Background: Glioblastoma multiforme and high-risk neuroblastoma are cancers with poor outcome. Immunotherapy in the form of neurotropic oncolytic viruses is a promising therapeutic approach for these malignancies. Here we evaluate the oncolytic capacity of the neurovirulent and partly IFNβ-resistant Semliki Forest virus (SFV)-4 in glioblastoma multiformes and neuroblastomas. To reduce neurovirulence we constructed SFV4miRT, which is attenuated in normal central nervous system (CNS) cells through insertion of microRNA target sequences for miR124, miR125, miR134.
Methods: Oncolytic activity of SFV4miRT was examined in mouse neuroblastoma and glioblastoma multiforme cell lines and in patient-derived human glioblastoma cell cultures (HGCC). In vivo neurovirulence and therapeutic efficacy was evaluated in two syngeneic orthotopic glioma models (CT-2A, GL261) and a syngeneic subcutaneous neuroblastoma model (NXS2). The role of IFNβ in inhibiting therapeutic efficacy was investigated.
Results: The introduction of miRNA target sequences reduced neurovirulence of SFV4 in terms of attenuated replication in mouse CNS cells and ability to cause encephalitis when administered intravenously. A single intravenous injection of SFV4miRT prolonged survival and cured four of eight mice (50%) with NXS2 and three of 11 mice (27%) with CT-2A, but not for GL261 tumor-bearing mice. In vivo therapeutic efficacy in different tumor models inversely correlated to secretion of IFNβ by respective cells upon SFV4 infection in vitro. Similarly, killing efficacy of HGCC lines inversely correlated to IFNβ response and interferon-α/β receptor-1 expression.
Conclusions: SFV4miRT has reduced neurovirulence, while retaining its oncolytic capacity. SFV4miRT is an excellent candidate for treatment of glioblastoma multiforme and neuroblastoma with low IFN-β secretion. Clin Cancer Res; 23(6); 1519–30. ©2016 AACR.
With Talimogene laherparepvec, the first oncolytic virus approved for treatment of cancer, oncolytic virus immunotherapy has a very potential future. Glioblastoma multiforme is a rapidly growing neoplasm that is very difficult to cure with the current standard therapies. Oncolytic viruses have been previously tested in the clinic, but without much therapeutic benefit. We have engineered a safe oncolytic Semliki Forest virus (SFV) detargeted using three central nervous system (CNS)-related microRNAs (SFV4miRT) that has potential to kill malignant gliomas and neuroblastomas in preclinical murine models. Because SFV4 is partly type-I IFN insensitive, the SFV4miRT virus could kill also tumors cells with moderate secretion of type-I IFNs. By using the newly established human glioblastoma cell culture (HGCC) resource (www.hgcc.se), we demonstrate that SFV4miRT could kill human glioblastoma cell cultures of all molecular subtypes (proneural, classical, mesenchymal, neural), which indicates the potential for its clinical translational.
Malignant tumors arising from the central nervous system (CNS) are among the most feared types of cancer. Glioblastoma multiforme is the most common, malignant form of primary brain tumor with poor prognosis and 1-year survival rate of only 35.7% (1). Surgical resection combined with radiotherapy and adjuvant temozolomide is the current standard of care therapy, which extends survival but is not curative (2). Glioblastoma multiforme is more common in adults but can also arise in children. Neuroblastoma is the most common extracranial solid cancer in children arising from neural crest cells. Low-risk neuroblastomas are associated with good outcome but children with high-risk neuroblastoma often relapse even after multimodal therapy (3). Conditionally replicating oncolytic viruses are an attractive option for treatment of cancers, as viruses have a natural ability to replicate inside dividing tumor cells and kill them and simultaneously induce an immune response against the tumor. Herpes simplex virus, adenovirus, and Newcastle disease virus have been evaluated in phases I and II clinical trials for treatment of glioblastoma multiforme. They show excellent safety data but have so far not resulted in successful tumor shrinkage and prolonged survival (4). Adenovirus has been used to treat children with neuroblastoma with encouraging results (5, 6).
Semliki Forest virus (SFV) is an Alphavirus belonging to the Togaviridae family (7). It is an enveloped, positive strand RNA virus, having high replication efficacy and ability to kill a variety of tumor cells (7–10). It has a natural neurotropism (11), making it an attractive candidate for use as an oncolytic immunotherapy agent to treat neuroblastoma and glioblastoma multiforme. Several strains of SFV have been described, including the neurovirulent strain SFV4 and the avirulent strain SFV-A(7)74 (referred to as A7/74 from now; ref. 7). The neurovirulence of SFV4 is mainly due to initial replication in neurons and oligodendrocytes, and subsequent spread to the rest of the brain resulting in encephalitis (12, 13). A7/74 is usually preferred for use as an oncolytic agent due to its natural avirulence in adult rodents. However, type-I IFN α/β (IFNα/β) responses elicited by the host in response to virus infection impede A7/74 efficacy (14, 15). The nonstructural proteins nsP3-nsP4 of SFV4 are involved in reducing STAT1 Tyr701 phosphorylation (P-STAT1) upon stimulation of virus infected cells with exogenous IFNβ, and it confers its ability to partly resist type-I antiviral defense (16). Thus, SFV4 can be a promising alternative to treat IFNα/β responsive tumors if its neurovirulence can be attenuated.
miRNA are small noncoding RNAs involved in posttranscriptional regulation of gene expression. miRNA expression is cell-type restricted, which has been taken advantage of when designing oncolytic virus to prevent their replication in healthy tissues (17–20). This is achieved by introduction of miRNA target sequences (miRT) in the viral genome, leading to that viral messenger RNA (mRNA; in the case of positive stand RNA viruses it is also genomic RNA) degradation in cells where the matching miRNA are expressed. Regarding miRNAs in the CNS, miR124 and miR125 are highly expressed in mouse brain (21), whereas miR134 is expressed in the dendrites present in the hippocampal neurons (22). Their expression in glioblastoma multiforme and neuroblastoma are partly unknown and investigated herein.
In this article we evaluate the efficacy of SFV immunotherapy with respect to type-I IFN response in patient-derived human glioblastoma cell cultures (HGCC) along with preclinical in vivo glioma and neuroblastoma models. We improved the safety of SFV4 immunotherapy by insertion of target sequences for miR124, miR125, and miR134 in its genome (SFV4miRT) to reduce neurovirulence in mice but retain replication potential and oncolytic capacity in glioma and neuroblastoma cells.
Materials and Methods
Quantitative real-time PCR
Total RNA from cell lines was isolated for miRNA detection using miRNeasy Mini Kit (Qiagen). To isolate RNA from uninfected mouse tissues: brain, spinal cord, and dorsal root ganglion (DRG) were excised, dissected, immersed in RNAlater (Ambion, Thermo Fisher Scientific) and homogenized before total RNA was prepared by TRizol extraction (Invitrogen). cDNA was synthesized using QuantiMir RT Kit (System Biosciences). miRNAs were quantified from cDNA by SYBR green based real-time PCR using the forward primers miR124.F: 5′-TAAGGCACGCGGTGAATGCC-3′, miR125.F: 5′-TCCCTGAGACCCTTTAACCTGTGA-3′, miR134.F: 5′-TGTGACTGGTTGACCAGAGGGG-3′, and the universal reverse primer provided in the QuantiMir RT Kit. Human (hU6.F: 5′-CGCAAGGATGACACGCAATTC-3′) and mouse (mU6.F: 5′-TGGCCCCTGCGCAAGGATG-3′) U6 snRNA was used to normalize expression levels of miRNAs of interest.
Total RNA from 15 mg mouse brain tissue infected with SFV4 or SFV4miRT was extracted using RNeasy-Mini Kit (Qiagen). RNA samples (1 μL) at the concentration of 100 ng/μL were used directly to determine viral genome copies, IFNα and IFNβ mRNA by real-time PCR analysis using iScript One-Step RT-PCR Kit (BioRad). Viral genomes were determined by using primers annealing to nsP1 encoding region: nsP1.F 5′-CGCCAAAAGATTTTGTTCCA-3′; nsP1.R, 5′-CCATCGTGGGTGGTTAATCT-3′. Primers used for murine IFNα detection were mIFNα.F: 5′-AGGACAGGAAGGATTTTGGA-3′, mIFNα.R: 5′-GCTGCTGATGGAGGTCATT-3′, for murine IFNβ detection were mIFNβ.F: 5′-CACAGCCCTCTCCATCAACT-3′, mIFN-β.R: 5′-GCATCTTCTCCGTCATCTCC-3′, and for house-keeping murine hypoxanthine-guanine phosphoribosyltransferase (HPRT1) detection were HPRT.F: 5′-CATAACCTGGTTCATCATCGC-3′, HPRT.R: 5′- GGAGCGGTAGCACCTCCT-3′. All primers were from Sigma-Aldrich. Relative expression of murine IFNα and murine IFNβ was calculated relative to HPRT1 expression levels. Data were evaluated using the 2−ΔΔCT method (23).
In vitro cell killing assay using GL261, CT-2A, NXS2, and HGCC cells
Mouse glioma cell lines GL261, CT-2A, and mouse neuroblastoma cell line NXS2 were seeded at densities 20,000 cells/well (96 well plate) and infected with SFV4, SFV4miRT, A7/74, or A7/74miRT at multiplicity of infection (MOIs) 0.001 to 10 plaque forming units (PFU)/cell. Cells infected with SFV4miRT were pretreated with murine IFNβ (mIFNβ; R&D Systems) at concentrations 1 ng/mL, 0.01 ng/mL, or only medium. Cell viability was analyzed at 3 days post virus infection using MTS aqueous cell titer reagent (Promega).
Human glioblastoma HGCC cell lines were seeded at density 5,000 cells/well (96-well plate) and infected with SFV4miRT at MOIs 0.01 to 1 PFU/cell 24 hours after seeding. Cells were pretreated with human IFNβ (hIFNβ; Peprotech) at concentrations 10, 1, 0.01 ng/mL, or only medium before virus infection. Cell viability was analyzed at 2 and 3 days postinfection using AlamarBlue Cell viability reagent (Thermo Fisher Scientific).
IFNβ ELISA and flow cytometry
Murine GL261, CT-2A, NXS2, and human HGCC cells were plated at density 50,000 cells/well (24-well plate) in 250 μL medium, infected with SFV4 at MOI-0.01 or 1 PFU/cell; control cells were uninfected. Supernatants were collected after 24 hours. mIFNβ was quantified using LEGEND MAX Mouse IFNβ ELISA Kit (BioLegend). IFNβ secretion by HGCC cells was quantified using VeriKine Human Interferon Beta ELISA Kit (PBL Assay Science). Murine cells were stained at 24 hours postinfection using biotin-labeled anti-mouse IFNα/β receptor (IFNAR)-1 antibody (BioLegend) and streptavidin-conjugated Alexa Fluor 488 (Thermo Fisher Scientific), whereas HGCC cells were stained with phycoerythrin (PE)-conjugated human IFNAR1 antibody (R&D Systems). Stained cells were analyzed in BD FACS Canto II (BD Biosciences).
Four to five weeks old outbred NMRI female mice (Janvier Labs) were injected intravenously or intraperitoneally with SFV4-nLuc or SFV4-nLucmiRT (1 × 107 PFU in 100 μL PBS/mouse). Mice were sedated using isoflurane (Isofluran Baxter, Baxter Medical AB); sites of virus replication were revealed by measuring nano-luciferase (nLuc) activity using Nano-Glo Luciferase Assay substrate (100 μl i.v., diluted 1:40 in 1× PBS; Promega) and NightOWL in vivo imaging system (Berthold Technologies). Some mice were sacrificed 3, 4, or 5 days post virus injection and brain samples were either snap frozen or fixed with cold 4% paraformaldehyde overnight. Fixed tissues were stored in 70% ethanol until paraffin embedment.
Four to five weeks old female, A/J mice (Taconic) were implanted with NXS2 cells (2 × 106 cells in 100 μL PBS) subcutaneously in the hind flank and were treated intratumorally (5 × 107 PFU in 50 μL) or intravenous (1 × 107 PFU in 100 μL) injections of either SVF4miRT or A7/74miRT on day 7 posttumor inoculation. Tumor growth was monitored by caliper measurements and tumor size was calculated using the ellipsoid volume formula (length × width × depth × π/6).
Four to five weeks old female C57BL/6NRj mice (Janvier Labs) were used for the orthotopic glioma models. GL261-Fluc (2 × 104 cells in 2 μL DPBS) or CT-2A-Fluc (5 × 104 cells in 2 μL DPBS) were injected intracranially 1 mm anterior and 1.5 mm right from bregma at 2.7 mm depth using a Hamilton syringe and stereotactic equipment (Agnthos AB). Tumor growth was measured as bioluminescence signal using D-luciferin (150 mg/kg i.p.; PerkinElmer) and NightOwl imaging. All mice that had stable luciferase signal 7 or 8 days after tumor inoculation received a single dose of 4 × 107 PFU or 1 × 108 PFU of SFV4miRT virus or PBS intravenously. Mice were sacrificed upon appearance of symptoms like paralysis, hunchback, loss of more than 10% of body weight, or notable distress.
Paraffin-embedded brain tissues were sliced into 6 μm sections and deparafinized. Antigen revival was done by heating the slides for 20 min at 121°C in antigen revival solution (Vector Laboratories). Sections were blocked with goat serum (Vector Laboratories) and stained with rabbit antibody (diluted 1:3,000 in PBS) against SFV structural proteins (kind gift form Dr. Ari Hinkkanen, University of Eastern Finland, Finland). Mouse brain cell types were detected using mouse anti-MAP2 (Sigma), mouse anti-GFAP (Sigma), and mouse anti-CNPase (Sigma) antibodies. IFNβ was detected using rabbit anti-IFNβ polyclonal antibody (Thermo Fisher Scientific). Primary antibody staining was detected by probing with goat-anti-rabbit-HRP or goat-anti-rabbit-AF647 or donkey-anti-mouse-AF555 secondary antibodies (Thermo Fisher Scientific). The sections were imaged in Zeiss AxioImager microscope (Zeiss).
One-way ANAOVA post hoc Tukey test for multiple comparisons was used for statistical comparison of means between more than two experimental groups in an experiment. Statistical comparison of Kaplan–Meier survival curves of mice treated with different viruses was performed by Log-rank (Mantel–Cox) test. Association of cell viability with IFNβ secretion and IFNAR1 expression was evaluated by fitting a linear regression model by backward selection. Associations with P-value <0.05 was considered as statistically significant. Results were analyzed using GraphPad-Prism 6 (Graph-Pad Software) and R statistical programming software.
Biosafety level and ethics declaration
Swedish Work Environment Authority has approved the work with genetic modification of SFV [ID number 202100-2932 v66a14 (laboratory) and v67a10 (mice)]. All experiments regarding modified SFV were conducted under Biosafety level 2. The local animal ethics committee in Stockholm (N164/15, N170/13) approved the animal studies. All human glioblastoma multiforme samples were collected in accordance with protocols approved by the regional ethical review board (2007/353) and after obtaining written consent by all of the patients.
Other related material and methods can be found in the supplementary document.
miRNAs miR124, miR125, and miR134 are expressed in healthy CNS cells but not in glioblastoma multiforme and neuroblastoma cells
Expression of miR124, miR125, and miR134 were examined in a variety of cells and tissues. All healthy murine tissues from the CNS (brain, spine, and DRG) expressed miR124 and miR125 whereas miR134 was expressed only in healthy mouse brain. In vitro cultured murine neural stem cells (NSC) had low expression of all three miRNAs, differentiated neurons expressed miR124, oligodendrocytes expressed miR134, and astrocytes had relatively low expression of all three miRNAs (Fig. 1A, miR expression values normalized to U6 snRNA; Supplementary Table S1). Most human and mouse neuroblastoma and glioblastoma multiforme cells had lower expression of all three miRNAs in comparison to healthy murine neural tissues. Exceptions were the U3034 HGCC cells, which expressed moderate levels of all three miRNAs, and U3024 with expression of miR134 (Fig. 1A). Non-CNS-related tumor cell lines HeLa (cervical cancer) and mel526 (melanoma) had low expression of all three miRNAs as expected (Fig. 1A).
miRT-detargeted SFV retains replication ability in tumor cells but is attenuated in in vitro differentiated murine CNS cells
To prevent neurovirulence of SFV, we inserted two copies of sequences complementary to each of miR124, miR125, and miR134 (hereafter referred to as target sequences, SFV4miRT or A7/74miRT) in the beginning of 3′UTR of the SFV genome that was modified to contain duplicated subgenomic (SG) promoter (Fig. 1B). The oncolytic capacity of SFV4miRT was compared to its wild-type counterpart (SFV4) in vitro. Neuroblastoma and glioma cells were infected with SFV4miRT or SFV4, using amount of virus corresponding to MOIs from 0.001 to 10 PFU/cell (titrated on BHK-21 cells). No difference was observed between viruses with and without miRTs to kill murine neuroblastoma NXS2 (Fig. 1C), murine glioma GL261 (Fig. 1D), or CT-2A cells (Fig. 1E). CT-2A cells were more susceptible to viral oncolysis, which is probably due to more susceptibility to virus infection whereas GL261 were more resistant to virus killing, which may reflect that these cells are poorly infected by SFV viruses. The same trend was observed when A7/74 and A7/74miRT strains were tested (Supplementary Fig. S3A–S3C). Also, all four viruses efficiently killed human neuroblastoma cell lines in a dose dependent manner (Supplementary Fig. S2A–S2D).
In vitro cultured NSC, differentiated neurons, astrocytes, and oligodendrocytes were infected with SFV4-GFP or SFV4-GFPmiRT to assess toxicity in healthy CNS cells. Virus replication (GFP expression as an indirect measure of viral mRNA translation) for the SFV4-GFPmiRT virus was largely attenuated in neurons, astrocytes, and oligodendrocytes but not in NSC (Fig. 1F and G). There was also a significant reduction in the percentage of relative GFP positive BHK-21 cells for the SFV4-GFPmiRT and A7/74-GFPmiRT infected cells when transfected exogenous with miR124, miR125, or miR134, whereas SFV4-GFP and A7/74-GFP were not affected by the addition of miRNAs (Supplementary Fig. S1).
SFV4miRT has reduced replication in adult mouse brain, neurovirulence, and type-I IFN response compared to SFV4
Infection and replication characteristics of SFV4miRT were assessed in vivo by infecting intravenously with 1 × 107 PFU of SFV4 and SFV4miRT. Mice infected with SFV4 start developing virus-related symptoms like mild paralysis at day 3 postinfection and had to be sacrificed due to severe neurological symptoms at day 5 (Fig. 2A and C). Mice infected with SFV4miRT did not develop any visible virus-related symptoms (Fig. 2A and D). Infected mice were sacrificed on day 3 or 5 postinfection and virus replication in mouse brain was quantified. SFV4-infected mice had very high viral RNA copy numbers at day 3 (>106 viral genome copies/g) and that increased even further by day 5 (>107 viral genome copies/g; Fig. 2B). The average viral genome copy number in the brain for SFV4miRT-infected mice was in the range of 103 viral genome copies per gram on day 3 and 104 viral genome copies on day 5 (Fig. 2B), which is around 1,000 times lower compared to the neurovirulent SFV4. Mice infected with nanoluciferase-encoding virus SFV4-nLuc either intraperitoneally (Fig. 2E) or intravenously (Fig. 2F) displayed wide distribution of virus infection at day-1 postinfection, which remained on day-4. SFV4-nLucmiRT infected mice had very low luminescence signal day-1 postinfection and luminescence signal was completely lost by day-4 except in the tail region of the mice, who were intravenously-injected in the tail vein (Fig. 2E and F). SFV4miRT induced significantly lower IFNα and IFNβ mRNA levels in the brain as compared to the wild type SFV4 counterpart both at 3 and 5 days postinjection (Fig. 2G and H). IFN-β protein levels were also verified by IHC in SFV4 and SFV4miRT-infected mouse brain at day 5 (Fig. 2I and J). IHC analysis also confirmed the observation made by qRT-PCR analysis of the IFNβ mRNA levels in the brain.
Brain sections from SFV4-infected, nontumor-bearing mice had positive staining for SFV proteins (Fig. 3B). Mice infected with SFV4miRT did not have any strong staining patterns for viral proteins above the background staining observed in PBS-treated mice (Fig. 2A and C). SFV-infected mouse brains were also costained for neuron-specific (MAP2), oligodendrocyte-specific (CNPase), and astrocyte-specific (GFAP) markers. SFV4 primarily infected neurons (Fig. 3E), seen as colocalized SFV (red) and MAP2 (green) staining, also infected oligodendrocytes to some extent (Fig. 3H) but clearly did not infect astrocytes (Fig. 3K). The SFV4miRT replication was attenuated in all the three cell types; neurons (Fig. 3F), oligodendrocytes (Fig. 3I), and astrocytes (Fig. 3L), and the staining pattern resembled that of brain from PBS-treated mice (Fig. 3D, G, and J). These data confirm that miRT-tagged virus replication was strongly attenuated in mouse brain.
SFV4miRT prolonged survival of mice bearing subcutaneous neuroblastoma tumors
Therapeutic efficacy of SFV4-miRT was evaluated in A/J mice implanted s.c. with syngeneic murine neuroblastoma NXS2. Tumors grew vigorously in the PBS-treated group and all mice were sacrificed before day 20. Tumor growth for individual mice (Fig. 4A and C) was significantly delayed and a significant survival benefit (Fig. 4B and D; 50% remained tumor free) was observed in mice treated with SFV4miRT irrespective of route of virus administration. None of SFV4-miRT treated animals developed encephalitis and the mice were sacrificed due to tumor growth (volume > 1,000 mm3). All mice that remained tumor free after day 60 were re-challenged with NXS2 cells and none of them developed tumors (data not shown), indicating that immunological memory was formed. The miRNA-detargeted IFN-sensitive SFV strain A7/74-miRT also showed similar therapeutic efficacy on curing mice bearing NXS2 tumors, regardless of administration route (Supplementary Fig. S3D–S3G).
SFV4miRT prolongs survival of mice bearing orthotopic CT-2A glioma but not GL261 glioma
Therapeutic efficacy of SFV4miRT was evaluated in two orthotopic murine gliomas, GL261 and CT-2A, in syngeneic C57BL/6 mice. All mice inoculated with GL261 had tumor growth at day 8, the time of treatment. GL261 tumor-bearing mice treated with SFV4miRT did not have statistically significant survival benefit compared to PBS treated group. Only one mouse in the SFV4miRT treatment group survived (Fig. 5A) and one mouse had tumor regression but later relapsed (Fig. 5C). Around 85% of mice inoculated with CT-2A cells developed tumors at day 8 after tumor implantation, and only those with tumors were included and treated. CT-2A tumor-bearing mice treated with SFV4miRT had a significant survival benefit and delayed tumor growth (3/11 mice survived, 27.3% survival) compared to the PBS-treated group (Fig. 5B). Representative pictures of mice with luminescence signal and immunohistochemistry from GL261 and CT-2A tumor-bearing mice treated with SFV4miRT are shown (Fig. 5C–H). CT-2A tumor-bearing mice treated with SFV4miRT had strong staining for SFV proteins (Fig. 5H) in the tumor, whereas the GL261 tumor bearing mice did not have any staining for SFV proteins (Fig. 5G).
Therapeutic efficacy of SFV4miRT treatment negatively correlates with IFNβ secretion by tumor cells upon virus infection
To analyze mechanisms leading to difference in susceptibility to SFV-killing in vitro and correspondingly very different therapeutic efficiencies in the three in vivo models, the abilities of NXS2, GL261, and CT-2A cell lines to mount type-I IFN response upon SFV4miRT infection was measured. Cells were infected with SFV4miRT at MOI-0.01 or 1 PFU/cell and 24 hours later their ability to secrete IFNβ and regulate IFNAR1 expression was analyzed. Uninfected cells and cells infected at MOI-0.01 secreted undetectable amounts of IFNβ. Upon infection at MOI-1 GL261 cells secreted high amounts of IFNβ whereas CT-2A cells secreted moderate amounts of IFNβ and NXS2 cells secreted undetectable amounts of IFNβ (Fig. 6A). There was no significant change in the surface expression of IFNAR1 upon infection of cells with SFV4miRT for both MOIs evaluated (Fig. 6B). The reduced expression of IFNAR1 observed on CT-2A cells at MOI-1 is due to cell death caused by virus infection (Fig. 6B).
Previous studies have revealed that SFV4 is partly resistant to antiviral effects of type-I IFN (16). Therefore, cell-killing assays were performed on NXS2, GL261, and CT-2A cells pretreated with high amounts of exogenous recombinant murine IFNβ. Lower concentration of IFNβ (0.01 ng/mL) was sufficient to reduce death of CT-2A cells at MOI-0.01 and 0.001 but had virtually no effect on survival of NXS2 cells (Fig. 6C). At a high virus dose (MOI-10), no protective effect of exogenous IFNβ was observed for NXS2 and CT-2A (Fig. 6C). GL261 cell line was more resistant to virus killing compared to NXS2 and CT-2A even in the absence of exogenous IFNβ. However, in the presence of exogenous IFNβ (0.01 and 1 ng/mL), GL261 cells were protected from SFV4miRT killing even at the highest virus dose (MOI-10; Fig. 6C).
SFV4miRT efficiently kills patient-derived primary HGCC in vitro
Oncolytic effect of SFV4miRT in 11 HGCC lines either in the presence or absence of exogenous human IFNβ was also evaluated. The cohort of cell lines used represented all four subtypes of glioblastoma multiforme and detailed information is mentioned in Supplementary Table S2. After 72 hours, most HGCC cell lines were efficiently killed in a dose-dependent manner by SFV4miRT (Supplementary Fig. S4). The efficacy of virus-mediated tumor cell killing decreased in the presence of exogenous hIFNβ in a concentration-dependent manner (Supplementary Fig. S4). Figure 6D represents mean-relative viability across different viral doses (MOI-0.01 to 1 PFU/cell) in the absence or presence of 1 ng/mL exogenous hIFNβ. The presence of IFNβ inhibited cell killing, with U3024 being an exception having unchanged cell viability when hIFNβ was added. We believe that this is attributed to the high miR134 expression of U3024, leading to virus genome degradation (Fig. 1A).
IFNβ secretion from the HGCC cell lines was measured before and after SFV4miRT infection. None of the uninfected HGCC lines secreted IFNβ (Fig. 6E). Only U3024 and U3213 secreted detectable levels of IFNβ when infected by SFV4miRT at low virus dose (MOI-0.01). However, at high virus dose (MOI-1), a majority of the HGCC lines (U3013, U3021, U3024, U3028, U3034, U3053, U3054, and U3213) secreted varied amounts of IFNβ (Fig. 6F). Most of the HGCC lines that secreted IFNβ were less effectively killed by SFV4miRT in the absence of exogenous hIFNβ (black squares above dotted line; Fig. 6D). Furthermore, the mean relative viability of cells after SFV4miRT infection correlated with the ability of cells to secrete IFNβ (P = 0.01164; Fig. 6G). The HGCC line U3034 secreted very low IFNβ upon virus infection, but had high cell viability (Fig. 6G, grey dot), which may be because U3034 had expression of all three miRNAs (Fig. 1A). Change in IFNAR1 expression was measured 24 hours after SFV4miRT infection by flow cytometry. All HGCC cells expressed basal levels of IFNAR1 (Fig. 6F, white bars) and had only feeble change in receptor expression after virus infection. Mean relative cell viability after SFV4miRT infection correlated to IFNAR1 expression (MFI values) after SFV4miRT infection (P = 0.00205; Fig. 6H). Taken together, efficacy of SFV4miRT to kill HGCC lines was negatively associated with the amount of IFNβ secreted by cells upon virus infection and levels of IFNAR1 expression.
Many oncolytic viruses have been tested and proven to be safe in the clinic for glioblastoma multiforme treatment (4, 24). Despite encouraging safety results there are many factors like virus delivery, anti-virus immune response, intratumoral spread of infection that still need to be addressed to significantly improve efficacy, and make oncolytic virus therapies for glioblastoma multiforme a reality (25, 26). An important factor impeding development of successful therapies for glioblastoma multiforme is lack of models closely representing the characteristics of tumor in patients. We utilized the HGCC biobank to test efficacy of oncolytic SFV therapy. HGCC lines are cultured in serum-free medium and retain the characteristics of their original tumors as compared to the traditionally used human glioblastoma multiforme cell models (27, 28). They also express NESTIN and SOX2, markers that are usually associated with glioblastoma multiforme stem cells. It is commonly believed that glioblastoma multiforme stem cells are not efficiently killed by conventional therapies (29, 30). Hence, efficacy of SFV and other potential therapies to kill HGCC lines may aid in better evaluating its clinical translational ability.
The SFV A7/74 strain has been successfully tested safe for use as an oncolytic agent in several preclinical models for glioblastoma multiforme and other cancer types (8, 31, 32). However, type-I IFN responsive gliomas are resistant to A7/74 virotherapy in murine models (14). The neurovirulent strain SFV4 is interesting because it has the ability to partly resist type-I antiviral defense (33) and this phenotype is important for treatment of IFNα/β responsive tumors (16). To attenuate neurovirulence of the lethal SFV4 strain, we inserted target sequences complementary to miR124, miR125, and miR134 into the SFV genome (Fig. 1B). The initial choice of these miRTs was based on their expression in CNS reported in the literature. miR124 is highly expressed in all brain regions, predominantly in the neurons except for the pituitary gland (21, 34, 35), miR125 is expressed in mid-hindbrain boundary and neurons in mouse (21, 36) and miR134 is a brain-specific miRNA, localized in dendrites and synaptic sites (22). We also confirmed that these miRNAs are highly expressed in normal mouse CNS cells, including in vitro-differentiated neurons, and are downregulated in several murine gliomas and neuroblastomas (Fig. 1A). Most HGCC lines also had low expression of these miRNAs, which makes these miRNAs suitable targets to be used to attenuate virus replication in normal CNS cells however, expression profiles in healthy human brain cells need to be confirmed before clinical translation. Previously, miRT124-tagged SFV4 was shown to possess significantly reduced neurovirulence (16, 20). But, a few mice still developed severe neurological symptoms, with virus replication detected in the oligodendrocytes of corpus callosum and spine (16, 20). The triple miRT de-targeted SFV4 virus (SFV4miRT) presented herein had attenuated replication in mouse brain cells specifically neurons, and oligodendrocytes (Fig. 3). SFV4 did not infect astrocytes (Fig. 3J–L), which is in accordance with previous reports (37). In immune competent mice SFV4 is also known not to infect any other organ except CNS (37), therefore we believe that SFV4miRT has an improved safety profile.
Insertion of miRTs did not inhibit virus replication in tumor cells lines that did not express any of the miRNAs (Fig. 1C–E and Supplementary Figs. S2 and S3). But in cell lines where one or more of these miRNAs were expressed (U3024 and U3034), virus replication was attenuated when administered at low MOIs (Supplementary Fig. S4). Nontumor-bearing brain samples from SFV4miRT-infected mice had no or very little IFN-α/β secretion compared to samples from mice infected with SFV4 (Fig. 2G–J). The levels of IFNα/β secretion also correlated with virus load in the brain, that is, more IFNα/β on day 5 postinfection, where the titers of the SFV4 virus in the brain were also higher (Fig. 2B, G–J). This finding is in accordance with previous reports, where miRT124 attenuated SFV4 was used (20) and confirms that IFNβ secretion depends on the efficiency of virus replication in the brain (38).
Successful biological outcomes of type-I IFN response depends on expression of all components in the IFN signaling pathway like IFNα/β itself, IFN receptor (IFNAR), signaling molecules JAK1, TYK2, STAT1, STAT2, and Interferon Stimulated Genes (ISG; ref. 39). In three murine tumors models tested, the neuroblastoma cell line NXS2 did not secrete IFNβ upon virus infection, but the two glioma cell lines CT-2A and GL261 secreted IFNβ upon virus infection, with GL261 secreting the highest amount of IFNβ (Fig. 6A). In the murine tumor models, IFNβ secretion levels (autocrine effect) negatively correlated with therapeutic efficacy in vivo (Figs. 4–6). NXS2-bearing mice benefitted the most (50% survival) from treatment with SFV4miRT (Fig. 4) followed by CT-2A tumor-bearing mice, then GL261 tumor-bearing mice (Fig. 5A and B). It should be noted that NXS2 model used was subcutaneous, whereas orthotopic brain models were used for GL261 and CT-2A, and therapeutic efficiencies can differ because of the location of the tumor. GL261-bearing mice treated with the IFNβ-sensitive strain A7/74miRT did not have any survival benefit (data not shown). This is in accordance with previously published data showing that even low amount of IFNβ is enough to inhibit replication of A7/74 strain in both GL261 and CT-2A models (14, 16). SFV4miRT efficiently killed most of the HGCC lines in vitro and efficacy depended on the amount of IFNβ secreted by the cells upon virus infection (autocrine effect; Fig. 6G), which is accordance with the murine cell line data.
In the presence of exogenous IFNβ, SFV4miRT killed NXS2 and CT-2A cells in a dose-dependent manner in vitro (Fig. 6C). However, GL261 were resistant to SFV4miRT oncolysis in the presence of exogenous IFNβ, even at the highest virus dose tested (Fig. 6C). As previously observed, in clones of mouse colon carcinoma, CT26WT and CT26LacZ differential expression of ISGs can contribute to dramatic difference in the efficacy of SFV therapy (40). ISG expression in CT-2A and GL261 is unknown, which may be a reason for the observed difference in oncolytic response. In HGCC lines, addition of exogenous IFNβ decreased SFV4miRT oncolysis, indicating that all HGCC lines are responsive to IFNβ (Fig. 6D). However, at high SFV4miRT dose, IFNβ did not inhibit viral oncolysis, confirming the ability of SFV4 to partly resist type-I antiviral defense (Supplementary Fig. S3), as previously reported (33).
The ability of tumors to initiate antiviral defense is an important factor that determines the therapeutic efficacy of SFV4miRT. This emphasizes the need to test oncolytic potential of SFV in immune competent models as compared to xenografts where excellent efficacy is observed (41). It may still be possible to use the avirulent, type-I IFN-sensitive A7/74 strain for treatment of non-IFNβ responsive tumors like NXS2 but our data stresses the need for an IFN-insensitive SFV strain for successful treatment of IFN-responsive tumors. Recent studies suggest that viral-vaccines combined with checkpoint inhibition (αPD-1 or αCTLA4 antibodies) is beneficial for treatment of aggressive tumors like GL261 (42). Oncolytic viruses are naturally immunogenic; combination of SFV4miRT with checkpoint inhibition for treatment of glioblastoma multiforme would be worth to investigate. In conclusion, our results show that, insertion of multiple miRTs reduces SFV4 neurovirulence and yields an improved safety profile of SFV4 as an antitumor agent.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M. Ramachandran, D. Yu, M. Dyczynski, A. Lulla, J. Leja-Jarblad, M. Essand
Development of methodology: M. Ramachandran, D. Yu, M. Dyczynski, L. Zhang, A. Lulla, V. Lulla, S. Saul, A. Merits
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Ramachandran, D. Yu, M. Dyczynski, S. Baskaran, V. Lulla, S. Saul, A. Dimberg, J. Leja-Jarblad
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Ramachandran, D. Yu, M. Dyczynski, S. Baskaran, S. Nelander, A. Merits, J. Leja-Jarblad, M. Essand
Writing, review, and/or revision of the manuscript: M. Ramachandran, D. Yu, M. Dyczynski, S. Baskaran, S. Saul, S. Nelander, A. Dimberg, A. Merits, J. Leja-Jarblad, M. Essand
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Ramachandran, D. Yu, M. Dyczynski
Study supervision: D. Yu, S. Nelander, J. Leja-Jarblad, M. Essand
The authors wish to thank Patrik Johansson (Uppsala University) for tips and guidance with statistical analysis, Berith Nilsson, Grammatiki Fotaki, Chuan Jin, and Tina Sarén (Uppsala University) for assistance in the lab, Dr. Miika Martikainen for fruitful discussions and reviewing the manuscript. We also thank Prof. Ari Hinkkanen (University of Eastern Finland) for providing us α-SFV antibodies and Dr. Markus Vähä-Koskela (Ottawa Hospital Research Institute) for providing the CT-2A cell lines.
The Swedish Children Cancer Foundation (PROJ12/082), the Swedish Cancer Society (CAN2013/373), and the Swedish Research Council (K2013-22191-01-3) supported this work (to M. Essand). Estonian Research Council (Eesti Teadusagentuur) and grant number 20-27 (to A. Merit). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
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.