Abstract
Zika virus (ZIKV) is largely known for causing brain abnormalities due to its ability to infect neural progenitor stem cells during early development. Here, we show that ZIKV is also capable of infecting and destroying stem-like cancer cells from aggressive human embryonal tumors of the central nervous system (CNS). When evaluating the oncolytic properties of Brazilian Zika virus strain (ZIKVBR) against human breast, prostate, colorectal, and embryonal CNS tumor cell lines, we verified a selective infection of CNS tumor cells followed by massive tumor cell death. ZIKVBR was more efficient in destroying embryonal CNS tumorspheres than normal stem cell neurospheres. A single intracerebroventricular injection of ZIKVBR in BALB/c nude mice bearing orthotopic human embryonal CNS tumor xenografts resulted in a significantly longer survival, decreased tumor burden, fewer metastasis, and complete remission in some animals. Tumor cells closely resembling neural stem cells at the molecular level with activated Wnt signaling were more susceptible to the oncolytic effects of ZIKVBR. Furthermore, modulation of Wnt signaling pathway significantly affected ZIKVBR-induced tumor cell death and viral shedding. Altogether, these preclinical findings indicate that ZIKVBR could be an efficient agent to treat aggressive forms of embryonal CNS tumors and could provide mechanistic insights regarding its oncolytic effects.
Significance: Brazilian Zika virus strain kills aggressive metastatic forms of human CNS tumors and could be a potential oncolytic agent for cancer therapy. Cancer Res; 78(12); 3363–74. ©2018 AACR.
Introduction
The recent outbreak of Zika virus (ZIKV), especially throughout South and Central Americas, revealed an unprecedented impact of gestational infection on neurodevelopment, resulting in severe central nervous system (CNS) development effects in neonates, such as microcephaly and other associated abnormalities (1). Recent studies showed that ZIKV prominently infects neural stem and progenitor cells (NPC) and disrupts key cellular processes, e.g., survival, proliferation, and differentiation (2, 3), leading to massive cell death and growth reduction.
Notably, aggressive CNS embryonal tumors with high incidence in infants are originated from NPC aberrations, affecting key cell signaling pathways that regulate neurogenesis, such as the mTOR/Wnt pathway (4). These tumors are comprised by cells with neural stem–like features, also known as cancer stem cells (CSC), which are highly tumorigenic and resistant to classical cancer therapies (5). CNS tumors enriched in stem-like cancer cells are very difficult to treat and usually associated with poor prognosis (6). Available therapies have a low efficiency and severe adverse effects that include endocrine, motor, and cognitive deficits.
Oncolytic viral therapy has emerged as an alternative approach to treat aggressive, fast-growing forms of cancer. Oncolytic viruses are defined as native or genetically modified viruses that are able to directly infect and lyse tumor cells (7). Because ZIKV infects preferentially NPC, we hypothesized that ZIKVBR could act as an oncolytic agent in particular against aggressive and metastatic human CNS embryonal tumors.
Materials and Methods
Human and animal samples
The study followed the International Ethical Guideline for Biomedical Research (CIOMS/OMS, 1985) and was approved by the Institutional Animal Experimentation Ethics Committee (CEUA-USP 290/2017; CEUA-Instituto Butantan 3473210817). A total of 66 animals were included in the present study. Animals with 30% weight loss and/or visible tumor and/or ataxia and/or freezing were subjected to euthanasia, and all efforts were made to minimize suffering. Human samples (peripheral blood and tumors) were obtained after written informed consent of the patients, according to protocol approved by the Internal Review Board (CEP-IB No. 121/2011).
Cell lines and cultures
Three embryonal CNS tumor cell lines: DAOY (medulloblastoma, ATCC HTB-186), USP13-MED (medulloblastoma, in-house established; ref. 8), and USP7-ATRT (atypical teratoid/rhabdoid tumor, in-house established); three non-CNS tumor cell lines: MCF-7 (breast cancer, ATCC HTB-22), HCT-8 (colorectal cancer, ATCC CCL-244), and DU-145 (prostate cancer, derived from brain metastasis, ATCC HTB-81); one control human-induced pluripotent stem cell (hiPSC, C2535, in-house reprogrammed)–derived NPCs and neurons were analyzed. Detailed information regarding the generation and characterization of hiPSCs, NPCs, and neurons is provided in Supplementary Materials. All commercial tumor cell lines were grown according to ATCC recommendations, and cell authentication was performed by high-resolution karyotype analysis. USP13-MED and USP7-ATRT were isolated and characterized as previously reported (8). After thawing, all cell lines were cultured under standard conditions (5% CO2 (g) at 37°C) up to 4 weeks (passages 1–4) and tested for Mycoplasma contamination by PCR (Cat. MP002; Sigma-Aldrich), before use in the described experiments.
Zika virus strain
ZIKVBR was donated by Dr. Pedro Vasconcelos, Instituto Evandro Chagas, Brazil. Viral stock was established in VERO cells, cultured in serum-free medium (VP-SFM, Thermo Scientific) after two serial passages at a multiplicity of infection (MOI) of 0.05, and supernatants harvested after 72 hours. Virus stocks were titrated by plaque-forming units (PFU) assay on VERO cells.
In vitro two-dimensional and three-dimensional infection with ZIKVBR
Cells (4.2 × 103 cells/cm2) in two-dimensional were exposed to ZIKVBR (MOI: 0.01, 0.1, 1, and 2) or Mock for 1 hour at 37°C, washed with culture medium, and maintained up to 72 hours. Tumorspheres (three-dimensional, 3D), generated and characterized as previously reported (8), were infected with MOI 1, 2, or Mock, with viral exposure for 2 hours to ensure complete sphere infection. ZIKVBR PFU and copy numbers were analyzed from supernatant media every 24 hours. Total viable cells and spheres area were daily evaluated with ImageJ software.
Functional ZIKVBR yielding analysis
Tumor cells were infected with ZIKVBR (MOI 1), based on prior PFU results of all tested cell lines. Seventy-two hours later, culture supernatants were harvested and used to infect human NPC (8.4 × 104 cells/cm2 infected with 1.26 × 106 ZIKVBR viral copies). After another 72 hours, NPC culture supernatants were harvested for PFU analysis in order to verify the efficiency of functional ZIKVBR yielding by each tumor cell line. The volumes used to infect NPC were calculated based on qRT-PCR results.
Immunofluorescence
Brain sections and fixed tumor cells (3.7% formaldehyde for 10 minutes) were treated for 30 minutes with 0.1% Triton X-100 in 1× PBS, prior to 2-hour incubation in 5% bovine serum albumin in 1× PBS, and then incubated overnight with the primary antibody (Supplementary Table S1) at 4°C. Brain sections and cells were then washed 3 times in 1× PBS, incubated with the secondary antibody for 2 hours (Supplementary Table S1), and washed again 2 times in 1× PBS. Cell nuclei were stained with 1 μg/mL DAPI for 2 minutes. Tissue sections and tumor cells were mounted on glass slides and cover slipped with VectaShield. All images were taken in confocal microscope (Zeiss LSM 800).
Cell death analysis
Staining for caspase-3/7 (CellEvent, Thermo Fisher) and propidium iodide (PI) was performed in 1 × 105 tumor cells, 72 hours after infection with ZIKVBR, following the manufacturer's protocol. Flow cytometry analysis was performed using FACS Aria II (BD) collecting 10,000 events per run. Analysis was carried using FlowJo software.
Wnt–β-catenin pathway modulation
Cell lines (USP7-ATRT, USP13-MED, and DAOY) were cultured as previously described, with addition of CHIR99021 (Wnt pathway inductor; 5 nmol/L, 50 nmol/L, 100 nmol/L, 500 nmol/L, or 1,000 nmol/L), IWP-2 (Wnt pathway inhibitor; 20 nmol/L, 100 nmol/L, 200 nmol/L, 1,000 nmol/L, or 2,000 nmol/L), or DMSO (dilution reagent for CHIR99021 and IWP-2), together with ZIKVBR (MOI 0.1). Seventy-two hours after viral infection, β-catenin expression (see Supplementary Table S1), cell death, and virus titration were evaluated.
Orthotopic metastatic xenograft model/ZIKVBR in vivo injection assay
Tumor cells expressing firefly luciferase were generated with pLV/Luc lentiviral vector, as previously described (9). The orthotopic metastatic model was performed as previously described in detail (10). After tumor establishment, according to cell line–dependent growth kinetics (Ti = 1 week for USP7-ATRT and 2 weeks for USP13-MED and DAOY), 2 × 103 ZIKVBR particles/2 μL were injected in the right lateral ventricle. Sham group was injected with DMEM, instead of tumor cells. The initial amount of animals per experimental group was 10. Few animals died during the surgery procedure and therefore could not be included in the experiment. Tumor development was assessed in vivo with IVIS Imaging System (PerkinElmer) as previously described (8). Metastasis classification followed M-Stage proposed by Chang's system (11). Tumor detection was confirmed by histologic and immunofluorescence analyses. Tumor remission in each experimental animal was evaluated by time course analysis of normalized bioluminescence levels (photon counts) and considered if respective bioluminescence levels in T1, T2, or T3/Ti < 1. Animal activity was monitored as described in Supplementary Materials.
Molecular analysis
Gene expression profiling was determined with Affymetrix GeneChip Human Gene 2.0 ST whole-transcript arrays (GEO accession number: GSE103935). Identification (12) and interactome analysis of differentially expressed genes were performed with Transcriptome Analysis Console (Affymetrix) and IPA Software (Ingenuity Pathway Analysis, Qiagen), respectively. Whole-exome sequencing was performed using Illumina's TrueSeq kits and the Illumina HiSeq sequencer. Candidate pathogenic mutations were determined by allele frequency < 0.05 in NIH 1000 Genomes Project and by SIFT and FATHMM algorithms. Array comparative genomic hybridization was performed using 60K whole-genome platform (Agilent Technologies) following the manufacturer's recommendation. Detailed description is provided in Supplementary Experimental Procedures.
Statistical analysis
All experiments were performed in triplicate, and three independent experiments were carried out. Data were analyzed by one-way and two-way ANOVA followed by Bonferroni post hoc test. The t test with two-tailed unpaired test was used for pairwise comparison. Clinical and pathologic parameters were analyzed by the Fisher exact test. Graphpad Prism software was used to perform all statistical analysis (version 6.0 GraphPad Software Inc.). Quantification of data is represented as mean ± SEM, and P value threshold was as follows: *, 0.05; **, 0.01; ***, 0.001; and ****, 0.0001.
Results
Oncolytic in vitro effects of ZIKVBR
In order to evaluate ZIKVBR oncolytic properties in different tumors, six human tumor cell lines were tested against various MOIs: 0.01, 0.1, 1, and 2. Three non-CNS cell lines, MCF-7, HCT-8, and DU-145, and three CNS embryonal tumor cell lines, USP13-MED (7), DAOY, and USP7-ATRT, were infected in monolayer cultures. Figure 1 and Supplementary Fig. S1A show that 72 hours postinfection (hpi), ZIKVBR induced significant cell death and/or growth reduction in DAOY, USP13-MED, and more pronounced death in USP7-ATRT, starting at MOI 0.1. At least 50% of the USP7-ATRT and USP13-MED cell populations were already infected starting at MOI 0.1 at 48 hpi. Under the same conditions, infection of DAOY cells was less effective, reaching a maximum of about 40% of infected cells at MOI 1 and 2. Complementary flow cytometry analysis (Fig. 2A and B) showed that ZIKVBR infection significantly increased the population of PI-positive cells, whereas caspase-3/7–positive cells remained low in all CNS tumor cell lines. Confocal microscopy analysis confirmed the presence of ZIKVBR particles within the cytoplasm (Fig. 2C). On the other hand, exposure to ZIKVBR had little or no effect on non-CNS tumor cell lines. These results indicate that ZIKVBR induces death involving rupture of the plasma membrane of CNS embryonal tumor cells.
Interestingly, viral titer quantification showed that all CNS embryonal tumor cells produced high amounts of infectious viral particles starting at 24 hpi, but, after 72 hpi, these cells started producing defective viral particles, indicative of host cell dysfunction (Supplementary Fig. S1B–S1D). MCF-7 and HCT-8 cells did not produce significant amounts of ZIKVBR (Supplementary Fig. S1E). Of note, DU-145 cells showed high viral titers during all the infection kinetics experiment, which supports ZIKVBR clinical findings of prolonged viral titers found in human semen samples (13).
ZIKVBR oncolytic effects in tumorspheres and normal neurospheres
All tumor cell lines susceptible to ZIKVBR infection were then cultured in a CSC-tumorsphere promoting system and tested in parallel with normal hiPSC-derived NPC. Figure 3A–F shows that ZIKVBR significantly disrupted CNS tumorspheres and produced high viral titers, particularly in USP7-ATRT cultures. Conversely, despite infected by ZIKVBR, DU-145 tumorspheres were not disrupted, reinforcing a selective oncolytic effect toward the CNS tumor cells. Compared with the cell cultures in monolayer, the 3D culture system showed a greater impact of ZIKVBR on cellular viability. Interestingly, normal NPC neurospheres were less affected by ZIKVBR as compared with CNS embryonal tumorspheres. Neurons derived from these normal NPC, however, were not infected by ZIKVBR (Fig. 3G). Of note, single cells from dissociated CNS embryonal tumorspheres were not capable of forming new tumorspheres after ZIKV infection (Supplementary Fig. S1F and S1G). For characterization of hiPSC, NPC, and neurons, see Supplementary Fig. S2. Together, these results suggest that proliferative tumor cells from CNS origin are more permissive to ZIKVBR infection and virus-mediated death than normal nervous system cells and tumors from other primary origins such as breast, colorectal, and prostate.
ZIKVBR effects in tumor development and metastasis
Next, we tested ZIKVBR oncolytic properties in an orthotopic xenograft animal model in BALB/c nude mice with DAOY, USP13-MED, and USP7-ATRT cell lines. A single dose of 2 × 103 infectious viral particles was injected in the right ventriculum after respective tumor engraftment period of DAOY, USP13-MED, and USP7-ATRT cell lines (Fig. 4A).
After few weeks of viral particles administration, striking results were observed. ZIKVBR induced tumor remission in 20 of 29 animals (USP7-ATRT: 8/10; USP13-MED: 8/9; DAOY: 4/10; Supplementary Fig. S3A–S3C), 7 of which achieved complete remission (two USP7-ATRT and six USP13-MED). Of note, an additional USP7-ATRT–bearing mice had complete remission 3 weeks after ZIKVBR injection, but it was not included in our analysis due to lack of imaging acquisition at the time of injection (Ti). Overall survival of USP7-ATRT tumor–bearing mice was significantly improved (P = 0.0046) by ZIKVBR treatment (Fig. 4B). Three cases of tumor relapse after complete remission were observed in ZIKVBR-treated animals with USP13-MED tumors. Figure 4C–G shows a marked reduction of USP7-ATRT and USP13-MED tumor growth ratio, whereas a poor response of DAOY tumors to ZIKVBR was detected in vivo, which is consistent with the relative lower rates of cell infection previously observed for this cell line.
In addition, in ZIKVBR-treated animals, 60% (3 of 5) of USP7-ATRT tumor–bearing mice had complete metastatic remission, and none of the USP13-MED tumor–bearing mice developed M3 metastasis (Table 1). USP7-ATRT and USP13-MED cell lines were the only capable of generating M3 stage metastasis (tumors in the spinal cord), which occurred in 66% (6 of 9) and 33% (3 of 9) of animals from the mock group, respectively (Table 1).
Tumor . | USP7-ATRT . | USP13-MED . | DAOY . | |||
---|---|---|---|---|---|---|
Ti . | Mock . | ZIKVBR . | Mock . | ZIKVBR . | Mock . | ZIKVBR . |
Activity gain | 0/4 | 2/4 | 1/4 | 0/4 | 3/4 | 3/4 |
Weight gain | 1/9 | 7/10a | 4/9 | 5/9 | 5/9 | 6/10 |
Weight loss | 8/9a | 4/10 | 6/9 | 4/9 | 2/9 | 7/10 |
Event M3 | 6/9 | 5/10 | 3/9 | 0/9 | 0/9 | 2/10 |
Tumor remission | 0/9 | 8/10b | 0/9 | 8/9b | 0/9 | 4/10 |
Tumor relapse | — | 2/8 | — | 4/9 | — | 4/4 |
M3 remission | 0/7 | 3/5 | 0/3 | — | — | — |
M3 recover | — | 0/3 | — | — | — | — |
Tumor . | USP7-ATRT . | USP13-MED . | DAOY . | |||
---|---|---|---|---|---|---|
Ti . | Mock . | ZIKVBR . | Mock . | ZIKVBR . | Mock . | ZIKVBR . |
Activity gain | 0/4 | 2/4 | 1/4 | 0/4 | 3/4 | 3/4 |
Weight gain | 1/9 | 7/10a | 4/9 | 5/9 | 5/9 | 6/10 |
Weight loss | 8/9a | 4/10 | 6/9 | 4/9 | 2/9 | 7/10 |
Event M3 | 6/9 | 5/10 | 3/9 | 0/9 | 0/9 | 2/10 |
Tumor remission | 0/9 | 8/10b | 0/9 | 8/9b | 0/9 | 4/10 |
Tumor relapse | — | 2/8 | — | 4/9 | — | 4/4 |
M3 remission | 0/7 | 3/5 | 0/3 | — | — | — |
M3 recover | — | 0/3 | — | — | — | — |
NOTE: Events expressed in ratios of number of animals/group/Ti. Weight gain and weight loss correspond to 10% to 30% alteration based on the initial animal body weight along time. M3 corresponds to advanced stage of tumor metastasis in the neural axis according to Chang's staging system (11).
Statistical analysis following the Fisher exact test.
a *, P < 0.05, statistical analysis following the Fisher exact test.
b ***, P < 0.001, statistical analysis following the Fisher exact test.
USP7-ATRT cells formed the most aggressive tumors in our experimental model, being the only tumor cell line causing 100% of death rate in the mock control group within 30 days of experimental follow-up. The same poor survival was observed in mice orthotopically injected in low density of cell (2 × 105 per animal). The AT/RT cell line was able to generate M2 and M3 metastatic tumors after 1 week (Supplementary Fig. S4A–S4G).
Histologic analysis confirmed necrotic tumor foci and ZIKVBR-positive tumor cells in brain tissue of USP7-ATRT and USP13-MED tumor–bearing mice (Supplementary Fig. S5A), but significant improvement in overall survival was detected only for animals bearing USP7-ATRT tumors (Supplementary Fig. S5B and S5C). In tumor-bearing mice, viral titer in brain, spinal cord, peripheral blood, and spleen indicated that ZIKVBR infection was more consistently persistent and restricted to host CNS at the time of euthanasia. Interestingly, sham animals (without tumor cell injection) that were only infected with ZIKVBR had persistent virus in peripheral blood, brain, spinal cord, and spleen and decreased survival rate as compared with those bearing tumors (Supplementary Fig. S5D–S5G). These results are in agreement with the previous observation of high noninfectious viral particle production by host tumor cells, especially in USP7-ATRT. The ZIKVBR-mediated cell lysis in USP7-ATRT is so intense that it rapidly destroys tumor sites hampering viral infection amplification.
NPC-like tumor cells are more prone to ZIKVBR-mediated oncolysis
Among the tumor cell lines tested, USP7-ATRT and USP13-MED displayed the highest levels of viral infection, USP7-ATRT being the most sensitive to ZIKVBR-mediated oncolysis. USP7-ATRT cells are highly invasive and proliferative, with a population doubling time (PDT) of 24.16 hours (Supplementary Fig. S6A–S6C). However, the higher USP7-ATRT sensitivity to ZIKVBR is unlikely related to cell proliferation rates because its PDT is equivalent to those from USP13-MED (24.5 hours) and DAOY (29.8 hours) cells (8). Also, both USP7-ATRT and USP13-MED (8) cell lines are enriched in highly tumorigenic CSC with increased resistance to chemotherapeutic agents and capability of generating tumorspheres (Supplementary Fig. S6D–S6H).
A more detailed analysis, integrating whole-exome sequencing, chromosomal copy-number aberrations, and global gene expression profiling, revealed TP53, CDKN1A, CTNNB1, and CCND1 as hotspots of an interactome map of affected proteins in USP7-ATRT cells (Fig. 5). Interestingly, the molecular profile of USP7-ATRT is consistent with SMARCB1/INI1-positive type of AT/RT (Supplementary Fig. S6I–S6O), known to be closely related to normal neural stem cells (14). In fact, USP7-ATRT cells express typical neural stem cell markers (Fig. 5A and B), and a comparative global gene expression profiling revealed a stronger similarity of USP7-ATRT cells with normal NPC than with other CNS embryonal tumors (Fig. 5C–F). In addition, USP7-ATRT cells and NPC are highly similar regarding the pattern of expression of TAM receptor genes and other genes associated with ZIKVBR cellular entry (15, 16, 17). Such neural stem cell molecular fingerprint is correlated with a higher sensitivity of USP7-ATRT cells to ZIKVBR-mediated oncolysis, as compared with the other tumor cell lines.
Wnt/β-catenin activity modulates ZIKVBR-induced oncolysis and efficient viral replication in tumor cells
USP7-ATRT was the most sensitive cell line to the oncolytic properties of ZIKVBR, and the comparative molecular analysis suggested that the Wnt pathway is hyperactive in this specific tumor cell line (Fig. 5G). Given the involvement of the Wnt pathway in normal neural development and tumorigenesis, we further investigated a possible role of Wnt/β-catenin signaling in ZIKVBR susceptibility.
All CNS embryonal tumor cell lines were treated with CHIR99021 or IWP-2, an activator and an inhibitor of Wnt/β-catenin signaling, respectively. Both small molecules were not cytotoxic in mock-infected cells at the concentrations used in this study for all cell lines tested (Supplementary Fig. S7A and S7B). The CHIR99021 treatment significantly increased ZIKVBR-mediated oncolysis of USP7-ATRT and USP13-MED cells, whereas IWP-2 treatment significantly attenuated such virus effect in USP7-ATRT (Fig. 6A and B; Supplementary Fig. S7C–S7F). Although unlikely, a possible cytotoxicity due to a synergistic effect of ZIKVBR and CHIR99021 cannot be ruled out. ZIKVBR-induced oncolysis was not affected by both treatments in DAOY cell line (Fig. 6C; Supplementary Fig. S7G and S7H). Western blotting analysis showed that CHIR99021 treatment efficiently modulated β-catenin expression in USP7-ATRT and USP13-MED, but not in DAOY (Fig. 6D and E). β-Catenin expression was altered by IWP-2 only in USP7-ATRT cell line. Basal β-catenin levels were significantly higher in USP7-ATRT compared with the other cell lines, confirming the previous global gene expression findings. CHIR99021 treatment also reduced efficient viral replication in USP7-ATRT cells (Fig. 6F).
Interestingly, production of infectious viral particles by host CNS tumor cells was not efficient. To address this question, USP7-ATRT, USP13-MED, and DAOY cells were first infected with MOI 1 ZIKVBR, and 72 hours after tumor cell infection, respective supernatants of these cell cultures containing the same amount of ZIKVBR copies were used to infect normal NPC at 72 hpi. In this cross-infection experiment, NPC infection with USP7-ATRT and USP13-MED supernatants did not produce functional ZIKVBR particles, whereas infection with DAOY supernatant produced relatively few functional ZIKVBR particles (Fig. 6G).
Discussion
Here, we compared six cell lines from five different human tumors (breast, prostate, colon rectal, and two different CNS embryonal tumors: ATRT and medulloblastoma), normal human NPC, and neurons. We show that ZIKVBR oncolytic effects are not generalized to all kinds of cancer, but more specific to CNS tumors. Interestingly, we observed that ZIKVBR kills CNS tumor cells more efficiently than normal NPC and that distinct CNS tumors may have differential sensitivity to ZIKVBR, indicating a selective oncolytic property. Notably, ZIKVBR injection improved overall survival, induced tumor remission, and effectively inhibited metastatic spread of human CNS tumor xenografts in athymic nude mice.
To our knowledge, this is the first study demonstrating oncolytic effects of ZIKVBR against human tumor cells in vivo. A very recent study by Zhu and colleagues (18) reported that ZIKV can also infect and inhibit mouse glioblastoma in syngeneic mice, improving survival. However, it is known that animal models not always recapitulate human pathologies. Furthermore, these authors also showed that ZIKV can destroy human glioblastoma cells in vitro, but at a relatively high MOI of 5, for much longer periods of time (2–4 weeks), as compared with the lower 0.01–2 MOI range tested in our study up to 72 hours.
The demonstration of selective oncolytic effects at low infection rates is highly relevant regarding safety in possible clinical trials. Some studies reported prolonged viral shed in human CNS (19), but very few cases were reported in the literature as lethal ZIKV-associated infection (20). Postnatal, infant, and adult infection rarely results in clinically relevant findings [e.g., conjunctivitis (21), Guillain–Barré syndrome (22), and encephalitis (20)]. Nonetheless, most ZIKV-infected individuals remain asymptomatic (23), and, in general, prognosis of ZIKV infection is largely uneventful with few or no intervention required.
In our preclinical study, conjunctivitis, ataxia, inflammation, and behavior abnormalities were not detected in any of the experimental animal subjects. Interestingly, upon ZIKVBR infection, overall survival of tumor-bearing mice was higher than that of sham animals. This higher mortality of control immunosuppressed mice raises an important safety issue and alert for possible adverse effects that cannot be ignored in future clinical development steps. Cross-infection experiments suggest that this effect may involve low production of functional viral particles by the infected tumor cells. The in vitro dose-escalation study at different time points also showed that infection with increasing virus MOI correlates with increasing tumor cell death and lower virus replication. This pattern was observed for the CNS tumor cell lines, with USP7-ATRT being the most affected cells. However, in this scenario, establishing optimal ZIKV dosing scheme required for efficient treatment of CNS tumors is paramount.
This lower virus replication in tumor cells can be explained by the fact that viruses use the host cell machinery to complete their replication. Thus, virus replication requires alive and fully functional host cells. In fact, death of host cells is a well-known defense mechanism that limits virus replication in infected cells, and some viruses encode cell death inhibitors to circumvent this defense and facilitate their own replication (24).
Among the CNS embryonal tumor cells lines, USP7-ATRT displayed the highest sensitivity to ZIKVBR, which may be partly explained by the fact that ATRT is originated from early neural stem cells and neuroprogenitors (25), whereas medulloblastoma is thought to originate from more mature granular neurons and their progenitors (26). A closer molecular similarity of USP7-ATRT with NPC than with the other tumor cell lines was indeed confirmed by an integrated molecular study, which also highlighted CTNNB1 as a key affected protein in USP7-ATRT cells.
Unbalanced regulation of Wnt signaling effectors, like CTNNB1, is common in AT/RT (27). The relevance of Wnt signaling pathway to normal neural development is also well established (28). Our functional study revealed that activation of Wnt signaling in USP7-ATRT increased ZIKVBR-induced tumor cell death, whereas inhibition of Wnt signaling decreased ZIKVBR-induced cell death. Future alternative complementary genetic approaches should also be considered to corroborate these findings obtained with small molecules. Interestingly, an independent study using human NPC obtained from discordant dizygotic twins for congenital Zika syndrome found a similar involvement of the Wnt signaling pathway regarding differential sensitivity of NPC to ZIKVBR infection (29). Translating these results to the clinical setting, patients with aggressive AT/RT and medulloblastoma should be good candidates for a future oncolytic therapy with ZIKVBR, because Wnt pathway activation plays an important role in the biology of both tumor types (27, 30), contributing to stemness and therapeutic resistance (31, 32). Importantly, observed metastasis remission also suggests that a possible treatment exploring ZIKVBR oncolytic effects could benefit patients with advanced and disseminated disease. Pursuit of attenuated forms of ZIKV and genetic modifications to avoid adverse effects and neutralization by the host immune system should facilitate a safe transition to clinical trials. Oncolytic viral therapy using modified Herpes simplex virus type I for melanoma has been approved by the FDA, and one clinical trial using H1-parvovirus therapy for glioblastoma is currently on phase I (33).
In conclusion, ZIKVBR has strong and specific oncolytic property against human CNS embryonal tumor cells, as demonstrated by both in vitro and in vivo assays. Our results show that ZIKVBR induces massive death of highly proliferative tumor cells at an MOI as low as 0.1. Selective infection of tumor cells results in low production of functional viral particles. Significant CNS tumor remission and spinal cord metastasis inhibition were also achieved after a single ZIKVBR infection in tumor-bearing mice with relatively low amount of virus particles. These effects were more prominent in tumors generated by neural stem–like cancer cells with high Wnt/β-catenin basal activity. It is reasonable to speculate that other signaling pathways might contribute to tumor cell–dependent susceptibility to ZIKA-induced oncolysis, which should be further investigated. Our findings provide novel mechanistic insights and open new perspectives for future investigations using ZIKVBR strains or engineered mimicking derivative approaches to treat patients affected by highly aggressive and metastatic CNS tumors lacking effective treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: C. Kaid, E. Goulart, L.C. Caires-Júnior, K, B.H.S. Araujo, A. Telles-Silva, M. Zatz, O.K. Okamoto
Development of methodology: C. Kaid, E. Goulart, L.C. Caires-Júnior, B.H.S. Araujo, A. Soares-Schanoski, H.M.S. Bueno, K.A. Telles-Silva, R.M. Astray, M. Zatz, O.K. Okamoto
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Kaid, E. Goulart, L.C. Caires-Júnior, A. Soares-Schanoski, H.M.S. Bueno, R.M. Astray, A.F. Assoni, A.F.R. Júnior, D.C. Ventini, A.L.P. Puglia, R.P. Gomes, M. Zatz, O.K. Okamoto
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Kaid, E. Goulart, L.C. Caires-Júnior, B.H.S. Araujo, A. Soares-Schanoski, H.M.S. Bueno, K.A.Telles-Silva, A.F. Assoni, M. Zatz, O.K. Okamoto
Writing, review, and/or revision of the manuscript: C. Kaid, E. Goulart, L.C. Caires-Júnior, B.H.S. Araujo, A. Soares-Schanoski, H.M.S. Bueno, K.A. Telles-Silva, M. Zatz, O.K. Okamoto
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Kaid, E. Goulart, L.C. Caires-Júnior, A. Soares-Schanoski, M. Zatz, O.K. Okamoto
Study supervision: C. Kaid, E. Goulart, L.C. Caires-Júnior, M. Zatz, O.K. Okamoto
Acknowledgments
The authors thank Drs. Patricia Semedo-Kuriki and Vivek Kumar for helping with cytometry assay and neuron differentiation, respectively. This study was supported by grants from FAPESP (CEPID number 2013/08028-1 and INCT to M. Zatz and O.K. Okamoto); C. Kaid is a fellow of CAPES (1379594); E. Goulart, A.F. Assoni, B.H.S. Araujo, and L.C. Caires-Júnior are fellows of FAPESP (2015/14821-1; 2016/09707-8; 2014/08049-1; 2017/16283-2).
Data and materials availability: Original data are curated and stored in the server of the Human Genome and Stem Cell Research Center (HUG-CELL). Requests for data may be sent to [email protected] and [email protected]
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.