Oncolysis of Medulloblastoma and PXA

To test the ability of PVSRIPO to effectively target, infect, and halt proliferation in pediatric brain tumor tissue, 2 discordant tumor types were selected for in vitro studies: medulloblastoma and PXA. PVR expression was found in both PXA and medulloblastoma cell lines (Fig. 4A). Of note, PXA 654 contained a BRAF V600E mutation while 2363 did not. Neither PXA cell line had a V600K mutation.

FIGURE 4.

PVSRIPO invades and propagates in PXA and medulloblastoma cells. (A) Western blots of HeLa, medulloblastoma (283 and 341), and PXA (645 and 2363) cells demonstrating CD155 expression; rpS6 serves as loading control. One-step growth curves were performed using a MOI of 10 demonstrating robust propagation in PXA (B) and medulloblastoma (C) cells plateauing at 6–8 hours. Mock infected cells demonstrated no plaque formation.

PVSRIPO efficiently infected and propagated in both medulloblastoma and PXA cell lines reaching a peak between 6 and 8 hours (Fig. 4B, C). This timing corresponded with halted cellular proliferation in all cell lines and significantly diminished growth in 2363 and D341 cells at 48 hours (p < 0.0001) (Fig. 5A, B). Expression of PARP, a validated marker of cell death (21), was found in D341, 645, and 2363 cell lines at 24 and 48 hours (Fig. 5C).

FIGURE 5.

PVSRIPO halts proliferation and kills PXA and medulloblastoma cells. MTT assays of PXA (A) and medulloblastoma (B) cells using a PVSRIPO MOI of 10 demonstrates halted proliferation at 8 hours, corresponding to the timing of peak viral propagation. Western blots of PXA (C) and medulloblastoma (D) using a PVSRIPO MOI of 10 demonstrates cell death as determined by PARP cleavage at 24 and 48 hours. Loss of loading control in 645 at 48 hours is due to complete destruction of all cellular components seen in some tumor cell types (21). *p ≤ 0.05, **p < 0.01.

DISCUSSION

In this work, we demonstrate robust expression of CD155 in a variety of pediatric low-grade and malignant tumors. These included both low-grade tumors such as PXA and malignant tumors including medulloblastoma, ATRT, PNET, and Grade III ependymoma. All samples tested were positive for CD155. Work is ongoing to determine if CD155 is preferentially expressed according to molecular subgroups of medulloblastoma and ependymoma. Our findings are concordant with CD155 expression in a variety of other solid tumors including colorectal carcinoma, lung adenocarcinoma, melanoma, breast cancer (22), and glioblastoma (GBM) (17).

To demonstrate proof-of-concept of PVSRIPO as a novel immunotherapy for pediatric brain tumors, we tested the ability of PVSRIPO to infect, propagate, and halt cellular proliferation in 2 disparate cancer cell types: PXA (both BRAF 600E mutant and wild type) and medulloblastoma. Our group has previously shown comparable growth curves in gliomas (19, 23) and PARP cleavage in melanoma, prostate, and breast cancer cell lines (21) Those in vitro results have been translated to demonstrate efficacy of PVSRIPO against malignant glioma in preclinical rodent models (23) and in a recently completed clinical trial of adult patients with recurrent GBM in which 20% of patients had durable responses over 1 year (11). PVSRIPO produces a range of immunogenic pathogen- and danger-associated patterns yielding profuse neutrophil invasion and chemokine/cytokine secretion within the tumor (24). This culminates in CD4+ and CD8+ T cell infiltration of inflamed tumors resulting in durable immunogenic responses. PVSRIPO also infects and activates dendritic cells and macrophages (which also express CD155), to facilitate antigen presentation and the production of antitumor immunity (21).

Further, using a large patient dataset, we show that CD155 transcript abundance is variable amongst medulloblastoma subtypes. When comparing PVR RNA expression across the 12 medulloblastoma subgroups, Group 3γ, WNTα, and WNTβ had the highest expression. These subgroups have the commonality of c-Myc overexpression (25). MYC expression is thought merely to reflect WNT signaling in the WNT subgroup (the group with the best prognosis) but is likely pathogenic in Group 3, given the association of MYC expression and poor prognosis (26). Work is currently underway to elucidate the putative association between c-Myc and CD155 expression.

For children with malignant brain tumors that receive adjuvant chemotherapy and radiation, severe ototoxicity occurs in over 18% of patients (27), endocrinopathies such as hypothyroidism and growth hormone insufficiency in over 50% of patients (28), and permanent cognitive decline in nearly all patients (29). Furthermore, 10% of patients develop secondary malignancies (30). Oncolytic viral immunotherapy is a novel therapeutic approach that has not been associated with significant morbidities (11). In this work, we describe rational criteria indicating a utility for PVSRIPO in the treatment of both malignant and low-grade pediatric brain tumors. Given the profound toxicity associated with adjuvant therapies of malignant pediatric brain tumors (particularly those without a targetable genetic mutation), and the lack of efficacy of traditional adjuvant therapy for lower-grade gliomas such as PXA, oncolytic virus-mediated immunotherapy is a promising novel therapeutic approach.

Conclusions

In summary, we demonstrate widespread expression of the PVR, CD155, on a spectrum of low-grade and malignant pediatric brain tumor specimens. As proof-of-concept, we demonstrate the ability of PVSRIPO to infect, propagate in, prohibit proliferation of, and kill low-grade and malignant pediatric brain tumors. Given similar successful translation of PVSRIPO to treat malignant gliomas in adult and pediatric patients, future in vivo work and prospective clinical trials will explore this possibility of PVSRIPO as an immunotherapeutic approach to treat pediatric brain tumors.

FUNDING

This study was supported by NIH (P50CA190991), Duke Health Fellow Fund, and Chetna & Meena Trust (EMT); American Brain Tumor Association, the Garron Family Cancer Center, Meagan’s Walk, the Brain Tumour Foundation of Canada, and the Collaborative Ependymoma Research Network (VR); Canadian Cancer Society Research Institute, Terry Fox Research Institute, Canadian Institutes of Health Research, National Institutes of Health, Pediatric Brain Tumor Foundation, and Garron Family Chair in Childhood Cancer Research (MDT); CA124756, CA190991, grants from the Lefkofsky Family Foundation and Hope and Gavin Wolfe (MG). The Preston Robert Tisch Brain Tumor Center biorespository is supported by the Pediatric Brain Tumor Foundation.

The authors have no duality or conflicts of interest to declare.

Supplementary Data can be found at http://www.jnen.oxfordjournals.org.